Charged Particle and Photon Interactions
Charged Particle and Photon Interactions
Charged Particle and Photon Interactions
2
+RO, then occurs in
f
10
14
sec in water
and alcohols [13]. The time needed for electron hydration is theoretically computed to be
f
0.2 psec [14] while experiment gives an upper limit of 0.3 psec [15]. The information
presented in Table 1 is not to be taken too literally. Each stage may contain many orders of
magnitude of time, but the usefulness of the time scale picture rests on the perception that
within an order of magnitude relatively few processes compete. As the species that exist at
the end of one stage serve as the input to the next stage, the qualication early is relative.
In radiobiological systems, the time of appearance of a change may be subjective depending
on the cell cycle, which is due to several dierent kinds of generations of biochemical and
biological transformations that must take place to render the radiation-induced transfor-
mation visible. Also, these processes can occur both in the directions of repair and ampli-
cation. Certain biochemical reactions, for example, that of OH with sugars, may occur in
the nanosecond time scale while others, such as those giving the O
2
eect, may take micro-
seconds. DNA strand breakage may be considered an early biological eect while other
damages can take
f
1 day to
f
40 years when genetic eects are considered. This should be
contrasted with
f
1 Asec needed for track dissolution in liquid water [16].
In the current millennium there is a strong need for interdisciplinary research involv-
ing an integrated outlook. In this sense various disciplines connected with radiation action
all start with radiation physics and radiation chemistry. From that point on one sees a
bifurcation. One way goes to radiation biochemistry to radiation biology and to medical
applications. Another way goes to engineering and industrial applications including, but
not limited to, polymerization, waste management, food preservation, etc.
3. FUNDAMENTAL PROCESSES IN THE PHYSICAL, PHYSICO-CHEMICAL,
AND CHEMICAL STAGES OF INTERACTIONS OF HIGH-ENERGY
CHARGED PARTICLES AND PHOTONS WITH MATTER
In the interactions of high-energy incident particles, i.e., photons, electrons, heavy charged
particles (or positive and negative ions), and other particles, with matter, the succession of
events that follow the absorption of their energies has been classied into three character-
istic temporal stages: physical, physico-chemical, and chemical stages [17] (see Sec. 2). The
physical stage consists of the primary activation of molecules in matter (in the case, for
example, it is composed of molecular compounds), due to the collision of high-energy
incident particles to form electronically excited or ionized states of molecules and ejected
electrons. The electrons thus formed may have sucient energy to further ionize the sur-
rounding molecules. At the end of the interactions, electrons are formed in a wide energy
range via cascading electronmolecule collision processes. These secondary electrons fur-
ther decrease their energies to the subexcitation energy range, leading to the formation of
reactive species, i.e., ions, excited molecules, free radicals, and low-energy electrons. These
species interact with each other or with stable molecules. Subexcitation electrons, which
are electrons below the rst excitation potential, degrade their energies by vibrational,
rotational, and/or elastic collisions to reach thermal energy. Electrons in the thermal and
epithermal ranges disappear predominantly by recombination, attachment, or diusion. It
is concluded therefore that a decisive step in the physical and physico-chemical stages of
the interactions of high-energy incident particles with matter is the collision of secondary
electrons with molecules in a wide energy range (see Chapters 25, 9, 17). The entire
feature of the fundamental processes of the interactions of high-energy incident particles
Copyright 2004 by Taylor & Francis Group, LLC
with molecules, i.e., the molecular processes in the physical and physico-chemical stages
(and further, the chemical stage which mainly consists of free radical reactions), are sum-
marized in Table 2 [18] (see Chapter 6).
Table 2 shows that molecules AB in collisions with electrons in a wide energy range
are directly ionized and excited into superexcited states [9] above their rst ionization
potentials (IP) and into excited states below IP. Superexcited states ABV are competitively
autoionized or dissociated into neutral fragments, i.e., free radicals or stable product
molecules (see Chapter 5). Electronically excited states AB* are also dissociated to neutral
fragments. Parent ions directly formed via ionization, or indirectly via autoionization, are
collisionally (and/or unimolecularly) stabilized or dissociated into fragment ions. Such
ions, particularly in the condensed phase, may be quickly neutralized with electrons by
geminate recombination or competitively converted to other ions via ionmolecule re-
actions (see Chapter 6), while in the gas phase such conversion is predominant compared
with recombination, depending on the pressure. In case of the addition of electronegative
solute molecules, negative ions are produced in electron attachment processes. Electrons
with characteristic energies are selectively captured by molecules to form negative ions (see
Chapter 6). It is generally accepted that molecular clusters, large aggregates of molecules,
and molecules in the condensed phase can capture electrons with large cross sections due
to electron attachment dynamics dierent from those in the gas phase for isolated single
molecules [19,20]. For polar molecules, electrons or ions are solvated with molecules to
form solvated electrons and ions, respectively (see Chapter 7). The recombination of
positive ions with electrons, particularly in the condensed phase, is classied as geminate
and bulk recombination processes (see Chapter 10). A relative importance of these two
processes is closely related with the transport mechanism of electrons in it, i.e., the mag-
Table 2 Fundamental Processes in the Physical, Physicochemical,
and Chemical Stages of the Interactions of High-Energy Charged
Particles and Photons with Molecules AB
AB
AB
+
+e
Direct ionization
AB
*
Excitation]
ABV
AB
+
+e
Autoionization
A+B Dissociation
AB
+
A
+
+B Ion dissociation
AB
+
+AB or S
Products Ionmolecule reaction
AB
+
+e
AB
*
Electronion recombination
AB
+
+S
+S
S
Electron attachment
e
+nAB
e
s
Solvation
AB
*
A+B Dissociation
BA Isomerization
AB+hr Fluorescence
AB
*
+S
AB+S
*
Energy transfer
AB
*
+AB
(AB)
2
Excimer formation
2A
A
2
Radical recombination
C+D Disproportionation
A+AB
A
2
B Addition
A
2
+B Abstraction
Copyright 2004 by Taylor & Francis Group, LLC
nitudes of the electron drift mobility or the electron conduction band energy in it. Phe-
nomenologically, it is more dependent on the properties of the medium molecules (see
Chapter 8). In some cases, particularly in admixed systems both in the gas and condensed
phases, energy or charge transfer processes may predominantly occur from molecules of
the major component in the system to added solute molecules (see Chapter 6).
As described above, a crucial step of the physical and physicochemical stages is the
collision of secondary electrons with molecules in a wide energy range. Therefore the infor-
mation on electronmolecule collision cross-section data, which must be correct, absolute,
and comprehensive, is of great importance and is helpful for understanding the essential
features of the fundamental processes in these two stages [21]. Such information is avail-
able both experimentally and theoretically in the gas-phase (see Chapter 3), but in some
cases also in the condensed-phase as well (see Chapter 9). Cross sections for the ionization
and excitation of molecules in collisions with electrons in the energy range greater than
about 10
2
eV are well elucidated quantitatively by the BornBethe theory [22]. This theory
is also helpful for calculating, at least roughly, these cross-section values further in the
lower energy range.
According to the optical approximation, which was shown by Platzman [23] to be
based on the BornBethe theory, a radiation chemical yield Gmay be estimated fromoptical
data, viz., photoabsorption cross sections, oscillator strength distributions, photoionization
cross sections, and photodissociation cross sections of molecules in a wide range of the
photon energy or wavelength, particularly in VUVSX region [9] (see Chapter 5). Basically,
this approximation should be applied only for a rough estimation of G values. Such a
limitation may be attributed to a supercial comparison of radiation chemical yields for
atomic systems with electron- atom cross-section data in a limited collision-energy range
which were available in earlier stages of the related research elds in the 1960s. Recently,
however, its applicability has been examined by comparing electron-collision cross sections
for molecules in a wide collision energy range between optically allowed and forbidden
transitions, providing the conclusion, at least tentatively, that this approximation may be
applicable to the estimation of G values more precisely than previously considered [24].
There is a great need to examine the applicability more systematically because of recent
progress in the experimental and theoretical investigations of electronmolecule collision
processes [21].
The experimental and theoretical investigations of the physical, physico-chemical,
and chemical stages of the interactions of high-energy incident particles with matter have
made a remarkable contribution to recent progress in fundamental studies of the static and
dynamic behavior of reactive species, i.e., electrons, ions, excited atoms and molecules, and
free radicals. Conversely, these are needed for a comprehensive understanding of recent
progress in the fundamental studies of reactive species, where a scientically reasonable
analysis of the interesting but complex mechanism of the interactions of high-energy inci-
dent particles with matter is involved. Thus it often provides newndings of reactive species
as well as their static or dynamic behavior. For that purpose, the matrix representation is
helpful for summarizing or surveying the diverse information on such fundamental studies
of a variety of reactive species [18].
4. FIELDS OF STUDY: RELEVANCE TO BASIC AND APPLIED SCIENCES
The elds of study undertaken in the present book are rather broad. However, the con-
tributors, recognized experts in their respective areas, have striven to ensure continuity by
Copyright 2004 by Taylor & Francis Group, LLC
connecting important concepts to adjacent areas. Thus, in one example, the logical ow
would be from physics to chemistry to biology and medicine, and in another example it
could be from physics to chemistry to industrial applications, including polymerization,
waste, and ue-gas treatment, etc. In choosing the subject matters it has been the objective
of the editors that comparable importance is given to fundamentals and applications.
While it has been well recognized by almost everyone that radiation eect starts with a
physical interaction, very often followed by chemical reactions, etc., the chemical stage is
sometimes understressed in radiobiological discussions [25]. In a similar vein, the applied
radiation scientist may have a right to think that aqueous radiation chemistry has been
overstressed by the physical chemists who claim it to be a better understood topic in
chemical kinetics [26]. In any case the editors have taken care so that each signicant area
of investigation would receive comparable attention.
Studies in radiation science and technology may be viewed from three angles,
namely, life, industry, and basic knowledge. According to some estimates [27], the
continual exposure to environmental radiation over geological periods is important and
must be taken into account. There is now a widespread belief that aging and, indeed, the
genetic evolution of man may be signicantly inuenced by ambient radiation. Under-
standing of these eects duly belongs to radiation biology, which depends on radiation
physics and radiation chemistry for basic information. Radiation is applied in industry in
various waysfor example, as reaction initiators, sustainers, and also as control mech-
anisms. Some of the important ones are included in this book, while others, such as curing
of paints, materials for textile nishing, applications in mining and metallurgy, etc., are left
out for obvious space limitation. Finally, the important contribution made by radiation
chemistry to experimental basic science should be appreciated in making available
excitations to states which would otherwise be inaccessible by thermal activation.
REFERENCES
1. Mozumder, A. Fundamentals of Radiation Chemistry; Academic Press: San Diego, 1999.
2. Lind, S.C. Radiation Chemistry of Gases; Reinhold Publishing Corporation: New York, 1961.
3. Burton, M. J. Phys. Colloid. Chem. 1947, 51, 611.
4. Bethe, H.A. Ann. Physik. 1930, 5, 325.
5. Spinks, J.W.T.; Woods, R.J. An Introduction to Radiation Chemistry. 3rd ed. Wiley-Inter-
science: New York, 1991.
6. Handbook of Radiation Chemistry; Tabata, Y., Ito, Y., Tagawa, S., Eds.; CRC Press: Boca
Raton, 1991.
7. Excess Electrons in Dielectric Media; Ferradini, C., Jay-Gerin, J.-P., Eds.; CRC Press: Boca
Raton, 1991.
8. Hatano, Y. Adv. At. Mol. Opt. Phys. 2000, 43, 231.
9. Hatano, Y. Phys. Rep. 1999, 313, 109.
10. Lentle, B.; Singh, H. Radiat. Phys. Chem. 1984, 24, 267.
11. Mozumder, A. Radiat. Res. 1985, 104, S-33.
12. Morrison, P. Symposium in Radiobiology: The Basic Aspects of Radiation Eects on Living
Systems; In Nickson, J.J., Ed.; Wiley: New York, 1952; 112.
13. Magee, J.L.; Chatterjee, A. Radiation Chemistry: Principles and Applications; Farhataziz,
Rodgers, M.A.J., Eds.; Chap. 5. VCH Publishers: New York, NY, 1987.
14. Mozumder, A. J. Chem. Phys. 1969, 50, 3153.
15. Wiesenfeld, J.M.; Ippen, E.D. Chem. Phys. Lett. 1980, 73, 47.
16. Turner, J.E.; Magee, J.L.; Wright, H.A.; Chatterjee, A.; Hamm, R.N.; Ritchie, R.H. Radiat.
Res. 1983, 96, 437.
Copyright 2004 by Taylor & Francis Group, LLC
17. Platzman, R.L. Vortex 1962, 23, 327.
18. Hatano, Y. In: Handbook of Radiation Chemistry; Tabata, Y. Ito, Y., Tagawa, S., Eds.; CRC
Press: Boca Raton, 1991.
19. Hatano, Y. In: Electronic and Atomic Collisions; Lorents, D.C. Meyerhof, W.E., Peterson,
J.R., Eds.; Elsevier: Amsterdam, 1986; 153 pp.
20. Hatano, Y. Aust J Phys 1997, 50, 615.
21. Inokuti, M., Ed. In: Atomic and Molecular Data for Radiotherapy and Radiation Research;
IAEA-TECDOC-799 IAEA: Vienna, 1995.
22. Inokuti, M. Rev. Mod. Phys. 1971, 43, 297.
23. Platzman, R.L. Radiat. Res. 1962, 17, 419.
24. Hatano, Y. Radiat Phys Chem, to be published in 2003.
25. Lea, D.E. In: Actions of Radiations on Living Cells; Second ed. University Press: Cambridge,
1956.
26. Allen, A.O. In: The Radiation Chemistry of Water and Aqueous Solutions; D. Van Nostrand:
Princeton, 1961.
27. Vereshchinskii, I.V.; Pikaev, A.K. In: Introduction to Radiation Chemistry (translated from
Russian); Daniel Davey: New York, 1964.
Copyright 2004 by Taylor & Francis Group, LLC
2
Interaction of Fast Charged Particles
with Matter
A. Mozumder
University of Notre Dame, Notre Dame, Indiana, U.S.A.
1. ENERGY TRANSFER FROM FAST CHARGED PARTICLES
1.1. General Features
Any momentum change of an incident particle constitutes a collision. In the interaction of
an incident particle with an atom or a molecule, the collisions are classied as elastic or
inelastic. In elastic collisions, energy and momentum are conserved between the external
degrees of freedom (e.g., translation) of the colliding partners. In inelastic collisions, en-
ergy is transferred into, or from, the internal degrees of freedom of the struck molecule,
designated as collisions of the rst and the second kind, respectively. Because collisions of
the second kind are important for epithermal and thermal particles alone, we will only be
concerned with the collisions of the rst kind in this chapter. Furthermore, akin to the
Grottus principle (see the Introduction) in photochemistry, only the energy absorbed by a
molecule can bring about radiationchemical change, or the ensuing radiobiological trans-
formation. Therefore the internal molecular excitation (electronic, vibrational, etc.) is a
necessary precursor of radiation-induced transformation. In this context, ionization is in-
cluded as an extreme form of electronic excitation.
Particles and waves are distinguished only in classical mechanics. In quantum me-
chanics they are interchangeable, giving rise to rst and second quantizations, respectively.
However, it is customary to group the incident radiations as light (electrons, muons, x-
and g-rays, etc.) or heavy (protons, a-particles, ssion fragments, etc.) particles. Photons,
classied as a separate group, can cause excitation and ionization in a molecule. Ionization
by photons can proceed by any of the three following mechanisms: photoelectric eect,
Compton eect, and pair production. In the photoelectric eect, the photon is absorbed and
the emergent electron inherits the photon energy reduced by the sum of the binding energy
of the electron and any residual energy left in the resultant positive ion (Einstein equation).
In the Compton eect, the photon is scattered with signicantly lower energy by a medium
electron, which is then ejected with the energy dierence. In the pair production process,
the photon is annihilated in a nuclear interaction, giving rise to an electronpositron pair,
which together carries the photon energy lessened by twice the rest energy of the electron.
This process therefore has an energetic threshold. With increase in photon energy, the
dominant interaction changes from photoelectric to Compton to pair production. For
Copyright 2004 by Taylor & Francis Group, LLC
60
Co-g rays, almost the entire interaction is induced by the Compton eect which, by the
KleinNishina theory, gives a nearly at electron energy spectrum. Thus the 1.2-MeV
photon generates a wide spectrum of Compton electrons having an average energy of
f0.6 MeV; consequently, the medium molecules respond as if being showered by these
fast electrons. Pair production is a characteristic of very high energy photons; again, to the
medium molecules, the eect is the same as a spectrum of fast electrons and positrons of
appropriate energy. Electrons of various energy are the most important sources as primary
radiations in the laboratory and in the industry [1], and also as secondary radiations in any
form of ionizing event. Other often-used irradiations include x-rays, radioactive radiations
(a, h, or g), protons, deuterons, various accelerated stripped nuclei, and ssion fragments.
X-rays dier from g-rays operationally, i.e., x-rays are generated by machines whereas g-
rays are produced in nuclear transformations.
At ultrarelativistic speeds (vfc, the speed of light), a radiative process, called the
nuclear bremsstralung, becomes signicant. Bethe and Heitler [2] estimated the ratio of
energy losses of a charged particle due to radiative (bremsstralung) and collisional (elec-
tronic) processes at an energy of E MeV approximately as EZ/800, where Z is the nuclear
charge. For electrons in water, this radiative process only becomes signicant above 100
MeV. For the vast majority of cases with incident velocity f0.99c down to about the
speed of a least-bound electron, the dominant energy transfer mechanism is by electrosta-
tic and electromagnetic interactions between the incident charged particle and the medium
electrons. Of these, the electromagnetic interaction becomes important only at relativistic
speeds [3]. Therefore in most cases of importance to radiation science, the principal
interaction of an incident-charged particle with the medium electrons is of electrostatic
origin.
For fast charged particles impinging on matter, the very rst discernible eect is the
electronic excitation of molecules or atoms. Because electron is the lightest particle, its
excitation occurs at the earliest time scale consistent with the uncertainty principle, i.e., in
f10
16
sec. Later, transformation of the deposited energy in the various internal degrees
of freedom occurs in their respective time scale, e.g., in f10
12
sec for vibration, etc. Some
details were presented elsewhere [4]. Molecular dissociation and low-grade heat appear at
longer time scales and, in some cases involving exothermal reactions, the level of low grade
heat can terminate above that of absorbed electronic energy. There is no evidence of local
rise of temperature for fast incident charged particles.
1.2. Radiation Physics and Radiation Chemistry
Although in nature there is no distinction between physical and chemical eects, in
practice it is protable, depending on the investigators choice of discipline, to view the
radiation eect either from the point of view of the incident particle or from that of the
medium molecules, designated as radiation physics and radiation chemistry, respectively.
Thus the study of the charge, rate of energy loss, range, and penetration, etc. of the
incident particle, any of which may change under the interaction, constitutes radiation
physics. Study of the matter receiving the absorbed energy to produce chemical changes,
charge separation, luminescence, etc. essentially belong to radiation chemistry. It is evident
that radiation chemistry is the link between radiation physics and radiation biology (see
Ch. 1, Sec. 2). Some investigators recognize this link explicitly [5,6], while others short
circuit the chemical part, working directly from the physics of energy deposition to
biological eects [7]. That necessitates the introduction of additional parameters, such as a
critical dose or concepts such as microdosimetry [8,9].
Copyright 2004 by Taylor & Francis Group, LLC
1.3. Condensed Matter: Localization of Deposited Energy
Fano [10] drew attention to the fact that, derived from the uncertainty relation, an energy
loss f15 eV from a high-speed (fc) particle cannot be localized within f90 nm. That
such energy losses in a condensed medium would give rise to collective excitations
(plasmons) was a source of some concern until recently. On the other hand, understanding
of radiation-induced transformations, operation of detectors, etc. would require that the
deposited energy be localized in a molecule before dissociation or ionization can occur.
Because no reasonable explanation has yet been proposed for the localization of energy
after the initial delocalization, rationalization has been advanced as the delocalization
applying only to the rst interaction [4]. After that, all energy losses are localized and
correlated on a track [11].
Nevertheless, certain collective excitations can occur in the condensed phase. These
may be brought about by longitudinal coulombic interaction (plasmons in thin lms) or by
transverse interaction, as in the 7-eV excitation in condensed benzene, which is believed to
be an exciton [12]. Special conditions must be satised by the real and imaginary parts of
the dielectric function of the condensed phase for collective excitations to occur. After
analyzing these factors, it has been concluded that in most ordinary liquids such as water,
collective excitations would not result by interaction of fast charged particles [13,14].
2. STOPPING POWER, LET, AND FLUCTUATIONS
2.1. Bohrs Theory and the Bragg Rule
The rate of energy loss of a charged particle per unit pathlength is called the linear energy
transfer (LET). The corresponding energy received by the medium, which remains in the
vicinity of the particle track, is designated as the stopping power. Therefore LET and
stopping power refer to the penetrating particle and the medium, respectively. The dif-
ference between these quantities arises because, even in the shortest time scale (f10
15
s),
some of the deposited energy may be removed from the track vicinity by fast secondary
electrons, by Cerenkov radiation, and, in the case of ultrarelativstic particles, by bremss-
tralung. Cerenkov radiation is seen as a faint bluish light when the speed of the penetrating
particle (usually an electron) exceeds the group velocity of light in the medium. Although
it contributes negligibly to the LET, it has importance as a time marker, because it is
emitted almost instantaneously. The term LET was rst coined by Zirkle et al. [15] in a
radiobiological context, replacing some earlier nebulous terminology. Both LET and
stopping power are average concepts and considerable uctuations are to be expected
around these values. The physical theories of energy loss rate should refer to LET, but
these are often called stopping power theories. The International Commission on Ra-
diation Units and Measurements recommends the use of the symbol L
A
to denote energy
transfers below a limit A. Thus L
100
would denote LET for energy losses of less than 100
eV and L
l
would include all energy losses.
The rst successful theory of stopping power is attributed to Bohr [16,17]. He takes
the simplest case of a fast, yet nonrelativistic, heavy particle of velocity v and charge ze,
where e is the electronic charge. Under this condition, the basic electrostatic interaction
can be treated as a perturbation, that is, the particle path can be considered to be virtually
undeected. In addition, the energy losses occur through quasi-continuous inelastic col-
lisions with medium electrons, so that the average stopping power is a good representation
of the statistical process of energy loss. All stopping power theories recognize these sim-
Copyright 2004 by Taylor & Francis Group, LLC
plications. Neglecting the binding energy of the electron and calling the distance of
closest approach of the incident particle as the impact parameter b, it may be shown that
the component of momentum transfer parallel to the particle trajectory vanishes by
symmetry and that, perpendicular to the trajectory, is given by 2ze
2
/bv [18]. The same
expression is obtained by multiplying the peak force ze
2
/b
2
with the so-dened duration
of collision 2b/v. Thus the energy transfer is given by Q = (2ze
2
/bv)
2
/2m = 2z
2
e
4
/mb
2
v
2
,
where m is the electron mass. The dierential cross-section for this process is given geo-
metrically as:
dr pdb
2
2pz
2
e
4
=mv
2
dQ=Q
2
;
which is just the Rutherford cross-section for scattering by free charges receiving the
energy transfer Q. For application to bound electrons which can be excited (or ionized),
for the same momentum transfer, to the nth state with energy E
n
, Bohr surmised the sum
rule A
n
f
n
E
n
ZQ, where Z is the atomic number and f
n
is the oscillator strength of that
transition (see later for proper denition). The stopping power dE/dx is the energy loss
to all the medium electrons per unit path of the incident particle, over all permissible
energy transfers. Therefore
dE=dx N
_
dr
n
f
n
E
n
2pz
2
e
4
NZ=mv
2
lnQ
max
=Q
min
: 1
For the maximum energy transfer, binding eect may be ignored, giving in the non-
relativistic case Q
max
= 2mv
2
. For the minimum energy transfer, Bohr argued that col-
lisions must be sudden so that an eective energy loss may occur. That is, the collision time
2b/v must be Vt/E
1
, where E
1
is atypical atomic transition energy and t is Plancks
constant divided by 2p. This gives b
max
= tv/2E
1
and Q
min
= 8z
2
e
4
E
1
/[(mv
2
)(t
2
v
2
)]. Sub-
stituting in Eq. (1), one obtains
dE=dx 4pz
2
e
4
N=mv
2
B; 2
where the expression within the parentheses in Eq. (2) is called the kinematic factor and B,
denoted as the stopping number, is given in Bohrs theory as Z ln[(2mv
2
/E
1
)(tv/4ze
2
)].
Although the impact parameter is not an observable quantity, Bohrs theory has a wide
range of validity if the typical transition energy E
1
is properly chosen. Equation (2) also
shows the typical structure of a stopping power equation as a product of a kinematic
factor and a stopping number. The kinematic factor is free from any target property,
which is solely contained in the stopping number. In most stopping power expressions, the
kinematic factor therefore remains the same as in Eq. (2).
As it stands, Bohrs theory is only applicable to atoms. To extend it to molecules and
mixtures, one invokes the Bragg rule, originally used for charged particle ranges in ion-
ization chambers. Simply stated, the Bragg rule equates the stopping number of the mol-
ecule (or the mixture) to the weighted sum of the stopping numbers of the constituting
atoms. Braggs additivity rule applies impressively within a few percent for many com-
pounds. However, the contribution of an atomic stopping number in dierent compounds
is not necessarily the same as that for the free atom. That this contribution remains the
same in each case indicates the similarity of that atomic participation in various com-
pounds. The apparent success of Braggs rule has been traced to two factors: (1) similarity
of atomic binding in dierent molecules, and (2) electronic transitions that have the most
oscillator strengths involve excitation energies far in excess of chemical binding. Where
such is not the case as, for example, in H
2
, NO and compounds of the lightest elements, the
rule does not hold well. There is evidence that the Bragg rule starts to break down at low
incident energies. The quantum mechanical theories of stopping power are also amenable
Copyright 2004 by Taylor & Francis Group, LLC
to this additivity rule in a similar fashion (see Sec. 2.2). However, this does not mean that
the ensuing chemical response of the molecule is necessarily related to those of the con-
stituent atoms. There are denite eects of chemical binding and state of aggregation
which alter the energy levels and the oscillator strengths of transitions. The Bragg rule
refers only to the stopping of the incident particle.
2.2. Bethes Theory and Extensions
The rst (and still the foremost) quantum theory of stopping, attributed to Bethe [19,20],
considers the observables energy and momentum transfers as fundamental in the inter-
action of fast charged particles with atomic electrons. Taking the simplest case of a heavy,
fast, yet nonrelativistic incident projectile, the excitation cross-section is developed in the
rst Born approximation; that is, the incident particle is represented as a plane wave and
the scattered particle as a slightly perturbed wave. Representing the Coulombic interaction
as a Fourier integral over momentum transfer, Bethe derives the dierential Born cross-
section for excitation to the nth quantum state of the atom as follows.
dr
n
2pz
2
e
4
=mv
2
jF
n
qj
2
dQ=Q
2
3
In Eq. (3), q is the magnitude of the momentum transfer, Q = q
2
/2m is the energy of
a free electron having that momentum q, and other kinematic parameters have the same
signicance as in the Bohr theory. The quantity, F
n
q S
j
n exp2pihq r
j
0
_
, called the
inelastic form factor, is the sum over all the target electron coordinates, of the matrix
elements of the exponential operator involving the vector momentum transfer q between
the ground and the nth excited state. Obviously, the stopping power would be obtained
from Eq. (3) as dE=dx NS
n
m
Q
max
Q
min
E
n
dr
n
, where N is the atomic number density, E
n
is
the transition energy, and Q
min
and Q
max
stand for the minimum and maximum values of
Q, respectively, consistent with a given excitation energy. It is convenient to divide the
range of Q into small and large momentum transfers [3]. In collisions involving small
momentum transfers, sometimes called soft or glancing collisions, Bethe shows that
jF
n
( q)j
2
!Qf
n
/E
n
, where f
n
is the corresponding dipole (optical) oscillator strength (vide
infra). This allows the contribution of such collisions to the stopping power to be written
(cf. Eq. (3)) as 2pz
2
e
4
N=mv
2
S
n
f
n
ln
Q
1
E
2
n
=2mv
2
, where Q
1
is the rather inconsequential
upper limit of Q for small momentum transfers, and Q
min
, for excitation energy E
n
, is
given kinematically as E
2
n
/ 2mv
2
. For large momentum transfer collisions, sometimes de-
signated as hard or knock-on collisions, Bethe directly proves the sum rule S
n
E
n
j Fq
n
j
2
ZQ. Inserted into Eq. (3), one then obtains the contribution of hard collisions to the
stopping power as (2pz
2
e
4
N/mv
2
)Z ln(2mv
2
/Q
1
), where Q
1
is taken as the lower limit of Q
for hard collisions and the kinematic maximum 2mv
2
is used for Q
max
. Noting the
ThomasKuhn sum rule
S
n f
n
Z, the total number of electrons in the atom, and adding
the contributions of soft and hard collisions, the total stopping power is then given by
dE=dx 4pz
2
e
4
NZ=mv
2
ln2mv
2
=I; 4
where we used Bethes denition of the mean excitation potential I of the atom through
Eq. (5).
Z ln I
n
f
n
ln E
n
5
The basic stopping power formula of Bethe has a structure similar to that of Bohrs
classical theory [cf. Eq. (2)]. The kinematic factor remains the same while the stopping
number is given by B =Zln(2mv
2
/I) for incident heavy, nonrelativistic particles. The Bethe
Copyright 2004 by Taylor & Francis Group, LLC
theory gives a meaning to the mean excitation potential I in terms of strengths and ener-
gies of atomic transitions and underscores the importance of the dipole oscillator strength.
It should be noted that, in the denition of the mean excitation potential [Eq. (5)], as well
as in the derivation of the ThomasKuhn sum rule, excitation to energetically inaccessible
states are also formally included. At a suciently high incident velocity for the Born ap-
proximation to be valid, it usually does not entail much error. However, it is a separate
approximation. Another approximation, called Bethes asymptotic cross-section, derives
from expanding the interaction cross-section at high incident energies and retaining only
the lead term [21]. A simplied derivation of Bethes formula has been given by Magee
[22].
Several extensions of the basic stopping power theory of Bethe can be made. For
compounds and mixtures, Braggs additivity rule may be applied with the resultant value
of I given by the geometrical average of the mean excitation potentials of the constituent
atoms over their electron numbers. In the case of incident electrons, Q
max
, ignoring atomic
binding, would be given by (1/4)mv
2
rather than by 2mv
2
, because the distinction between
the incident and the secondary electron can only be made on the basis of energy.
Incorporating this modication, Bethe gives Belectron Z lnmv
2
=2
e=2I
_
, where e
is the base of natural logarithm. Thus the electron stopping power at the same velocity is
always a little less than that of a heavy particle, sometimes by as much as 20%.
At relativistic speeds, the maximum energy transfer increases to 2mv
2
/(1b
2
), where
b = v/c. Also, electromagnetic eects have to be considered in addition to electrostatic
interaction. These result in the addition of a term Z[ b
2
+ln(1b
2
)] to the stopping
number. This extra stopping, called the relativistic rise, is proportional to b
4
for small
values of b. Its eect is <0.1% for v<510
9
cm sec
1
, but it can be very signicant when
vfc. Fermi showed that the electronic stopping power would diverge were it not for the
mutual polarization screening of medium electrons. This eect, therefore, is more
important in the condensed phase and is called a density correction. This corrective
factor, added to the stopping number, is denoted by Zd/2, where d =ln[ tx
2
p
/I
2
(1b
2
)]1
and x
p
= (4kNe
2
Z/m)
1/2
, denoted as the plasma frequency [3]. Another correction is usu-
ally needed when the incident particle is not faster than the speed of atomic electrons, as
required for the validity of the Born approximation. This is more important for the K
electrons, but it is also sometimes needed for the L electrons in heavy elements. Denoting
the total so-dened shell correction as C, the result is a little reduction of the stopping
number by that amount. Combining the eects of relativity, density, and shell corrections,
the Bethe stopping number of a heavy particle is given by
B Zln 2mv
2
=I b
2
ln1 b
2
C=Z d=2: 6
The corresponding stopping number for the electron is given by
Belectron Z=2
_
ln mc
2
b
2
E=2I
2
I b
2
_
2
1 b
2
_
1 b
2
_
ln 2
1 b
2
1=8
_
1
1 b
2
_
_
2
2C=Z d
_
7
where E is the electron energy at velocity bc. Because electrons can acquire relativistic
speed at moderate energies, say >20 keV, the required correction is more important for
electrons.
Slow positive ions tend to capture electrons from the medium when their speeds fall
below z
1
v
0
, where z
1
is the bare nuclear charge and v
0
is Bohrs velocity. At rst, the
Copyright 2004 by Taylor & Francis Group, LLC
captured electron is promptly lost in a subsequent charge-changing collision. The cycle of
capture and loss continues for a while and an equilibrium charge is established. The
incident charge in the stopping number should then be changed from z to z
e
, given by
hz
2
i
1/2
, where the angular brackets indicate averaging with respect to charge-changing
collision cross-sections. After the rst electron is rmly captured, another cycle of electron
capture and loss may continue until a second electron is rmly captured, and so on. The
net eect of charge changing collisions is to lessen the stopping power a little because z
e
is
always <z. Even so, electron capture and loss constitute minor energy loss processes by
themselves.
Despite the apparent similarity of the Bohr and the Bethe stopping power formulae,
the conditions of their validity are rather complimentary than the same. Bloch [23] pointed
out that Born approximation requires the incident particle velocity v>>ze
2
/h, the speed of
a 1s electron around the incident electron while the requirement of Bohrs classical theory
is exactly the opposite. For heavy, slow particles, for example, ssion fragments penetrat-
ing light media, Bohrs formula has an inherent advantage, although the typical transition
energy has to be taken as an adjustable parameter.
2.3. Dipole and Generalized Oscillator Strengths: Sum Rules
From the inelastic form factor (see Sec. 2.2), Bethe [19] denes a generalized oscillator
strength f
n
( q) u E
n
F
n
( q)
2
/Q, for excitation with an energy loss E
n
, simultaneously with a
momentum transfer q, in close association with the (optical) dipole oscillator strength. The
latter may be obtained experimentally from the absorption coecient a(v), viz. f
n
= (mc/
pnVc
2
)ma(v)dv, where nV is the medium refractive index and the integration is carried over
the frequency v for the same transition. Theoretically, it is related to the dipole moment
matrix element for the transition from the ground to the nth state through the Einstein
Acoecient as f
n
8pmv=e
2
hg
n
hnje S
j
x
j
j0i
2
. Here g
n
is the degeneracy of the excited
state (the ground state is taken as nondegenerate), x
j
is the x component of the position of
the jth electron, and the summation is carried over all the atomic electrons. Expanding the
exponential in the inelastic form factor [see Eq. (3) et seq.] for a small momentum transfer,
and noting the orthogonality of the wave functions, one obtains lim
q!0
f
n
q f
n
, thus
establishing a relationship between the generalized and dipole oscillator strengths. Fig. 1
shows such a procedure for the electron impact excitation in helium at an incident energy
of 500 eV [24]. Platzman [25] advocated the construction of a complete optical spectrum
( f
n
), and therefrom, that of the excitation spectrum ( f
n
/E
n
) [cf. Eq. (3) and the denition of
the generalized oscillator strength], from a few experimental determinations augmented by
the various sum rules. However, the results in the cases of methane and water were not
very satisfactory, probably because of the unavailability of a sucient number of dipole
sums of adequate accuracy.
Both generalized and dipole oscillator strengths satisfy sum rules, which can be used
to unravel their character, although their calculation from rst principles is not easy except
for the lightest elements. For the dipole oscillator strength, a sum may be dened by
Sl
n
E
l
n
f
n
; 8
where the sum is over all the states of excitation, including integration over continuum
states (here and in later discussion). S(0) = Z, the number of electrons in the atom or
molecule. This is called the ThomasKuhn sum rule. Bethe [19] has shown that the same
rule is satised by the generalized oscillator strength for any momentum transfer,
Copyright 2004 by Taylor & Francis Group, LLC
i.e., S
n
E
l
n
F
n
q Z, for any q. Considering the oscillator strength as a normalized distri-
bution, S(l) in Eq. (8) may be taken as the lth moment of the transition energy. While
S(l) diverges for l z 3 in systems where the ground state electron density is nonvanishing
at the nucleus [21], for 4 Vl V2, the sums are related to important physical properties
[26]. S(4) may be experimentally obtained from the refractive index and the Verdet
constant, the latter referring to the rotation of the plane of polarization per unit thickness
per unit magnetic eld parallel to the propagation direction. S(2) = a/4, where a is the
polarizability. S(2) = (16pZ/3)times the average electron concentration at system center.
S(1) is related, apart from a constant ground state energy, to momentum correlation of
electrons [26]. S(1) equals the total dipole matrix element squared, M
2
tot
S
i; j
x
i
x
j
_
,
which appears as a lead term in the total inelastic collision cross-section [27].
Another class of sum rules may be dened as logarithmic moments, or by formally
dierentiating S(l) with respect to l [see Eq. (8)], yielding
Ll
n
E
l
n
f
n
ln E
n
Sl
d
dl
Sl: 9
L(0)=Z ln I, where I is the mean excitation potential appearing in Bethes stopping power
equation [Eq. (4)]. L(2) is proportional to the logarithm of average excitation energy,
which is also involved in the Lamb shift [26]. L(1) has been shown to be an optical
Figure 1 Generalized oscillator strength for the 1
1
S!2
1
Ptransition in heliumobtained by Lassettre
et al. [24], with 500-eV incident electrons. The abscissa is the square of the momentum transfer
in atomic unit (1.9910
19
g cm/s). Note that, according to the Lassettre procedure, lim
q!0
f
n
q f
n
,
the optical oscillator strength, is attained irrespective of the validity of the Born approximation.
However, if that approximatiom remains valid, then the curve of f
n
( q) would be independent of
incident energy.
Copyright 2004 by Taylor & Francis Group, LLC
quantity [27]. Oscillator strength sum rules have been used with success, limited chiey to
the lightest atoms, in amending uncertain oscillator strengths and in calculating a range of
physical properties including the refractive index and the Verdet constant [26]. Thus, for
example, the dipole oscillator strength of the transition 2
1
Pp1
1
S in helium was corrected
by Miller and Platzman [28] to 0.277, against an earlier calculated value of 0.19.
2.4. Special Features of the Condensed Phase
One aspect of the condensed phase regarding the delocalization of the deposited energy
has been alluded to in Sec. 1.3. Here we will consider the modications on the oscillator
strength and the mean excitation potential, due to condensation, which would enter in
Bethes stopping power theory [see Eqs. (47)].
For polyatomic molecules such as water, the oscillator strength is, in general,
continuously distributed because of any combination of dissociation, ionization, etc.
The stopping power is then determined by the dierential oscillator strength distribution
(DOSD). Phase eects occur naturally through the DOSD. In the gas phase, light
absorption and/or inelastic scattering experiments can be used to obtain the DOSD,
and a fairly complete determination has been made for water vapor [29]. In the condensed
phase (solid or liquid), absorption measurement in the far-UV and beyond is experimen-
tally very dicult. In such cases, an indirect method, depending on electromagnetic
relationships, can be sometimes used if reectance measurements are available at the
vacuumliquid (or vacuumsolid) interface over a wide range of energies. Denoting the
reectance at energy E as R(E), the phase angle is obtained from the KramersKronig
relation, /E E=pPm
l
0
ln REVdEVEV
2
E
2
1
, where P is Cauchys principal value
for the integral. From these, the real and imaginary parts of the refractive index are
obtained as n = (1R)/(1+R2R
1/2
cos /) and k = (2R
1/2
sin /)/(1R2R
1/2
cos /),
respectively. The real and imaginary parts of the dielectric function are then given by
e
1
= n
2
k
2
and e
2
= 2nk, with Im(1/e) = e
2
/(e
2
1
+e
2
2
). The DOSD f V(E) is nally calcu-
lated from the equation Im(1/e) = (h
2
e
2
NZ/2m) f V(E)/E. Here f V(E)dE is the oscillator
strength contained for transitions with energy between E and E+dE. In a certain sense
within the context of the Bethe theory, the function Im(1/e) is more fundamental for the
interaction in the condensed phase.
There is an upper limit of experimental reectance measurement, usually around 26
eV. With present-day synchrotron sources, this limit is being progressively extended.
Nevertheless, in the global integration of the KramersKronig relation, a long extrap-
olation is needed. In the original experiments of Heller et al. [30] on liquid water, an
exponential and a power-law extrapolation beyond the upper limit of experiments were
used. Later, LaVerne and Mozumder [31], requiring transparency in the visible region and
the correct number of valence electrons (8.2 for water), opted for the power-law with index
3.8, close to the theoretical limit of 4.0 for valence electrons. The contribution of the K
electrons remains the same as in the gas phase and are, therefore, simply added. The so-
determined DOSD for liquid water is shown in Fig. 2, and compared with that of the gas
phase given by Zeiss et al. [29], and with that of gaseous cyclohexane, obtained by
Koizumi et al. [32] by using synchrotron UV-absorption measurement. In water, there are
two main eects of condensation. First, there is a loss of structure in the condensed phase.
Second, there is an upward shift of the excitation energy. The peak in the DOSD changes
from f18 eV in the gas phase to f21 eV in the liquid phase. This means that collisions in
the liquid phase are more dicult. Using the oscillator distribution, the computed values
of I, the mean excitation potential is 74.9 and 71.4 eV for liquid and gaseous water,
Copyright 2004 by Taylor & Francis Group, LLC
respectively [31]. This dierence may not seem critical in determining the stopping power.
However, the nature and energy of the excited states are dierent in the two phases,
yielding dierent chemical outcomes (see later).
2.5. RangeEnergy Relation
The range of an incident particle is dened as the average crooked path length between the
initial and nal energies, while penetration refers to the vector distance between the
starting and ending points. The dierence between these quantities is attributed to
scattering, mostly elastic. Obviously, range increases with initial energy. The rangeenergy
relation is important for the identication and energy measurement of charged particles, a
tool that is routinely used in nuclear physics. In radiation chemistry, it provides a basis for
charged particle tracks and the reactions on them. Various tables of ranges of light and
heavy particles are now available, of which an earlier collection [33] and a later
compilation by Ziegler [34] are noteworthy.
Range, obtained by (numerical) integration of the stopping power as, R m
E
i
E
0
dE=
dE=dx, is called the continuous slowing down approximation (CSDA) range. Here E
i
is the initial energy, E
0
is the energy where the particle is considered stopped, and the
approximation refers to the stopping power being a continuous function of energy,
ignoring random uctuations. For electronic stopping, E
0
should be the energy of the
rst excited state, but a higher value is often adopted for fast incident particles, because, at
Figure 2 Dipole oscillator strength distribution in gaseous water [29, curve, A], in liquid water [31,
curve, B] and in gaseous cyclohexane [32, curve, C]. Data in liquid water are obtained from an
analysis of UV-reectance and that in cyclohexane, from synchrotronUV absorption. The Thomas
Kuhn sum rule is satised approximately in each case.
Copyright 2004 by Taylor & Francis Group, LLC
low energies, the stopping power formula is not reliable and, in any case, it is superseded
by subelectronic processes. Such an arbitrary value of E
0
becomes inconsequential if the
incident energy is suciently high.
Often, the rangeenergy relation is displayed as a power law, R~E
m
, where the index
m would approach 2 if the slow variation of the stopping number B with particle velocity is
ignored [see Eq. (4) et seq.]. For protons and He ions of a few hundred MeV, m
approaches 1.8, decreasing somewhat with decreasing energy. The index m also depends
on the medium traversed. For electrons in water, it diminishes gradually from f1.7 to
f0.9 as the initial energy is decreased from a few thousand to a few hundred eV. Low-
energy electron range in water is of great importance to the spur theory of radiation
chemistry, but it is also rather dicult to ascertain and often greatly inuenced by elastic
scattering. This eld is being continually improved and track simulation techniques have
been employed [35], which give not only the mean range, but also its distribution (see the
next section). However, very little is known about elastic scattering in liquid water, forcing
many investigators to use gas phase data for that purpose. Typical values of computed
ranges in water are f550 nm for 5-keV electrons, f0.5 mm for 5-MeV protons, and
f0.04 mm for 5-MeV a-particles. Ranges of slow heavy ions and ssion fragments are
nearly proportional to energy, because the increase of stopping power in relation to
slowing down is largely compensated by a lowering of eective charge as a result of
electron capture, thus providing an almost constant stopping power. The typical density-
normalized range of Ne ions of 10 MeV/amu in Al is 40 mg/cm
2
.
2.6. Discussion of Stopping Power, Range, and Fluctuations
The most important factor in Bethes stopping power equation, which embodies the
aggregate eect of transition energies and oscillator strengths, is the mean excitation
potential, I [see Eqs. (4) and (5)]. Its direct calculation requires the knowledge of the
ground and all excited state wave functions. Such a calculation for atomic hydrogen gives
I = 15 eV, whereas in atomic binding in dierent molecules, a value of 18 eV is preferred
to satisfy the Bragg rule [3]. For other atoms, application of the ThomasFermi model by
Bloch shows that the spectral distribution of the oscillator strength has universal shape in
all atoms, if the transition energy is scaled by Z. This means I~Z, with the constant of
proportionality being f12 eV. A detailed examination of the ThomasFermiDirac model
gives I/Z = a+bZ
2/3
, with a = 9.2 and b = 4.5 as best adjusted values, and where I is
expressed in eV. This equation agrees rather well with range experiments. However, Fano
[3] points out that it should not be construed as a validation of the statistical model of the
atom. I, being a logarithmically averaged quantity, can be well approximated without an
accurate knowledge of the distribution. An error yI in determining I only appears as a
relative error of f(1/5)yI/I in the evaluation of range. In practice, I is often treated as an
adjustable parameter to be xed by range measurement of fast protons or a-particles. With
known I values of constituent atoms, the mean excitation potential of a molecule may be
obtained by the application of the Bragg rule (see Sec. 2.1). Thus if the molecule has n
I
atoms of atomic number Z
I
, and mean excitation potential I
I
, then the overall mean
excitation potential of the molecule would be given by the relation Z ln I = Sn
i
Z
i
ln I
i
,
where Z = Sn
i
Z
i
is the total number of electrons in the molecule.
The stopping power of water, over a wide span of energy, for various incident
particles has been discussed elsewhere [Ref. [4], Sec. 6], with special consideration for
eects at low and very high energies. Generally speaking, the electronic stopping power
shows a peak at a relatively low energy (f100 eV for incident electrons and f1 MeV for
incident protons), as a result of the combined eect of the velocity denominator in the
Copyright 2004 by Taylor & Francis Group, LLC
kinematic factor and the logarithmic velocity dependence in the stopping number [c.f., Eq.
(4) et seq.].
The usual stopping power formula for low-energy electrons (BornBethe approx-
imation) requires corrections on two accounts. First, the integrals dening the total
oscillator strength and the mean excitation potential must be truncated at the maximum
energy transfer (c(1/4)mv
2
), resulting in the replacement of Z and I by Z
e
and I
e
,
respectively[36]. These quantities are lower than their respective asymptotic values and
their eects partially cancel each other. Second, the eect of the nonvanishing of the
momentum transfer for inelastic collisions becomes signicant. When a quadratic exten-
sion of the generalized oscillator strength is made in the energy-momentum plane still
using the DOSD, the net eect is expressible as a reduction of the electron stopping
number by an amount Z
e
(e/2E)ln(4e E/I ), where e is the base of natural logarithm, E is
incident electron energy, e is average energy loss in an encounter, and I is a mean energy
dened by Z
eff
e ln I m
0
e
max
e ln efede [36]. Here e
max
is the maximum transferable ener-
gy. In track simulation procedure, it is more convenient to work with basic cross-sections
(with which the stopping power is calculated) and keep a detailed account of energy loss,
distance traversed, etc. The stopping power may then be computed from denition by
averaging over a large number (typically f10
4
) of simulations. The results of such a
calculation by Pimblott et al. [35] is shown in Fig. 3 for water in the gaseous and liquid
phases. The dierence, of course, is directly traceable to that in the DOSD as discussed
earlier.
Even if the statistical uctuation of energy loss is ignored, there would be a dierence
between the CSDA range and the penetration, because of large-angle elastic scattering (see
Sec. 2.5). The distribution of the nal position vector of the penetrating particle can be
obtained in terms of its energy-dependent stopping power and the dierential cross-section
of scattering [36]. Actual calculation in the general case is complex, but it simplies con-
siderably when the number of scatterings are >>1. The distribution can then be shown as
gaussian and spheroidal, i.e., W(r) =(2pA)
1
(2pB)
1/2
exp[{(x
2
+y
2
)/2A+(zhzi)
2
/2B}],
where A=hx
2
i and B=hz
2
ihzi
2
, and the angular brackets refer to mean values. The
mean square radial penetration is then given by r
2
_
2pm
l
0
r
4
drm
p
0
Wrsin hdh, where h is
the angle of the position vector relative to the z-axis, the initial direction of motion. The
root mean square (rms) range, dened by hr
2
i
1/2
is often the more signicant quantity.
Fig. 4 shows the CDSA and rms ranges in gaseous water as functions of electron energy
according to LaVerne and Mozumder [36]. In these calculations, stopping power was com-
puted by the procedure detailed above. The scattering cross-section was tted to swarm
and beam data by adopting Molieres modication of screened Rutherford scattering. It is
obvious that the rms range is always shorter than the CSDArange, the dierence becoming
more signicant at lower energies. The following general observations can be made for the
penetration distribution. (1) The memory of the initial direction is maintained by the
electron until it has lost f80% of the starting energy, after which the distribution becomes
spherical. (2) The electron penetrates a certain distance with only a few scatterings and then
diusive motion sets in. (3) The number of elastic scatterings needed to give a spherical
distribution increases with energy, being f15 and f74, respectively, for electrons of 1 and
10 keV energy in water.
Because of the statistical nature of collision processes, charged particles starting with
the same energy do not travel the same distance after losing a xed amount of energy, nor
do they have the same energy after traveling a xed distance, respectively giving rise to
pathlength straggling and energy straggling. Fano [3] adds a third kind of distribution of
distance for particles that have dropped below a certain energy at the last collision, which
Copyright 2004 by Taylor & Francis Group, LLC
is not necessarily the same as pathlength straggling, because of the discrete nature of the
energy loss process. While all these distributions are related, the most important
distribution is range straggling when the nal energy is reduced to insignicance. Bohr
[17] considers a group of energy losses and a large number of loss processes in each group,
thereby obtaining a gaussian distribution by virtue of the central limit theorem. This
means (d/dx)hDE
2
i =4pe
4
z
2
NZ, where hDE
2
i is the mean square uctuation in energy loss
of particles after traversing a distance x. The corresponding pathlength and range
distributions are also gaussian. These considerations apply mainly to heavy particles of
relatively high energy penetrating virgin media. The ratio of the root mean square range
dispersion to the mean range is typically a few percent for a heavy particle, but may be a
few tens of percent for electrons of moderate energy. Straggling is always more important
for electrons of any energy because of the large fractional energy that an electron can lose
in a single encounter. Also, if the sample penetrated is thin, the resultant energy
Figure 3 Density-normalized stopping power (MeV
cm
2
/g) of gaseous and liquid water as a
function of electron energy according to track simulation by Pimblott et al. [35]. There is a noticeable
phase eect, while a peak is seen at f100 eV in both phases.
Copyright 2004 by Taylor & Francis Group, LLC
distribution of emergent particles of any kind will not be gaussian, but will show a high
energy tail. Such LandauVavilov distribution has been extensively tabulated [33], having
experimental justication. Its application to radiation biology derives from the fact that
the radiation eect in a cell depends on the energy deposited in it, while the cell size is often
small compared to the particle range [37].
A collision-by-collision approach was introduced by Mozumder and LaVerne [38]
for range straggling of low-energy electrons without the consideration of large-angle
elastic scattering. The probability density of pathlength x for the rst collision is given by
P
1
(x) = K
1
exp(x/K), where the mean free path K is computed from the same dier-
ential cross-section of energy loss as used in the stopping power (vide supra). The integral
of the energy-loss weighted dierential cross-section hei gives the mean energy loss in a
single inelastic encounter. The ratio, hei/K, gives the spatial rate of energy loss for that
encounter, while the true stopping power is the mean value of the ratio of energy loss to
path length. To this extent, the procedure is an approximation. To have a pathlength
distribution P
n
(x) in n collisions, the electron must have had a distribution P
n1
(xy) in
(n1) collisions and a distribution P
1
( y) in the last collision. This generates a convolution,
P
n
x m
x
0
P
n1
x yP
1
ydy. By repeatedly using the Laplace transform method, one
can evaluate the integral with the result: P
n
x S
n
i1
K
1
i
S
n
i
expx=K
i
, where S
n
i
C
n
j1
1 K
j
=K
i
1
j p 1. Note that the mean free paths are energy-dependent. Therefore
a concurrent use of the stopping power is required to give the electron energy after each
collision. To obtain the range distribution, the total number of collisions is so adjusted
that the mean nal energy becomes immaterial. Thus, as shown in Fig. 5, Mozumder and
LaVerne [38] give the range distribution of 200-eV electrons in N
2
by convoluting the
pathlength distributions between energies 200 to 110 eV and between energies 110 to 31
eV. In this case, the nal energy has been taken to be twice the ionization potential. Notice
Figure 4 Continuous slowing down (CSDA) range and root mean square (rms) penetration in
gaseous water according to LaVerne and Mozumder [36]. See text for details.
Copyright 2004 by Taylor & Francis Group, LLC
that, in these calculations, the fundamental input still remains to be the DOSD. The
following remarks may be made about the range distribution of low-energy electrons. (1)
For electron energies below about 5 keV, CSDA>mean>median range. CSDA range is
the least reliable because it considers energy losses in a continuous manner within the
mean free path. (2) At low energies, the distribution is skewed toward longer ranges. The
skewness parameter approaches the gaussian limit of unity above f10 keV, almost
independent of the material traversed. (3) The most probable density-normalized range
for a 1-keV electron in H
2
O, N
2
and O
2
is 6.1, 7.1, and 8.4 Ag/cm
2
, respectively. However,
relative straggling, given by the ratio of FWHM of the range distribution to the most
probable value, is nearly independent of the medium, being 1.4, 0.7, and 0.4 at energies of
100, 500, and 2000 eV, respectively.
So far, the discussion presented here is based either on penetration deriving from
large angle elastic scattering, or on the statistical nature of inelastic collisions giving
straggling. As yet, there has not been any analytic theory that combines both, but Monte
Carlo simulation, including some form of realistic dierential cross-section of elastic
scattering, shows promise [35].
3. SECONDARY IONIZATION
3.1. Cross-Sections
Ionizations produced by electrons generated in the primary interaction of the incident fast
charged particle are called secondary. The denition may be extended to all successive
generations if sucient energy is available. In this sense, secondary ionization is very
Figure 5 Range distribution in nitrogen for 200-eV electrons obtained via the convolution method
at 110 eV [38]. The nal energy is about twice the ionization potential. See text for details.
Copyright 2004 by Taylor & Francis Group, LLC
important and it is a ubiquitous feature of all fast primary charged particles. Typically,
f25% of all ionizations are produced in secondary processes. Of course, experiments
count all ionizations, irrespective of whether they are primary or secondary.
The importance of total ionization yield derives from dosimetry and from providing
a standard against which other radiationchemical yields may be discussed [25]. For the
complete description of impact ionization, a vefold dierential cross-section is required in
principle; one for ejected electron energy and two each for the angular distributions of the
ejected electron and the incident particle. Such a detailed description is rarely needed. In
radiation chemical track structure, a cross-section doubly dierential in the energy and
angle of the ejected electron is sucient. Moreover, the Born approximation, against
which measurements and numerical calculations may be compared, starts to break down
as the incident energy is decreased. The onset for this breakdown occurs at a greater
energy if higher order dierential cross-sections are considered. A special issue of
Radiation Research [39] contains articles written by several experts on the experimental
and theoretical aspects of impact ionization, including higher order dierential cross-
sections. The calculation of the impact ionization cross-section for incident velocity v
proceeds, in the BornBethe approximation, in the same manner as that for energy loss
(see Sec. 2.2). In eect, energy loss is written as W
i
= E+I
I
, where E is the secondary
electron energy and I
I
is the ionization potential for the ith orbital from which the electron
is ejected. Combining the contribution from all the orbitals in the atom or molecule, the
dierential ionization cross-section for ionization may be given as follows [40].
dr=dE 4pa
2
0
z
2
R
2
=T
i
W
1
i
df=dW
i
ln4c
i
T=R 10
In Eq. (10), R is the Rydberg energy (13.6 eV), a
0
, the Bohr radius (0.053 nm), T=mv
2
/2,
df/dW
i
is the dierential dipole oscillator strength for the ionization process, and c
i
is a
dened additional contribution of the generalized oscillator strength [21]. Note that this
cross-section is basically given by the dipole oscillator strength, while there is a weak
dependence on the generalized oscillator strength through c
i
. Platzman has advocated the
use of the ratio of ionization cross-section to the Rutherford cross-section, Y(T,W) =(dr/
dE)(T/4pa
0
2
z
2
)(W/R)
2
, and its plot vs. R/W is called the Platzman plot [see Eq. (19) et seq.
of Chap. 3]. Here W=E+B
1
and B
1
is the lowest binding energy of the electron in
the molecular orbital. The usefulness of such a plot rests on: (1) lim
W!l
Z, the total
number of electrons, (2) autoionization and Auger processes show up as dips or peaks, (3)
for small values of E, Y resembles the shape of W(df/dW), and (4) in proton impact, a
peak is seen at ejected electron speed equal to the speed of the proton. Furthermore, the
area under the Platzman plot is proportional to the total ionization cross-section, the
constant of proportionality being T/4pa
0
2
R. Kim [40] has made repeated use of the Platz-
man plot in a careful analysis of secondary electron spectra, in some cases correcting some
experimental values. Another plot, (T/R)(dr/dE) vs. ln(T/R), called the Fano plot, is
useful in extrapolation and interpolation of experimental data. It approaches a straight
line at high T.
The above considerations need relativistic correction at vfc, which may be
performed in a straightforward manner. More importantly, Eq. (10) assumes that the
ionization process is direct, i.e., once a state above the ionization potential is reached,
ionization occurs with a certainty. Platzman [25] points out that in molecules, this is not
necessarily so and superexcited states with energy exceeding the ionization potential may
exist, which will dissociate into neutral fragments with a certain probability. For example,
in water in the gas phase, ionization occurs with a sharp threshold at the ionization
potential (I.P.)=12.6 eV, but only with an eciency of 0.4. Beyond the I.P., the ionization
Copyright 2004 by Taylor & Francis Group, LLC
eciency, determined by the ratio of ionization to the photoabsorption cross-section, varies
with the energy until f18 eV, beyond which it approaches unity [41]. In such cases, the
cross-section formula given in Eq. (10) must be suitably modied to account for the
ionization eciency. There is a very denite phase eect on the ionization eciency,
because of which the existence of superexcited states in liquid water has recently been
questioned [42]. The typical values of proton impact ionization cross-section in the gas
phase, for H, He,H
2
, N
2
, CO
2
, NH
3
, and CH
4
are 1.12, 0.93, 1.96, 5.33, 8.11, 4.62 and 6.84
A
2
, respectively, at 100 keV, and 0.19, 0.21, 0.38, 1.46, 2.13, 1.30, and 1.66, respectively, at 1
MeV. Rudd etal. [43] have given an empirical formula for the total ionization cross-section
due to a heavy ion at energy U, viz. r
t
1
=r
l
1
+r
h
1
, where r
h
(A
2
) =3.52(R/U)(A ln[1+U/
R]+B) and r
t
(A
2
) =3.52C(U/R)
D
. This form has the correct asymptotic behavior given by
the BornBethe cross-section at high energy. The parameters A, B, C, and D have been
tabulated, and a similar formula for incident electrons is also available. The corresponding
secondary electron spectrum involving some 10 adjustable parameters has been empirically
found to be very useful. It is discussed in some detail in Chap. 3 [Eqs. (29)(35)].
3.2. The Degradation Spectrum
The complexity of electrons of dierent energy and generation actually present in an
irradiated system, together with their primary counterparts, if dierent from the electron,
is represented by the degradation spectrum, which is dened to be the normalized energy
spectrum of these particles owing through the surface of a small cavity inside the
medium. Spencer and Fano [44] show that a convenient measure of this spectrum is given
by y(T
0
,T), such that y(T)dT is the total mean distance traveled by electrons of all
generations (primary, secondary, etc.) between energies T and T+dT, with T
0
being the
initial energy at unit ux. The degradation spectrum serves a purpose similar to the
distribution function in statistical mechanics, in the sense that the yield of any primary
species x is given by N
x
T
0
Nm
T
0
I
x
r
x
TyT
0
; TdT, where r
x
is the cross-section for the
production of x with an energetic threshold I
x
(e.g., for ionization). Note that y needs to be
evaluated only once for each initial energy, after which it may be used for all products.
Although y depends on T
0
, some approximate scaling property of the electron degradation
spectrum has been noted [45].
Calculation of the degradation spectrum has proved to be dicult in all but the
simplest cases of H, He, and H
2
, because of the lack of accurate energy-loss cross-sections
(mainly nonionizing) over a wide span of energy. Furthermore, to calculate the yield of a
particular species, its own production cross-section must be known over a wide interval of
energy, which is rarely available. In the continuous slowing down approximation (CSDA),
the contribution to the degradation spectrum of a single electron is the reciprocal of the
stopping power. Therefore after knowing the ionization cross-section and the electron
stopping power, one can compute the partial degradation spectrum generation by
generation. Finally, the total degradation spectrum may be obtained by summing up.
Such a procedure has been followed by Kowari and Sato [46] for gaseous water. Apart
from demonstrating the methodology, the actual results are of limited use because of
various dubious cross-sections used. More sophisticated methods employ the Spencer
Fano equation [45], or the Fowler equation [47]; however, the problem of accurate cross-
sections still remain in most cases. Another numerical approach by Turner et al. [48] relies
on Monte Carlo simulation. Their computed spectra show considerable dierence between
the gas and liquid phases of water, especially in the low energy region, which is a direct
consequence of the peculiar kinds of cross-sections used by them in these phases. One
feature is common to all electron degradation spectra. It has large values at the high
Copyright 2004 by Taylor & Francis Group, LLC
starting energy and also at very low energies. The former is attributed to low stopping
power and the latter is attributed to the accumulation of electrons of dierent generations.
Therefore the spectrum is expected to show a drop at some intermediate energy, often near
where the stopping power goes through a broad peak. It should be noted that, like the
stopping power and the range, the degradation spectrum also refers to mean values, and
considerable uctuation may be expected, especially at higher energies.
3.3. The W Value
The reciprocal of the ratio of the mean number of ionizations n
i
produced by the complete
absorption of a primary particle of energy T is dened as the (integral) W value. Ioniza-
tions produced by all secondary electrons are counted in it. A dierential, or track segment
x value, may be similarly dened, so that 1/x=(d/dT)(T/W). If W is independent of T,
so is x and vice versa. Generally, W depends weakly on T, unless the initial energy is very
low. Its dependence on the nature of the incident radiation (electrons, alpha-particles, etc.)
is also minor. These factors constitute the basis of dosimetry, by collecting and measuring
the total ionization. It also means that the Bragg curve, or the variation of total ionization
with incident particle energy, is a reliable measure of the relative stopping power. Thus the
W value is mainly a property of the medium, determined by the cross-sections of ioniza-
tion and of other nonionizing energy-loss processes. As such, the W value often depends
on the phase.
Platzmans [49] analysis for the W value may given as W = hE
ion
i+v
ex
hE
ex
i+hE
s
i,
where hE
ion
i, hE
ex
i, and hE
s
i are the average energies for ionization, excitation, and
subexcitation electrons, respectively, and v
ex
is the relative number of excitations to
ionizations. It provides a basis for understanding the insensitivity of W on particle energy,
because the quantities involved depend not so much on absolute cross-sections for
ionization and nonionizing energy loss, as it depends on their ratio.
An early theory of the W value was proered by Spencer and Fano [44], based on the
degradation spectrum. Another method, the Fowler equation, was employed by Inokuti
[47] for electron irradiation, based on the approximation that there is only one ionization
potential and that the ionization eciency is unity. These restrictions can be relaxed. The
main result of Inokutis analysis may be given as follows.
WT W
l
1 U=T
1
; where U EV I: 11
In Eq. (11), W
l
is the asymptotic W value at very high energy, I is the ionization potential,
U is a so-dened energy parameter, and EV is interpreted as the average energy transfer
in ionizing collisions where the ejected electron is subionization, i.e., incapable of further
ionization. Thus, U is the excess of EV over the ionization potential. An expression of
EVmay be given in terms of the dierential probabilities of ionization with energy loss
between E and E+dE, viz. EV m
2l
l
Edp
i
E; T=dEdE=m
2I
I
dp
i
E; T=dEdE, where these
probabilities are simply the ratios of the relevant ionization cross-sections to the total
energy loss cross-section. Eq. (11) impressively applies for various media over a wide span
of energy, except perhaps at very low energy.
Another numerical method devised by LaVerne and Mozumder [50] has been applied
to gaseous water under electron and proton irradiation. Considering a small section of the
track, the Wvalue due tothe primaryparticle only, maybe writtenas x
P
=S(E)/Nr
i
(E), where
S(E) is the stopping power, N is the molecular density, and r
i
(E) is the total ionization
cross-section at energy E. The number of secondary electrons produced, in all generations,
per unit track length is given byNm
e
m
I
dr
i
=dee=Wede, where W(e) is the integral W value
Copyright 2004 by Taylor & Francis Group, LLC
for electrons at energy e, dr
i
(e)/de is the dierential cross-section for producing secondary
electrons of energy e, and e
m
is the maximum secondary electron energy often well
approximated by (EI)/2. Combining, the dierential x value at electron energy E is given
by xE SE=Nr
i
m
e
m
I
dr
i
=deeW
1
ede. From this equation, the overall W value
may be calculated through integration by using the stopping power, ionization cross-
section, and the integral W values obtained at lower energies. A building-up principle has
been used by dividing the incident energy into several intervals, I3I, 3I7I, 7I15I, and so
on. In the rst interval, no secondary ionization is possible and x(E) = x
P
(E), although
an integration is necessary to compute W(E). In subsequent intervals, the W(E) values of
previous intervals may be used together with stopping power and ionization cross-section
obtained either from experiments or tabulations. By careful analysis of experiments and
compilations, LaVerne and Mozumder [50] made the following conclusions. (1) Various
experiments on the ionization cross-section are basically in agreement with each other, but
those on the excitation cross-section are not, which is the major source of uncertainty in
the calculation of W value. (2) For electron irradiation, an asymptotic W
l
is reached at 1
keV, but the computed value (34.7 eV) exceeds measurement [51] (29.6 eV) by 15%,
attributable to errors in inelastic cross-sections. (3) For proton irradiation, W
l
is reached
at 500 keV and the computed value (28.9 eV) is in good agreement with experiment [51]
(30.5 eV). In both cases, x
P
greatly exceeds W, underscoring the importance of secondary
ionization.
Cole [52] measured the W value in air for electron energies from 5 to 20 eV, while the
electrons were completely absorbed in the ionization chamber. Later, Combecher [53]
extended the measurements of W(E) to several gases including water vapor. Fig. 6 shows
the variation of W(E) with electron energy in water vapor, as measured by Combecher,
Figure 6 Comparison of experimental (. . .) variation of integral W value for water vapor [50] with
theoretical calculation ( ___ ) and with an empirical model (- - -). Note that the accuracy of the
theoretical calculation is limited by an inherent error in inelastic collision cross-sections, while in
the empirical model, U = 12.6 eV has been adjusted to obtain best agreement with the experiment.
The experimentally determined asymptotic limit, W
l
= 29.6 eV is indicated by the arrow.
Copyright 2004 by Taylor & Francis Group, LLC
and compares it with the calculation of LaVerne and Mozumder and also with Inokutis
result, obtaining U = 12.6 eV, close to the ionization potential, for best adjustment.
The eect of condensation on the W value has been treated elsewhere in some detail
[54]. Except for liqueed rare gases, where the W value in the liquid is somewhat less than
that in the gas phase, there is an operational problem in the denition of Wthus
requiring long extrapolation. In some hydrocarbons such as cyclohexane, the admittedly
long extrapolation of scavenging yield gives a W value not far from the level obtained in
the gas phase (f26 eV for cyclohexane). In most polar liquids, W
liq
(22 F 1 eV for water)
is considerably less than W
gas
(31 F 1 eV for water). The exact reason for the additional
yield of ionization in the liquid phase is not clear, but a signicant contributing factor
could be the conversion of excited states into ionization by interaction with neighboring
ground state molecules in the liquid phase.
As for the stopping power and the degradation spectrum, the W value is also a mean
quantity and ionization uctuation is expected because of the various ways a deposited
energy can be partitioned into ionizing and nonionizing events. The Fano factor represents
such uctuations, and it is dened as the ratio of mean square uctuation of ionization to
the average number of ionizations, i.e., F = (hn
2
i
ihn
i
i
2
)/hn
i
i, where n
I
is the number of
ionizations detected in a sample. For a Poisson distribution, F = 1; for other distributions,
it is <1. For the operation of a radiation detector, a smaller Fano factor means better
energy resolution. With solid state detectors using Si or Ge, Fano factors of f0.05 can be
achieved. Other liqueed rare gas detectors have Fano factors in the range f0.1 to f0.2.
4. CONCLUSION
In this chapter, we have attempted to describe the interaction of fast charged particles with
matter from the point of view of the incident radiation, i.e., basically radiation physics. We
have discussed such topics as energy transfer from incident radiation to ionized and
excited states of medium molecules, as well as theories of stopping power and secondary
ionization eects. In many of the succeeding chapters, topics which are more chemical in
nature will be found, i.e., the response of the matter receiving the transferred energy. In
this sense, this chapter may be considered as preparatory groundwork for the succeeding
contributions.
REFERENCES
1. Tabata, Y.; Ito, Y.; Tagawa, S., Eds. CRC Handbook of Radiation Chemistry. Boca Raton,
FL: CRC Press, 1991, chaps. 2 and 3.
2. Bethe, H.A.; Heitler, W. Proc. Roy. Soc. A 1934, 146, 83.
3. Fano, U. Annual Review Nuclear Science 1963, 13, 1.
4. Mozumder, A. In Advances of Radiation Chemistry; Burton, M.; Magee, J.L., Eds.; Wiley-
Interscience: New York, 1969, chap. 1.
5. Ward, J.F. In Advances in Radiation Biology; Lett, J.T.; Adler, H., Eds; Academic Press: New
York, Vol. 5,181239.
6. Land, L.L.; Hanrahan, R.J. In Environmental Applications of Ionizing Radiation; Cooper, W.J.;
Curry, R.D.; OShea, K.E.; Eds.; John Wiley: New York, 1998, chap. 24.
7. Woods, R.J.; Pikaev, A.K. Applied Radiation Chemistry. John Wiley: NewYork, 1994, chap. 10.
8. Rossi, H.H. In Radiation Dosimetry, Attix, F.H.; Roesch, W.C., Eds.; Academic Press: New
York, 1968, 4392.
Copyright 2004 by Taylor & Francis Group, LLC
9. Kellerer, A.M.; Rossi, H.H. In Proceedings of 2nd Symposium on Microdosimetry, Ebert, H.G.,
Ed.; Euratom: Brussels, 1969, 843853.
10. Fano, U. Comparative Eects of Radiations; Burton, M.; Kirby-Smith, J.S.; Magee, J.L., Eds.;
John Wiley: New York, 1960, 1421.
11. Mott, N.F. Proc. Roy. Soc. A 1930, 126, 79.
12. Killat, U. J. Phys. C 1974, 7, 2396.
13. Fano, U. Revs. Mod. Phys. 1992, 64, 313.
14. LaVerne, J.A.; Mozumder, A. Radiat. Res. 1993, 133, 282.
15. Zirkle, R.E.; Marchbank, D.F.; Kuck, K.D. J. Cell. Comp. Physiol. 1952, 39 (suppl. 1), 75.
16. Bohr, N. Philos. Mag. 1913, 25, 10.
17. Bohr, N. Kgl. Dan. Vied. Selskab. Math.-Fys. Medd. 1948, 18, 9.
18. See, for example, Kuppermann, A. J. Chem. Ed. 1959, 36, 279.
19. Bethe, H.A. Ann. Physik. 1930, 5, 325; 1933, 24, 273.
20. Bethe, H.A. Z. Physik. 1932, 76, 293.
21. Inokuti, M. Rev. Mod. Phys. 1971, 43, 297.
22. Magee, J.L. Annu. Rev. Phys. Chem. 1961, 12, 389.
23. Bloch, F. Ann. Physik. 1933, 16, 285.
24. Lassettre, E.N.; Skerbele, A.; Dillon, M.A. J. Chem. Phys. 1969, 50, 1829.
25. Platzman, R.L. In Radiation Research, Silini, G., ed. Amsterdam: North Holland, 1967,
20.
26. Nicholls, R.W.; Stewart, A.L. In Atomic and Molecular Processes, Bates, D.R., Ed.; Academic
Press: New York, 1962, 47.
27. Inokuti, M.; Kim, Y.-K.; Platzman, R.L. Phys. Rev. 1967, 164, 55.
28. Miller, W.F.; Platzman, R.L. Proc. Phys. Soc. (London) A 1957, 70, 299.
29. Zeiss, G.D.; Meath, W.J.; Macdonald, J.C.F.; Dawson, D.J. Radiat. Res. 1975, 63, 64.
30. Heller, J.M.; Hamm, R.N.; Birkho, R.D.; Painter, L.R. J. Chem. Phys. 1974, 60, 3483.
31. LaVerne, J.A.; Mozumder, A. J. Phys. Chem. 1986, 90, 3242.
32. Koizumi, H.; Sinsaka, K.; Hatano, Y. Radiat. Phys. Chem. 1989, 34, 87.
33. Studies in Penetration of Charged Particles in Matter. National Academy of SciencesNational
Research Council: Washington, DC, 1964, (publication 1133).
34. The Stopping and Ranges of Ions in Matter; Ziegler, J.F., Ed.; Pergamon Press: Elmsford, New
York, 19801985, 16.
35. Pimblott, S.M.; LaVerne, J.A.; Mozumder, A. J. Phys. Chem. 1996, 100, 8595.
36. LaVerne, J.A.; Mozumder, A. Radiat. Res. 1983, 96, 219.
37. Maccabee, H.D.; Raju, M.R.; Tobias, C.A. Phys. Rev. 1968, 165, 469.
38. Mozumder, A.; LaVerne, J.A. J. Phys. Chem. 1985, 89, 930.
39. Various, Radiat. Res. 1975, 64, 1204.
40. Kim, Y.-K. Radiat. Res. 1975, 64, 96.
41. Haddad, G.N.; Simpson, J.A.R. J. Chem. Phys. 1986, 84, 6623.
42. Mozumder, A. Phys. Chem. Chem. Phys. 2002, 4, 1451.
43. Rudd, M.E.; Kim, Y.-K.; Madison, D.H.; Gallagher, J.W. Rev. Mod. Phys. 1985, 57, 965.
44. Spencer, L.V.; Fano, U. Phys. Rev. 1954, 93, 1172.
45. Fano, U.; Spencer, L.V. Int. J. Radiat. Phys. Chem. 1975, 7, 63.
46. Kowari, K.; Sato, S. Bull. Chem. Soc. Jpn. 1978, 51, 741.
47. Inokuti, M. Radiat. Res. 1975, 64, 6.
48. Turner, J.E.; Paretzke, H.G.; Hamm, R.N.; Wright, H.A.; Ritchie, R.H. Radiat. Res. 1982,
92, 47.
49. Platzman, R.L. Int. J. Appl. Radiat. Isot. 1961, 10, 116.
50. LaVerne, J.A.; Mozumder, A. Radiat. Res. 1992, 131, 1.
51. Christophorou, L.G. Atomic and Molecular Radiation Physics; Wiley-Interscience: London,
1971.
52. Cole, A. Radiat. Res. 1969, 38, 7.
53. Combecher, D. Radiat. Res. 1980, 84, 189.
54. Mozumder, A. Fundamentals of Radiation Chemistry; Academic Press: San Diego, 1999.
Copyright 2004 by Taylor & Francis Group, LLC
3
Ionization and Secondary Electron Production
by Fast Charged Particles
Larry H. Toburen
East Carolina University, Greenville, North Carolina, U.S.A.
1. INTRODUCTION
From the rst studies of cathode rays leading to the discovery of the electron in 1897 by
J.J. Thomson, there has been intense interest in understanding the characteristics of the
interactions of fast charged particles with matter. Early studies of electron impact
excitation of gases led to important constraints on evolving models of atomic structure
as well as contributing to the discovery of x-rays and natural radioactivity. With the dis-
covery of energetic alpha particles emanating from natural radioactivity, Rutherford and
his coworkers were able to investigate heavy charged-particle scattering in materials and
gain new insights into the physical nature of atomic structure and of charged particle
scattering. These studies also stimulated new questions regarding the mechanisms of ener-
gy loss by heavy charged particles in matter.
The rst decades of the 20th century saw great advances in our knowledge of charged
particle interactions with matter. With the discovery of nuclear ssion came enhanced
availability of radionuclides and nuclear radiations, including fast ssion fragments, neu-
trons, alpha and beta particles. Coincidently, with the development of nuclear technology
came the need to better understand the physical, chemical, and biological interactions of
radiation with matter. The potential exposure of materials and people to nuclear radiation
emphasized a need to understand the underlying mechanisms leading to damage by ra-
diations of dierent types. The early part of the 20th century was also a period of rapidly
expanding technology enabling one to produce and accelerate electrons and ions and to
explore new applications of radiation in science and industry. Understanding of the in-
teraction of radiation with matter has important applications in numerous research and
applied elds of endeavor, including the sterilization of food and medical instruments,
medical diagnosis and treatment of disease, and developments of nuclear energy and radio-
active waste management (see the Preface and Chap. 26). In this chapter, we will explore
the fundamental mechanisms for the interaction of fast charged particles with matter, with
focus on their signicance to the subsequent chemistry initiated by the absorption of ioniz-
ing radiation.
Ionizing radiation, as the term implies, denes those radiations that interact with
matter by the production of charged particles, namely electrons and residual positive ions.
Copyright 2004 by Taylor & Francis Group, LLC
Energetic electromagnetic radiation, i.e., x-and g-rays, interact with matter predominately
via production of photoelectrons and Compton electrons, and, if the photon energy is
greater than about 1 MeV, the production of electronpositron pairs. These primary
electrons can have a broad spectrum of energies from subexcitation, or very slow
electrons, to what are generally called fast charged particles. Here fast implies particles
with velocities much larger than bound electrons of the medium. As we will see, electrons
and fast heavy charged particles, e.g., protons and alpha particles, predominantly interact
with the bound electrons of the medium producing secondary electrons via target
ionization. For completeness, we note that the interaction of neutrons with matter also
leads to local energy deposition by charged particles; neutrons primarily interact through
elastic nuclear scattering, resulting in the production of energetic recoil charged particles
characteristic of the absorbing medium. Ionization resulting from the slowing of energetic
ions and electrons leads to the production of charged and/or neutral molecular fragments,
reactive radicals, and other excited chemical species that foster subsequent chemical
reactivity in the absorbing medium. Because all ionizing radiation leads to electron
production, the dierent chemical yields observed for dierent types of radiation must
depend on the detailed spatial distributions of initial events produced in the physical
process of electron slowing. All radiation-induced chemistry begins with the stochastic
physical processes involved in energy deposition by charged particles.
Investigation of the interactions of charged particle with matter has traditionally
been viewed as comprised of studies falling into two fundamental classes: Class I studies
focus on the fate of the particle, and class II studies focus on the fate of the absorbing
medium [1] (cf. Sec. 2.1.2). The former is most commonly represented by the studies of
charged-particle stopping powers, whereas the latter includes studies of media quantities
such as the average energy per ion pair (W value), or more specically, the fundamental
processes of ionization and excitation that contribute to spatial structure in the patterns of
energy deposition. In the following, we will rst discuss the stopping of fast charged
particles through the general theory of stopping power, its strengths and limitations, and
then look more closely at the individual processes that contribute to energy loss and are
incorporated within the general concepts of stopping power.
2. STOPPING POWER
When a fast charged particle passes through matter, it loses energy by interactions in-
volving momentum transfer to the bound electrons of the media. If the particle slows to
the point that insucient momentum can be transferred to excite these bound electrons to
vacant states, energy loss will then occur through elastic collisions with the target atom, or
the target matrix, as a whole. This latter energy-loss mechanism, termed nuclear scat-
tering, dominates for ion energies less than a few keV per atomic mass unit (keV/u), i.e.,
for ions near the extreme end of their range. For high-energy or fast particles, i.e., particles
with velocities large relative to the velocities of bound electrons, inelastic collisions involv-
ing electron excitation and ionization dominate the energy loss process. At intermediate
energies, the competition between elastic and inelastic processes depends on the atomic,
molecular, and condensed-phase properties on the absorbing medium and, for heavy
charged particles, additional channels of electron capture and loss can become important
mechanisms for energy loss.
Stopping power has been the subject of study for more than a century beginning with
the pioneering work of Thompson and Rutherford. With the discovery of nuclear ssion
Copyright 2004 by Taylor & Francis Group, LLC
and subsequent developments in accelerator technology, energetic heavy ions with rel-
atively high energies became readily available and the experimental and theoretical study
of stopping power made steady advances (see, e.g., Refs. 24; and the excellent review of
stopping power for protons and alpha particles published in 1992 by the International
Commission on Radiation Units and Measurements [5]). These studies have been an im-
portant means of testing our theoretical understanding of charged particle interactions
with matter as well as being of practical use in many scientic and applied elds of ap-
plication. Because stopping power can be measured with relatively high accuracy using
modern charged particle detection techniques, of order 1%, we can rene our theoretical
understanding of the penetration of charged particles in matter to a high degree by com-
parisons of theoretical and experimental results.
Many textbooks describe derivations of the dominant features of energy loss incor-
porated in stopping power (see, e.g., Refs. 6 and 7). The primary energy loss mechanism
for a fast charged particle is the interaction between the coulomb charge of the moving
charged particle and that of electrons of the absorbing medium. This interaction is gen-
erally approximated as a transverse momentum impulse delivered to the bound electron by
the coulomb charge of the passing charged particle integrated over the interaction times
and impact parameters. A detailed derivation of stopping power has been presented else-
where (see Sec. 2.2). The important concept from stopping power is that the energy loss by
the fast particle is to the bound electrons of the medium. Increasingly complete theories of
stopping power, from detailed rst-order approximations to full quantum calculations, are
described in a number of reviews (see, e.g., Ref. 5 and references therein). The expression
for the mass stopping power for a charged particle of velocity v and atomic number z in a
medium of atomic number Z and atomic mass A is given in ICRU report 49 [5] as
1
q
dE
dx
0:30708
b
2
z
2
Z
A
Lb
MeV cm
2
g
_ _
; 1
where L(b) is obtained from the sum of three terms
Lb L
0
b zL
1
b z
2
L
2
b: 2
The parameters in Eq. (1) illustrate the primary dependence of stopping power on the
square of the projectile charge z, the velocity of the projectile b, and the density of target
electrons Z/A. The mean excitation energy I, a major target parameter, is incorporated in
the rst term of L(b) given as
L
0
b
1
2
ln
2mc
2
b
2
W
m
1 b
2
_ _
b
2
lnI
C
Z
d
2
3
with
W
m
2mc
2
b
2
1 b
2
1 2m=M1 b
2
1=2
m=M
2
1
: 4
Equation (3) incorporates relativistic eects, eects of target density, and corrections to
account for binding of inner-shell electrons, as well as the mean excitation energy; C/Z is
determined from the shell corrections, d/2 is the density correction, W
m
accounts for the
maximum energy that can be transferred in a single collision with a free electron, m/M is
the ratio of the electron mass to the projectile mass, and mc
2
is the electron rest energy. If
the value in the bracket in Eq. (4) is set to unity, the maximum energy transfer for protons
Copyright 2004 by Taylor & Francis Group, LLC
is only overestimated by 0.1% at 1 MeV, and by 0.23% at 1000 MeV [5]. The shell cor-
rections are included to reect the fact that, as the ion velocity decreases, the assumption
of the particle being fast can fail for interactions involving inner shell electrons; thus
contributions from inner-shell electrons decrease and corrections to the Bethe theory are
required. The density correction accounts for the fact that a moving charge tends to po-
larize the transport medium thereby reducing the stopping power slightly. The shell correc-
tions are predominately a low-energy phenomenon whereas the density correction occurs
for high energies. The second term in Eq. (4) is required to account for the observation
that stopping power for negatively charged particles is smaller that that for positively
charged particles at the same velocity. This correction is often called the Barkas correction
after his measurements determined that the range of negative pions was somewhat larger
than that for positively charged pions. This correction is often simply referred to as the z
3
correction because it depends on the cube of the particle charge. The last term in Eq. (4) is
needed to bring agreement between the rst-order theory of Bethe and a full quantum
mechanical result of Bloch valid for high-energy particles. All of these additional cor-
rection factors have come about because of approximations made in the theoretical
development of stopping power theory and because the ability to make accurate measure-
ments of stopping power enables one to identify subtle dierences between theory and
experiment. The reader is directed to ICRU Report 49 [5] and to Chap. 2 for details in-
cluding the full description of stopping power and the terms identied in Eqs. (1)(4).
The most signicant target parameters for the determination of stopping power
are the electron density and the mean excitation energy I found in the logarithmic term of
Eq. (3). The mean excitation energy is related to the oscillator strength distributions of the
absorber and is, in general, determined from tting experimental data after all corrections
to the rst-order theory have been included. Once values of I have been determined for
elemental constituents of molecules or materials, one can obtain I for a composite
material by
lnI
i
n
i
Z
i
lnI
i
i
n
i
Z
i
; 5
where n
i
, Z
i
, and I
i
represent the atomic concentration, the atomic number, and the mean
excitation energy of the ith component of the target molecule or compound. This tech-
nique has been extensively used by Porter et al. (see Refs. 59 and references therein) to
determine stopping power for composite materials such as mylar and Al
2
O
3
. It is this
formula for determining the mean excitation energy for compounds from individual ele-
ments that also leads to additivity among stopping powers. One can obtain the stopping
power for a compound from a weighted sum of the stopping powers of the constituent
elements.
The discussion of stopping power above has focused on interactions by bare fast
charged particles. As fast positively charged particles slow, they can capture and sub-
sequently lose electrons; thus they can move through material with a somewhat reduced
average charge. Electrons bound to the projectile screen the projectile charge, reducing
the interaction strength, thereby reducing the stopping power. This reduction is generally
incorporated in stopping power theory by assuming a reduced value of the projectile
charge, z
e
in Eq. (1). This reduced eective charge is dened as the square root of the
ratio of the stopping power for the screened projectile to that of the bare projectile. It must
be emphasized that this eective charge of stopping power theory is not the same as the
Copyright 2004 by Taylor & Francis Group, LLC
average charge state of the moving ion. The dierence exists because interactions leading
to energy loss occur at a somewhat dierent range of impact parameters than those as-
sociated with the capture and loss of electrons, and the screened nuclear charge of the
moving particle is strongly dependent on the impact parameter. The eective charge of
stopping power theory is an average over all individual stochastic events contributing to
the stopping power, each of which is weighted by an eective charge z
e
(E) that depends
on the energy loss E occurring in that event [10]; this is discussed in more detail later in this
chapter. The dierence between the eective and average charge is illustrated in Fig. 1
where the eective/average charge of a proton is plotted vs. the ratio of the projectile ve-
locity to the bound target electron velocity. Data shown in Fig. 1 were derived from stop-
ping power measurements of Yarlagadda et al. [11], the eective charge formula of Barkas
derived for stopping power [12], and the average charge obtained from charge-transfer
measurements [13]. Note that only for large velocities of the projectile, where the projectile
in essentially bare, are the average and eective charges the same.
The importance of the various interactions that contribute to stopping power is
illustrated for a range of energies involved in stopping a fast proton in Fig. 2. These data,
taken from the compilation by Uehara et al. [14], clearly illustrate the importance of target
ionization as the leading contributor to energy loss by protons: ionization by protons
occurs at the higher energies and ionization by H
o
at the lower energies. Energy loss
processes involving electron capture by protons and electron loss by H
o
are also important
at particle energies less than about 100 keV; or energies less than a few hundred keV/u for
heavier charged particles (not shown). Excitation is observed to be a small contribution to
Figure 1 Comparison of the eective charge of a proton obtained from the work of Yarlagadda et
al. [11], the eective charge obtained from the empirical formula of Barkus [12], and the average
charge
_
q determined from charge transfer cross section tabulated by McDaniel et al. [13]. The solid
line is a guide through the data of Yarlagadda et al.
Copyright 2004 by Taylor & Francis Group, LLC
total energy loss at all energies and never exceeds about 10% even at its maximum con-
tribution. However, it should be emphasized that contributions to energy loss attributed to
excitation are relatively uncertain having been estimated from an empirical model based
on limited data [15]; there is little actual data available for excitation by protons or other
charged, or neutral, heavy particles.
There are also many channels of energy loss that have not been included in Fig. 2
because they either contribute only a small amount to the total energy loss, or because we
lack reliable data to assess their contribution. However, it is possible that the special
nature of some of these less-probable interactions might have weighted important in the
evolution of subsequent chemical reactions. For example, multiple ionization of the target
atom or molecule involving the simultaneous ejection of two or more electrons in a single
event has been documented for ion collisions with rare gas atoms (see, e.g., Refs. 16 and
17, and references therein). Although these interactions are relatively rare events, the
emission of two or more electrons correlated in time and space might have important
consequences on the subsequent chemistry [18]. Unfortunately, there are few data avail-
able to guide our understanding of the relevance of such interactions in chemical systems.
Energy loss by ionization of inner-shell electrons is also missing from Fig. 2. Ionization of
the K-shell of oxygen in the water molecule happens only rarely compared to outer shell
interactions, but inner shell ionization followed by Auger electron emission can lead to
multiply ionized targets with unique chemical properties. When estimating the importance
of energy deposition events, the subsequent chemical reactions under consideration can
oer the dening criteria as to whether the rare events have weighted importance. Also
lacking from Fig. 2 is information on the contributions leading to dierent molecular end
points such as dissociative excitation, dissociative ionization, autoionization, etc. Surely
such processes occur and are relevant to understanding the subsequent chemical activity
Figure 2 Contributions to the total stopping cross section S
T
from ionization r
i
by H
+
and H
o
,
electron capture r
EC
by H
+
, electron loss r
EL
by H
o
, and excitation r
EX
by protons in water. The
data are from a compilation by Uehara et al. [14]. The ICRU recommendation S
T
-ICRU is from
Ref. 5.
Copyright 2004 by Taylor & Francis Group, LLC
following energy deposition by fast charged particles, but again there is little information
regarding these processes for heavy charged particles that can be used to dene their im-
portance. On the other hand, a great deal is known regarding the interactions of electrons
with molecules and the production of various nal molecular states (see, e.g., Ref. 19), this
will be addressed below.
The discussion of stopping power presented above was not meant to be rigorous, but
to illustrate that extensive study has been devoted to both experimental and theoretical
features of stopping power, and that these results provide benchmarks against which
stochastic models of energy loss can be tested. Stopping powers are known with extreme
accuracy and any alternative full description of the stopping of charged particles, e.g.,
stochastic trackstructure models, MUST be consistent with known stopping powers (cf.
Chap. 4). Second, we note that the stopping power gives essentially no information on the
fate of the energy deposited, it is a particle-centered parameter. Thus stopping power, no
matter how accurate, is of limited use in generating detailed information as to what chem-
istry might follow energy deposition by ionizing radiation.
To model the potential chemical reactions that might follow energy deposition, it is
necessary to better understand the local spatial characteristics of the initial products of
ionization and excitation, e.g., are products within reaction distances, is recombination
likely prior to chemical reaction, etc. This information is contained in the collection of
many small events clustered along the path of the particle, and to understand the sub-
sequent chemistry, one must describe the basic interactions that accompany energy depo-
sition on a spatial scale relevant to the reaction of interest (see Chap. 4). For fast protons
and electrons, the important fundamental interactions are fairly straightforward. Fast ions
deposit about 90% of their energy in ionizing collisions with about 10% of the energy
deposited as excitation. Thus if we understand ionization, the production and slowing
down of secondary electrons, we can understand how most of the energy is deposited; but
be aware that determining the fate of the energy in the absorbing medium requires in-
formation on the dierential ionization cross sections. For fast electrons and other bare
particles, e.g., protons and alpha particles, there are many sources of experimental and
theoretical data; the same is not true for heavier particles or for slow particles that carry
bound electrons.
3. W VALUES
As has been emphasized, energy deposition by energetic charged particles is a stochastic
process involving large numbers of events, each event involving small quantities of energy
transfer. All detection devices used in the laboratory provide some form of averaging,
whether we are counting numbers of target ions created, the number of electrons or
photons emitted, yields of radicals, etc. Stopping power is the stochastic average over all
energy loss processes of the projectile. On the other hand, the W value is a media-centered,
class II, quantity representing the average energy required to produce an ion pair by the
passage of a charged particle (cf. Sec. 2.3.3). Thus W is given by
W
E
J
; 6
where E is the total amount of energy absorbed in a volume of gas and J is the total
number of ions produced in that volume. This quantity has been, as was stopping power,
Copyright 2004 by Taylor & Francis Group, LLC
the subject of extensive investigation for more than a century (see, e.g., Ref. 20 and
references therein) and provides a media-centered quantity of relatively high accuracy for
testing of quantities obtained from stochastic averages of energy deposition events. As was
the case for stopping power, this average provides little information on the actual events
produced during the absorption of radiation energy and thereby is of little guidance in
estimating the chemical reactions that might follow energy deposition. On the other hand,
the study of W values provided the rst approach to the analysis of energy deposition in
terms of a description of the target as a mixture of oscillators whose frequencies are
characteristic of the spectral absorption frequencies of the target [21]. In addition, as an
accurately measurable target property, it provides another benchmark that any stochastic
calculation must return when averaged over sucient numbers of interactions.
4. CHARGED PARTICLE TRACK STRUCTURE
Both stopping power and W values give averages over all energy deposition events making
up the path of the charged particle in the medium of interest. As averages, they tell us little
of the individual processes that can aect the subsequent chemical reactions initiated by
the absorption of ionizing radiation. Historically, the term linear energy transfer (LET)
was dened to help describe information on the local densities of energy transferred to
the medium per unit length of particle track (class II), as distinct from the energy lost
dened by stopping power (class I). This was a rst step in better understanding the eect
of local energy density on chemical yields. The quantity LET diers from stopping power
in that excitation or ionization resulting in emission of photons that might carry energy
from the local area of interest are not included in LET. However, the use of LET as a
radiation parameter has been found inadequate for many applications because particles of
the same LET can produce quite dierent yields of chemical products [22]. An extension of
the LET concept to dene a restricted LET
D
to either (1) restrict the energy deposited to be
within a certain radius of the charged particle track, or (2) to restrict the magnitude of the
energy deposited in any single event, was made to further investigate the eects of local
energy deposition. These denitions both eectively restrict the energy of secondary elec-
trons that are included in the analysis of events along the charged particle track. Although
restricted LET has had some success in radiation biology [23,24], it has found little ap-
plication in radiation chemistry.
The importance of the stochastic nature of energy deposition in nonhomogeneous
reaction kinetics was emphasized in early studies of molecular yields by Mozumder and
Magee [25,and,referentherein,therein]. They realized that the density of energy deposition
would randomly vary along a charged particle track and that these variations would be a
function of the primary electron energy. Their rst eort to incorporate the statistical
distribution of energy along charged particle tracks was to dene regions called spurs in
which energy deposition of up to 100 eV would occur stochastically distributed along a
particle track. Later, these entities were augmented with regions called blobs, in which
energy depositions from 100 to about 500 eV were dened, and with short tracks, which
would include energy deposition from 500 eV to about 5 keV. These entities represent
regions of signicantly dierent chemical reactivity brought about by the statistical nature
of energy deposition. The probabilities for energy deposition associated with each of these
track entities were either determined analytically or by the use of Monte Carlo techniques.
Their estimates of the frequency of such entities were somewhat crude at the time owing to
the lack of detailed information on the interaction of electrons in the medium of interest,
Copyright 2004 by Taylor & Francis Group, LLC
but their results were successful in illustrating the importance of spatial variations in en-
ergy deposition on the subsequent radiation chemistry. These pioneering studies have led
to a large number of new Monte-Carlo-based models of energy deposition that maintain,
to varying degrees, the stochastic nature of the energy deposition process (see, e.g., Sec. 4.4
and Refs. 2634). The basic premise in the application of charged particle track structure
models is that one must include accurate information on the initial spatial distribution of
events involving ionization and excitation occurring along the path of a charged particle if
accurate models of the subsequent chemistry are to be developed. A motivating force in
the development of these stochastic models has been their application in radiation biology
where the chemistry and biochemistry following energy deposition by ionizing radiation
ultimately leads to biological damage.
To develop complete descriptions of the process of energy deposition along a
charged particle track requires an extensive database of interaction cross sections for the
primary particle as well as those initiated by secondary electrons produced in ionizing
collisions. Some of the more important interaction processes for heavy charged particles
were shown in Fig. 2. Not explicit in the depiction of cross sections shown in Fig. 2 is the
importance of dierential cross sections for electron production (ionization) that enable
one to derive a three-dimensional description of the spatial patterns of energy deposition.
To develop detailed models of the spatial characteristics of charged particle tracks requires
detailed information on the energy and angular distributions of electrons produced by
primary electrons or ions, along with similar cross sections for all secondary electrons. In
the case of fast heavy charged particle track structure, dierential cross sections for
electron production for partially screened projectile nuclei are also needed as they capture
electrons during slowing. The initial positions of all products of excitation, ionization,
and, where appropriate, charge transfer, will determine whether secondary electrons and
the corresponding residual media ions will recombine or proceed to initiate chemical
reactions.
In the following sections, the interaction mechanisms contributing to the stochastic
description of the energy loss process are discussed, as well as the trends in cross sections
as a function of the incident particle energy and charge state. To develop a full stochastic
description of a charged particle track useful for chemical and radiological applications, it
is important that the available cross section data fulll what has been called by Inokuti
[35] as the trinity of requirements; that the cross sections be absolute, correct, and
comprehensive. Where gaps in data exist, one must often use theoretical techniques to
bridge the gaps, in which case it is important to understand the limitations of the theories
and/or models being used and to strive to test the methodology with experimental data
wherever possible.
5. IONIZATION
The primary mechanisms for energy loss by ionizing radiation lead to ionization of the
atoms and molecules of the absorbing medium. The subsequent slowing of the electrons
from these primary ionizing events leads to the spatial relationships among the nal
physical products, e.g., ions, excited states, molecular fragmentation, etc., that are of
utmost importance to subsequent chemical reactivity. For suciently fast ions and
electrons, there are many similarities in the interaction process and some obvious dier-
ences. Electrons lose their energy primarily through coulomb interactions with bound
target electrons, in much the same way as heavy ions. However, electrons can lose all their
Copyright 2004 by Taylor & Francis Group, LLC
initial kinetic energy in a single collision with a bound electron whereas a heavy ion is
restricted to a maximum energy loss to a free electron E
max
given by
E
max
4m
1
m
e
m
1
m
e
2
E; 7
where m
1
is the incident particle mass, m
e
is the electron mass, and Eis the energy per atomic
mass unit u of the incident particle. Furthermore, electrons can undergo large angle scatter-
ing in collisions with media electrons, leading to extremely torturous paths, whereas heavy
ions move in a relatively straight line until very near the end of their range. For heavy ions
or neutral particles with energies less than a few keV/u, elastic scattering from the atom as
a whole, referred to as nuclear scattering, can dominate energy loss. At the other extreme,
very fast electrons can also lose energy by emission of radiation, bremsstrahlung, if they are
abruptly accelerated in collisions with target nuclei. The ratio of energy loss by radiation,
(dE/dx)
Rad
relative to that by ionization and excitation (dE/dx)
Col
is given approximately
by
dE=dx
Rad
dE=dx
Col
ZE
800
; 8
where Z is the atomic number of the target nucleus and E is the electron energy in MeV.
Slowing in water, Z c 8, a 10-MeV electron would lose somewhat less than 10% of the
energy loss per unit path length via radiation. In principle, heavy charged particles can also
lose energy through radiation; however, the probability of radiative energy loss is propor-
tional to the inverse of the square of the mass of the accelerated charge. Thus the probability
of radiative energy loss by a proton is a few million times smaller than that for an electron,
and for heavier ions the probability of radiative energy loss is even smaller.
Both electrons and heavy charged particles lose energy by coulomb interactions with
target electrons in basically the same manner. Collisions can be divided into two general
categories, those representative of distant, or soft, collisions and those characteristic of
close, or hard, collisions. For soft collisions, the energy transfer can be described as
occurring essentially with the entire atom or molecule leading to excitation of bound and
continuum states of the target. In hard collisions, the energy is directly transferred to a
bound electron in close knock-on collisions. Soft collisions resemble excitation and
ionization produced by photons; that is, the moving charged particle can be viewed as the
source of a continuum spectrum of photons with excitation and ionization of the target
produced by virtual photon exchange to the oscillators of the target. Thus character-
istics of soft collisions are similar to photo-excitation resulting in ejected electron spectra
determined from the oscillator strengths of the target and with angular distributions of
ejected electrons exhibiting dipole-like characteristics. On the other hand, the character-
istics of hard collisions are normally described by binary-encounter models of momentum
and energy transfer. Models of excitation and ionization also dier for electron and heavy
charged particle impact because electron exchange must be included in models for incident
electronsthis occurs because of the indistinguishability of incident and ejected electrons.
Ionization models specic to electron and heavy particle impact will be independently
discussed in the following sections of this chapter.
5.1. Ionization by Electrons
The fundamental mechanisms through which electrons interact with atoms and molecules
have been reviewed is detail by several authors; for additional details, see Ma rk et al. [19],
Copyright 2004 by Taylor & Francis Group, LLC
ICRU Report 55 [36], and references therein. Fast electrons predominantly interact via the
coulomb force with the bound electrons of the medium resulting in ionization, i.e., leading
to formation of free electrons and residual positive ions. The residual ions can be left in
any of a wide variety of nal states ranging from their ground ionic state, to states
resulting from simultaneous ionization and excitation, and/or dissociation of molecular
constituents of the target material. The interaction of a primary electron e
p
with the
diatomic molecule AB can precede via many channels, e.g., ionization can occur by any of
the following reaction pathways
e
p
AB !
AB
e
p
e
s
single ionization 9:1
AB
n
ne
s
e
p
n times multiple ionization 9:2
AB** e
p
! AB
e
p
e
s
autoionization 9:3
AB
* e
p
e
s
! Ionization plus !
A
B e
p
e
s
fragmentation 9:4
AB
2
e
p
2e
s
autoionization 9:5
AB
e
p
e
s
hm radiation 9:6
A
B e
p
e
s
dissociative ionization 9:7
A
e
p
ion pair formation 9:8
where e
s
is the secondary, or ionized, electron. For fast electrons, the most probable
reaction is that described by Eq. (9.1) with ionization occurring predominately with
valence electrons; however, in general, ionization can involve electrons bound in either
inner or outer electronic shells of the molecule or constituent atoms. There are also a
number on nonionizing channels that can be excited by an incident electron. These include
e
p
AB !
AB* e
p
simple excitation 9:9
A B e
p
neutral dissociation 9:10
A* B e
p
dissociation into excited neutrals 9:11
AB
4U
3
1
E
3
1
T W
3
_ _ 17
Copyright 2004 by Taylor & Francis Group, LLC
where U is the average kinetic energy of an electron with binding energy B given by U
h p
!
2
i=2m and p
!
is the momentum operator for the electrons in the subshell under con-
sideration.
Although these semiclassical approaches to electron impact ionization tend to be
reasonably accurate at predicting the cross sections for large momentum transfer, when
integrated with respect to energy loss, the total ionization cross section derived suers
from having an incorrect asymptotic high-energy dependence on the incident electron
energy. The energy dependence of the total cross section varies as T
1
, whereas
experimental evidence, and more complete quantum mechanical descriptions of the energy
loss process, suggests that the asymptotic dependence should be T
1
lnT.
The rst quantum mechanical approach to describe the interaction of charged
particles with atomic systems was that of Bohr, who considered the interaction in terms
of an impulse approximation resulting in the transfer of energy and momentum from the
incident charged particle to the atom or molecule. This perturbation approach was further
developed by Bethe in the 1930s; an extensive review of this work has been published by
Inokuti [40]. The Bethe theory provides the basis for a quantum mechanical understanding
of collisions of fast charged particles with atomic and molecular targets. Within the Bethe
theory, the dierential cross section for energy loss can be expressed [41] as
dr
dW
4pa
2
o
Z
2
T
N
k1
R
2
E
k
df
k
dW
ln
4RT
E
2
k
_ _
b
k
W O
E
k
T
_ _
; 18
where df
k
/dW is the partial optical oscillator strength for ejection of an electron with
energy W by photoionization of the kth subshell of the target, and E
k
is the total energy
transfer given by W+B
k
. Expressed in this way, the rst term in Eq. (18) represents the
contributions to the cross sections from soft collisions, the second term represents the hard
collisions contributions, and the last term includes contributions of higher order in powers
of T
1
. The optical nature of soft collisions is clearly illustrated by the relationship to
optical oscillator strengths. As expressed in Eq. (18), the contributions from exchange to
electron impact excitation and ionization are not included. A more complete description of
the use of the Bethe theory for the study of energy loss by electrons, including electron
exchange, is given by Dingfelder et al. [42].
The strength of the Bethe approximation is that it properly treats soft collisions,
thereby giving the proper energy dependence T
1
lnT for the total ionization cross
sections. A weakness of the Bethe approximation is the computational diculty,
including the need for accurate wave functions for all initial bound and nal continuum
states. Still it can be a powerful tool for the calculation of cross sections representative of
the target structure through the use of appropriate measured target oscillator strengths
to evaluate matrix elements otherwise dicult to obtain (see Ref. 42 and references
therein).
By careful inspection of the relationships of the semiclassical close collision
approximations and Bethe formula, one can obtain simple and accurate information on
ionization cross sections. By the method rst proposed by Platzman, and used extensively
by others, it is instructive to form the ratio of the dierential cross sections [measured
r(W,T) or calculated dr(W,T)/dW)] to the Rutherford cross section. This ratio, called Y,
is mathematically dened as
Y
drW; T
dW
T
4pa
2
o
E
2
R
2
19
Copyright 2004 by Taylor & Francis Group, LLC
and has the property that if plotted as a function of R/E, the integral is proportional to the
total ionization cross section
_
YT; WdR=E r
ion
T
T
4pa
2
o
R
:
20
Therefore when experimental data are plotted as Y vs. R/E, the relative importance of
electrons of dierent energies as contributors to the total ionization cross section is clearly
apparent and the relative magnitude of the dierential cross sections must be such that the
area under the curve is equal to the total ionization cross section [43]. The latter point is
particularly important because total ionization cross sections can generally be obtained
more accurately than dierential cross sections. Because Eq. (20) is a sum over energy loss
to all electrons, in the limit of large secondary energy the ratio Y tends to a value equal
to the number of electrons in the target. If dr(W,T)/dW given by the Bethe formula of
Eq. (18) is inserted in Eq. (19) and terms in higher-order T are considered negligible, one
can focus on the low-energy electrons, soft collisions, dominated by the rst term in the
Bethe theory
Yi
N
k1
E
k
df
k
dW
ln
4RT
E
2
k
_ _
: 21
If we now plot Y as a function of R/E, owing to the small variation of the logarithm term
with low-energy ejected electron energy, the ratio Y is expected to have the same general
structure with ejected electron energy as
N
k1
E
k
df
k
dW
, i.e., as the optical oscillator strength
multiplied by the energy loss.
An example of the utility of the Platzman Plot from the work of Kim [44] is shown
in Fig. 6 where single dierential cross sections for ionization of N
2
by electrons are
shown. Here the low-energy ejected electron spectra have the characteristic shape of the
optical oscillator strengths multiplied by the energy loss, Edf/dW, and the magnitude of
the cross sections are determined such that the areas under the curves are proportional to
the total ionization cross section. At the lowest incident electron energies, the optical
nature of the spectra tends to disappear as close, nondipole, interactions dominate. Note
also that as the ejected electron energy increases, the Rutherford ratio approaches a value
equal to the number of active, outer shell, electrons in the molecule. This type of plot
provides an excellent means to evaluate the internal consistency and accuracy of
experimental cross sections. The cross sections for ionization of water vapor by electron
impact measured by Bolorizadeh and Rudd [38] are shown in Fig. 7 along with the
distribution obtained from the dipole oscillator strengths [45]. Note that the low-energy
electrons do not decrease as would be expected by the shape of Edf/dW, indicating that
some extra electrons are probably being detected from scattering events in the exper-
imental system. However, the high-energy electrons ejected from fast collisions do
approach a value near 10, in agreement with the number of electrons in the water
molecule. Note that the very rapid rise in the ratio at the extreme high energy end of
each spectra is an eect of the indistinguishability of electrons and results from the elastic
scattering contribution to the spectrum.
Because the semiclassical theories can be used to calculate dierential cross sections
with relative ease for close collisions between the incident charged particle and the bound
electron, and the Bethe theory provides a straightforward method to describe low-energy
electrons ejected in distant collisions, it is only natural to combine the best characteristics
of the two approaches to derive a comprehensive description of electron impact ionization.
Copyright 2004 by Taylor & Francis Group, LLC
Figure 6 A Platzman plot of the ratio of the secondary electron cross sections for ionization of
N
2
by electrons to the comparable Rutherford cross sections. Q is the energy loss, Q = W+B, and is
equivalent to the parameter E used in the text. (From Ref. 44.)
Figure 7 The ratio of measured single dierential secondary electron production cross sections to
the corresponding Rutherfordcross sections for ionizationof water vapor by electrons. (FromRef. 38.)
Copyright 2004 by Taylor & Francis Group, LLC
Kim and Rudd [39] have developed such a model, the binary encounter dipole (BED)
model, in which the dierential ionization cross sections are given by
drW; T
dW
4pa
2
o
N
Bt u 1
R
B
_ _
2
N
i
=N 2
t 1
1
w 1
1
t w
_ _ _
22
2 N
i
=N
1
w 1
2
1
t w
2
_ _
lnt
Nw 1
dfw
dw
_
where t = T/B, w = W/B, u = U/B, N is the number of electrons in the subshell, df/dW
is the dierential dipole oscillator strength of the electron in this subshell with binding
energy B, and N
i
is obtained from the dierential oscillator strength distribution by
N
i
_
l
0
df w
dW
dW: 23
This relationship becomes very useful because dipole oscillator strengths are available for
a wide selection of atoms and molecules. The authors also provide a means of modifying
this equation for use when oscillator strengths are not available; the interested reader is
directed to their work [39] for additional details. The strength of their work is also evident
in that it provides a functional relationship for the determination of total cross section as
well. By integration of the single dierential cross sections, one obtains
r
i
4pa
2
o
N
t u 1
R
B
_ _
2
Dtlnt 2
N
i
N
_ _
t 1
t
_ _
lnt
t 1
_ _ _ _
; 24
where
Dt N
1
_
t1=2
0
1
w 1
dfw
dw
dw: 25
Both the single dierential and total ionization cross sections derived from these
models have been very successful in reproducing reliable experimental results. Because of
the dominance of the dipole term for low-energy ejected electron energies, one would
expect that the normal dipole distribution, as a function of angle for these electrons and
the high-energy electrons, will have an angular distribution described by the kinematic
constraints of the binary encounter theory.
5.2. Ionization by Protons and Other Bare Ions
As with the study of cross sections for electron impact discussed above, the decades of the
1970s and 1980s saw rapid advances in our understanding of ionization by fast protons
and other heavy charged particles; see, e.g., reviews presented in Refs. 10, 36, and 46.
Experimental studies have focused on the measurement of doubly dierential cross
sections, dierential in ejected electron energy and emission angle, for protons with
energies from a few tens of keV to several MeV. Measurements have also been conducted
for He
+
and He
2+
impact; however, these data are not nearly as comprehensive, and a few
measurements were made for ions of C, O, Ne, and Ar. As with electron impact,
measurements of electron emission cross sections are conducted for atoms and molecules
in gas phase owing to the very short range of low-energy electrons and residual ions in
condensed phase material.
Copyright 2004 by Taylor & Francis Group, LLC
The process of ionization by protons, and other bare ions, is essentially the same as
that described above for incident electrons, except that heavy ions and their secondary
electron are distinguishable, removing the need to consider exchange processes, and these
projectiles, being much heavier than the secondary electrons, will travel predominantly in
a straight line. One also must be more careful in describing ionization for heavy charged
particles because ionization of the target atom or molecule can occur with either the
release of a free electron, as with electron impact, or by the charged particle capturing an
electron from the target. In the latter case, a residual ion is produced, but no free electron
is released; thus one needs to specify whether ionization measurements are measuring free
electrons or residual ionsfor ionization by heavy ions, these cross sections are dierent.
As with electron impact, one can see a division of energy loss events occurring
between either from soft or hard collisions. The spectra of electrons ejected from H
2
and
N
2
at 20j with respect to an incident 1-MeV proton is shown in Fig. 8; these data are from
the work of Toburen et al., Refs. 47 and 48, respectively. Note the rather clear separation
between the low-energy peak in the cross sections at near zero ejected electron energy and
the high-energy peak near 3000 eV; the high-energy peak is the so-called binary encounter
peak. The electron energy W and emission angle h observed for the high-energy peak are
related to the kinematics of the system as
W
4MmE
p
M m
2
cos
2
h; 26
where M is the incident ion mass, m is the electron mass, and E
p
is the incident ion energy
in MeV/u. The shape and width of the binary encounter peak is determined by the velocity
distribution of the initially bound electron, i.e., the Compton prole. The largest dier-
ences between the two spectra shown in Fig. 8, other than the N
2
cross sections being
Figure 8 Electron emission at 20j with respect to the forward direction of the proton for ion-
ization of H
2
and N
2
by fast proton impact. The data for ionization of H
2
are from Ref. 47 and those
for N
2
are from Ref. 48.
Copyright 2004 by Taylor & Francis Group, LLC
about an order of magnitude larger, are the contributions to the spectra at about 360 eV
from Auger electrons emitted following inner-shell ionization of the nitrogen atom, and
the larger contribution of electrons in the energy region from about 200 to 1500 eV in the
N
2
spectrum relative to the H
2
spectrum. Cross sections in this intermediate-energy region
are somewhat enhanced by contributions from electrons ejected from inner shells of the
nitrogen molecule, in contrast to the spectrum from hydrogen where only a single shell is
involved. Because the experiments generally do not determine the initial state of the ejected
electron, these spectra are an average over all initial bound states.
An example of the extent of the data obtained in typical measurements of doubly
dierential cross sections is shown in Fig. 9 for ionization of CH
4
by 1-MeV protons. This
example from the work of Lynch et al. [49] also serves to illustrate the features of the
spectra of electrons ejected from molecules by fast protons. This three-dimensional display
shows the isotropic nature of the ejection of low-energy electrons and the sharply dened
angular dependence of the binary encounter peaks in the ejection of high-energy electrons.
The ridge formed by the binary encounter electrons across the dierent angles is
commonly referred to as the Bethe ridge [40] and the peak energies are approximately
given by the expression provided by Eq. (26). One can also see a ridge running across the
surface for electron energies of approximately 250 eV. This ridge results from Auger
electron emission following ionization of the K shell of carbon in the CH
4
molecule. For
Figure 9 Energy and angular distribution of cross sections for electron emission from CH
4
following 1.0-MeV proton impact. (From Ref. 49.)
Copyright 2004 by Taylor & Francis Group, LLC
low atomic number elements, such as carbon, the uorescence yield is very small (x
K
c
0.0024 for carbon), meaning that the cross sections for Auger electron emission are
essentially identical to the cross sections for K-shell ionization. Thus secondary electron
spectra, such as these shown in Fig. 9, provide a good source of information on inner-shell
ionization cross sections.
The theoretical descriptions of the ejected electron spectra for heavy ion impact are
basically the same as that for electron impact discussed above, except that the theory is
simplied for heavy ions because exchange forces are not an issue. One can write the
equivalent of Eq. (17) for the binary encounter approximation to the single dierential
ionization cross sections for bare heavy ion impact [36] as
drE; T
dE
4pa
2
o
Z
2
1
R
2
TE
2
1
4U
3E
_ _
27
for E
z E z 0, and
drE; T
dE
4pa
2
o
Z
2
1
R
2
TE
2
U
E
4
3
T
U
_ _
3=2
1
6
E
U
1
_ _
1=2
1
_ _ _ _
3
28
for E
+
z E z E
, where E
F
= 4T F4(TU)
1/2
, Z
I
is the charge of the incident ion, and the
other parameters were dened above for Eq. (17). For heavy ions, as dened above for
electrons, the kinetic energy T of the incident ion is given as T = m
e
v
2
p
/2, where m
e
is the
electron mass and v
p
is the incident particle velocity; thus T is the energy of the incident
particle in terms of the equivalent velocity electron. The agreement between experiment
and results of binary encounter theory can be somewhat improved by integrating over a
Fock distribution of orbital electron velocities, rather than a simple hydrogenic distribu-
tion, as was shown by Rudd et al. [50]; however, for light atoms, the hydrogenic
distribution is also a good approximation.
More detailed information on the characteristics of secondary electron emission can
be obtained by consideration of the angular distributions of the double dierential cross
sections for electron production by fast ions. A comparison of the results of the binary
encounter approximation, the quantum mechanical Born approximation, and experimen-
tal results is shown in Fig. 10 for ionization of helium by 2-MeV protons. These cross
sections clearly illustrate the isotropic nature of low-energy electron emission compared
with the sharply peaked angular distributions for high-energy electrons. The theoretical
results shown in Fig. 10 include a plane wave Born approximation (PWBA) of Madison
[51] in which the author used HartreeFock initial discrete and nal continuum wave
functions, and the binary encounter approximation (BEA) of Bonsen and Vriens [52] in
which they average over the initial velocity distribution of the bound electrons. Notice that
the BEA is more sharply peaked in angle than the quantum mechanical result. It should be
noted that the PWBA results using hydrogenic wave functions provides essentially the
same results as the BEA. Although the PWBA calculations of Madison are in much better
agreement with experimental data than the BEA, there are still discrepancies for small
angles and ejected electron velocities less than and near that of the proton velocity, i.e.,
equivalent to electron energies less than or equal to E
e
= m
e
v
p
2
/2. For the particle energy
shown in Fig. 10, this occurs at electron energies of approximately 1 keV and less. This
enhancement of the cross sections for small angles of emission has been attributed to an
ionization mechanism called charge transfer to continuum states of the proton, an
additional ionization channel not included in the PWBA calculation [53].
Copyright 2004 by Taylor & Francis Group, LLC
Figure 10 Angular distributions of electrons ejected from helium by 2-MeV protons. The
experimental data are from Ref. 54; theoretical results are the binary encounter approximation of
Bonsen and Vriens [52]; and the plan wave is the Born calculation of Madison [51].
Copyright 2004 by Taylor & Francis Group, LLC
As was discussed for electron impact, the classical Rutherford formula, Eq. (14), and
Bethe theory, Eq. (18), provide excellent tests of the asymptotic features of the singly
dierential cross sections for electron emission in fast proton, or other bare ions, in
collisions with atoms and molecules. The Platzman plot, a plot of the ratio of measured
single dierential cross sections to the Rutherford cross section as a function of the R/E, is
shown in Fig. 11 for ionization of helium by protons [54]. Note that from small values of
R/E (energetic ejected electrons), the ratio tends to a value of 2 as expected from the
number of electrons in the atom, and the spectra of low-energy ejected electrons has the
general shape expected from Edf/dW as predicted by the Bethe theory. The use of the Ru-
therford ratio was also proposed by Kim [55] as an excellent method to test the absolute
accuracy of ionization cross sections by plotting the Rutherford ratio as a function of E/R,
in which case the ratio should reach a asymptotic value for ejection of high-energy
electrons equal to the number of electrons in the atom. This use of the quantity Y is shown
in Fig. 12 for ionization of helium by protons. The dashed line in Fig. 12 is the predicted
ratio for emission of high-energy electrons. The experimental results presented by Manson
et al. [54] are measured absolute cross sections, and they are in excellent agreement with the
expected cross sections based on the Rutherford theory. There is some evidence from this
ratio that the experimental results might actually be as much as 10% larger than expected
from Rutherford theorythis dierence is well within the stated 25% uncertainties of the
absolute values of the experimental data, suggesting that a renormalization to the
asymptotic Rutherford value might be advised. From the data shown in Fig. 12, we see
that, for the asymptotic form of the ratio Y to be reached requires incident proton energies
greater than about 500 keV. At lower energies, the Rutherford plateau is not reached by
the most energetic of the ejected electrons.
Doubly dierential cross sections for electron emission following proton impact have
been measured for a wide variety of dierent atomic and molecular target systems (for
Figure 11 A Platzman plot, the ratio of the experimental single dierential cross section for
electron emission from helium by 1-MeV protons to the corresponding Rutherford cross sections
plotted as a function of R/E. The experimental cross sections are from Ref. 54 and the dierential
oscillator strength is taken from Ref. 43.
Copyright 2004 by Taylor & Francis Group, LLC
details, see reviews of these data in Refs. 10, 36, 46, and 56) providing an opportunity to
explore the eects of target and projectile parameters on the interaction cross sections. For
molecules composed of second-row elements, where only K and L shells are involved, the
cross sections for ejection of fast electrons is observed to be approximately proportional to
the number of electrons in the atom or molecule. This is illustrated in Fig. 13 where the
single dierential cross sections for ionization of a number of molecules are plotted per
eective target electron for ionization by 1-MeV protons; hydrocarbon data are from
Refs. 49 and 57 and the oxygen-containing molecular data are from Ref. 58. The number
of eective electrons used here is simply the number of outer-shell electrons, and this is
determined by subtracting the number of K-shell electrons from the total number of
electrons in the molecule. Note that the dierences in the spectra scaled in this manner
occur only in the regions of Auger electron emission and for electron energies less than
about 50 eV. The Auger electron peak energies depend on the K-shell binding energies and
occur at about 250 eV for carbon, 365 eV for nitrogen, and 540 eV for oxygen; for the
energy resolution used in these measurements, about 4%, the ne structural details of the
spectra are not observed [59]. The dierences in the spectra for low-energy electron
emission reect the dierences in oscillator strength distributions for the dierent
molecules as is expected from the Bethe theory.
It is also interesting to examine the angular distributions of these doubly dierential
cross sections for low-Z molecules. Angular distributions of cross sections scaled by the
number of outer-shell electrons for selected energies of electrons ejected by 1-MeV protons
are shown in Fig. 14. Scaled in this manner, cross sections for these molecules are in
excellent agreement with one another except for those of molecular hydrogen. The
hydrogen cross sections are much more sharply peaked in emission angle, and in this
Figure 12 The ratio of the measured single dierential cross section for ionization of helium by
protons to the corresponding Rutherford cross sections plotted as a function of the ejected electron
energy. The solid line represents the expected high-energy behavior of the ratio; it should approach
the number of electrons in the atom. The measurements are from Manson et al. [54].
Copyright 2004 by Taylor & Francis Group, LLC
way reect the results of theoretical calculations that employ hydrogenic wave functions,
as was also observed in the data shown in Fig. 10.
The relative success of the binary encounter and Bethe theories, and the relatively
well established systematic trends observed in the measured dierential cross sections
for ionization by fast protons, has stimulated the development of models that can extend
the range of data for use in various applications. It is clear that the low-energy portion of
the secondary electron spectra are related to the optical oscillator strength and that the
ejection of fast electrons can be predicted reasonable well by the binary encounter theory.
The question is how to merge these two concepts to predict the full spectrum.
Miller et al. [60,61] examined the structure of the BetheBorn formula and observed
that the second term of Eq. (18), the hard collisions term, is independent of the projectile
properties. In addition, for many target atoms and molecules, the optical oscillator
strength has been measured, making evaluation of the rst term readily accessible. Thus
if the higher-order terms in E/T can be ignored and experimental data are available for
single dierential cross sections, even for a very limited incident ion energy range, it should
be possible to extract the hard collisions term from the experimental data. Because the
hard collision contribution is independent of particle energy, it can be used along with the
rst term in the Bethe formula to obtain dierential cross sections throughout the range of
validity of the Bethe formula. This model was shown to be very useful in predicting the
Figure 13 Singly dierential cross sections for ionization of several molecular targets by 1-MeV
protons are plotted as a function of the ejected electron energy. The cross sections are scaled by the
eective number of target electrons in each molecule; the eective number of electrons is dened as
the total number of molecular electrons minus those of the K-shell. (From Refs. 49, 56, and 58.)
Copyright 2004 by Taylor & Francis Group, LLC
single dierential cross sections where limited data existed, and in improving the accuracy
of low-energy electron cross sections where experimental artifacts can render measure-
ments unreliable. Their model uses the optical oscillator strengths to determine the shape
of low-energy electron cross sections where experimental accuracy can be lacking.
However, the model was limited to fast protons by the high-energy nature of the theory.
The most comprehensive model of secondary electron emission by bare charged
particles has been developed by Rudd [62] for secondary electron production by protons;
Figure 14 Doubly dierential cross sections for ionization of several low-Z molecular targets by 1-
MeV protons plotted for selected ejected electron energies as a function of the emission angle. (From
Refs. 47, 48, 5658.)
Copyright 2004 by Taylor & Francis Group, LLC
this model is described in detail in ICRU Report 55 [36] and in an excellent review by
Rudd et al. [63]. Rudds model has the advantage of being applicable to the full range of
incident proton and ejected electron energies. It has the proper asymptotic dependence on
proton energy at both low and high energies. The major shortcoming is that it must be t
to sucient experimental data to x the models 10 parameters. For those target atoms
and molecules where parameters have been xed, it is a powerful tool because of its
analytic capabilities for application in elds such as radiation chemistry and biology that
need secondary electron spectra to evaluate subsequent chemical reactivity. Rudds
formula is given as
drW; T
dW
S=BF
1
F
2
W1 W
3
1 expaW W
c
=v
; 29
where S=4pa
o
2
N(R/B)
2
and W
c
is the cuto energy given by
W
c
4v
2
2v R=4B; 30
N is the number of electrons initially bound with energy B, v is the ion velocity, and W is
the ejected electron energy. The functions F
1
and F
2
are given as
F
1
v L
1
H
1
31
and
F
2
v L
2
H
2
=L
2
H
2
; 32
with
H
1
A
1
ln1 v
2
=v
2
B
1
=v
2
; 33
L
1
C
1
v
D
1
=1 E
1
v
D
1
4
; 34
H
2
A
2
=v
2
B
2
=v
4
; 35
and
L
2
C
2
v
D
2
: 36
The 10 parameters A
1
, B
1
, C
1
, D
1
, E
1
, A
2
, B
2
, C
2
, D
2
, and a are given in both Refs. 36 and
63 for 10 dierent atomic and molecular targets. An indication of the accuracy of the
cross section provided by the Rudd model is illustrated in Fig. 15, where cross sections for
ionization of N
2
by protons is compared to measurements [48,6466] over a wide energy
range. The major advantage of this model is that it is accurate for both low-energy and
high-energy protons where most models have been developed for only high-energy
ions.
5.3. Ionization by Dressed Ions
Perhaps the greatest challenge for the detailed investigation of energy deposition by heavy
charged particles occurs at particle velocities where the capture of electrons become
probable and cross sections must be determined for the interaction of dressed ions with
target electrons. As an ion captures electrons, these electrons can partially shield the ion
nucleus from coulomb interactions with the electrons of the medium. For example, a
proton with a bound electron (a moving hydrogen atom) looks like a neutral particle from
a large distance and might be expected to have little probability of interacting with
Copyright 2004 by Taylor & Francis Group, LLC
electrons of the target medium. However, this is far from what is actually observed; the
moving hydrogen atom can actually be as eective as a proton in producing ionization of
the target. To explain this phenomenon, we recall that the hydrogen electron in its ground
state is bound with a spherically symmetric probability distribution having its maximum at
the Bohn radius. As we approach the hydrogen atom on a microscopic scale, this charge
distribution provides somewhat less than a full charge screening of the nucleusthe unit
of electronic charge is distributed over the orbital volume of the ground state of the atom.
In fact, if we were inside the Bohr radius, we would expect little or no screening of the
nucleus by the bound electron. This distance-dependent shielding of the nucleus by bound
electrons occurs for all dressed particles and results in ejection of low-energy ionization
electrons, those produced in large impact parameter (soft) collisions, with a reduced
probability owing to a more eective screening of the projectile charge at large distances.
On the other hand, the ejection of high-energy electrons, those produced in small impact
parameter (hard) collisions, will show little eect of screening by the projectile electron(s)
because the impact parameter can be inside of some, or all, of the distributions of bound
electrons. In addition, the electrons bound to the moving particle must be considered as
active participants in the collision. They can also interact with target electrons via their
coulomb charge and can either be ionized in the collision leading to free electrons in the
media frame of reference with velocity comparable to the projectile (elastic scattering), or
they can interact with the bound electrons of the media producing excitation or ionization
(inelastic scattering) of the target. The interactions of projectile electrons with target
Figure 15 The ratio of calculated and measured single dierential cross sections for electron
emission from N
2
following proton impact to the corresponding Rutherford cross sections. The solid
line is the result of the model of Rudd [62]; experimental data are from Rudd (o) [64], Crooks and
Rudd (.) [65], Toburen () [48], and Stolterfoht (5) [66].
Copyright 2004 by Taylor & Francis Group, LLC
electrons is complicated by the convolution of the initial velocity distributions resulting
from their binding, with the velocity of the moving projectile, and conversion of these
vectors to the medium (laboratory) frame-of-reference.
An example of the features of the spectrum of secondary electrons emitted in H
o
impact on water molecules from the work of Bolorizadeh and Rudd [67] is shown in Fig.
16. Compared to the simple spectrum of electrons emitted by proton impact shown as the
solid line in Fig. 16 the spectrum from H
o
impact has an additional peak centered at an
electron energy of approximately 82 eV. This broad peak is from the superposition of the
spectrum of electrons stripped (elastically scattered) from the projectile on the spectrum of
electrons ejected from the target. Because the stripped projectile electrons originate as
bound electrons in the rest frame of the moving projectile, their laboratory energy is given
approximately by W = m
e
E
p
/M
p
and the width of the peak is determined by the Compton
prole of electrons in the projectile frame, but also transformed to the laboratory frame-
of-reference. The results shown in Fig. 16 clearly illustrate that the cross-sections for
Figure 16 Ionization of water vapor by 150-keV H
o
particles. The e
and H
o
impact data are from
Refs. 38 and 67, and the parameters for the Rudd model are from Ref. 36. The dotted line is
discussed in the text under the section on eective charge.
Copyright 2004 by Taylor & Francis Group, LLC
electron production by fast neutral hydrogen atoms can actually be larger than those
for the comparable velocity proton. Owing to screening by the projectile electron, we
might expect low-energy electrons emitted from the target in soft collision to be suppressed
relative to the bare proton (as predicted by the dashed line in Fig. 16). However, for
ionization by H
o
, this eect of screening is compensated by contributions from elastic
scattering of projectile electrons (the electron loss peak at 82 eV in this case) and by the
very low-energy electrons produced by inelastic scattering of projectile electrons from
target electrons resulting in ionization of the target. The dashed lines in Fig. 16 are from
the cross sections for inelastic energy loss by electrons with energies representative of those
in the elastic peak, the data come from the work of Bolorizadeh and Rudd [38]. To predict
the full extent of the low-energy inelastic scattering contribution to the spectra requires
integration of the inelastic cross sections over the weighted distribution of projectile
electron velocities.
For more complex projectiles and targets, additional features become apparent in
the spectra of ejected electrons. Doubly dierential cross sections for ionization of CH
4
by
0.3 MeV/u C
+
ions (3.6 MeV total projectile energy), taken from some of the authors
unpublished work, are shown in Fig. 17 to illustrate the increasing complexity of dressed-
ion collisions compared to that for bare ions represented in Fig. 9 by proton impact. In
Fig. 17, the cross sections are seen to have their maximum at very low energy ejected
electrons as was the case for proton impact, and, although their shape is not as sharp and
Figure 17 Ionization of CH
4
by 0.3-MeV/u C
+
ions. (These are unpublished data of Toburen
discussed in Ref. 56.)
Copyright 2004 by Taylor & Francis Group, LLC
discrete for this intermediate velocity ion as for the higher velocity proton impact, there is
a broad distribution of binary encounter electrons at the kinematically predicted forward
angles. In addition, we see evidence of K-Auger emission following the inner-shell
ionization of the target carbon atom as a ridge in the surface at approximately 250 eV;
this is most clearly seen for large angles. At small angles, the Auger electron yield becomes
insignicant relative to electrons directly ejected from outer shells. The broad peak at
about 160 eV, most strongly seen for small angles, results from electrons stripped from the
projectile, and Auger electrons resulting from excitation of the K-shell of the projectile are
observed as the Doppler-shifted Auger peaks superimposed on the continuum spectrum at
the high-energy edge of the binary encounter peaks. Although electrons stripped from the
projectile are predominantly observed ejected into small angles, there is also evidence of
their presence in the spectra at large angles.
Most of the features of the secondary electron spectra induced by impact of dressed
ions can be simply explained by the combined eects of the individual interactions of the
incident ion and its bound electrons with the target. However, a quantitative description is
complicated by the eects of electron binding and nuclear screening by bound electrons.
As discussed above for the calculation of stopping power for dressed ions, the eects of
bound projectile electrons on target ionization can be included through use of an eective
projectile charge. However, it is clear that a useful eective charge must be a function of
energy loss to be able to adequately describe the eects of electron screening on the
secondary electron emission spectra induced by dressed ions. To simply multiply the
spectrum of electrons induced by bare ions by a constant factor, such as the eective
charge determined from stopping power, would not provide a realistic estimation of the
spectra of electrons ejected in collisions involving dressed ions. This will be explored in
detail in the next section of this chapter.
6. EFFECTIVE CHARGE
As we have seen, the spectra of electrons ejected from a target atom or molecule by dressed
ions and neutrals are quite dierent from those induced by bare projectiles. These
dierences cannot be explained by use of a simple eective charge as was dened by
stropping power theory. For fast relatively simple dressed ions, such as H
o
or He
+
, the
separation of target and projectile ionization is relatively straightforward. The single
dierential cross sections for ionization of water vapor by 2-MeV He
+
and He
2+
ions
from the work of Toburen and Wilson [68] are shown in Fig. 18. For comparison, the data
for He
2+
ion impact has been scaled by z
2
(divided by 4) to represent the spectrum
expected of a bare singly charged ion (e.g., similar to proton impact at the same velocity as
the He
+
ion); this is shown as the solid line. The cross sections for He
+
impact are in good
agreement with the singly charged ion impact for electrons ejected with low energies (soft,
or distant, collisions), whereas the high-energy ejected electron cross sections are in
excellent agreement with the He
2+
results showing that screening is inecient in these
close collisions. The dashed line in Fig. 18 is simply a free-hand interpolation of target
ionization in the energy region containing the peak from electrons stripped from the
projectile. From experimental data such as these, one can determine the eective nuclear
charge as a function of energy loss from the formula
z
eff
E z
drE; q
dE
_
drE; z
dE
_ _
1=2
; 37
Copyright 2004 by Taylor & Francis Group, LLC
where dr(E,q)/dE is the measured dierential cross section for target ionization by an
incident ion of charge q and dr(E,z)/dE is the measured dierential cross section for
energy loss E by the bare ion of charge z. An example of z
e
(E) as a function of energy loss
obtained for ionization of He by He
+
ions is shown in Fig. 19. We must take care that the
portion of the electron spectrum that includes electrons from projectile stripping is not
included in the ratio, or that the stripping contribution is not included because they can
otherwise dominate the ratio. One can estimate the target ionization as was performed in
the case of the interpolated target contribution shown as the dashed line in Fig 18, or
simply use the portion of the electron spectra beyond the energy limitations of the stripped
projectile electrons. The origins of the theoretical curves shown in Fig. 19 will be discussed
in some detail below.
The theoretical basis for an eective charge z
e
(E) associated with collisional energy
loss has been investigated by several authors (see, e.g., Refs. 6972). Within the Born
approximation, the doubly dierential cross section for ejection of an electron with energy
W by a He
+
ion can be written as
A
2
r
AWAX
_
K
max
K
min
2
1
1 Ka
o
=4
2
2
_ _
2
IKdK; 38
where K is the momentum transfer, a
o
is the Bohr radius, and I(K) is a complicated
function of radial matrix elements and phase shifts [70]. The second term in the brackets
accounts for the eective screening of the nuclear charge by the bound helium electron.
Manson and Toburen have evaluated this expression using HartreeSlater wave functions
for initial discrete and nal continuum states for ionization of He by He
+
ions. Notice
that, within the Born approximation, the screening is a function of momentum transfer
and not energy transfer. The eect on energy transfer is obtained by an appropriate
Figure 18 Ionization of water vapor by 2-MeV He
+
and He
2+
ions. The dashed line is an estimate
of the contribution of target ionization by He
+
obtained by interpolation from regions beyond the
vicinity of electrons contributed by electron loss by the He
+
ion. (From Ref. 68.)
Copyright 2004 by Taylor & Francis Group, LLC
Figure 19 The upper portion of the gure shows the eective charge of the He
+
projectile as a
function of the ejected electron energy that is obtained from the experiment [68], Born theory [70],
and the model of Toburen et al. [71]. The lower portion of the gure is the same data, but plotted in
terms of the impact parameter. The impact parameter is obtained using the relationship between
projectile velocity, energy loss, and impact parameter dened by the Massey criterion. (See the text
and Ref. 71 for details.)
Copyright 2004 by Taylor & Francis Group, LLC
integral over momentum transfer. However, the eective charge for target ionization
calculated for electron emission at 60j using the Born approximation is in good agreement
with the measured single dierential cross sections shown in Fig. 19.
An important nding of the calculations using the Born approximation was that
there is a sizable contribution of target excitation (including ionization) that accompanies
projectile ionization. Doubly dierential cross sections for emission of 219 eV electrons
from He by 2 MeV He ions are shown in Fig. 20 with the results of PWBA calculations.
The peak in the distribution of electrons ejected at 0j in the laboratory frame of reference
is dominated by target ionization simultaneously occurring with projectile ionization. This
is in contrast to the binary encounter peak at about 70j that is attributed to target
ionization with projectiles left in the ground state, and to electrons ejected at large angles
(backward angles) where simultaneous excitation appears only of minor importance.
Figure 20 Doubly dierential cross sections for ejection of electrons of 219 eV (16 Ry) from He by
2-MeV He
+
ions. The points are measured cross sections and the calculated results are: line A
projectile ionization, target remains in the ground state; line Bprojectile ionization with
simultaneous target excitation; line Ctarget ionization, projectile remains in the ground state;
and line Dtarget ionization with simultaneous projectile excitation. (From Ref. 70.)
Copyright 2004 by Taylor & Francis Group, LLC
The evaluation of the I(K) term in Eq. (38) becomes increasingly dicult for more
complex target and projectile systems. This encouraged the development of a phenom-
enological approach for use with such systems. In the method proposed by Toburen et al.
[71], one considers the screening of the projectile by its bound electrons as a function of the
distance R from the nucleus as
z
eff
R z SR; 39
with
SR
i
N
i
_
R
0
w
i
r j j
2
r
2
dr; 40
where w
i
is the normalized radial wave function for the N
i
electrons residing in the ith
subshell of the projectile. To relate the energy loss E with the radial distance R, the Massy
adiabatic criterion R
ad
= v
p
/E is used, where v
p
is the incident ion or neutral particle
velocity and all parameters are in atomic units. Using this model, we obtain the solid line
in Fig. 19 in good agreement with both the experiment results and the Born calculation.
This model was also used to obtain the dotted line in Fig. 16 for target ionization of water
by atomic hydrogen impact. In the case of the neutral projectile, the model predicts zero
cross section for ionization involving zero-energy ejected electrons, i.e., for electrons
ejected by very distant collisions; this reects the lack of coulomb eld from the neutral
particle as observed at suciently large distances.
As noted above, the calculations including eects of screening by heavy ions and
interactions with complex targets, i.e., molecular targets, are dicult to accomplish by
fundamental theories such as is represented by the Born approximation. However, it is
possible to determine approximate cross sections by combining the model of Rudd,
described above for ionization by protons, with the screening model presented in Eqs. (39)
and (40). Single dierential cross sections for ionization of neon by protons and C
+
ions
of 0.1 MeV/u are shown in Fig. 21 along with these model predictions. The dotted line is
the result of Rudds model for ionization of neon by 0.1-MeV protons; it is in excellent
agreement with measured proton impact ionization cross sections for neon of Crooks and
Rudd [65] shown as the lled circles. A simple z
2
multiplication of the proton cross
sections, i.e., classical z
2
scaling, is shown by the solid line with lled points; notice the
large overestimation of the low-energy electron cross sections by this naive scaling
technique, although the high-energy portion of the spectrum is in relatively good agree-
ment with the measured cross sections. Applying the screening model for z
e
given by Eqs.
(39) and (40) above, evaluated using hydrogenic wave function for the electrons bound to
the carbon projectile, one obtains the solid line shown in Fig. 21. This application of the
screening model returns cross sections in much better agreement with the unpublished
measurement of Toburen throughout the energy range of the ejected electrons. There are
still dierences between measured and predicted cross sections in the intermediate electron
energy region, but the dierences have been reduced from factors of 30 or more at the
lowest ejected electron energies to a factor of about two at intermediate energies. The
remaining dierences might be due to overestimates of the amount of contribution from
continuum electron capture derived from scaling of the bare proton spectrum, or to simple
inaccuracies in the model.
From the discussion of projectile charge and its eect on the spectra of secondary
electron production, it is clear that the eective charge of the projectile is a function of the
ejected electron energy, i.e., projectile energy loss. From this it is evident that the single
Copyright 2004 by Taylor & Francis Group, LLC
value for eective charge obtained from stopping power theory cannot predict the dis-
tribution of secondary electrons that, through subsequent collisions in slowing, produce
the spatial patterns of energy deposition important to subsequent radiation chemistry.
7. CHARGE TRANSFER
As positively charged particles slow and reach velocities comparable to bound electron of
the medium, they capture and/or lose electrons. These processes can result in ionization of
the medium, with no free electron released (electron capture), or release of a free electron
from the projectile, with no concomitant target ion being produced (electron loss);
however, as pointed out above, these processes often occur with simultaneous ionization
of both the projectile and the target. The capture and lose of electrons by the projectile can
be an important energy loss mechanism for the incident particle (see Fig. 2) accounting for
as much as 1520% of the total energy-loss cross sections at their maximum. Although the
Figure 21 Cross sections form ionization of neon by 0.1-MeV/u protons and singly-charged
carbon ions. The experimental data for protons in from the work of Crooks and Rudd [65] and the
C
+
results are from some unpublished work of Toburen [56]. The results of Rudds model were
obtained using the parameters given in Ref. 36.
Copyright 2004 by Taylor & Francis Group, LLC
state-to-state electron transfer from target to projectile might not involve signicant
energy loss by the projectile (in some cases, an energy gain can be experienced), the kinetic
energy that must be gained by the captured electron to reach the velocity of the moving ion
can be signicant. For electron loss by a moving particle, the energy loss by the projectile
is simply the binding energy of the stripped electron, assuming energy is not contributed
by the velocity distribution of the target electrons that might aid in stripping the projectile
electron. With regard to the eects on subsequent chemistry, pure electron capture leaves a
residual ion, but no free electron and electron loss contributes a relatively energetic
electron to the track of the charged particle, with no residual target ion; again, in reality,
simultaneous excitation of the target or projectile often accompanies these processes.
Figure 22 Cross sections for electron capture by protons r
10
and electron loss by hydrogen atoms
r
01
for a number of molecular targets are shown scaled by the number of outer-shell target electrons.
(From Ref. 73.)
Copyright 2004 by Taylor & Francis Group, LLC
Because of the importance of electron capture and loss in a number of applied elds,
such as fusion plasma, atmospheric physics, and radiological physics, there has been
substantial eort contributed to the study of these events for simple collision systems, such
as protons and neutral hydrogen particles in simple gas targets. Less is known of the cross
sections for helium ions and neutrals, and only a few data exist for heavier ions (for a
recent review of the charge transfer data, see Ref. 73 and references therein). Data for
electron capture and loss cross sections for proton collisions with a few selected molecular
targets are shown in Fig. 22; these data are taken from the review of Toburen [73]. In this
example, the cross sections have been scaled by the number of target electrons using the
relatively naive assumption that the molecule simply provides a cluster of electrons
from which the moving ion can either capture or be ionized. Notice that this cross section
scaling is reasonably appropriate for fast collisions, i.e., ion energies larger than about 100
keV (and perhaps somewhat lower energies for electron loss), although target structure is
seen to play an important role for electron capture by low-energy protons. From the data
shown in Fig. 22, one can obtain an appreciation for the broad range of particle energies
and target species that have been investigated for protons and neutral hydrogen. A similar
plot for other ions would be much less complete; fewer targets have been investigated and
a more limited range of particle energies explored.
8. MULTIPLE IONIZATION
Interactions involving electrons and heavy ions can lead to multiple ionization of the
target atom or molecule. For example, ionization of an inner shell of an atom can result in
subsequent emission of one or more Auger electrons as the excited atom relaxes to the
ground state. This leads to two or more electrons, correlated in time and space, being
emitted in the immediate region of the original point of ionization. The correlation
involving multiple electrons produced by a single ionizing event can have a strong
inuence on subsequent chemical reactivity of the products of energy deposition.
Although inner-shell ionization cross sections are well known, the consequence of multiple
ionization is generally neglected in models of chemical kinetics because inner-shell
ionization cross sections are generally several orders of magnitude smaller than outer
shell ionization, i.e., multiple ionization via inner-shell ionization is a rare event.
On the other hand, the study of Auger electron spectra has been a productive means
of gaining information on the extent that multiple ionization occurs simultaneous with the
production of inner-shell ionization, i.e., the simultaneous ejection of outer-shell electrons
accompanying inner-shell ionization. When outer-shell vacancies exist in an atom or
molecule that is also ionized in an inner shell, the Auger transitions initiated in the inner
shells are shifted in energy by the presence of these outer-shell vacancies and these shifts
can be detected in the Auger spectra [74,75]. In fact, for collisions of high-energy heavy
ions, the degree of multiple ionization can provide such extreme shifts in the Auger line
energies that the normal spectral shapes are replaced with a statistical distribution of
transition energies [76]. Although Auger spectra provide indirect information on the
amount of multiple outer-shell ionization that accompanies inner-shell ionization, it
cannot give a good indication of the overall yields of multiple ionization because a sizable
fraction of the multiple ionization does not involve inner-shell ionization processes, but
occurs from direct ionization of the outer shell electrons.
As discussed above, heavy charged particles, particularly multiply charged heavy
ions, have a relatively large probability for inducing multiple ionization involving outer, as
Copyright 2004 by Taylor & Francis Group, LLC
well as inner, shells. For ions that carry several electrons interacting with a target atom, or
molecule that also have several bound electrons, multiple ionization may be the rule rather
than the exception. Such a collision can be represented as
A
q
B ! A
qj
B
i
i je
; 41
where the collision results in the capture of j electrons by the projectile and ionization of i
electrons from the target. Because of the multiple interactions between the nuclei and
electrons of the projectile and target, the theoretical treatment of such systems is generally
approached statistically [77].
The most detailed experimental information describing multiple ionization has been
obtained through the study of recoil target ion states in coincidence with selected projectile
states. To fully investigate the system would also require measurements of the electron
spectra in coincidence with post collision projectile and target charge states; however, such
experiments are extremely dicult. Therefore the general experimental focus has been on
the measurements of recoil charge state in coincidence with the selected projectile charge
state. Results of a coincidence study by DuBois [17,78] of the multiple ionization of a neon
target by proton and helium ions are shown in Fig. 23. The signicantly greater degree of
double and triple ionization of neon by He ions relative to that by protons is clearly
evident. At a few hundred keV, the contribution of double ionization reached more than
25% of that for single ionization; this is much greater than could be explained by
contributions from Auger cascades following inner shell ionization. For higher Z target,
the contributions from multiple ionization are even greater, reaching 50% for ionization of
Kr by He
+
. Only limited data exist for studies of multiple ionization of molecular targets,
and these studies are further complicated by the fact that molecules undergo a many
Figure 23 Cross sections for single, double, and triple ionization of neon by singly charged hy-
drogen and helium ions. (From Refs. 17 and 78.)
Copyright 2004 by Taylor & Francis Group, LLC
possible avenues of dissociation following ionization. In addition, coulomb explosions
can occur following multiple ionization that can disrupt molecular bonds and have severe
consequences on the subsequent radiation chemistry. Clearly more data is needed to
evaluate this important mechanism of ionization for heavy charged particles.
There is a wealth of fundamental information available on the interaction of fast
charged particles with atoms and molecules. Secondary electron distributions resulting
from ionizing collisions are well understood for incident bare projectiles and a few simple
dressed ions and neutrals. Even where a quantitative understanding of the collision is not
feasible, the basic features of the interactions are understood, making it possible to make
educated guesses as to the important parameters of the processes of energy loss. Still our
understanding and application of many of the known processes is limited. We are
particularly challenged to quantitatively describe the secondary electron spectra for
dressed ions and neutrals. We have only limited information on multiple ionization and
its consequences in radiation chemistry. There is a wealth of information in hand and a
plethora of questions remaining to answer.
REFERENCES
1. Inokuti, M. In Electronic and Atomic Collisions; Oda, N., Takayanagi, K., Eds.; North Holland
Publishing Co.: Amsterdam, 1980; p 31.
2. Bohr, N. Phys. Rev. 1940, 58, 654.
3. Bohr, N. Phys. Rev. 1941, 59, 270.
4. Bohr, N. Det Kgl. Danske Videnskabernes Selskab Mathematisk-fysiske Meddelelser XVIII, 8.
Hafner Publishing: New York, 1948.
5. ICRU Report 49. Stopping Powers and Ranges for Protons and Alpha Particles; International
Commission on Radiation Units and Measurements: Bethesda, MD, 1992.
6. Evans, R.D. The Atomic Nucleus; McGraw-Hill Book Company, Inc.: New York, 1955.
7. Turner, J.E. Atoms, Radiation, and Radiation Protection; John Wiley & Sons, Inc.: New York,
1995.
8. Porter, L.E. Int. J. Quantum Chem. 1997, 65, 997.
9. Porter, L.E. Nucl. Instrum. Methods 2000, 170, 35.
10. Toburen, L.H. IAEA-TECDOC-799; International Atomic Energy Agency: Vienna, 1995;
p 47.
11. Yarlagadda, B.S.; Robinson, J.E.; Brandt, W. Phys. Rev. B 1978, 17, 3473.
12. Barkas, W.H. Nuclear Research Emulsions-I. Techniques and Theory; Academic Press: New
York, 1963.
13. McDaniel, E.W.; Flannery, M.R.; Thomas, E.W.; Ellis, H.W.; McCann, K.T.; Manson, S.T.;
Gallagher, J.W.; Rumble, J.R.; Beaty, E.C.; Roberts, T.G. Compilation of Data Relevant to
Nuclear Pumped Lasers, Technical Report H-78-1; US Army Missile Research an Development
Command: Redstone Arsenal, Alabama, 1979; Vol. 5, p 1917.
14. Uehara, S.; Toburen, L.H.; Wilson, W.E.; Goodhead, D.T.; Nikjoo, H. Radiat. Phys. Chem.
2000, 59, 1.
15. Miller, J.H.; Green, A.E.S. Radiat. Res. 1973, 54, 343.
16. DuBois, R.D.; Toburen, L.H. Phys. Rev. A 1988, 38, 3960.
17. DuBois, R.D. Phys. Rev. A 1989, 39, 4440.
18. Ferradini, C.; Jay-Gerin, J.-P. Radiat. Phys. Chem. 1997, 51, 263.
19. Ma rk, T.D.; Hatano, Y.; Linder, F. IAEA-TECDOC-799; International Atomic Energy
Agency: Vienna, 1995; p 163.
20. Srdoc , D.; Inikuti, M.; Krajcar-Bronic , I. IAEA-TECDOC-799; International Atomic Energy
Agency: Vienna, 1995; p 547.
Copyright 2004 by Taylor & Francis Group, LLC
21. Fano, U. Phys. Rev. 1946, 70, 44.
22. Miller, J.H.; Wilson, W.E. Radiat. Phys. Chem. 1989, 34, 129.
23. Bartels, E.R.; Harder, D. Radiat. Prot. Dosim. 1990, 31, 211.
24. Blohm, R.; Harder, D. Radiat. Prot. Dosim. 1985, 13, 377.
25. Mozumder, A.; Magee, J.L. J. Chem. Phys. 1966, 45, 3332.
26. Miller, J.H. Radiat. Res. 1981, 88, 280.
27. Paretzke, H.G.; Turner, J.E.; Hamm, R.N.; Ritchie, R.H.; Wright, H.A. Radiat. Res. 1991,
127, 121.
28. Chatterjee, A.; Holley, W.R. Int. J. Quantum Chem. 1991, 39, 709.
29. Kraft, G.; Kra mer, M.; Scholz, M. Radiat. Environ. Biophys. 1992, 31, 161.
30. Pimblott, S.M.; LaVerne, J.A.; Mozumder, A. J. Phys. Chem. 1996, 100, 8595.
31. Cobut, V.; Frongillo, Y.; Patau, J.-P.; Goulet, T.; Fraser, M.-J.; Jay-Gerin, J.P. Radiat. Phys.
Chem. 1998, 51, 229.
32. Wilson, W.E.; Nikjoo, H. Radiat. Environ. Biophys. 1999, 38, 97.
33. Emetzoglou, D.; Papamichael, G.; Kostarelos, K.; Moscovitch, M. Phys. Med. Biol. 2000, 45,
3171.
34. Friedland, W.; Bo, W.; Jacob, P.; Paretzke, H. Radiat. Res. 2001, 155, 703.
35. Inokuti, M. Physical and Chemical Mechanisms in Molecular Radiation Biology; Glass, W.A.,
Varma, M.N., Eds.; Plenum Press: New York, 1991; p 29.
36. ICRU Report 55. Secondary Electron Spectra from Charged Particle Interactions; International
Commission on Radiation Units and Measurements: Bethesda, MD, 1995.
37. Weigold, E. Aust. J. Phys. 1990, 43, 543.
38. Bolorizadeh, M.A.; Rudd, M.E. Phys. Rev. A 1986, 33, 882. Bolorizadeh, Doctoral Disser-
tation; University of Nebraska, Lincoln, 1984.
39. Kim, Y.-K.; Rudd, M.E. Phys. Rev. 1994, 50, 3954.
40. Inokuti, M. Rev. Mod. Phys. 1971, 43, 297.
41. Miller, J.H.; Wilson, W.E.; Manson, S.T.; Rudd, M.R. J. Chem. Phys. 1987, 86, 157.
42. Dingfelder, M.; Hantke, D.; Inokuti, M.; Paretzke, H.G. Radiat. Phys. Chem. 1998, 53, 1.
43. Kim, Y.-K. Phys. Rev. A 1983, 28, 656.
44. Kim, Y.-K. RADIATION RESEARCH Biomedical Chemical and Physical Perspectives;
Academic Press, Inc.: New York, 1975.
45. Paretzke, H.G. Simulation von Elektronenspuren im Energiebereich 0.0110 keV in Was-
serdampf; Gesellschaft fu r Strahlen-und Umweltforschung mbH: Mu nchen, 1989.
46. Stolterfoht, N.; DuBois, R.D.; Rivarola, R.D. In Electron Emission in Heavy Ion-Atom
Collisions, Springer Series on Atoms + Plasmas; Toennies, J.P., Ed.; Springer: Berlin, 1997.
47. Toburen, L.H.; Wilson, W.E. Phys. Rev. A 1972, 5, 247.
48. Toburen, L.H. Phys. Rev. A 1971, 3, 216.
49. Lynch, D.J.; Toburen, L.H.; Wilson, W.E. J. Chem. Phys. 1976, 64, 2616.
50. Rudd, M.E.; Gregoire, D.; Crooks, J.B. Phys. Rev. A 1971, 3, 1635.
51. Madison, D.H. Phys. Rev. A 1973, 8, 2449.
52. Bonsen, T.F.M.; Vriens, L. Physica 1970, 47, 307.
53. Rudd, M.E.; Macek, J. Case Stud. At. Phys. 1972, 3, 47.
54. Manson, S.T.; Toburen, L.H.; Madison, D.H.; Stolterfoht, N. Phys. Rev. A 1975, 12, 60.
55. Kim, Y.-K. Radiat. Res. 1975, 61, 21.
56. Toburen, L.H. In Physical and Chemical Mechanisms in Molecular Radiation Biology; Glass,
W.A., Varma, M.N., Eds.; Plenum Press: New York, 1991; p 51.
57. Wilson, W.E.; Toburen, L.H. 0.3-2.0 MeV. Phys. Rev. A 1975, 11, 1303.
58. Toburen, L.H.; Wilson, W.E. J. Chem. Phys. 1977, 66, 5202.
59. Siegbahn, K.; Nordling, C.; Johansson, G.; Hedman, J.; Hede n, P.F.; Gelius, U.; Bergmark, T.;
Werme, L.O.; Manne, R.; Baer, Y. ESCA Applied to Free Molecules; North Holland Publishing
Co.: Amsterdam, 1969.
60. Miller, J.H.; Toburen, L.H.; Manson, S.T. Phys. Rev. A 1983, 27, 1337.
61. Miller, J.H.; Wilson, W.E.; Manson, S.T.; Rudd, M.E. J. Chem. Phys. 1987, 86, 157.
Copyright 2004 by Taylor & Francis Group, LLC
62. Rudd, M.E. Phys. Rev. A 1988, 38, 6129.
63. Rudd, M.E.; Kim, Y.-K.; Madison, D.H.; Gay, T.J. Rev. Mod. Phys. 1992, 64, 441.
64. Rudd, M.E. Phys. Rev. A 1979, 20, 789.
65. Crooks, J.B.; Rudd, M.E. Phys. Rev. A 1971, 3, 1628.
66. Stolterfoht, N. Z. Phyzik 1971, 248, 92.
67. Bolorizadeh, M.A.; Rudd, M.E. Phys. Rev. A 1986, 33, 893.
68. Toburen, L.H.; Wilson, W.E.; Popowich, R.J. Radiat. Res. 1980, 82, 27.
69. Briggs, J.S.; Taulbjerg, K. Top. Curr. Phys 1978, 5, 105.
70. Manson, S.T.; Toburen, L.H. Phys. Rev. Lett. 1981, 46, 529.
71. Toburen, L.H.; Stolterfoht, N.; Ziem, P.; Schneider, D. Phys. Rev. A 1981, 24, 1741.
72. McClure, J.H.; Stolterfoht, N.; Simony, P.R. Phys. Rev. A 1981, 24, 97.
73. Toburen, L.H. Radiat. Environ. Biophys. 1998, 37, 221.
74. Watson, R.L.; Toburen, L.H. Phys. Rev. A 1973, 7, 1853.
75. Stolterfoht, N.; Schneider, D. Phys. Rev. A 1975, 11, 721.
76. Jamison, K.A.; Woods, C.W.; Kauman, Robert; Richard, P. Phys. Rev. A 1975, 11, 505.
77. Olson, R.E.; Ullrich, J.; Schmidt-Bo cking, H. Phys. Rev. A 1989, 39, 5572.
78. DuBois, R.D.; Toburen, L.H.; Rudd, M.E. Phys. Rev. A 1984, 29, 70.
79. DuBois, R.D. Phys. Rev. A 1989, 39, 4440.
Copyright 2004 by Taylor & Francis Group, LLC
4
Modeling of Physicochemical and Chemical
Processes in the Interactions of Fast Charged
Particles with Matter
Simon M. Pimblott and A. Mozumder
University of Notre Dame, Notre Dame, Indiana, U.S.A.
1. INTRODUCTORY REMARKS
In Chapter 2 of this book, the energy transfer interactions of fast charged particles with
matter were discussed from the viewpoint of the incident particle. This chapter will describe
and examine the consequences of energy deposition in matter with the goal of providing an
understanding of the physico-chemical and chemical processes involved.
Energy deposition in a molecule is a necessary precursor to radiation-induced
transformations (see Sec 1.1 for a similar statement for photochemistry). In most situations,
especially in condensed media, the deposited energy is localized and available for physical
and chemical transformations. A sequence of correlated and localized energy loss events,
frequently in electronic form, may be termed a track. The incident particle generating the
track may suer occasional changes in direction, due mainly to elastic scatterings. This
eect becomes increasingly important as the particle slows down. At rst, the lost energy
appears in electronic form then generations of transformations may take place before a
particular radiation eect is observable. When there is sucient deposited energy,
ionization ensues and, frequently, secondary ionization and excitation will also occur.
These correlated events are localized in the vicinity of the primary interaction. For low
linear energy transfer (LET) radiations, the primary energy loss events are well separated.
The cluster of ionizations and excitations associated with each primary event is called a
spur, and on a certain timetable and in a certain chemical sense these clusters develop
independently. For high-LET radiations, the spurs of a track coalesce to form a more
continuous, cylindrical distribution. The thrust of this chapter is concerned mainly with
spur and track reactions. However, in conformity with the general aim and purpose of this
book, it should be remembered that the arguments are not limited to radiation chemistry
but can and should be extended to other applications (such as radiation biology and certain
industrial applications, etc.). One restriction is that the discussion will be chiey concerned
with condensed phase phenomena, although sometimes reference will be made to gas-phase
eects for the sake of comparison.
Copyright 2004 by Taylor & Francis Group, LLC
The organization of this chapter is as follows. In the following section, Sec. 4.2, the
elastic and inelastic interaction cross sections necessary for simulating track structure (ge-
ometry) will be discussed. In the next section, ionization and excitation phenomena and
some related processes will be taken up. The concept of track structure, from historical idea
to modern track simulation methods, will be considered in Sec. 4.4, and Sec. 4.5 deals with
nonhomogeneous kinetics and its application to radiation chemistry. The next section (Sec.
4.7) describes some application to high temperature nuclear reactors, followed by special
applications in low permittivity systems in Sec. 4.8. This chapter ends with a personal per-
spective. For reasons of convenience and interconnection, it is recommended that appro-
priate sections of this chapter be read along with Chapters 1 (Mozumder and Hatano), 2
(Mozumder), 3 (Toburen), 9 (Bass and Sanche), 12 (Buxton), 14 (LaVerne), 17 (Nikjoo), and
23 (Katsumura).
2. INTERACTION CROSS SECTIONS
In the interaction of fast charged particles with matter, the averaged macroscopic quantities
related to radiation physics can be evaluated from cross-sectional information that is of
limited complexity. For instance, the stopping power, which is the energy-weighted integral
over the inelastic collision cross section, can be computed fairly accurately using fairly ap-
proximate treatments (see Sec. 4.2.2). Bohrs theory is a prime example of such an approach.
On the other hand, chemical and other eects may depend on a specic excited or ionized
state of a molecule arising from a particular energy deposition event and on the local track
geometry. Elucidation of such eects therefore requires more accurate cross-sectional infor-
mation. Furthermore, because of the generation of copious low-energy secondary electrons,
the elastic and inelastic collision cross sections for these interacting particles are always sig-
nicant, even if the incident primary particle is highly energetic.
As an incident radiation particle slows down, various track processes occur with
sometimes several processes occurring simultaneously. Initially, electronic excitation dom-
inates inferior vibrational and rotational excitation. All the time, elastic collisions delineate
the track geometry. For the purpose of track simulation, it is often convenient to think in
terms of a collision mean free path, which is inversely related to the cross section. The basic
requirement for realistic simulation is then a comprehensive set of inelastic and elastic col-
lision cross sections for energetic (heavy) ions and for electrons over a wide span of energy.
In most cases, a doubly dierential cross section in energy and angle is sucient, but in cer-
tain cases, where detailed track geometry is not critical, a cross-section dierential in energy
is adequate.
2.1. Cross-Sections for Track Structure Simulation
Despite the fact that Bohrs stopping power theory is useful for heavy charged particles such
as ssion fragments, Rutherfords collision cross section on which it is based is not accurate
unless both the incident particle velocity and that of the ejected electron are much greater
than that of the atomic electrons. The quantummechanical theory of Bethe, with energy and
momentum transfers as kinematic variables, is based on the rst Born approximation and
certain other approximations [1,2]. This theory also requires high incident velocity. At
relatively moderate velocities certain modications, shell corrections, can be made to extend
the validity of the approximation. Other corrections for relativistic eects and polarization
screening (density eects) are easily made. Nevertheless, the BetheBorn approximation
Copyright 2004 by Taylor & Francis Group, LLC
gradually breaks down at electron energies of about 500 to 1000 eV in most media of low
atomic number.
The basic input information used in most cross-section calculations is the dierential
dipole oscillator strength distribution (DOSD) introduced in Sec. 2.2.2 and 2.2.3. For
polyatomic molecules, this is a continuous density distribution f(e) such that the dipole
oscillator strength for transition energies between e and e+de is given by f(e)de. In terms of
the DOSD, the total number of electrons receiving any energy transfer, the eective mean
excitation potential, the mean energy loss during a collision, and the mean excitation energy
for straggling [3] are given by
Z
e
_
e
max
e
0
fede
Z
e
lnI
e
_
e
max
e
0
feln ede
Z
e
e
_
e
max
e
0
efede
Z
e
elnI
_
e
max
e
0
efelnede
1
respectively [4]. In the above integrals e
0
is the lowest excitation potential of the molecule and
e
max
is the maximum energy transfer that is kinematically permissible. Because the incident
and ionized electrons are indistinguishable, e
max
virtually equals one-half of the incident
energy E, if that is large compared with the binding energy B. Otherwise, it should be taken
as
f
(E+B)/2. With known values of Z
e
, I
e
, q, and I, the electron stopping power at
relatively low incident energies may be expressed in the form [5]:
Selectron Z
e
ln
_
mv
2
I
e
~ e=8
_
_
e
2E
ln
4 eE
I
_ _
2
When E is suciently large that almost all the oscillator strength is contained within e
max
,
Z
e
and I
e
approach the limiting values of Z and I, respectively, in which case Bethes
nonrelativistic formula is recovered. Comparing Eq. (2) with Eq. (2.2.7), which includes
relativistic, shell, and density corrections, reveals that, as is stands, Eq. (2) does correct for
the energetically inaccessible states in the Bethe theory and therefore may be used at inter-
mediate electron energies. However, it is still limited to nearly zero momentum transfer in
collision kinematics. Ashley [6,7] has extended the use of the DOSDspectrumto account for
the nonvanishing momentum transfer by a quadratic approximation in the energymomen-
tum plane. It is better represented by the imaginary part of the reciprocal of the dielectric
response function e( q,x), where ,q and ,x are the momentum and energy losses, respec-
tively, in the collision process. Ashleys procedure involves replacing Im(1/e( q,x)) by
Im(1/e(0,x,q
2
/2m)). The relationship between the reciprocal dielectric response func-
tion and the DOSD has been discussed in Sec. 2.2.4 and is reproduced here with a slight and
obvious change in notations, viz. Im(1/e(0,x)) =(h
2
e
2
NZ/2m)f(e)/e, with e =,x. Ashleys
approximation has been well documented in a number of cases, including C and the metals
Al and Cu, and has been found to be accurate. Its most frequent application has been to
liquid water for relatively low energy electrons and protons [8,9].
Copyright 2004 by Taylor & Francis Group, LLC
Following Ashleys procedure and using the DOSD, f(e), the inelastic mean free path
K
i
for electronic excitation and ionization may be given by [8]
K
1
i
vN
_
e
1
feLe=Ede 3
here v = 2pe
4
/mv
2
, s = (12a)
1/2
and
La a
1
ln
1 a s1 a s
1 a s1 a s
_ _
2
1 a
F arcsin
s
1 a
_ _
;
1 a
1 a
_ _
Here F denotes an elliptic integral of the rst kind in Legendres normal form [10]. The
corresponding stopping power equation may also be expressed in a closed form[8] and when
expanded in powers of e/E , the lead term generates Bethes nonrelativistic result for the
stopping power for an electron. For the purpose of track simulation, the desired quantity is
the cumulative inelastic cross section of an event with energy loss less than e, which is given
by
rE; e N
1
K
1
i
v=E
_
e
e
0
f eVMe=E; eV=EdeV
_ _
4
in Ashleys approximation, where
Mb; a
1
a
ln
b a1 a b
b1 b
_ _
2
1 a
F arcsin
1 a 2b
1 a
_ _
;
1 a
1 a
_ _
The density of the actual energy loss is, of course, the derivative of the expression in Eq. (4)
The Ashley approximation greatly enhances the validity of the computed cross section for
electronic excitation. However, as only the DOSD is used, excitations to states that are
dipole forbidden are not included. This may not be a serious limitation unless the incident
energy is very low. An additional advantage of the approximation is that it is conveniently
used in the condensed phase.
2.2. Vibrational and Elastic Collision Cross Sections
For vibrational excitation under electron impact in the gas phase, experimental measure-
ments are generally available and are usually more reliable than calculation. In the case of
gaseous water, an almost complete compilation by Hayashi [11] is available in the all-
important lower energy regime. Hayashi decomposed the total vibrational collision cross
section into three components: r
v
1
,
v
3
for the v
1
and v
3
stretching modes, which are virtually
indistinguishable; r
v
2
for the v
2
bending mode; and r
w
for the balance. For polyatomic
molecules, rotational excitation involves very little energy relative to electronic or vibra-
tional excitation. In any case, experimentally these events cannot be distinguished from
elastic collision and are included with them in cross sections. Cross-sectional data for
vibrational excitation are of very limited availability in the condensed phase. Because inter-
nuclear forces are much stronger than the intermolecular forces, it is generally believed that
condensation does not eect vibrational cross sections a great deal and gas-phase values are
frequently used to represent the liquid in simulations. A detailed description of the cross
sections for electrons in liquid water is presented in Ref. 9, where the relevant cross sections
have been quantied for incident electron energy in the interval 10 eV to 100 keV.
Copyright 2004 by Taylor & Francis Group, LLC
Elastic collision cross sections are important for track simulation and dierential cross
sections are needed to calculate angular deviation in the track trajectory. Pimblott et al. [9]
have given an elaborate analysis for gaseous water and compared the results with the ex-
periments of Katase et al. [12]. In brief, the total elastic cross section
r
el
2p
_
p
0
dr=dhsinhdh
is well represented by the Moliere form for the dierential cross section, viz. dr/dh=
(Z
2
+Z)e
4
/( p
2
v
2
(1cosh+2g
2
)), where Z is the number of electrons in the molecule, p is the
electron momentum at velocity v, and the screening parameter, g, is a carefully chosen
parameter. Grosswendt and Waibels [13] choice for g is
g Z
2=3
1:64 0:0825 lnmc
2
T
_
=TT 2
where T is electron kinetic energy in units of mc
2
. This represents the total elastic cross
section well but, unfortunately, does not match the experimental dierential cross sectionfor
water. The analytical formula gives too little scattering in the forward direction, and it results
in a nearly isotropic distribution. Therefore Pimblott et al. [9] chose to t the dierential
experimental elastic cross sectionwithafourth-order polynomial a
0
+a
1
h+a
2
h
2
+a
3
h
3
+a
4
h
4
,
in which the a-coecients themselves were tted, as a function of energy, with another
fourth-order polynomial. The nally tted cross section agreed with experiment quite well. It
should be emphasized that the tting was done for the gas phase of water. The use of these
cross sections to represent liquid water is questionable, but there are no experimental
measurements available. The approximation has been justied by the belief that condensa-
tion does not aect the elastic collision cross section a great deal as these processes are
Figure 1 Electron energy dependence of the inelastic collision probability, r
inelastic
/(r
inelastic
+
r
elastic
), in liquid water (A) and the probability of an inelastic collision causing ionization, r
ion
/
(r
ion
+r
ex
), in gaseous (B) and liquid (C) water.
Copyright 2004 by Taylor & Francis Group, LLC
dominated by interaction of the electron with nuclear charges. While this may be true for
polyatomic molecules, it is well known that the elastic collision cross sections of atomic
liquids can be very dierent fromthose in the gas phase due to correlation eects represented
by the structure factor. In any event, it is customary to use the gas-phase elastic cross sections
in the condensed phases of molecular compounds for track simulation purposes.
For the purpose of track simulation, it is necessary to know whether the next realized
collision would be inelastic, and if it is, then either the collision would result in ionization or
in the production of a molecular excited state. The probabilities of these processes are given
by r
inelastic
/(r
elastic
+r
inelastic
), where r
elastic
and r
inelastic
are the cross sections for elastic and
inelastic collisions, and r
ion
/(r
ex
+r
ion
), where the cross sections for molecular excitation
and ionization are denoted by r
ex
and r
ion
, respectively. These probabilities for water in the
gaseous and liquid phases are shown as a function of electron energy in Fig. 1. The
probability of an inelastic process is virtually independent of the phase, although there is a
strong dependence on electron energy. On the other hand, the ionization probability de-
pends both on the phase and on the electron energy, with more ionization occurring in the
liquid phase at any electron energy.
3. IONIZATION, EXCITATION, AND RELATED PHENOMENA
The rst excitation potential of a polyatomic molecule is generally unambiguous. For
ordinary molecules it usually corresponds to transitions in the UV or in the far UV. The
excited molecule, either in the rst or in a higher excited state, may undergo prompt
dissociation, or fast internal conversion to states of the same multiplicity or convert,
relatively slowly, to states of dierent multiplicity by intersystem crossing. Kahsas rule
applies in most cases, signifying that the conversion to the state of lowest energy, singlet or
triplet, is very fast and further transformations occur relatively slowly. Occasionally, the
excited molecule will undergo a radiative transition resulting in the emission of a photon.
This process generally has a low probability, except for aromatic molecules where the
probability can be signicant. (Of course, the probabilities of the various outcomes, once the
excited molecule is formed, must add up to unity.) When the excited state is formed by the
absorption of a single photon, the probabilities are referred to as quantumyields. If the state
results from the absorption of n photons then the quantum yield is, by denition, that
respective probability divided by n.
Excited states can be formed by a variety of processes, of which the important ones are
photolysis (light absorption), impact of electrons or heavy particles (radiolysis), and, espe-
cially in the condensed phase, ion neutralization. To these may be added processes such as
energy transfer, dissociation from super-excited and ionized states, thermal processes, and
chemical reaction. Following Brocklehurst [14], it is instructive to consider some of the direct
processes giving excited states and their respective inverses. Thus luminescence is the inverse
of light absorption, super-elastic collision is the inverse of charged particle impact excitation,
and collisional deactivation is the inverse of the thermal process, etc.
The term ionization may refer to dierent processes depending on the context. For
radiation eects in the gas phase, it usually implies the removal of the least bound electron to
innity. Such a theoretical denition is not feasible in the condensed phase and it is necessary
use a heuristic or operational procedure. Thus, in liquid hydrocarbons, one may use the
electron scavenging reaction or a conductivity current to quantify the electrons liberated
from molecules. It has only been possible to extrapolate the conductivity current at a low
irradiation dose and at a relatively low external eld to saturation in the cases of liqueed
Copyright 2004 by Taylor & Francis Group, LLC
rare gases (LAr, LKr, and LXe) so that an unequivocal meaning to ionization may be
accorded [15].
The ionization potential is the minimum energy required to remove an electron from a
molecule, and as such it may be dened uniquely in the gas phase. Even so, as rst pointed
out by Platzman, ionization is not a certainty when that amount of energy is supplied to the
molecule. In super-excited states, ionization for polyatomic molecules is in eective
competition with dissociation into neutral fragments. At the same time, a small amount of
ionization may occur at energies a little below the ionization potential by a variety of
secondary processes. This blurring of the boundaries gives rise to the concept of ionization
eciency, which is the experimental ratio of the photoionization cross section to the
photoabsorption cross section. This eciency can be given a probabilistic interpretation
if it is assumed that once an excited state is formed its eventual fate is independent of the
process by which it was formed. In this sense, once the ionization eciency g
i
of a given state
is known, the cross sections for ionization and excitation may be given by g
i
r and (1g
i
)r,
respectively, where r is the cross section of production of that state either by photo-
absorption or by the impact of charged particles. This, of course, is true only when there is no
other channel for energy disposal, such as nonradiative transition to the ground state. In the
case of water in the gas phase, where the ionization potential is well established at 12.6 eV,
the ionization eciency, as a function of photon energy, shows a series of maxima and
minima until it asymptotically approaches unity at
f
20 eV [16]. This dependence implies
that the eciency of ionization depends not only on energy but also on the quantum nature
of the state reached by the absorption of a specied amount of energy.
Closely associated with the ionization potential, but distinct from it, are the W-value
and the appearance potential. The minimumenergy necessary to observe the production of a
particular ion out of a given molecule is called its appearance potential (see Ref. 2, p. 72). As
this is an experimental quantity without reference to the state originally produced, there has
been considerable confusion in the literature between the ionization potential and the
appearance potential in the liquid phase. In Sec. 4.3.2, a clarication is attempted for the
special case of liquid water. The W-value refers to the gross average energy needed to create a
single ionization in the medium, whether primary or secondary. It is greater than the
ionization potential, because of energy wastage in various nonionizing events, viz., pro-
duction of excited states, delivering energy to subexcitation electrons and to the resultant
positive ions, etc. A more detailed discussion will be found in Sec. 2.3.3.
3.1. Photoionization and Electron Impact Ionization
Photoionization is the process by which an electron is removed to generate a positive ion by
the absorption of light. Arelated process in which an electron is removed froma negative ion
to produce a neutral entity by light absorption is called photo detachment. The inverse of
photoionization is radiative capture of an electron. Applying microscopic reversibility to the
direct and inverse processes, the respective cross sections of photoionization (r
i
) and
radiative capture r
r
may be related by detailed balance as, r
i
/r
r
=(2x
f
/x
i
)(mc
2
T/h
2
m
2
),
where m is the frequency of light, and x
f
and x
i
are the statistical weights of the nal and
initial states, respectively. The factor 2 represents the spin degeneracy. Asimilar relationship
may be expected between the photo detachment and electron attachment cross sections.
These are very useful when one or another process is not easily amenable to experimental
investigation. It should be noted that the ionization cross sections are usually smaller than
the respective photoabsorption cross sections to which they are related by the ionization
eciency.
Copyright 2004 by Taylor & Francis Group, LLC
Ionization by a charged particle or by electron impact is a special inelastic collision
process in which molecular ionization occurs and the ejected electron carries the residual
energy. The basic cross section is still given by the BetheBorn theory as was used in Sec. 2.2
for stopping power and in Sec. 2.3 for secondary ionization. It should be noted that the
complete description of the impact ionization process requires a vefold dierential cross
section: one in the ejected electron energy and two each for the angular distributions of the
ejected electron and the primary particle. Such detailed information is rarely available
experimentally. Furthermore, the approximations used in the theory break down more
easily when higher order dierential cross sections are used. In radiation science, one is
frequently concerned with the total number of ionizations. As such, a singly dierential cross
section in ejected electron energy is sucient. For track simulation, a cross-section further
dierential in the angle of the ejected electron is necessary. However, in the more important
case when the energy of the secondary electron is much greater than the ionization potential,
a nearly classical relation between the ejected electron energy and angle can be invoked. In
any case, it should be emphasized that the ionization cross section, dierential in energy,
such as represented by Eq. (2.10), implies that ionization is a certainty once the deposited
energy exceeds the ionization potential. Because of the existence of super-excited states,
these cross sections must be multiplied by the ionization eciency factor. Presently, the only
useful source for this ionization eciency is experiment in the far UV. Even then, it is still
necessary to assume that the eciencies remain valid for the dipole-allowed states reached
by charged particle impact.
Under electron impact ionization, when the energy transfer greatly exceeds the orbital
ionization energy, the process resembles a Rutherford-type collision between two nearly free
electrons. On this Mott [17] imposed the condition of indistinguishability of the outgoing
electrons and obtained in the nonrelativistic limit
dr=dE
Mott
4pa
2
0
R
2
=T
_ _
i
n
i
W
2
i
W
1
i
T E
1
T E
2
_ _
5
where the same symbols are used as in Eq. (2.10). The second termwithin the brackets on the
right-hand side of Eq. (5) represents the interference between direct and exchange scatter-
ings. A manifestly relativistic treatment of the process has been given by Moller [18], how-
ever, with the caveat that in the extreme relativistic case and for a large angle scattering,
energy loss in the radiative process may not be negligible [19].
A wide variety of chemical reactions can occur following ionization or excitation of
a molecule in both gaseous and condensed phases. These may be of uni-molecular or bi-
molecular nature, initiated by electrons, ions or by the transformations of excited or ionized
molecules. These reactions include, but are not limited to, dissociation, elimination of atoms
and smaller molecules (H, H
2
, etc.), transfer of H
+
, H, H
2
, H
, and H
2
, fragmentation, ion-
molecule reaction, luminescence and energy transfer, neutralization, chain reaction, con-
densation, and polymerization, etc. These reactions will not be reviewed in this chapter but
may be found elsewhere in this book. A brief summary is also found in Chapters 4 and 5 of
Ref. 2. In the next section, some features of yields and mechanisms following excitation and/
or ionization in the liquid phase are discussed with special reference to water.
3.2. Condensed Phase: Yields and Mechanisms
The diculty of providing a theoretical denition of ionization in the liquid akin to that in
the gas phase has already been discussed. In practice, any available reaction that mimics that
Copyright 2004 by Taylor & Francis Group, LLC
of a free electron can be taken as an operational denition of ionization. These include
scavenging, attachment (dissociative or nondissociative), and other reactions of the
electron, as well as generation of solvated electrons in polar liquids and of free electrons
in nonpolar liquids. In some hydrocarbon liquids, careful extrapolation of the yield of
scavenged electrons gives a total ionization yield which is not much dierent from the
corresponding ionization yield in the gas phase; however, this is not always true. Extrap-
olation of the free-ion yield at innite external electric eld strength is generally uncertain,
depending on an assumed form of the initial electron-ion separation distance, although in
many cases such extrapolated ionization yields have been rationalized with those obtained in
scavenging experiments.
In highly polar liquids, such as water, the situation is altogether dierent and until
recently, the mechanisms and the yields of ionization and excitation were controversial.
Although it has been known for a long time that the fates of the lower excited states of water
are the same in all phases, viz. dissociation into H+OH, or into H
2
+O, the calculated yields
in the liquid phase exceeded experimental determinations by as much as a factor of four
[20,21]. An ad interim rationalization in terms of fast cage recombination lacks realism in
viewof the open structure of liquid water. As for ionization in liquid water, a very small yield
is seen at 6.5 eV[22,23] and this threshold has been variously called ionization potential [23],
threshold potential [24], or appearance potential [21]. Sander et al. [24,25] have postulated
that the mechanism of ionization by which low energy processes give e
aq
aq
OH H
3
O
aq
In the sense that the hydrated electron may be formed by the reaction of an excited state with
a ground state water molecule, the above reaction may have some validity [27]. In some of
the earlier work the initial yields were directly determined from scavenger studies. Later
works, using mostly two-photon photolysis, have been extended to higher energies and have
yielded more accurate photodissociation and photoionization yields. These have been
recently summarized in Ref. 31. Briey, the time dependencies were determined in the pico-
and subpicosecond time scales from which the initial and escape yields were found using
either analytical or Monte Carlo methods for the recombination kinetics. Fig. 2 shows
the variation of the quantum eciencies of photodissociation (g
d
) and photoionization (g
i
)
in liquid water as a function of photon energy in which some earlier data are incorpo-
rated. The quantum eciency of nonradiative transition to the ground state, computed as
Copyright 2004 by Taylor & Francis Group, LLC
(g
nr
)=1g
d
g
i
, is also shown. It is signicant below 9 eV but was ignored in earlier
expositions. The photodissociation quantum yield is very low below 6.70 eV because of the
high probability of nonradiative transition to the ground state and again very lowabove 10.0
eV because of the high ionization probability. It therefore exhibits a peak at
f
8 eV. The
ionization probability in liquid water has been postulated to reach a near certainty at a
photon energy 11.7F0.2 eV (see Fig. 2 ). Therefore, in accordance with the analysis of
Sander et al. [24,25], the ionization potential in liquid water may be dened as the minimum
energy needed to generate a quasi-free electron in the band gap, and a value of 11.7 eVcan be
assigned to it [31]. Notice that, according to this denition, the existence of the superexcited
state is denied in liquid water as the ionization potential is dened with respect to the
ionization eciency approaching unity and this quantity does not have maxima or minima.
An approximate calculation of the ionization yield in liquid water for high energy electron
radiolysis, using the data of Fig. 2 and a simplied Bethe theory, gives a G-value (yield for
100 eV absorbed energy) of 1.0 at rst collision and a total G-value of 2 obtained from it by
heuristic energy balance [31]. The latter is in good agreement with a recent experimental re-
evaluation of Bartels and co-workers [32]. The computed dissociation yield, however, falls
short of the experimental value by more than an order of magnitude. Therefore not only is
there no need for invoking cage recombination, there is actually a need for an additional
channel for dissociation into neutral fragments. Such a process, called dissociative electron
attachment (DEA), is well known in the gas phase and appears as a compound resonance in
low temperature ice in the energy interval
f
515 eV [33]. The immediate product of DEA is
H
+OH, where H
carries most of the excess energy and is likely to detach the electron
quickly. The net result is an electron degraded in energy and a dissociated water molecule.
An estimate of the total dissociation yield ( G
f
0.8) based on this hypothesis seems to be
correct in the order of magnitude [31].
Figure 2 Variation of the quantum eciencies of photodissociation (g
d
), of photoionization (g
i
),
and of nonradiative transition to the ground state, g
nr
=1 g
d
g
i
, in liquid water as a function of
photon energy.
Copyright 2004 by Taylor & Francis Group, LLC
4. TRACK STRUCTURE
4.1. Historical Perspective
The spatial distribution of the energy loss events of a charged particle is usually referred to as
a track. This conceptual picture of a track is the backbone of the theoretical description of
radiation chemistry. Tracks are considered to have a transitory existence and exist so long as
permitted by the diusion and fast reactions of radiation-produced intermediates (ions,
electrons, and radicals). A large body of radiation-physical and radiation-chemical phe-
nomena requires track models for their elucidation, including (1) LET variation of product
yields; (2) energy loss in primary excitations and ionizations; (3) yield of escaped ions; (4)
radiation-induced luminescence; and (5) particle identication.
The statistics of the energy loss events and their spatial distribution along a track, and
its branches, denes the structure of a radiation track. The ensuing chemistry depends on
incident particle energy and particle quality (LET). In 1966, a rudimentary Monte Carlo
method was used by Mozumder and Magee [35] for the statistics of energy loss events by
high energy electrons in water, employing simplied Bethe cross section derived from a
dipole oscillator distribution, which in turn was synthesized from then known moments and
experimental constraints. Their calculation divided energy deposition events into three
groups as follows: (1) spurs (spherical entities, up to 100 eV); (2) blobs (spherical or el-
lipsoidal, 100500 eV); and (3) short tracks (cylindrical, 5005000 eV). This arbitrary clas-
sication has proven to be tremendously useful and extremely robust in modeling radiation-
chemical kinetics. The track structure calculation showed that the energy partition between
the three track structure entities is a strong function of incident electron energy, dividing
approximately as the ratio of 0.64:0.12:0.24 in the spur, blob, and short track fractions for
60
Co-g irradiation. Recently, a more sophisticated calculation based on improved cross-
section methodology employing a realistic dipole oscillator distribution for liquid water has
provided an improved estimate of the ratio, 0.75:0.12:0.13, between the spur, blob, and short
track fractions [36].
The reaction of the radiation-induced radicals and ions comprising a track is signi-
cant in the condensed phases and the observed chemistry reects the competition between
the diusive relaxation of the track structure and the intra-track reaction. By contrast, in the
gas phase the reactants are widely dispersed at any reasonable dose and LET, and the in-
termingling of inter- and intra-track reactants gives a homogeneous character to the reac-
tions. The rst nonhomogeneous diusionreaction treatment of the radiation chemistry
of liquid water was proposed by Samuel and Magee [37] to explain the relative forward yield
as a function of LET. (Forward yield means the yield of observable molecular product
against radical yield, where the latter is estimated as the yield of scavenger reaction at a small
scavenger concentration.) Their simple analysis used a one-radical model, which made no
distinction between H and OH. Explicit consideration of reaction with homogeneously
distributed scavengers within the context of a one-radical-species model was later intro-
duced by Ganguly and Magee [38].
4.2. Track Structure Simulation
Track structure simulation has found application in many areas of radiation research since
the pioneering studies of Mozumder and Magee [35]. These studies all employ essentially the
same type of approach, a collision-to-collision modeling of the trajectory of the primary
radiation particle and of its daughter secondary electrons, with the most signicant dier-
ence between dierent calculations being the interaction cross sections used to describe the
Copyright 2004 by Taylor & Francis Group, LLC
various inelastic and elastic processes. The attenuating medium is usually treated as a
homogeneous continuum. The basic methodology for fast electrons is as follows: The
primary electron has an energy E and is at point z traveling in a dened direction. The
distance the radiation particle travels before its next collision, Dz, has a Poisson distribution
with a mean free path,
total
, which is dened by the total cross section, r
total
, for elastic and
inelastic processes, i.e.,
L
total
qr
total
1
qr
elastic
r
inelastic
1
6
where q is the number density of molecules. Consequently, the distance between two
consecutive collisions is obtained by the inversion method sampling from the probability
distribution function
PDz; L
total
1 exp Dz=L
total
7
and employing a uniformly distributed random number, U, in the range (01). Having
determined the distance traveled between collisions and knowing the direction of travel, the
position of the primary electron is modied to z
new
. The elastic and inelastic cross sections
(r
elastic
and r
inelastic
) comprising the total cross section are now used to determine the na-
ture of the collision by comparing a second uniformly distributed random number with
the probability of an inelastic collision, i.e., r
inelastic
/(r
elastic
+r
inelastic
). If the event is in-
elastic, the type of inelastic event is determined in a similar manner from the cross sections
for ionization (r
ionization
), excitation (r
excitation
), and the inferior energy loss processes
(r
vibration
), i.e.
0 <U <r
ionization
=r
inelastic Z
ionization
r
ionization
=r
inelastic
<U <r
ionization
r
excitation
=r
inelastic Z
excitation
r
ionization
r
excitation
=r
inelastic
<U <1
Z
vibration
The magnitude of the energy transfer, c, is calculated and the primary energy is reduced to
reect the loss, E
new
=Ec. For elastic events and for inelastic events that do not result in
ionization, the new trajectory of the electron is determined and the simulation proceeds in
the same manner. Following an elastic event, the direction of travel is determined from the
dierential elastic cross section, r
elastic
V (h,E), while excitation events are often assumed not to
alter the trajectory of the electron. If an inelastic collision causes an ionization event, the
ionization channel is determined using the cross sections appropriate for the dierent
ionization processes and the energy lost to the sibling positive ion is calculated. The
trajectories of the primary and the secondary electron are calculated from the kinematics.
When the energy of the daughter electron is smaller than a pre-dened cut-o energy, c
stop
,
this electron is not considered further, and the attenuation of the parent electron is
continued. If the energy of the secondary electron is greater than c
stop
, then its energy is
degraded until it is smaller than c
stop
before continuing the simulation of the parent electron
trajectory. The trajectory of the primary electron is followed until its energy drops below a
pre-dened cuto, c
nal
. Individual track simulations have many uses, for instance, in the
modeling of radiation chemical kinetics or quantifying radiation-induced DNA damage.
However, repeated realization of many dierent tracks is necessary to obtain statistically
signicant track averaged quantities such as stopping powers, ranges, and penetrations. The
mean values for many of these macroscopic parameters are available from alternative
calculations, but track structure simulation provides the best means for performing
sensitivity analyses and for accessing the distribution of a radiation parameter about its
mean value.
Copyright 2004 by Taylor & Francis Group, LLC
The simulation of a heavy ion track structure employs essentially the same method-
ology as described for energetic electrons; except that
the eects of elastic and inferior inelastic collisions are not signicant for ener-
getic heavy ions,
electron capture and loss by the primary ion must be considered, and
k
ij
c
i
c
j
k
mn
c
m
c
n
8
where D
i
and c
i
are the diusion coecient and spatially dependent concentration of
species i. The rst term of the right-hand side of the equation represents the diusive
contribution to the evolution of c
i
, while the second and third terms represent removal
and production of i by reaction. The set of coupled equations is solved either by analytic
approximation or numerically.
The most frequently used analytic, approximate models for the fast kinetics in
radiolysis are based on an approximation suggested by Jae and known as prescribed dif-
fusion. The initial spatial distribution of the radiation-induced particles, their concentration
prole, is assumed to be Gaussian, and the kinetic analysis invokes the approximation that
reaction only aects the number of particles and not the form of the nonhomogeneous
spatial prole, which is therefore always Gaussian. The prescribed diusion approximation
relies upon the Greens function solution for the diusion equation being a Gaussian whose
variance develops as a linear function of time. This assumption is incorrect if the kinetics
is nonlinear, which is the case in real scenarios. In his pioneering study, Jae considered a
cylindrically symmetric columnar system, appropriate for high-LET radiation. In conven-
tional studies considering the kinetics of low-LET radiolysis, the system is usually described
in terms of a spherically symmetric typical isolated spur. The concentration prole of species
i is
c
i
r
N
i
2pr
2
i
3=2
expr
2
=2r
2
i
9
where N
i
is the number of i particles in the spur and r
i
2
is the variance of the Gaussian prole
and is time dependent, r = (r
2
(t = 0)+2Dt)
1/2
. Schwarz [42] extended the methodology to
a fully functional multi-radical model. By considering the species e
aq
, H+OH, H
3
O
+
, and
H
2
and employing an appropriate reaction scheme with realistic diusion and reaction rate
coecients, he was able to give a satisfactory (although not entirely accurate) description of
the variation of molecular and radical yields with LET, scavenger concentration, and pH.
In addition to the prescribed diusion approximation, several dierent numerical
treatments have been developed for solving the set of coupled diusion-reaction equations
(Eq. 9) [4345]. Most recent studies have been performed using a technique in which a
typical, average spur is used as a representation of the electron track. The spur is divided
into concentric shells as shown in Fig. 3. Each shell is suciently thin that the concentration
of reactants within the shell can be regarded as constant. Diusion takes place between
adjacent shells and reactions within a shell are modeled using deterministic rates. The
resulting set of coupled equations describing the diusion and the kinetics is solved using the
FACSIMILE implementation of the Gear algorithm [46]. The parameters dening the spur
for aqueous solutions have been optimized to match measured experimental yields in simple
scavenger studies [45]. The optimum energy of the spur, e
spur
, is 62.5 eV, which given the
Copyright 2004 by Taylor & Francis Group, LLC
approximate nature of this model is about the mean energy loss predicted by track structure
calculations.
5.3. Stochastic Kinetics
Stochastic diusion-kinetic treatments address the two fundamental aspect of the radiolysis,
i.e., the structure of the radiation track and the chemistry of the resulting spatially
nonhomogeneous distribution of radiation induced reactants. The track structure, describ-
ing the initial positions of the reactants, is usually prescribed by a simulation, as described
above, employing a collision-to-collision methodology with cross sections appropriate for
liquid water, then a diusion-kinetic simulation using either a computer intensive random
ight approach or the more elegant, but approximate, Independent Reaction Times
methodology is employed to model the radiation chemical kinetics.
Random Flight Simulation
The modeling of the radiation chemistry by diusion-kinetic simulation relies upon the
generation of reaction times frominitial coordinate positions. In a randomight simulation,
this calculation is performed in a brute force manner by mimicking the diusion and
encounter of the reactive particles. Several variants of the random ight simulation method
have been developed at dierent levels of sophistication, but all follow the same general
approach [4750]. Each simulation begins from a realizable conguration, generated at
random from the assumed spatial prole, either a spur or an idealized track structure. The
diusive trajectories of the particles are then simulated using a discretization of the
appropriate stochastic dierential equations. Reaction is modeled by overlap at the end
of a time step and/or by estimating the probability of encounter during a time step using a
bridging process. Each realization is continued until reaction is complete or a pre-dened
cut-o time is attained. Reaction kinetics is obtained by simulation of
f
10
4
spatial
congurations using dierent initial random number seeds. Each diusive jump has two
components. The rst represents the motion caused by the bueting of the solvent and the
Figure 3 Division of the typical spur into concentric shells in the deterministic treatment of spur
kinetics using the FACSIMILE algorithm.
Copyright 2004 by Taylor & Francis Group, LLC
second reects the drift caused by any force F. The jump is calculated using an Euler-type
discretization of the stochastic dierential equation, which gives change in a particles
position of
yr
i
2D
i
yt
_
N
3
0; 1
D
i
F
i
k
B
T
dt
where D
i
is the diusion coecient of the particle, yt is the length of the time step, and N
3
is a
three-dimensional normal random vector with unit variance. This simulation methodology
has been validated, for several types of idealized system, including radical spurs and ionic
spurs in both high and low permittivity solvents [47,48,51].
For neutral reactants the force F=0, and therefore [47]
dr
i
2D
i
dt
_
N
3
0; 1
When the particles are charged, the strong inter-ion forces modify the diusion. The
(Coulombic) force on ion i is then [48]
F
i
k
B
T
j
z
i
z
j
r
c
r
3
ij
r
ij
where z
i
is the charge on ion i, and r
ij
is the inter-ion vector. The Onsager distance, r
c
,
is the distance at which the potential energy of the Coulombic interaction of two unit
charges has magnitude k
B
T.
In the presence of an applied electric eld, E, the expression for the force on an ion is
further modied to [52]
F
i
k
B
T
j
z
i
z
j
r
c
r
3
ij
r
ij
z
i
eE
where e is the electron charge.
The majority of simulations reported in the literature use a xed time step, yt [48,50];
however, this treatment is computationally inecient. When the particles are well separated
the probability of encounter is small. Sophisticated methods have been developed, which
allow more ecient computation by incorporating variable time steps [47,49,51]. In these
treatments, the time step, yt, is determined by the proximity of the particles. The time step is
selected by one of two methods:
(i) so that the pair with the shortest inter-particle separation has a very small
encounter probability [47], or
(ii) so that the change in the inter-particle drift of every pair during the time step is
small [49,51].
In both treatments, the minimum time step becomes increasingly small as two particles
approach so a minimum time step, typically
f
10 fsec, has to be employed to ensure that
the simulation does not stall.
If a reaction is diusion-controlled then reaction occurs on encounter of two particles.
In a simulation employing nite time steps, a pair of reactants may encounter during a time
step and then separate before the end of the time step. This pair should have reacted, but
the reaction is not registered. Consequently, the modeled kinetics underestimates both the
rate and the amount of reaction. This problem of encounter during a time step can be
Copyright 2004 by Taylor & Francis Group, LLC
overcome using the conditional encounter probability for an interpolating bridging
process [53]. Two forms of bridging process have been developed: the Bessel bridge, which
assumes that the separation behaves as a Bessel process, and the Brownian bridge in which
the separation between the two particles is approximated by a Wiener process.
While many of the important reactions in radiation and photochemistry are fast, not
all are diusion-limited. The random ight simulation methodology has been extended to
include systems where reaction is only partially diusion-controlled or is spin-controlled
[54,55]. The technique for calculating the positions of the particles following a reecting
encounter has been described in detail, but (thus far) this improvement has not been incor-
porated in realistic diusion kinetic simulations. Random ight techniques have been suc-
cessfully used to model the radiation chemistry of aqueous solutions [50] and to investigate
ion kinetics in hydrocarbons [48,50,5658].
Independent Reaction Times Modeling
The IRT model has been developed in detail in a series of papers of Green, Pimblott and co-
workers and has been validated by comparison with full random ight simulations
[47,49,51]. The IRT treatment of the radiation chemistry relies upon the generation of
randomreaction times from initial coordinate positions from pair reaction time distribution
functions. A simulation, such as a random ight calculation, starts with the initial spatial
distribution of the reactants. The separations between all the pairs of particles are evaluated
Figure 4 Schematic representation of the IRT method for a four-radical spur.
Copyright 2004 by Taylor & Francis Group, LLC
and overlapping pairs are allowed to react. Reaction times for all the surviving pairs are
calculated from the reaction time distribution functions for the pairs as if they were in
isolation. (This approximation is an application of the independent pairs approximation that
is implicit in the DebyeSmoluchowski treatment of diusion limited kinetics.) The resulting
ensemble of times is then used to determine the times of subsequent reactions. When a
reaction occurs, any reactive products are positioned using an independent diusion
treatment [47]. New reaction times are determined from the appropriate rst passage time
distribution function and a modied ensemble of times is created. The simulation proceeds
until a predened time (usually 1 Asec) is reached or until no reactive particles remain. The
kinetics of the system is obtained by repeated realization using a dierent initial cong-
uration and dierent random number seeds. This technique is demonstrated schematically
for a four-radical spur in Fig. 4.
The IRT method was applied initially to the kinetics of isolated spurs. Such calcu-
lations were used to test the model and the validity of the independent pairs approximation
upon which the technique is based. When applied to real radiation chemical systems, iso-
lated spur calculations were found to predict physically unrealistic radii for the spurs,
demonstrating that the concept of a distribution of isolated spurs is physically inappro-
priate [59]. Application of the IRT methodology to realistic electron radiation track struc-
tures has now been reported by several research groups [6064], and the excellent agreement
found between experimental data for scavenger and time-dependent yields and the pre-
dictions of IRT simulation shows that the important input parameter in determining the
chemical kinetics is the initial conguration of the reactants, i.e., the use of a realistic ra-
diation track structure.
6. APPLICATION TO RADIATION CHEMICAL STUDIES
Investigation of the eects of radiation type and energy on radiation chemical kinetics using
a combination of experimental and stochastic simulation methods provides the most direct
insight into the nonhomogeneous diusion and reaction processes occurring in a radiation
track, which ultimately determine the damage caused by radiation. While the chemistry
induced by low-LET radiation has received considerable experimental and theoretical at-
tention, similar comparisons on the eects of high-LET radiation are lacking.
6.1. Gamma and Electron Radiolysis
The irradiation of water is immediately followed by a period of fast chemistry, whose short-
time kinetics reects the competition between the relaxation of the nonhomogeneous spatial
distributions of the radiation-induced reactants and their reactions. Avariety of gamma and
energetic electron experiments are available in the literature. Stochastic simulation methods
have been used to model the observed short-time radiation chemical kinetics of water and
the radiation chemistry of aqueous solutions of scavengers for the hydrated electron and the
hydroxyl radical to provide fundamental information for use in the elucidation of more
complex, complicated chemical, and biological systems found in real-world scenarios.
Hydrated Electron Kinetics
There is a large amount of experimental data on the radiation chemical yield of e
aq
,
including direct absorption measurements of the time dependence of the yield in deaerated
Copyright 2004 by Taylor & Francis Group, LLC
water [6673] and a variety of scavenger studies that have been used to determine the eect of
scavenging capacity, s (equal to the product of the scavenger concentration and the rate
coecient for the scavenging reaction, k[S]) on the scavenged yield of e
aq
[45]. Extrapolation
of the recently re-evaluated time dependence of e
aq
)
f
23 nm [42].
Figure 5 Decay kinetics of the hydrated electron. Experimental data: (....) [32]; ( ) [73].
IRT kinetics: ( ) r(e
aq
)=7.5 nm.
Copyright 2004 by Taylor & Francis Group, LLC
Precursors to the Hydrated Electron
The scavenging capacity dependence of the scavenged yield of e
aq
and of e
aq
scavengers. The
yields of N
2
from N
2
O solutions correspond to predictions from the Laplace transform of
the decay kinetics of e
aq
+NO
3
) reaction does
not have an eect on the amount of electrons scavenged. At higher concentrations,
k(e
aq
+S)[S]
f
10
9
sec
1
, the scavenging of e
pre
scavenger N
2
O and for NO
3
) [79]; nitrous oxide(E) [80], (z) [81], (Y) [82]. IRT simulation: nitrate( );
selenate( ); nitrous oxide( ).
Copyright 2004 by Taylor & Francis Group, LLC
the reaction of e
pre
with SeO
4
2
plays in determining the amount of scavenging. Only in very
dilute solution is the reaction of e
aq
with SeO
4
2
dominant.
6.2. Light and Heavy Ion Radiolysis
Stochastic simulation methods have recently been developed for modeling heavy ion track
structures using collision cross sections appropriate for liquid water [83]. Realistic diusion-
kinetic calculations on the eects of high-LET radiations such as alpha particles and
accelerated light and heavy ions are scarce [8486], and the only system that has been
investigated thoroughly is the dependence of the Fricke dosimeter on radiation ion and
energy [86]. Fig. 7 compares the results of stochastic IRT simulations for track segments
with experimentally determined dierential yields [8789], showing excellent agreement for
energetic electrons and for nonrelativistic ions, including
1
H,
4
He,
12
C, and
20
Ne. The data
reveal a signicant eect of particle type and energy, which reects the competition between
intra-track reaction of the radiation-induced radicals, diusion, and scavenging. This
competition is modied by changes in the ion track structure. Examination of the underlying
Monte Carlo track structure simulations shows that the radial energy loss proles are similar
for ions with the same velocity/charge ratio and that
f
40% of the energy is initially
deposited within a water diameter of the track axis. The simulations demonstrate that LET
of the ionizing radiation is a poor parameter for characterizing the Fricke dosimeter, and the
Figure 7 Eect of ion energy on the response of the Fricke dosimeter. The solid lines and lled
points are the experimental data. ( ) labeled curves for
1
H,
4
He,
12
C, and
20
Ne ions [87]; ()
2
D(at E/2) [88]; (E)
4
He
2+
[88]; (z)
12
C [89]; (Y)
20
Ne [89]. The dashed lines joining the open points
represent stochastic IRT simulations for heavy ions, and the dotted line refers to calculations for the
complete slowing down of energetic electrons.
Copyright 2004 by Taylor & Francis Group, LLC
Figure 8 Eect of
1
H ion energy on the yield of N
2
from saturated N
2
O solution. Experimental
data: track average yield (
) [90]. IRT simulation: track segment yield (o); track average yield
( ).
Figure 9 Eect of radiation type and of scavenging capacity for e
aq
on the yield of NH
3
and N
2
from glycyl-glycine and N
2
O solutions, respectively.
Copyright 2004 by Taylor & Francis Group, LLC
observed chemistry is predicted more precisely by the square of the ratio of the particle
charge to its velocity.
There are almost no radiation chemical studies of the yields of the primary radiation-
induced species made with ions of well-dened energy. To develop a realistic, predictive
model for the eects of radiation it is important to chart the interplay between the various
physical and chemical factors aecting track structure and track kinetics for all of the
primary radiation-induced species, e
aq
, OH, H
2
, and H
2
O
2
. For instance, modeling of the
track chemistry of a heavy ion necessitates the simulation of track segments for ions of
dierent energy, followed by integration, as track segment yields are very dierent from the
track average yield except at high ion energies. A typical calculation of this kind is shown in
Fig. 8, which examines the production of N
2
from the
1
H ion radiolysis of aqueous nitrous
oxide solutions [90]. Astudy of this type is necessary for each calculated result in a scavenger
concentration study, such as Fig. 9, which compares recent experimental yields with
stochastic IRT predictions for the eect of the concentration of glycyl-glycine on the yield
of NH
3
and of nitrous oxide on the yield of N
2
from aqueous solutions irradiated with
60
Co
gamma rays and 5 MeV He ions [91,92]. The potential interplay of calculation with
experiment suggested by the data shown in Figs. 79 has many benets. In addition to
the accurate prediction of the eect of radiation quality on radiation-induced chemistry,
realistic kinetic modeling will enable the elucidation of the underlying physical and chemical
processes by analysis of the simulated track structures and the modeled interplay of diusion
and reaction.
7. APPLICATION TO HIGH-TEMPERATURE NUCLEAR REACTORS
There is a large amount of data on nonhomogeneous track chemistry of energetic electrons
at room temperature, and the track structure and diusion-limited kinetics are well param-
eterized. This wealth of knowledge contrasts with the limited information about the eects
of radiation on aqueous solutions at elevated temperatures. The majority of the studies at
elevated temperatures have been performed at AECL, Canada [93], or at the Cookridge
Radiation Laboratory, University of Leeds, UK [94]. These two groups have focused on
measuring the rate coecients of the reactions of the radiation-induced radicals and ions of
water. The majority of the temperature dependencies can be tted with an empirical
Arrhenius-type expression, k=A exp(E
a
/k
B
T ). This type of parameterization provides a
satisfactory estimate of the rate coecient, but it should not be taken to have any
mechanistic implications. Radiation-induced radicals are very reactive, and so most of their
reactions are close to diusion-controlled at room temperature. At elevated temperature,
however, deviations from diusion-control are apparent. Consequently, the temperature
dependence has to be described using two components, k
1
obs
=k
1
di
+k
1
react
, one for the
encounter process, k
di
, and one for the reaction process, k
react
. The temperature depend-
encies of the diusion coecients of the radiation-induced ions and of the water molecule
are partially known; however, the diusion-coecients of the neutral radicals are poorly
characterized even at room temperature. In the diusion-kinetic calculations, it has been
necessary to assume that these species have diusion coecients with the same temperature
dependence as the self-diusion of water.
The temperature dependencies of the yields of e
aq
, H, H
2
, OH, and H
2
O
2
following g
irradiation have been investigated using radical scavengers. The yields of the principal
reducing, e
aq
, and oxidizing, OH, radicals are quite well known and increase with increasing
temperature. There are some discrepancies between dierent experimental studies for H
2
;
Copyright 2004 by Taylor & Francis Group, LLC
however, the yield is either independent of temperature or increases slightly with increasing
temperature (<30% at 300jC). The yield of H atom is not accessible directly and can only
be obtained in conjunction with H
2
. The sum of the yields of H and H
2
is more or less
independent of temperature. The yield of H
2
O
2
appears to decrease with increasing
temperature in contrast to H
2
; however, there is no reliable information above 150jC. A
number of deterministic and stochastic calculations have investigated the gamma and heavy
ion radiolysis of water at elevated temperatures [9599]. For the most part these studies have
focused on the observable yields rather than the underlying track structures. The principal
eect of elevated temperature on the track structure of energetic electrons in water is due to
changes in the inter-event mean-free path. This property depends inversely upon the density
of water so increasing temperature increases the separation between intra-track energy loss
events. Consequently, (i) the dose distribution of the track is more diuse at elevated
temperatures than that at roomtemperature, and (ii) the spurs (clusters of radiation-induced
reactants) comprising the nonhomogeneous spatial distribution of reactants are more
isolated. Fig. 10 compares the dose distribution of a 1-keV electron at 25jC and at
300jC. The very apparent spreading of the track structure is signicant as chemistry in
clusters emphasizes stochastic eects and should therefore lead to increased recombination
Figure 10 Comparison of the dose distribution of a 1-keV electron at 25jC (dashed contours) and
at 300jC (solid contours). Dose in eV/nm
3
.
Copyright 2004 by Taylor & Francis Group, LLC
relative to molecular product formation. The reaction time for a pair of radiation-induced
reactants depends upon their separation, their transport properties, and the rate of reaction
upon diusive encounter. In addition, physical parameters, such as the relative dielectric
coecient and the density of water, have to be considered in calculations as they aect the
diusion-reaction kinetics by modifying the inter-ionic forces and the rates of processes
involving the solvent, i.e., water. Stochastic (and deterministic) radiation chemical kinetic
calculations have been made for gamma radiolysis of water over the temperature range 25jC
to 300jC, and the predictions have been compared with available experimental data for
gamma radiolysis. Agreement is found for all of the radiation-induced species and for the
combined yield of H and H
2
. For the radicals e
aq
and OH, the calculations quantitatively
reproduce the measured yields and their temperature dependence. The predictions of
stochastic IRT calculations, employing simulated fast electron tracks, for the yield of OH
radicals are compared with the available experimental data in Fig. 11 [100,101].
8. LOW PERMITTIVITY SOLVENTS
In contrast to liquid water, a detailed mechanistic understanding of the physical and chem-
ical processes occurring in the evolution of the radiation chemical track in hydrocarbons is
not available except on the most empirical level. Stochastic diusion-kinetic calculations for
lowpermittivity media have been limited to simple studies of cationelectron recombination
in aliphatic hydrocarbons employing idealized track structures [5658], and simplistic
deterministic calculations have been used to model the radical and excited state chemistry
[102]. While these calculations have been able to reproduce measured free ion yields and end
product yields, respectively, the lack of a detailed mechanistic model makes it very dicult
Figure 11 Eect of temperature on the yield of OH radicals from gamma radiolysis. Experiments:
(E) Kent and Sims [100], (n) Elliot et al. [101] Calculation: (solid line) IRT modeling using track
structure simulation.
Copyright 2004 by Taylor & Francis Group, LLC
to predict the eects of dierent types of radiation or to speculate on the eects of radiation
on uninvestigated systems.
Dierential dipole oscillator strengths are available for a large variety of gaseous
hydrocarbons, and for a limited number of liquids and solids. These data have allowed the
calculation of the condensed phase energy loss properties, including the cross sections for
inelastic collisions, of electrons and positrons in n-hexane, cyclohexane and benzene, and
in polyethylene and polystyrene [103]. Cross sections for elastic collisions have also been
evaluated using partial wave methods [104]. Recently, these inelastic and elastic cross sec-
tions have been successfully used in the stochastic diusion-kinetic simulation of the electric
eld dependence of the o-positronium yield in n-hexane [58], showing that positronium for-
mation involves reaction of the thermalized positron with electrons produced along the last
few hundred nanometers of its track-end, corresponding to several kiloelectronvolts of en-
ergy attenuation. More complex studies of radiation chemical eects in hydrocarbons are
necessary.
9. PERSPECTIVE
Over the last 15 years, stochastic methodologies for modeling the radiation chemical kinetics
of aqueous systems have made signicant progress, advancing from isolated spur studies to
58
Ni track chemistry simulations. The IRT methodology for modeling radiation and
photochemical kinetics is now used worldwide. Development of realistic models for track
chemistry is crucial to predicting radiation eects in complicated practical applications and
where experiments cannot be performed. However, the calculations for both low- and high-
LETradiation still suer froma number of deciencies. The physical, physico-chemical, and
chemical processes that occur in the radiation track on a subpicosecond time scale are not
well characterized despite recent advances into this time domain in laser irradiation studies.
A multi-faceted approach employing experiments in conjunction with Monte Carlo kinetic
modeling and electronic structure calculations is clearly desirable to elucidate fundamental
radiation chemical processes from their earliest stages.
REFERENCES
1. Inokuti, M. Revs. Mod. Phys. 1971, 43, 297.
2. Mozumder, A. Fundamentals of Radiation Chemistry; Academic Press: San Diego, Ch.2.
3. Fano, U. Annu. Rev. Nucl. Sci. 1963, 13, 1.
4. LaVerne, J.A.; Mozumder, A. J. Phys. Chem. 1985, 89, 4216.
5. LaVerne, J.A.; Mozumder, A. Radiat. Res. 1983, 96, 219.
6. Ashley, J.C.; Williams, M.W. Report RADC-TR-83-87, Rome Air Force Development Center;
Griss Air Force Base: NY, 1983; 5 pp.
7. Ashley, J.C. J. Electron Spectrosc. Relat. Phenom. 1988, 46, 199.
8. Pimblott, S.M.; LaVerne, J.A. J. Phys. Chem. 1991, 95, 3907, and references therein.
9. Pimblott, S.M.; LaVerne, J.A.; Mozumder, A. J. Phys. Chem. 1996, 100, 8595.
10. See, for example, Gradshteyn, I.S.; Ryzhik, I.M. Tables of Integrals, Series and Products;
Academic Press: New York, 1980.
11. Hayashi, M. Atomic and Molecular Data for Radiotherapy; International Atomic Energy
Agency: Vienna, 1989; 193 pp.
12. Katase, A.; Ishibashi, K.; Matsumoto, Y.; Sakae, T.; Maezono, S.; Murakami, E.; Watanabe,
K.; Maki, H. J. Phys. B 1986, 19, 2715.
Copyright 2004 by Taylor & Francis Group, LLC
13. Grosswendt, B.; Waibel, E. Nucl. Instrum. Methods 1978, 155, 145.
14. Brocklehurst, B. Radiat. Res. Rev. 1970, 2, 149.
15. Doke, T. Portugal Phys. 1981, 12, 9.
16. Haddad, G.N.; Simpson, J.A.R. J. Chem. Phys. 1986, 84, 6623.
17. Mott, N.F. Proc. R. Soc. A 1930, 216, 259.
18. Moller, C. Z. Physik 1931, 70, 786.
19. Mott, N.F.; Massey, H.S.W. The Theory of Atomic Collisions; Third Edition; O.U.P.:
London Ch. XXII. 3, 1965.
20. Kaplan, I.G.; Miterev, A.M.; Sukhonosov, V.Ya. Radiat. Phys. Chem. 1990, 36, 493.
21. Pimblott, S.M.; Mozumder, A. J. Phys. Chem. 1991, 95, 7291.
22. Bernas, A.; Goulet, T.; Jay-Gerin, J-P.; Ferradini, C. Proc. 7th Tihany Symposium on Ra-
diation Chemistry; Dobo, J., Nykos, L., Schiller, R., Eds.; Budapest, 1991; 17 pp. And the
following discussion on p. 23.
23. Nikogosyan, D.N.; Oraevsky, A.A.; Rupasov, V.I. Chem. Phys. 1983, 77, 131.
24. Sander, M.U.; Luther, K.; Troe, J. Ber. Bunsen-Ges. Phys. Chem. 1993, 97, 953.
25. Sander, M.U.; Luther, K.; Troe, J. J. Phys. Chem. 1993, 97, 11489.
26. Keszei, E.; Jay-Gerin, J.-P. Can. J. Chem. 1992, 70, 21.
27. Bartels, D.M.; Crowell, R.A. J. Phys. Chem. A 2000, 104, 3349.
28. McNesby, J.R.; Tanaka, I.; Okabe, H. J. Chem. Phys. 1962, 36, 605.
29. Stief, L.J.; Payne, W.A.; Klemm, R.B. J. Chem. Phys. 1975, 62, 4000.
30. Sokolev, U.; Stein, G. J. Chem. Phys. 1966, 44, 2189 (ibid, 44, 3329 (1966)).
31. Mozumder, A. Phys. Chem. Chem. Phys. 2002, 4, 1451.
32. Bartels, D.M.; Cook, A.M.; Mudaliar, M.; Jonah, C.D. J. Phys. Chem. A 2000, 104, 1686.
33. Simpson, W.C.; Orlando, T.M.; Parenteau, L.; Nagesha, K.; Sanche, L. J. Chem. Phys. 1998,
108, 5027.
34. Mott, N.F.; Frame, J.W. Proc. Camb Philos. Soc. 1931, 27, 511.
35. Mozumder, A.; Magee, J.L. J. Chem. Phys. 1966, 45, 3332.
36. Pimblott, S.M.; LaVerne, J.A.; Mozumder, A.; Green, N.J.B. J. Phys. Chem. 1990, 94, 488.
37. Samuel, A.H.; Magee, J.L. J. Chem. Phys. 1953, 21, 1080.
38. Ganguly, A.K.; Magee, J.L. J. Chem. Phys. 1956, 25, 129.
39. Jae, G. Ann. Phys. IV 1913, 42, 303.
40. Wilson, C.T.R. Proc. R. Soc. A 1911, 85, 285.
41. Wilson, C.T.R. Proc. R. Soc. A 1912, 87, 277.
42. Schwarz, H.A. J. Phys. Chem. 1969, 73, 1928.
43. Trumbore, C.N.; Short, D.R.; Fanning, J.E.; Olson, J.H. J. Phys. Chem. 1978, 82, 2762.
44. Burns, W.G.; Sims, H; Goodall, J.A.B. Radiat. Phys. Chem. 1984, 23, 143180.
45. LaVerne, J.A.; Pimblott, S.M. J. Phys. Chem. 1991, 95, 3196.
46. Chance, E.M.; Curtis, A.R.; Jones, I.P.; Kirby, C.R. Report AERE-R 8775; AERE: Harwell,
1977.
47. Cliord, P.; Green, N.J.B.; Oldeld, M.J.; Pilling, M.J.; Pimblott, S.M. J. Chem. Soc., Faraday
Trans. 1986, 82, 26732689.
48. Bartczak, W.M.; Hummel, A. J. Chem. Phys. 1987, 87, 5222.
49. Cliord, P.; Green, N.J.B.; Pilling, M.J.; Pimblott, M. J. Phys. Chem. 1987, 91, 44174422.
50. Turner, J.E.; Hamm, R.N.; Wright, H.A.; Ritchie, R.H.; Magee, J.L.; Chatterjee, A.; Bolch,
W.E. Radiat. Phys. Chem. 1988, 32, 503510.
51. Green, N.J.B.; Pimblott, S.M. J. Phys. Chem. 1990, 94, 2922.
52. Pimblott, S.M. J. Chem. Soc., Faraday Trans. 1993, 89, 3533.
53. Green, N.J.B. Mol. Phys. 1988, 65, 1399.
54. Pimblott, S.M.; Green, N.J.B. J. Phys. Chem. 1992, 96, 9338.
55. Bolton, C.E.; Green, N.J.B. J. Phys. Chem. 1996, 100, 8807.
56. Siebbeles, L.D.A.; Bartczak, W.M.; Terrissol, M.; Hummel, A. J. Phys. Chem. A 1997, 101,
1619.
57. Bartczak, W.M.; Hummel, A. Radiat. Phys. Chem. 1997, 49, 675.
Copyright 2004 by Taylor & Francis Group, LLC
58. Alba Garc a, A.; Pimblott, S.M.; Schut, H.; van Veen, A.; Siebbeles, L.D.A. J. Phys. Chem. B
1997, 106, 1124.
59. Cliord, P.; Green, N.J.B.; Pilling, M.J.; Pimblott, S.M.; Burns, W.G. Radiat. Phys. Chem.
1987, 30, 125.
60. Zaider, M.; Brenner, D.J. Radiat. Res. 1984, 100, 245.
61. Green, N.J.B.; Pilling, M.J.; Pimblott, S.M.; Cliord, P. J. Phys. Chem. 1990, 94, 251.
62. Hill, M.A.; Smith, F.A. Radiat. Phys. Chem. 1994, 43, 265.
63. Cobut, V.; Frongillo, Y.; Patau, J.P.; Goulet, T.; Fraser, M.J.; Jay-Gerin, M.J. Radiat. Phys.
Chem. 1998, 51, 229; Frongillo, Y.; Goulet, T.; Fraser, M.J.; Cobut, V.; Patau, J.P.; Jay-Gerin,
J.-P. Radiat. Phys. Chem. 1998, 51, 245.
64. Pimblott, S.M.; LaVerne, J.A. J. Phys. Chem. A 1997, 101, 5828.
65. Watanabe, R.; Saito, K. Radiat. Phys. Chem. 2001, 62, 217.
66. Buxton, G.V. Proc. R. Soc. London A 1972, 328, 9.
67. Wol, R.K.; Bronskill, M.J.; Aldrich, J.E.; Hunt, J.W. J. Phys. Chem. 1973, 77, 1350.
68. Jonah, C.D.; Hart, E.J.; Matheson, M.S. J. Phys. Chem. 1973, 77, 1838.
69. Fanning, J.E. Evidence for spurs in aqueous radiation chemistry. Ph.D. Thesis, University of
Delaware, 1975.
70. Jonah, C.D.; Matheson, M.S.; Miller, J.R.; Hart, E.J. J. Phys. Chem. 1976, 80, 1267.
71. Sumiyoshi, T.; Tsugaru, K.; Yamada, T.; Katayama, M. Bull. Chem. Soc. Jpn. 1985, 58, 3073.
72. Chernovitz, A.C.; Jonah, C.D. J. Phys. Chem. 1988, 92, 5946.
73. Shiraishi, H.; Katsumura, Y.; Ishigure, K. Radiat. Phys. Chem. 1989, 34, 705.
74. Crowell, R.A.; Bartels, D.M. J. Phys. Chem. 1996, 100, 17940.
75. Pimblott, S.M. unpublished results.
76. Konovalov, V.V.; Raitsimring, A.M.; Tsvetkov, Y.D. Radiat. Phys. Chem. 1988, 32, 623;
Konovalov, V.V.; Raitsimring, A.M. Chem. Phys. Lett. 1990, 171, 326.
77. Hyder, M.L. J. Phys. Chem. 1965, 69, 1858.
78. Draganic, I.G.; Draganic, Z.D. J. Phys. Chem. 1973, 77, 765.
79. Pastina, B.; LaVerne, J.A. J. Phys. Chem. A 1999, 103, 209.
80. Dainton, F.S.; Logan, S.R. Trans. Faraday Soc. 1965, 61, 715.
81. Head, D.A.; Walker, D.C. Can. J. Chem. 1967, 45, 2051.
82. Jha, K.N.; Ryan, T.G.; Freeman, G.R. J. Phys. Chem. 1975, 79, 868.
83. Pimblott, S.M. unpublished results.
84. Brenner, D.J.; Zaider, M. Radiat. Prot. Dosim. 1985, 13, 127.
85. Herve du Penhoat, M.-A.; Meesungnoen, J.; Goulet, T.; Filali-Mouhim, A.; Mankhetkorn, S.;
Jay-Gerin, J.-P. Chem. Phys. Lett. 2001, 341, 135.
86. Pimblott, S.M.; LaVerne, J.A. J. Phys. Chem. A 2002, 106, 9420.
87. LaVerne, J.A.; Schuler, R.H. J. Phys. Chem. 1987, 91, 5770.
88. Sauer, M.C.; Hart, E.J.; Naleway, C.A.; Jonah, C.D.; Schmidt, K.H. J. Phys. Chem. 1978, 82,
2246.
89. Christman, E.A.; Appleby, A.; Jayko, M. Radiat. Res. 1981, 85, 443.
90. Sims, H.E.; Ashmore, C.B.; Tait, P.K.; Walters, W.S. Proceedings of the 1998 JAIF
International Conference on Water Chemistry in Nuclear Power Plants; Kashwazaki, 1998.
91. Yoshida, H. Radiat. Res. 1994, 137, 145.
92. LaVerne, J.A.; Yoshida, H. J. Phys. Chem. 1993, 97, 10720.
93. Elliot, A.J. Rate constants and G-values for the simulation of light water over the range 0
300jC. Report AECL-11073, A.E.C.L. Chalk River, Canada, 1994.
94. Elliot, A.J.; McCracken, D.R.; Buxton, G.V.; Wood, N.D. J. Chem. Soc., Faraday Trans. 1990,
86, 1539.
95. LaVerne, J.A.; Pimblott, S.M. J. Phys. Chem. 1993, 97, 3291.
96. Swiatla-Wojcik, D.; Buxton, G.V. J. Phys. Chem. 1995, 99, 11464.
97. Swiatla-Wojcik, D.; Buxton, G.V. J. Chem. Soc. Faraday Trans. 1998, 94, 2135.
98. du Penhoat, M.A.H.; Goulet, T.; Frongillo, Y.; Fraser, M.J.; Bernat, P.; Jay-Gerin, J.-P. J.
Phys. Chem. A 2000, 104, 11757.
Copyright 2004 by Taylor & Francis Group, LLC
99. B
egusova , M.; Pimblott, S.M. Radiat. Prot. Dosim. 2002, 99, 73.
100. Kent, M.C.; Sims, H.E. Report AEA-RS-2302; AERE: Harwell, 1992.
101. Elliot, J.A.; Chenier, M.P.; Ouellette, D.C.; Koslowsky, V.T. J. Phys. Chem. 1996, 100, 9014.
102. LaVerne, J.A.; Pimblott, S.M.; Wojnarovits, L. J. Phys. Chem. A 1997, 101, 1628.
103. Pimblott, S.M.; LaVerne, J.A.; Alba-Garcia, A.; Siebbeles, L.D.A. J. Phys. Chem. B 2000, 104,
9607.
104. Salvat, F. Radiat. Phys. Chem. 1998, 53, 247.
Copyright 2004 by Taylor & Francis Group, LLC
5
Interaction of Photons with Molecules:
Photoabsorption, Photoionization,
and Photodissociation Cross Sections
Noriyuki Kouchi and Y. Hatano
Tokyo Institute of Technology, Tokyo, Japan
SUMMARY
The absolute values of the photoabsorption, photoionization, and photodissociation cross
sections are key quantities in investigating not only the interaction of photons with mo-
lecules but also the interaction of any high-energy charged particle with matter. The
methods to measure these, the real-photon and virtual-photon methods, are described and
compared with each other. An overview is presented of photoabsorption cross sections
and photoionization quantum yields for normal alkanes, C
n
H
2n + 2
(n = 14), as a func-
tion of the incident photon energy in the vacuum ultraviolet range and of the number of
carbon atoms in the alkane molecule. Some future problems are also given.
1. INTRODUCTION
The interactions of photons with molecules are classied into absorption, scattering, and
pair production. In this chapter, photons of moderate energies, particularly in the vacuum
ultraviolet range, are discussed, and therefore only the absorption process is considered.
The absorption of a single photon by a molecule in the electronically ground state changes
its electronic state from the ground state 0 to a nal excited or ionized state j. Its transition
probability is expressed in terms of the optical oscillator strength f
j
[1,2].
The optical oscillator strength f
j
is expressed as
f
j
E
j
=RM
2
j
1
where E
j
is the excitation energy in the transition of 0!j, R is the Rydberg energy (R=
(1/2)(atomic units)=13.6 eV), and M
j
2
is the dipole matrix element squared in the transi-
tion of 0!j in atomic units. Note that f
j
is dimensionless. A set of E
j
and f
j
characterizes a
discrete spectrum. To discuss a continuous spectrum, one expresses the oscillator strength
in a small range of the excitation energy between E and E + dE as (df/dE)dE, and calls
df/dE the oscillator-strength distribution, or, more precisely, the spectral density of the
Copyright 2004 by Taylor & Francis Group, LLC
oscillator strength. Thus df/dE has the dimension of the reciprocal of energy. The quantity
df/dE is also dened for the discrete transition 0!j as
df=dE
j
f
j
d E
j
E
_ _
2
The integral of df/dE over a total space of E gives the total number Z of electrons in the
molecule under consideration,
m df=dE dE Z: 3
Equation (3) is known as the ThomasKuhnReiche (TKR) sum rule.
The oscillator-strength distribution df/dE is proportional to the cross section r
for the absorption of a photon of energy E, the so-called photoabsorption cross section.
Note that the excitation energy is equivalent to the photon energy. Explicitly, one may
write
rE 4p
2
aa
2
0
df
dE=R
4
where a is the ne structure constant and a
0
is the Bohr radius. Equation (4) is more
conveniently written as
rE 1:098 10
16
df
dE
_ _
5
where r(E) is expressed in cm
2
and df/dE in eV
1
.
The partial oscillator-strength distribution of a channel n, (df/dE)
n
, is also dened
and is related to the cross section of channel n following the absorption of a photon of
energy E, r
n
, by
r
n
E 4p
2
aa
2
0
df
dE=R
_ _
n
4V
and more conveniently by
r
n
E 1:098 10
16
df
dE
_ _
n
5V
where r
n
(E) is expressed in cm
2
and (df/dE)
s
in eV
1
. The channel n refers to, for example,
the ionization of a specied electron, formation of a specied product, and so on. The
summations of r
n
(E) and (df/dE)
n
over all channels give r(E) and df/dE, respectively, as
rE
n
r
n
E 6
df
dE
n
df
dE
_ _
n
7
Thus the quantum yield of channel n, g
n
(E), the number of times of channel n per
absorption of a photon of energy E, is expressed as
g
n
E
r
n
E
rE
df=dE
n
df=dE
8
Copyright 2004 by Taylor & Francis Group, LLC
The dipole matrix element squared for channel n in atomic units, M
2
n
, is given by
M
2
n
_
R
E
_ _
df
dE
_ _
n
dE 9
Note that E in the right hand side of Eq. (9) is not in atomic units.
As described in the Introduction of Chap. 1, a decisive step in the physical and physi-
cochemical stages of the interactions of any high-energy incident particles with matter,
in case it is composed of molecular compounds, is the collision of secondary electrons
with molecules in a wide energy range. Cross sections for the ionization and excitation of
molecules in collisions with electrons in the energy range greater than about 10
2
eV are well
elucidated quantitatively by the BornBethe theory [3]. This theory is also helpful for cal-
culating, at least roughly, these cross-section values further in the lower energy range.
According to the optical approximation, which was shown by Platzman [4] to be
based on the BornBethe theory, a radiation chemical yield of channel n, G
n
, may be
estimated from optical data as shown in the following equation:
G
n
100=W M
2
n
=M
2
i
_ _
10
where W is the mean energy (in eV) for the production of an ion pair, and M
2
n
and M
i
2
are
the dipole matrix elements squared for channel n and ionization, respectively. Because this
equation indicates that a radiation chemical yield is estimated from optical cross-section
data, it is concluded, therefore, that photoabsorption, photoionization, and photodisso-
ciation cross-section data of molecules are of great importance in understanding not only
the interactions of photons with molecules, but also those of any high-energy incident
particles with molecules [5].
There have been remarkable advances in synchrotron radiation research and related
experimental techniques in the range from the vacuum ultraviolet radiation to soft X-ray,
where the most important part of the magnitudes of these cross-section values is observed,
as shown below. Therefore, it is also concluded that synchrotron radiation can bridge a
wide gap in the energy scale between photochemistry and radiation chemistry. Such a
situation of synchrotron radiation as a photon source is summarized in Fig. 1 [5,6].
Fig. 1 shows the wavelengths of electromagnetic radiation from the infrared to the g-
ray ranges and corresponding photon energies. Characteristic X-rays,
60
Co and
137
Cs g-
rays, and vacuum ultraviolet light from discharge lamps are indicated by the arrows. The
shaded areas indicate the ranges for which photon sources, apart from synchrotron
radiation, are available and correspond to photochemistry and radiation chemistry.
In this chapter, we discuss the photoabsorption and photon-induced processes in
terms of r and r
n
(or g
n
), respectively.
In Fig. 2, the photoabsorption (r), photoionization (r
i
), and photodissociation (r
d
)
cross sections of CH
4
measured by Kameta et al. [7] are shown as a function of the incident
photon energy in the range 1024 eV. The electron conguration of the ground electronic
state of CH
4
in T
d
symmetry is [8,9,10,11]:
1a
1
2
..
inner shell
2a
1
2
..
inner valence
1t
2
6
..
outer valence
The vertical ionization potentials of the (1t
2
)
1
and (2a
1
)
1
ionic states are indicated in
Fig. 2, while that of the (1a
1
)
1
ionic state is far away from the range in Fig. 2. Kameta
et al. examined their photoabsorption cross sections r in terms of the TKR sum rule for
the oscillator-strength distribution df/dE, Eq. (3), following the conversion of r to df/dE,
Copyright 2004 by Taylor & Francis Group, LLC
Figure 1 Synchrotron radiation (SR) chemistry as a bridge between radiation chemistry and photo-
chemistry. The oscillator-strength distribution df/dE is shown as a function of wavelength k and
photon energy E. Note the relation E k = 1.24 10
3
, where E is measured in eV and k in nm. The
intensity of synchrotron radiation is also roughly shown, as are the energies of photons from several
line sources. VUV, EUV, SX, and HX stand for vacuum ultraviolet, extreme ultraviolet, soft X-rays,
and hard X-rays, respectively. (From Refs. [5,6].)
Figure 2 The photoabsorption (r), photoionization (r
i
), and photodissociation (r
d
) cross sections
of CH
4
as a function of the incident photon energy measured via the double ionization chamber and
synchrotron radiation as mentioned in Section 2.1. The values of r in the range below the rst
ionization potential were measured by the photon-beam attenuation method, using the ionization
chamber as a conventional gas cell. The bandpass was 0.1 nm, which corresponds to the energy
width of 32 meV at the incident photon energy of 20 eV. The vertical ionization potentials of the
ionic states involved are also indicated by the vertical bars [11]. (From Ref. [7]. Reprinted with
permission from Elsevier Science.)
Copyright 2004 by Taylor & Francis Group, LLC
Eq. (4). The result is shown in Table 1. The value of mdf=dEdE should be equal to the
number of electrons in CH
4
and the result yields 9.90, as shown in Table 1, only 1% less
than 10. Let us stress from Fig. 2 and Table 1 that the maximum value of df/dE lies
around the ionization potential for the removal of an outer-valence electron and the value
of m
24 eV
11 eV
df=dEdE amounts up to approximately 45% of the total integral. A similar
feature is also seen for a wide range of molecules [1316], and thus let us focus on the
range of the energy of incident photons for the excitation and ionization of valence
electrons as in Fig. 2, i.e., the vacuum ultraviolet range.
We note that the photoionization cross sections (r
i
) of CH
4
is much smaller than the
photoabsorption cross sections (r) just above the rst ionization potential, as seen in Fig. 2.
The values of r
i
then increase as the incident photon energy increases and become almost
equal to those of r around 16 eV. The small dierence between r and r
i
is seen around 22
eV, i.e., just below the (2a
1
)
1
ion state. This feature is shown in a more straightforward
way in the photoionization quantum yields (g
i
) of CH
4
as a function of the incident photon
energy, as seen in Fig. 3 [g
i
=r
i
/r following Eq. (8)]. The signicant deviation of g
i
from
unity around the ionization potentials of the outer-valence electrons is seen for a wide range
of molecules [1316]. This experimental fact of interest is attributed to the excited states of
molecules embedded in their ionization continua. The existence and importance of such
excited states were rst predicted by Platzman in the early 1960s in a pioneering theoretical
work on the interaction of ionizing radiation with matter, and they were named super-
excited states [4,18]. Let us consider a molecule, AB, absorbing a photon of the energy
higher than the rst ionization potential of AB. Then the following processes occur
AB hv !AB
: direct ionization 11
!AB** : superexcitation 12
!AB
: autoionization 13
!AB : neutral dissociation 14
!other processes 15
where AB denotes not only a diatomic molecule but also, more generally, a polyatomic
molecule. The most important feature in the dynamics of a superexcited molecule, AB**, is
the competition between the autoionization [process (13)] and neutral dissociation [process
(14)], because it cannot be seen in a superexcited atom. The processes other than auto-
ionization and neutral dissociation, such as emission of uorescence and ion-pair for-
Table 1 Oscillator-Strength Distribution in CH
4
Energy Range (eV)
Integral of the
Oscillator-Strength Distribution
z23.8 5.15
a
10.9523.8 4.47
b
8.5510.95 0.28
c
Total range 9.90
a
From Ref. [12].
b
From Ref. [7].
c
From Ref. [13].
Source: Ref. [7].
Copyright 2004 by Taylor & Francis Group, LLC
mation, are possible although their contribution is not really signicant. Thus the dierence
between r and r
i
gives the cross section for the neutral dissociation [process (14)] in the
energy range above the rst ionization potential, which is refereed to as the photo-
dissociation cross section in this chapter, r
d
:
r
d
r r
i
16
1 g
i
r 17
The recent stage for the study of the spectroscopy and dynamics of the superexcited
molecules was comprehensively discussed in Ref. [5], and let us stress that the formation of
the superexcited molecules and their decay processes produce a wide range of interesting
structures in the r, r
i
, and r
d
curves as a function of the incident photon energy.
In this chapter, we aim at giving an overview of the photoabsorption (r), photo-
ionization (r
i
), and photodissociation (r
d
) cross sections of molecules in the vacuum ultra-
violet range, where as stated the signicant part of the photoabsorption cross sections lies.
2. EXPERIMENTAL METHODS TO MEASURE THE PHOTOABSORPTION,
PHOTOIONIZATION, AND PHOTODISSOCIATION CROSS SECTIONS
2.1. Double Ionization-Chamber Method (Real-Photon Method)
The double ionization-chamber method [19] provides an excellent means of measuring the
photoabsorption cross sections of atoms and molecules in the range of the incident photon
Figure 3 The photoionization quantum yields (g
i
) of CH
4
as a function of the incident photon
energy measured via the double ionization chamber and synchrotron radiation as mentioned in
Section 2.1. The bandpass was 0.1 nm, which corresponds to the energy width of 32 meV at the
incident photon energy of 20 eV. The vertical ionization potentials of the ionic states involved are
indicated by the vertical bars [11] along with the rst adiabatic ionization potential by the arrow [17].
(From Ref. [7]. Reprinted with permission from Elsevier Science.)
Copyright 2004 by Taylor & Francis Group, LLC
energy higher than their rst ionization potentials, and bears several advantages over the
photon-beam attenuation method as follows (see also Fig. 4).
1. The light intensities need not be measured to obtain the photoabsorption cross
sections (r), as will be shown later using Eq. (20). Therefore we do not need to
consider the absorption of light by gas that euses through the entrance and exit
apertures of the chamber.
2. Photoionization quantum yields (g
i
) and, therefore, photoionization cross sec-
tions (r
i
) are obtained together with the photoabsorption cross sections (r).
Fig. 4 shows the illustration of a double ionization chamber. We describe the process
of measuring the photoabsorption cross sections as follows. I
0
denotes the incident photon
ux coming into the chamber lled with atoms or molecules of the number density n, I
1
and I
1
V denote the photon uxes entering and leaving plate 1, respectively, and I
2
and I
2
V
denote the photon uxes entering and leaving plate 2, respectively. The ion currents i
1
and
i
2
collected by plates 1 and 2, respectively, are expressed as
i
1
eg
i
I
0
exprnL
1
1 exprnd 18
i
2
eg
i
I
0
exprnL
2
1 exprnd 19
where L
1
, L
2
, and d are lengths as dened in Fig. 4, and e is the elementary charge. Thus
we obtain
r
ln i
1
=i
2
n L
2
L
1
20
According to Eq. (20), we measure only i
1
, i
2
, and the pressure of a gas in the chamber to
obtain the absolute values of the photoabsorption cross sections (r), and then we obtain
the values of g
i
I
0
from r following Eq. (18) or Eq. (19). If we use a rare gas as reference,
of which g
i
is unity in the whole range above its rst ionization potential, then I
0
is
obtained. Thus the relation between I
0
and the signal from the incident photon detector,
Figure 4 The outline of the double ionization chamber. (From Ref. [19].)
Copyright 2004 by Taylor & Francis Group, LLC
which is not shown in Fig. 4, is determined as a function of the incident photon energy
by using a suitable rare gas as reference. When this relation is known, the quantities i
1
, i
2
,
I
0
, and n are obtained together to give the absolute values of both photoabsorption cross
sections (r) and photoionization quantum yields (g
i
) following Eqs. (1820). From r
and g
i
, the photoionization cross sections (r
i
) and photodissociation cross sections
(r
d
) are easily derived, as mentioned in Section 1: r
i
= g
i
r [Eq. (8)] and r
d
= (1 g
i
)r
[Eq. (17)].
We note that the relation between I
0
and the signal from the incident photon detector
is inuenced by absorption of photons due to gas eusing from the entrance aperture of the
chamber, and thus the ionization chamber needs to be equipped with a suitable window to
prevent gas eusion while measuring the accurate values of g
i
[20,21]. The window is not
needed when aiming to measure only the photoabsorption cross sections (r) as stated
above. However, the use of window is recommended even in measuring r alone, because it
prevents the pressure gradient inside the chamber caused by eusion of gas. It also sup-
presses the contribution from higher-order components that become mixed in the mono-
chromatized photon beam from the continuous-wavelength light source such as
synchrotron radiation.
The important element in the double ionization-chamber method is a photon source
in the vacuum ultraviolet range. Discharge lamps were used in the 1960s and 1970s, and
are still used today. Since 1980s, however, synchrotron radiation has been widely used in
the eld covered by this chapter because of its broadband tunability in the range from the
vacuum ultraviolet radiation to soft X-ray, which is strongly required in measuring the
cross sections as a function of the incident photon energy.
2.2. Fast Electron Impact Dipole-Simulation Method (Virtual-Photon
Method) and Comparison with Real-Photon Method
In electron impact experiments of atoms and molecules for incident electron energies ex-
ceeding 1000 eV, for example, the generalized oscillator-strength distribution, dF(E,K)/dE,
is derived from the electron scattering cross sections dierential with respect to the energy
loss of electrons, E, and the solid angle for scattered electrons according to the BornBethe
approximation, where K is the momentum transfer [3]. The energy loss is equivalent to
excitation energy or photon energy, and thus the symbol E is used to denote the energy
loss as well. The most important characteristics of the generalized oscillator-strength
distribution is [3]:
lim
K!0
dFE; K
dE
df
dE
21
It follows fromthe above discussion that the fast electron scattering experiments at forward
direction is able to simulate the photoabsorption cross sections over a wide range of the
energy loss.
Brion and co-workers [14,2225] have extensively used and applied such an experi-
mental approach by using fast electrons as the virtual-photon source. Brion et al. [14,25]
pointed out some expected characteristics of the use of synchrotron radiation in compar-
ison with virtual photons in studies of the photoionization and photoexcitation of mole-
cules, and claried some necessary assumptions to virtual photons instead of real photons.
It should be noted here, as described below, that these two methods, real- and virtual-
Copyright 2004 by Taylor & Francis Group, LLC
photon experiments, have complementary roles with each other in understanding the
essential features of the interaction of photons with molecules and obtaining the absolute
values of the cross sections (r, r
i
, and r
d
) [7,5,26].
The fast electron scattering (dipole) approach to real-photon experiments using the
virtual-photon source is called the electron impact dipole-simulation method. The dipole
(e, e) method simulates total photoabsorption while coincidence techniques are used for
measurements of the electronic partial photoionization [dipole (e, 2e)] and ionic photofrag-
mentation [dipole (e, e+ion)] cross sections. These virtual-photon experiments are sum-
marized in Table 2 [14] together with the corresponding real-photon experiments. These
simulation experiments have provided a large body of data in comparison with real-photon
experiments of absolute cross sections [14,25]. The real-photon experiments using synchro-
tron radiation, in which a lot of progress in experimental techniques is being made, have
been comprehensively compared with the simulation measurements [5,16], which is briey
described below.
In most cases, except those in earlier comparative studies between the real-photon
method and the dipole-simulation method, the absolute cross-section values obtained by
both methods agree with each other [27]. Comparison of obtained cross-section values be-
tween the two methods were discussed in detail [27, 2, and references therein] and sum-
marized in conclusion [5]. It should be noted, at least briey, that it is essentially dicult to
accurately obtain the absolute values of photoabsorption cross sections (r) in the dipole-
simulation experiments, and it is necessary to use indirect ways in obtaining those values as
the application of the TKR sum rule, Eq. (3), to the relative values of the cross sections
obtained partly with theoretical assumptions. Moreover, in some cases, in relatively earlier
dipole-simulation experiments, particularly of corrosive molecules upon their electron
optics with poorer energy resolutions, serious discrepancy fromthe real-photon experiments
was clearly pointed out in the obtained absolute values of photoabsorption cross sections
[5,20,2528].
As for the absolute values of photoionization quantum yields (g
i
), a situation of the
dipole-simulation experiments in comparison with the real-photon experiments is much
more serious and controversial because their absolute scales are determined by the assump-
tion that the photoionization quantum yields should be unity around 20 eV photon energy
in addition to the above-mentioned diculty in obtaining absolute cross-section values in
the dipole-simulation experiments.
In case of simple diatomic molecules such as H
2
and N
2
, sharp peaks are observed with
large photoabsorption cross sections, especially in the lower energy range. In real-photon
experiments, if the peak shape is narrower in energy than the bandpass of the incident
Table 2 Photon and Electron Impact Experiments
Photon Experiment Equivalent Electron-Impact Experiment
Total photoabsorption Electron-energy-loss spectroscopy, dipole (e, e)
Total photoionization Dipole (e, 2e) or (e, e+ion) (from sums of partials)
Photoelectron spectroscopy Electron energy loss-ejected electron coincidence,
dipole (e, 2e)
Photoionization mass spectrometry Electron-ion coincidence, dipole (e, e+ion)
Source: Ref. [14].
Copyright 2004 by Taylor & Francis Group, LLC
photon beam, the measured photoabsorption cross section suers from the serious line
saturation eect [29]. In such cases, we should refer to the result by the virtual-photon
experiments or the dipole-simulation experiments, because the integrated photoabsorption
cross sections over a range of the incident photon energy for a transition give the correct
value of the oscillator strength of that transition even if the measured shape is aected by
the energy resolution [29,30]. However, it should be noted that the results of the dipole-
simulation experiments are obtained with rather poor energy resolutions, i.e., poorer than
several tens of meV, in comparison with the real-photon experiments, especially in the lower
energy range. It means that some important spectral features in the cross-section curves may
be sometimes missed in the dipole-simulation experiments.
The real-photon method is essentially more direct and easier compared to the dipole-
simulation method in obtaining absolute values of photoabsorption cross sections (r),
photoionization cross sections (r
i
), and photoionization quantum yields (g
i
). In the real-
photon method, however, there is a practical need to use the big and dedicated facilities of
synchrotron radiationwhere, in many cases, one should change the beam lines equipped
with dierent types of monochromators depending on used photon-wavelengthsand to
develop some specic new experimental techniques in the range from the vacuum ultra-
violet radiation to soft X-ray.
In conclusion, as summarized elsewhere [5,26], these two methods have comple-
mentary roles in understanding the interaction of photons with molecules, as well as in
obtaining the absolute values of the cross sections (r, r
i
, and r
d
).
3. PHOTOABSORPTION AND PHOTOIONIZATION OF MOLECULES
OF CHEMICAL INTEREST, NORMAL ALKANESAN OVERVIEW
In this section, we present an overview of the photoabsorption cross section (r) and the
photoionization quantum yields (g
i
) for normal alkanes, C
n
H
2n+2
(n = 14), as a function
of the incident photon energy in the vacuum ultraviolet range, and of the number of
carbon atoms in the alkane molecule, because normal alkanes are typical polyatomic
molecules of chemical interest. In Fig. 5, the vertical ionization potentials of the valence
electrons, which interact with the vacuum ultraviolet photons, in each of these alkane
molecules are indicated to show how the outer- and inner-valence orbitals associated with
carbon 2p and 2s orbitals, respectively, locate in energy [7].
3.1. Photoabsorption Cross Sections of Normal Alkanes
In Fig. 6 [7], the photoabsorption cross sections (r) of CH
4
, C
2
H
6
, C
3
H
8
, and n-C
4
H
10
are
compared, which were measured by our group by using the double ionization chamber
and synchrotron radiation, as described in Section 2.1. Those in the range below the rst
ionization potentials were measured by the photon-beam attenuation method using the
ionization chamber as a conventional gas cell. The following features can be noted [7].
1. Each cross section has a maximum around 1316 eV, and the observed maxima
shift to the higher energies with increasing the number of carbon atoms, i.e.,
from CH
4
to n-C
4
H
10
, while the rst ionization potentials shift to the lower
energies as shown in Fig. 5. This shift of the maxima seems to be amenable to
the trends in the ionization potentials of the deepest outer-valence orbitals of
each molecule.
Copyright 2004 by Taylor & Francis Group, LLC
2. In the energy range below the peak maximum, the r curves show some undu-
lations, which are attributed to either excitation, superexcitation, or ionization
of the outer-valence electrons.
3. Around 1823 eV, vibrational structures are clearly observed for CH
4
, C
2
H
6
,
and C
3
H
8
, which were assigned to the superexcited Rydberg states converging to
the (2a
1
)
1
, (2a
2u
)
1
, and (4a
1
)
1
ion states, respectively. These superexcited
Rydberg states move to lower energies as the number of carbon atoms increases
along the order of the ionization potentials of the 2a
1
electron in CH
4
, 2a
2u
electron in C
2
H
6
, and 4a
1
electron in C
3
H
8
as shown in Fig. 5.
It is dicult to give a simple explanation of the gross features in the r curves in Fig. 6
with existing theoretical knowledge. In this context, it should be noted that Nakatsukasa
and Yabana [31] calculated the absolute values of the photoabsorption cross sections (r)
as a function of the energy of up to 40 eV for SiH
4
(silane), C
2
H
2
(acetylene), and C
2
H
4
(ethylene) based on the time-dependent local density approximation, and obtained good
agreement with the results yielded by real-photon [20,32,33] and virtual-photon experi-
ments [24,28,34]. It is also interesting, in terms of the gross features mentioned above, that
the r curves of cyclic alkanes seem to behave dierently from those of normal alkanes,
e.g., C
3
H
8
(propane) vs. C
3
H
6
(cyclopropane) [35].
Figure 5 Ionization potentials of the valence electrons of CH
4
, C
2
H
6
, C
3
H
8
, and n-C
4
H
10
[8,9,
10,11], which are the vertical ionization potentials of the ionic states produced with the removal of
an electron from each valence orbital. The point groups for the molecules are also shown as well as
the notation of the orbitals based on them. (From Ref. [7]. Reprinted with permission from Elsevier
Science.)
Copyright 2004 by Taylor & Francis Group, LLC
3.2. Photoionization Quantum Yields of Normal Alkanes
In Fig. 7 [7], we compared the photoionization quantum yields (g
i
) of CH
4
, C
2
H
6
, and
C
3
H
8
, which were measured by our group using the double ionization chamber and
synchrotron radiation, as described in Section 2.1. The photon energies are considered in
two ranges, as follows, in terms of the behavior of the g
i
curves as a function of the inci-
dent photon energy [7]:
1. The energy range of the outer-valence electrons, where the g
i
curves rise up
gradually and then reach unity.
2. The energy range of the inner-valence electrons, i.e., higher than approximately
17 eV, where the g
i
curves show small but discernible deviations from unity. The
superexcited Rydberg states mentioned in Section 3.1 play an important role in
these deviations.
The g
i
curves rise up from the rst adiabatic ionization potentials and then reach
unity at photon energies, slightly higher than the ionization potentials of the deepest
outer-valence electrons. The energy dierences between the onsets of the g
i
curves and
the photon energies where the g
i
curves reach unity are approximately 3 eV for CH
4
, 5
eV for C
2
H
6
, and 6 eV for C
3
H
8
, respectively, reecting the dierence between the ion-
ization potentials of the shallowest and deepest outer-valence orbitals shown in Figs. 5
and 7.
Figure 6 The photoabsorption cross sections (r) of CH
4
, C
2
H
6
, C
3
H
8
, and n-C
4
H
10
as a function
of the incident photon energy measured via the double ionization chamber and synchrotron
radiation, as mentioned in Section 2.1. Those in the range below the rst ionization potentials were
measured by the photon-beam attenuation method, using the ionization chamber as a conventional
gas cell. The energy resolution of the incident photon energy at 20 eV is as follows: 32 meV for CH
4
,
29 meV for C
2
H
6
, 29 meV for C
3
H
8
, and 116 meV for n-C
4
H
10
. (From Ref. [7]. Reprinted with
permission from Elsevier Science.)
Copyright 2004 by Taylor & Francis Group, LLC
4. CONCLUSIONS AND FUTURE PROBLEMS
The features in C
1
C
4
normal alkanes discussed in Section 3 seem to be generalized to a
wide range of molecules, and thus we conclude that the major part of the photoabsorption
cross sections of molecules (r) is associated with the ionization and excitation of the outer-
valence electrons. Hence, there is a strong need to measure the absolute values of r in the
vacuum ultraviolet range, particularly in the range of the incident photon energy 1030 eV,
which is covered by the normal incidence monochromator used to monochromatize
synchrotron radiation. The photoionization (r
i
) and photodissociation (r
d
) cross sections,
Figure 7 The photoionization quantum yields (g
i
) of CH
4
, C
2
H
6
, and C
3
H
8
as a function of the
incident photon energy measured via the double ionization chamber and synchrotron radiation as
mentioned in Section 2.1. The energy resolution of the incident photon energy at 20 eV is as follows:
32 meV for CH
4
, 29 meV for C
2
H
6
, and 29 meV for C
3
H
8
. The vertical ionization potentials of the
ionic states involved are indicated by the vertical bars for each molecule [11]. (From Ref. [7].
Reprinted with permission from Elsevier Science.)
Copyright 2004 by Taylor & Francis Group, LLC
which are partial photoabsorption cross sections, are also key quantities for a full under-
standing of the interaction of photons with molecules in the vacuum ultraviolet range. The
structures seen in the cross-section curves as a function of the incident photon energy are
attributed to the superexcited states or shape resonances. In conclusion, the absolute
values of r, r
i
, and r
d
as a function of the incident photon energy in the vacuum
ultraviolet range are quantities to be measured with a foremost priority in investigating
the interaction of photons with molecules from a chemical point of view. However, they
have been measured only for a limited number of molecules [1316]; in particular, the
quantity and quality of the data sets for polyatomic molecules are still insucient. In this
context, we note that it is more dicult to measure the correct and absolute values of r
i
and r
d
than r, because the incident photon ux is needed in measuring the photo-
ionization quantum yields (g
i
) with the double ionization-chamber method, and the
coincidence measurements should be carried out in the dipole-simulation experiments
with the assumption that g
i
is unity around 20 eV photon energy (see also Section 2).
Finally, some future problems that need to be solved from the view point of this
chapter are summarized below.
1. The eect of temperature on the absolute values of r, r
i
, and r
d
as a function of
the incident photon energy. All the cross-section data shown in this chapter were measured
for molecules in the gas phase at room temperature and thus, the target molecules do not
lie in a single energy level as an initial level. This means that the measured cross sections
seem to be dependent on gas temperature, which is important in various applications of
the cross-section data.
2. The measurement of the absolute values of the partial photodissociation cross
sections as a function of the incident photon energy. As mentioned in Section 1, the
interesting structures due to superexcited states appear in the r, r
i
, and r
d
curves as a
function of the incident photon energy. They are much more noticeable in the rd curves
than in the r and r
i
curves because no contribution from ionization is involved in the rd
curves (e.g., see the r, r
i
, and r
d
curves around 22 eV in Fig. 2). It follows that the
measurement of partial photodissociation cross sections provides us with a more
powerful tool to investigate the superexcited molecules than that of r
d
. Most of the
experiments along this idea, however, have yielded just the relative cross sections. We have
recently been successful in measuring the absolute cross sections for the emission of
uorescences by excited CH radicals and hydrogen atoms in the photoexcitation of
methane as a function of the incident photon energy in the range 12.6541 eV, and found
an unexpectedly large contribution from the doubly excited states in terms of the
independent electron model [36]. The dynamics and spectroscopy of doubly excited
molecules are the subject of current interest from the viewpoint of the few-body problem
in quantum mechanics [37], and thus the absolute values of the partial photodissociation
cross sections should be extensively measured.
3. The extension of target molecules from those in the gas phase to the condensed
phase to investigate the eect of density on the cross sections [5]. The extension to clusters
and transient species such as radicals is also important.
ACKNOWLEDGMENTS
The authors wish to thank Drs. Kosei Kameta and Masatoshi Ukai for their eorts in
measuring the absolute values of the cross sections cited in this chapter.
Copyright 2004 by Taylor & Francis Group, LLC
REFERENCES
1. Inokuti, M. Photochem. Photobiol. 1986, 44, 279.
2. Hatano, Y.; Inokuti, M. In Atomic and Molecular Data for Radiotherapy and Radiation
Research. Inokuti, M., Ed.; IAEA: Vienna, 1995, (Chapter 5).
3. Inokuti, M. Rev. Mod. Phys. 1971, 43, 297.
4. Platzman, R.L. Radiat. Res. 1962, 17, 419.
5. Hatano, Y. Phys. Report 1999, 313, 109.
6. Hatano, Y. In Radiation Research, Fielden, E.M.; Fowler, J.F.; Hendry, J.H.; Scott, D., Eds.;
Taylor & Francis: London, 1987; 35 p.
7. Kameta, K.; Kouchi, N.; Ukai, M.; Hatano, Y. J. Electron Spectrosc. Relat. Phenom. 2002,
123, 225.
8. Kimura, K.; Katsumata, S.; Achiba, Y.; Yamazaki, T.; Iwata, S. Handbook of He I Photo-
electron Spectra of Fundamental Organic Molecules. Japan Scientic Societies: Tokyo, 1981.
9. Potts, A.W.; Streets, D.G. J. Chem. Soc., Faraday Trans. II 1974, 70, 875.
10. Bieri, G.; Burger, F.; Heilbronner, E.; Maier, J.P. Helvetica. Chim. Acta 1977, 60, 2213.
11. Bieri, G.; A
O
2
!
k
1
O
2
* 1
O
2
* !
k
2
O
2
e
2
O
2
* O
2
!
k
3
O
2
O
2
3
O
2
* M!
k
4
O
2
M 4
The vibrational characteristics of the negative ion O
2
) and O
2
(X
2
P
g
) are shown in Fig. 1. It can be seen
that the lowest resonance involves the vibrational levels, O
2
(v=0) and O
2
(vV=4), at low
electron energies and the resonance energy is about 0.08 eV.
Because all electron decays for O
2
M mixtures in the above-mentioned experimental
conditions show pseudo-rst-order behavior, each decay curve gives an electron lifetime
s
0
, which is related to molecular number densities [O
2
] and [M] as:
s
0
O
2
1
k
1
1
k
M
M 5
when [O
2
]<<[M], where k
M
(=k
1
k
4
/k
2
) is the overall three-body rate constant. Based on
Eq. (5), Shimamori and Fessenden [58] determined the value of k
1
=4.8 10
11
cm
3
/sec and
the value of k
M
for each stabilization partner, which is listed in Table 1 with the value
obtained by other workers. The autoionization lifetime of O
2
N
2
mixtures at very high gas pressures (<30 atm), and observed that the attachment rate in
the O
2
C
2
H
4
system showed saturation at very high pressures, but the rates in O
2
C
2
H
6
and O
2
N
2
mixtures continued to increase steeply with increasing pressures of C
2
H
6
and
Copyright 2004 by Taylor & Francis Group, LLC
N
2
, respectively. They reported the value of k
1
to be much larger than that obtained by
Shimamori and Hatano [61] and, correspondingly, to have a considerably shorter lifetime
of O
2
* (see Table 2). Such inconsistencies have been demonstrated more clearly and
analyzed by the recent work of Kokaku et al. [52,53]. Kokaku et al., in a joint research
between Tokyo and Notre Dame, have studied, using the microwave technique combined
with the pulse radiolysis method, O
2
CO
2
and several O
2
hydrocarbon systems at
pressures from <100 to 1000 Torr and found that the BB mechanism can account for
the data in the low-pressure range but fails to explain the results at higher pressures. The
data for O
2
C
2
H
4
mixtures are shown in Fig. 2 as an example, where the high-density part
of the data agrees well with the data obtained by Goans and Christophorou [101] using a
swarm technique. This research has claried a signicant discrepancy between the values of
k
1
and 1/k
2
obtained by the pulse radiolysis microwave technique and those obtained by the
swarm technique. Thus, it has become obvious that a consistent interpretation in terms of
only the BB mechanism is not possible in a wide range of pressures of M, and additional
mechanisms must be considered to explain the high-pressure data. One of the strong
Figure 1 Potential energy curves for O
2
and O
2
O
2
M !
k
5
O
2
M
* 7
O
2
M
* !
k
6
O
2
Me
8
O
2
M
* M!
k
7
O
2
M
M 9
Table 1 Three-Body Attachment Rate Constants k
M
for the Reaction e
+O
2
+M!O
2
+M at
Room Temperature (From Refs. 10 and 11.)
M
10
30
(cm
3
/sec) Authors Reference M
10
30
(cm
3
/sec) Authors Reference
He 0.033 Shimamori and Hatano [61] C
2
H
2
1.3 Kokaku et al. [53]
0.03 Chanin et al. [83] 1.5 Christophorou [105]
0.07 van Lind et al. [88] 1.7 Shimamori and Hatano [61]
Ne 0.023 Shimamori and Hatano [61] 0.9 Toriumi and Hatano [68]
Ar 0.05 Shimamori and Hatano [61] C
3
H
8
3.2 Kokaku et al. [52]
Kr 0.05 Shimamori and Hatano [61] 3.3 Shimamori and Hatano [61]
Xe 0.085 Shimamori and Hatano [61] n-C
4
H
10
4.2 Shimamori and Fessenden [58]
H
2
0.48 Shimamori and Hatano [59] 4.5 Kokaku et al. [53]
D
2
0.14 Shimamori and Hatano [59] 5 Shimamori and Hatano [61]
N
2
0.06 Chanin et al. [83] n-C
5
H
12
7.9 Shimamori and Hatano [61]
0.085 Shimamori and Hatano [60] neo-C
5
H
12
7 Kokaku et al. [52]
0.09 Shimamori and Fessenden [58] 8.0 Shimamori and Hatano [61]
0.1 Crompton et al. [84] n-C
6
H
14
8.1 Shimamori and Hatano [61]
0.11 Hegerberg and Crompton [86] C
6
H
6
8.5 Shimamori and Hatano [61]
van Lind et al. [88] 18 Bouby and Abgrall [81]
0.15 Goans and Christophorou [101] CO 1.31 Shimamori and Fessenden [58]
Hackam and Lennon [85] CO
2
3 Bouby and Abgrall [81]
McCorkle et al. [102] Pack and Phelps [90]
0.26 Young et al. [93] Warman et al. [92]
0.10 Toriumi and Hatano [70] Bouby et al. [82]
O
2
1.7 Young et al. [93] 3.2 Kokaku et al. [53]
2.0 Pack and Phelps [90] 3.23 Crompton et al. [84]
2.1 van Lind et al. [88] 3.5 Hegerberg and Crompton [86]
Nelson and Davis [89]
Parlant and Fiquet-Fayard [97]
2.15 Crompton et al. [84]
2.2 Shimamori and Fessenden [54] H
2
O 14 Bouby et al. [82]
Shimamori and Hatano [59] Pack and Phelps [90]
Warman et al. [92] Stockdale et al. [91]
2.3 Hackam and Lennon [85] 15.2 Bouby and Abgrall [81]
Shimamori and Hatano [60] H
2
S 9 Bouby et al. [82]
2.6 Hurst and Bortner [87] 10 Bouby and Abgrall [81]
2.8 Chanin et al. [83] NH
3
6.8 Bouby et al. [82]
2.2 Hegerberg and Crompton [86] 7.5 Bouby and Abgrall [81]
CH
4
0.34 Shimamori and Hatano [61]
C
2
H
4
1.5 Goans and Christophorou [101]
1.7 Bouby and Abgrall [81] CH
3
OH 8.8 Bouby et al. [82]
2.0 Kokaku et al. [52] 9.6 Bouby and Abgrall [81]
2.3 Hurst and Bortner [87] 11 Shimamori and Hatano [61]
2.5 Bouby et al. [82] C
2
H
5
OH 18 Shimamori and Hatano [61]
3 Shimamori and Hatano [61] CH
3
COCH
3
27 Bouby et al. [82]
3.1 Stockdale et al. [91] 35 Bouby and Abgrall [81]
3.4 Stockdale et al. [91]
1.3 Toriumi and Hatano [69]
Copyright 2004 by Taylor & Francis Group, LLC
Table 2 Lifetime of O
2
* (X
2
P
g
, vV=4) (From Refs. 10 and 11)
Lifetimes (10
12
sec) Authors Reference
Theory 300 Herzenberg [80]
170 Koike and Watanabe [96]
72 Koike [94,95]
88 Parlant and Fiquet-Fayard [97]
Experiment 100 Shimamori and Hatano [61]
63 Shimamori and Fessenden [58]
2 Christophorou [98]
Goans and Christophorou [101]
McCorkle et al. [102]
91 Toriumi and Hatano [68]
66 Toriumi and Hatano [70]
Figure 2 Dependence on C
2
H
4
density of the eective two-body rate constant of thermal electron
attachment in O
2
C
2
H
4
mixtures at room temperature. (From Ref. 52.) The dashed curve represents
the expected contribution from the BB mechanism.
Copyright 2004 by Taylor & Francis Group, LLC
where K
eq
is the equilibrium constant for vdW molecule formation. One should note here
that the density of vdW molecules is determined by K
eq
[O
2
][M], and the value of K
eq
can be
estimated by the theoretical treatment of Stogryn and Hirschfelder [103]. Several experi-
ments have provided evidence for the existence of vdW molecules in the gas phase [78,104].
In the kinetic treatment made by Kokaku et al. [52,53], the estimated values of k
5
are (2
20) 10
9
cm
3
/sec depending on M, where it is highly attractive that all the values for k
5
are
much larger than the value of k
1
(=4.8 10
11
cm
3
/sec). This result suggests that in the case
of vdWmolecules, the initial electron capture mechanismdiers substantially fromthe case
of isolated molecules. A recent study by Shimamori and Fessenden [58] has veried clearly
the presence of the vdW mechanism. They have measured the temperature dependence of
three-body rate constants in pure O
2
, O
2
N
2
, and O
2
CO mixtures. The result for O
2
is
shown in Fig. 3. According to the theory of Herzenberg [80], the three-body rate constant,
which corresponds to the experimentally obtained k
M
, can be expressed as:
k
M
2
3
h
2
2pmk
B
3=2
nk
L
1
T
3=2
exp
E
0
k
B
T
10
Figure 3 The temperature dependence of the three-body rate constant of O
2
. (From Ref. 58.) The
broken curve shows the temperature dependence of the rate constant calculated from Herzenbergs
theory. The solid curve shows a calculated rate constant, which involves both the contributions
from the broken curve and the rate constant due to electron attachment to van der Waals molecule
(O
2
)
2
.
Copyright 2004 by Taylor & Francis Group, LLC
where h, m, and k
B
have their usual meanings; k
L
is the Langevins rate constant; n is the
stabilization eciency; T is the absolute temperature; and E
0
is the resonance energy.
Equation (10) predicts a simple decrease in the rate constant with reduced temperature. The
expected curve for O
2
calculated from Eq. (10), assuming n to be unity, is drawn in Fig. 3.
Because an extra contribution that increases with lowered temperature is evident, electron
attachment to the vdW molecule (O
2
)
2
has been proposed to account for this. Similarly, the
importance of electron attachment to (O
2
N
2
) and (O
2
CO) has also been demonstrated.
Because the temperature dependence of k
5
is given by:
k
5
2
3
2ph
2
2pmk
B
3=2
C
5
1
T
3=2
exp
E
r
k
B
T
11
where E
r
and C
5
are, respectively, the energy and the width for the resonance attachment
process (Eq. (7)), both pressure-dependent and temperature-dependent experiments [68
70] have given important rate parameters for the BB mechanism such as the rate constant
for the initial electron attachment to O
2
(k
1
), the lifetime s of O
2
* (i.e., the resonance width), have been also obtained from this experiment
and summarized in Table 3 [10]. The value of k
5
in Table 3 is again much larger than the
above-mentioned k
1
values. The resonance energy for e
+(O
2
M)
!(O
2
M)
* is much
smaller than that for e
+O
2
!O
2
2
) k
5
(10
11
(cm
3
/sec)
(O
2
N
2
) 20 800 0.71 2500 3000
(O
2
C
2
H
6
) 30 450 0.89 1700 1100
(O
2
C
2
H
4
) 45 270 0.92 1100 380
O
2
88 (E
0
) 10 (C
1
) 0.47 570 3 (k
1
)
Copyright 2004 by Taylor & Francis Group, LLC
stabilized by the polarization interaction between O
2
M
systems [10,11]. Figure 4 shows that near the equilibrium intermolecular distance, the
eective resonance energy of the O
2
(vV=4)M systems as
a function of intermolecular distance. (From Refs. 10 and 11.)
Copyright 2004 by Taylor & Francis Group, LLC
the incident electron. The values of r
h
are also listed in Table 3. At extremely low-energy
electron collision, such as electron attachment to vdW molecules, a small vdW molecule
is supposed to collide with large electron clouds, of which the cross section is
determined by the size of the de Broglie wave length of incident electrons. With decreasing
resonance energy, therefore, the attachment cross section should increase. It should be
noted that the maximum value of empirically obtained cross-section values for dissociative
attachment processes at low energy is reasonably explained by the de Broglie wavelength
of incident electron [105,106].
The resonance width C
5
is expressed by the Wigners threshold rule. In the case of
isolated O
2
, the resonance state O
2
* (X
2
P
g
, vV = 4) can couple with only one electronic
partial wave with an angular momentuml=2. In the case of vdWmolecules, intermolecular
interaction may couple with additional partial waves such as p-wave and s-wave with low
energy. If a third-body molecule distorts the orbital of O
2
(X
2
P
g
), new attachment
channels can open with lower angular momentum of electrons and the resonance width
may increase.
It has been necessary to make a quantitative calculation of these eects using precise
wave functions of the O
2
M system. Huo et al. [107] made such calculations on the O
2
N
2
system and compared their result with experiments. They were successful in explaining the
large enhancement in the attachment rate constant for vdW molecules using SCF wave
functions corresponding to two geometries, T-shape and linear, for (O
2
N
2
) vdW molecules.
The large enhancement in the attachment rate constant was clearly elucidated quantitatively
in this theoretical calculation by the eect of additional vibrational structures of the vdW
molecule on the attachment process, the symmetry breaking, which allows the molecule to
attach a p-wave electron, and the lowering of resonance energy due to a deeper O
2
N
2
potential in comparison with O
2
N
2
potential, as shown schematically in Fig. 4.
An interesting approach [62,63] to this problem, the use of
18
O
2
instead of
16
O
2
,
further substantiated the electron attachment to vdW molecules. For the BB mechanism,
the isotope eect may be expected to appear as a change in the rates of initial attachment
and autoionization channels, which are caused by a decrease of the resonance energy for
18
O
2
in comparison with
16
O
2
.
As mentioned in the beginning of this section, electron swarm data were reported
[98102] in the O
2
C
2
H
6
, O
2
C
2
H
4
, and O
2
N
2
systems up to about 10
21
molecules/cm
3
(3 10
4
Torr) and simply elucidated only by the BB mechanism. An attempt [52,68,69]
was made, therefore, to elucidate the high-pressure swarm data by the combination of the
BB mechanism and the vdW mechanism. The electron swarm data are well explained up to
about 4 10
20
molecules/cm
3
by the combination of both mechanisms. It is obvious that
the contribution from the vdW molecule is dominant in these density ranges. The large
deviations in the data from the combination of the two mechanisms at densities higher
than 4 10
20
molecules/cm
3
may indicate electron attachment to large vdW molecules
such as O
2
(C
2
H
6
)
2
, or may require additionally some collective properties of these hydro-
carbon molecules to explain the density eect of electron attachment in this region. An
attempt [108,109] was made to explain theoretically such density eects in the whole den-
sity range using the statistical model.
Electron attachment to O
2
has been investigated in supercritical hydrocarbon uids
at densities up to about 10
22
molecules/cm
3
using the pulsed electric conductivity
technique [110], and the results have been explained in terms of the eect of the change
in the electron potential energy and the polarization energy of O
2
14
He2
3
S M!HeM
15
The ionizing reaction occudrs over a certain range of intermolecular distance R, where
He(2
3
S) and M can be regarded as intermediate quasi-molecules.
According to the theoretical investigations by Nakamura [126128] and Miller [129],
the collisional energy dependence of these processes can be calculated if the interaction
potential V(R) for the system He(2
3
S)M and the autoionization rate C(R)/h from the
intermediate quasi-molecule [He(2
3
S)M] to the resulting quasi-molecular ion [HeM]
+
e
are known. For example, by the classical formula of Miller [129], we have theoretical cross
sections for Penning ionization as:
rE 2p
Z
l
0
b 1 exp 2
Z
l
R
0
CR
t
dR
v
db 16
where R
0
is the classical turning point, v is the relative velocity, and l is the reduced mass.
The interaction potentials and the autoionization widths have been obtained in atomic
scattering experiments only for some simple target molecules, such as rare gas atoms and
hydrogen molecules. The proposed mechanism of autoionization via electron exchange is:
He2
3
S1 M2 !He2 M
1
He2
3
S1 M2 !He2M
1
where (1) and (2) denote the excited electron of the metastable He atom and the electron of
M, respectively. The empirical forms of V(R) and C(R) commonly employed as the most
reliable for theoretical analyses [130133] are:
CR A expaR 17
V*R B expbR 18
where A, B, a, and b are constants and R is an intermolecular distance.
Copyright 2004 by Taylor & Francis Group, LLC
The de-excitation rate constants of He(2
3
S) and Ne(
3
P
0
,
3
P
1
, and
3
P
2
) by various
atoms and molecules were obtained at room temperature using a pulse radiolysis method
[125,134136].
Experimental details for the cross-section measurements were presented in the
literature. Briey, after the irradiation by electron beam pulse for a few nanoseconds,
the time-dependent absorption for the atomic line transition Rg**!Rg*+hv was
measured to observe the time-dependent population of the excited rare gas atoms Rg*.
The population of excited Rg* was determined using an absorption law for the atomic
lines, where the broadening of the absorption prole due to the thermal Doppler eect and
due to the attractive interatomic potentials was reasonably taken into consideration. The
time-dependent optical emission from energy transfer products, such as:
He2
3
S N
2
!He1
1
S N
2
B
2
R
g
;
N
2
B
2
R
g
!N
2
X
2
R
g
hv
was also measured to monitor the population of Rg* [125,126].
An attempt has been made to correlate the rate constant values obtained with vari-
ous molecular parameters such as ionization potentials and polarizabilities. A relatively
good correlation has been obtained for He(2
3
S) [134], as shown in Table 4 between the de-
excitation probability P (=k
M
/k
C
) and the excess energy DE [=E(He
*
) IP
M
], where k
M
,
k
C
, E(He*), and IP
M
are the experimentally obtained rate constant, the calculated gas
kinetic rate constant, the excitation energy of He(2
3
S), and the ionization potential of the
target atom or molecule M, respectively. However, the reason for the correlation has not
been well understood. A similar experiment was made [135,136] also for the de-excitation
of Ne(
3
P
0
,
3
P
1
, and
3
P
2
) by atoms and molecules. Two new features of the obtained results
have been demonstrated in this experiment: one is the J-dependent de-excitation cross sec-
tions, the other is a comparison between a theory of the de-excitation of optically allowed
resonant state and the experimental result of Ne(
3
P
1
), which is partly allowed because of
a weak spinorbit coupling. A pulse radiolysis is very advantageous to obtain not only
absolute de-excitation rate constants or cross sections but also their collisional energy
dependence. Avelocity-averaged absolute cross section r
M
is obtained as a function of mean
collisional energy E from the temperature dependence of an absolute rate constant. The
temperature dependence of the rate constants for the de-excitation of He(2
3
S) by atoms and
molecules has been measured [122] in the temperature range from 133 to 300 K (Fig. 6).
From the above classical formula of the Penning ionization cross section, the collisional
Table 4 Derived Parameters for De-excitation of He(2
3
S) (From Ref. 137.)
M a/b P DE (eV)
IPM=IPHe2
3
S
p
N
2
1.4 0.05 4.24 1.81
CO 1.1 0.07 5.81 1.71
Ar 1.0 0.07 4.06 1.82
Kr 1.0 0.07 5.82 1.71
NO 0.6 0.16 10.56 1.39
O
2
0.6 0.18 7.75 1.59
C
2
H
4
0.3 0.31 9.31 1.48
CO
2
0.1 0.39 6.04 1.70
Copyright 2004 by Taylor & Francis Group, LLC
energy dependence of its cross section is readily given, if the interaction potential V*(R) for
He(2
3
S)M and the autoionization rate C(R)/ t from He(2
3
S)M to He(M
+
e
) are
obtained, by the following simple equation:
r
T
/ E
a=b1=2
or kT / T
a=b
19
where r
T
and k(T) are the total Penning ionization cross section and the corresponding
rate constant, respectively. The slope of loglog plots of k(T) vs. T gives the values of a/b
(Fig. 7) [137]. The obtained a/b value for each molecule M listed in Table 4 increases with
decreasing value of P, where P is a de-excitation probability per collision, or 1/P is an
eective collision number for energy transfer. Because Pis not so relatively dierent for each
M, Table 4 shows clearly that the bigger a (i.e., the shorter range interaction between
He(2
3
S) and M) gives the smaller P(i.e., the less ecient energy transfer fromHe(2
3
S) to M).
This conclusion satises the exterior electron density model by Ohno et al. [138].
3.2. De-excitation of Excited Rare Gas Atoms in Resonant States
De-excitation of the excited rare gas atoms in the resonant states has been studied less
extensively than that of the metastable atoms. This is due to experimental diculties
caused by the short lifetimes of the resonant atoms. There have been reported, however,
several theoretical formulations [139,140] based on a long-range dipoledipole interaction
Figure 6 Collisional energy dependence of the cross sections for de-excitation of He(2
3
S) by (n)
N
2
; (O) Ar; (
y
) CO; ( w ) Kr (left panel); (D) O
2
; (z) NO; (.) C
2
H
4
; and (o) CO
2
(right panel).
(From Ref. 137.)
Copyright 2004 by Taylor & Francis Group, LLC
for the de-excitation cross section of radiative atoms. It is, therefore, necessary to compare
the theory with the results of the resonant or the lowest radiative state atoms. It was re-
ported [141,142] that the collisional energy dependence of the de-excitation cross sections
of He(2
1
P), Ne(
3
P
1
), Ar(
1
P
1
), and Ar(
3
P
1
) by atoms and molecules, using a pulse radiolysis
method, is very advantageous to obtain the absolute values of the de-excitation cross sec-
tions of the resonant atoms as well as the metastable atoms. The experimental technique
of the time-dependent optical atomic line absorption is almost the same as that for the
metastable states.
A comparison, as shown in Fig. 8, between the experimental results of the collisional
energy dependence of the de-excitation cross section of He(2
1
P) by Ar and the theoretical
ones calculated from the WK theory [140] and the KW theory [139] makes clearly
possible for the rst time to compare in detail the experimental results with the theoretical
ones [141]. Previous comparisons between experiments and theories have been made only
for a value of the rate constant or the cross section, respectively, at a particular temper-
ature of collisional energy, usually at room temperature. The results of temperature-
Figure 7 Loglog plot of k
M
for He(2
3
S) vs. T. (From Ref. 137.) The same symbol indicates the
same quenching gas as in Fig. 6.
Copyright 2004 by Taylor & Francis Group, LLC
dependent measurement using the pulse radiolysis method supports evidently the modied
new theory, the KW theory [139], which means that the theory should take the bent tra-
jectory into consideration.
A classical expression for the cross section for collisional de-excitation of He(2
1
P) is
also derived from the formula by Eq. (16). However, the autoionization widths C(R) for
Penning ionization by resonant atoms are not identical to the empirical form of Eq. (18)
for electron exchange. Instead, a direct transition due to a dipoledipole interaction is
proposed to govern this Penning ionization [126,139,140,143], that is,
He2
1
P1 M2 !He1 M
2
He2
1
P1 M2 !He1M
2:
Figure 8 Experimental de-excitation cross sections of He(2
1
P) by Ar, Kr, and Xe (denoted by
circles). (From Ref. 144.) (), ( ), and ( ) represent r
WK
for Ar, Kr, and Xe. The ()
curve shows r
KW
for Ar.
Copyright 2004 by Taylor & Francis Group, LLC
According to the dipoledipole interaction, C(R) is given by the equation:
CR
2pl
2
He
l
2
M
R
6
1 3cos
2
h 20
where l
2
He
and l
2
M
are squared matrix elements for transitions of He! He(2
1
P) and
M!M
+
+e
at 58.4 nm, respectively; h is the angle between the intermolecular axis and
the direction of the atomic 2p polarization. The values of l
2
He
and l
2
M
are obtained from
optical oscillator strengths for He and the relevant atoms and molecules. Using Eq. (20)
together with an appropriate treatment for the p-state polarization during a collision,
Watanabe and Katsuura [140] (WK) proposed a formula of the Penning ionization cross
section as:
r
WK
13:88
l
2
He
l
2
M
tv
2=5
21
In the WK theory, the direction of the polarization axis of He(2
1
P) is carefully con-
sidered, but only rectilinear trajectories for relative motion are considered. The assumption
of the rectilinear trajectories has been generally employed in the theories for higher-energy
heavy- particle collisions, where the forward scattering is dominant. However, because the
relative collisional energy is comparable to the depth energy of the interaction potential well
in the thermal collisions, the assumption of the rectilinear trajectories can give rise to a
considerable disagreement with experimental cross sections, which will be discussed in this
section. The maximum limit for the cross section has been obtained by Katsuura [143] also
by assuming linear trajectories together with the perturbed rotating atom approximation
(i.e., the atomic polarization of He(2
1
P) is directed toward M throughout a collision).
The cross-section measurements using the pulse radiolysis method provided the
opportunity to examine the applicability and inapplicability of the WK theory systemati-
cally. Fig. 8 shows the experimental de-excitation cross sections of He(2
1
P) by Ar, Kr, and
Xe [144]. The values of the cross sections are, on the whole, extremely large in comparison
with simple energy transfer cross sections between neutral atoms and molecules, and
closely approach gas-kinetic close collision cross sections [125,134]. As compared with the
results in Section 3.1, the cross sections are more than one order of magnitude larger than
those for the metastable atoms. The cross section for Xe is larger than that for Kr in the
present energy region, whereas the cross section for Ar is smaller than that for Kr.
The cross sections show a decrease with increasing collisional energy, which is in
marked contrast to the de-excitation cross sections of He(2
3
S) atoms by the same
quenching atoms, as presented in Fig. 6. However, this behavior is rather similar to other
quenching molecules, such as CO
2
and C
2
H
4
, where the cross sections are fairly large, on
the order of tens of squared angstroms.
Relatively good agreement between the experimental and theoretical cross sections is
obtained for the absolute values and also for the collisional energy dependence of the cross
sections. In detail, the experimental values for Kr and Xe are larger than r
WK
and the
dierence increases at lower energies. On the other hand, the experimental cross section for
Ar is smaller than r
WK
and the dierence increases at higher energies. It is known that the
values of l
M
2
are almost the same for Ar, Kr, and Xe, but the value for Xe is a little smaller
than those for Ar and Kr, so that the cross-section values should be in the order of
r
Xe
<r
Ar
<r
Kr
(see r
WK
in Fig. 8). However, the experimental values are r
Ar
<r
Kr
<r
Xe
.
The dierences between r
Ar
and r
Kr
and between r
Kr
and r
Xe
are much greater than those
calculated fromEq. (21). Thus, the absolute values, the order, and the energy dependence of
Copyright 2004 by Taylor & Francis Group, LLC
the experimental cross sections are found to be dierent from those predicted by r
WK
. A
similar consideration was also made for molecular target in He(2
1
P)H
2
collision [145].
To make the comparison a little more obvious, Fig. 9 shows the cross-section ratios
of the experimental values of r
M
to r
WK
plotted against the collisional energy. The ratio
for Xe is larger than unity and decreases with increasing collisional energy. The value for
Kr is also larger but decreases almost to unity with increasing energy. The ratio for Ar is
smaller than unity and continues to decrease with increase in energy, a result that is in
good agreement with a classical trajectory calculation by Kohmoto and Watanabe [139]
(r
KW
). The ratios presented in Fig. 9 suggest that the inuence of the classical motion of
the colliding particles is important in determining the absolute cross sections.
Fig. 10 shows the relation between the experimental values r
M
and the theoretical
predictions r
KW
, together with the experimental cross sections previously reported for the
de-excitation of He(3
1
P) [146,147]. The dierence between the values for He(2
1
P) and
He(3
1
P) is explained by the dierence in the optical oscillator strengths for the two states.
The experimental values can generally be explained by r
WK
based on the long-range
Figure 9 Ratios of experimental cross section to WK cross section for de-excitation of He(2
1
P) by
Ar, Kr, and Xe (circles). (From Ref. 144.) () represents the ratio of r
KW
/r
WK
for Ar.
Copyright 2004 by Taylor & Francis Group, LLC
dipoledipole interaction. However, each of the cross sections is characteristically dier-
ent from the theory, which seems to be much beyond any possible systematic error.
The value of r
M
for Ar is smaller than the theory, but for CO
2
, SF
6
, or NO, r
M
exceeds r
WK
. The positive dierence for diatomic molecules other than H
2
and D
2
is more
than that for rare gas atoms, which is also observed for He(3
1
P).
The result for He(2
1
P)Ar in Fig. 9 shows that the long-range attractive part of
the interaction potentials has a close relation with the polarizabilities of the molecules, M.
Fig. 11 is a plot of r
M
/r
WK
vs. a
M
, the polarizability of M. As for Ar, Kr, Xe, CH
4
, and
SF
6
, some correlation is observed. From the slope of the r
M
/r
WK
vs. a
M
plots, the eect of
bent trajectories on the Penning ionization cross sections can be qualitatively evaluated
Figure 10 Comparison of r
M
for He(2
1
P) to r
WK
(From Ref. 144.): (o) present results; (.) cross
section for He(3
1
P) at 300 K (From Ref. 146.); and (E) cross section for He(3
1
P) at 600 K. (From
Ref. 147.)
Copyright 2004 by Taylor & Francis Group, LLC
because, for molecules with large polarizabilities, the attractive potential extends to large
intermolecular distances, whereas for molecules with smaller polarizabilities, the potential
becomes repulsive sooner at relatively smaller intermolecular distances. On the other hand,
for CO
2
and diatomic molecules, there is no same correlation between the cross-section
ratios and the polarizabilities observed for the rare gas atoms and spherical molecules,
although some other correlations characteristic of these molecules might be distinguished.
The eect of bent trajectories due to the attractive interaction on the absolute cross
section is further claried by the classical trajectory calculation shown in Fig. 12. The cross
section was calculated for He(2
1
P)Ar, He(2
1
P)Kr, and He(2
1
P)Xe collisions assuming
h=0 (perturbed rotating atom approximation [143]) in Eq. (20) for the autoionization
widths, employing the interaction potentials determined in cross-beams experiments for
He(2
1
S)M collisions, and using Eqs. (14)(16) for the general procedure [144]. The cal-
culated cross sections in general are strongly dependent on the collisional energy. However,
Figure 11 Cross section ratios of He(2
1
P) vs. a
M
. (From Ref. 142.) The horizontal lines show the
limits of a
M
values if seen in parallel or perpendicular direction to the molecular axis.
Copyright 2004 by Taylor & Francis Group, LLC
the energy dependence is dierent in each case, thus changing the relative order of the
absolute magnitudes.
The order in the high-energy region is, as expected from the rectilinear calculations
[125,128], r
Xe
<r
Ar
<r
Kr
. However, below 50 meV, the calculation reproduces the experi-
mental order of r
Ar
<r
Kr
<r
Xe
. The result demonstrates that the eect of entangled
trajectories decreases with the increase in the collisional energy. The behavior of the
experimental cross sections has been further examined by a classical trajectory calculation
employing the procedure of KW [139]. The problem of whether the direction of the polar-
ization axis of He(2
1
P) is conserved or not during a collision has also been examined. For Ar,
an atomic target of small polarizability, the experimental cross section is in good agreement
with the KW-type calculation for a spatially xed orientation. For Xe, which has a large
atomic polarizability, r
M
is more closely predicted by the perturbed rotating atom
Figure 12 Calculated Penning ionization cross sections for collisions: ( ) He(2
1
P)Ar, (),
He(2
1
P)Kr, and () He(2
1
P)Xe, assuming classical trajectories for relative motion. (From
Ref. 144.) A perturbed rotating atom approximation is employed in the evaluation of the magnitudes
of the dipoledipole interaction (h = 0; see Eq. (20)). Theoretical values by Katsuura at 500 meV
employing rectilinear trajectories are also indicated. (From Ref. 143.)
Copyright 2004 by Taylor & Francis Group, LLC
approximation. Therefore, in addition to the bent trajectories, the rotation of the polar-
ization axis is also probably an important factor in determining the absolute magnitude of
the cross sections. This includes an important collision physics, namely the angular
momentum of polarized He(2
1
P) during a collision, which should be quantized relative to
the interatomic axis or should remain in the initial xed space.
As for a total decay width of the transient autoionizing state of [HeM]* formed
during a collision, the direct transition due to the dipoledipole interaction and the tran-
sition with the electron exchange should also be involved simultaneously. Because the
outermost electron orbitals of the target rare gas atoms, CH
4
and SF
6
, are fully occupied,
the electron density outside their van der Waals sphere is supposed to be quite low. How-
ever, most of the linear molecules considered here have some amount of electron density
outside their hard sphere, so that the contribution from a short-range interaction (i.e.,
electron exchange), which is neglected in WK, might also be included in the total de-
excitation cross section.
In the present discussion, the total autoionization width for Penning ionization of
He*M or for decay of the transient molecular autoionizing state of [HeM]* is assumed to
be divided into two components: a dipoledipole part and an electron exchange part. The
maximum width possible is assumed to be the sum of the two components [142], that is,
C
total
C
dd
C
ex
22
where C
total
is the total decay width of [HeM]*, and C
dd
and C
ex
are the partial widths due
to the dipoledipole interaction and the electron exchange interaction, respectively.
Substitution of Eq. (22) into the classical formulation of the Penning ionization cross
section, as shown by Eq. (16), gives the theoretical cross section for simultaneous
contributions from both the dipoledipole interaction and the electron exchange.
First, in cases where the transition (ionization) rate is small, the total cross section
will be given by:
r
total
r
dd
r
ex
23
where r
dd
and r
ex
are the components of the cross section due to the dipoledipole inter-
action and the electron exchange interaction, respectively. Fig. 13 gives plots of (r
M
r
WK
)
vs. r
m
, the de-excitation cross section of the metastable helium atoms.
The values of r
m
are assumed to represent the exchange part of the total de-excitation
cross section for He(2
1
P). The values of (r
M
r
WK
) are a measure of the cross section due
to interactions other than the dipoledipole interaction (i.e., the electron exchange inter-
action). The plots in Fig. 13 are shown for linear molecules and also for Xe, CH
4
, and SF
6
.
The values of (r
M
r
WK
) for Ar and Kr are negative and are not included in the gure.
A good correlation is found for H
2
, D
2
, N
2
, CO, O
2
, and CO
2
. A better, or rather more
proportional, correlation is obtained for r(2
1
S) than for r(2
3
S), demonstrating the con-
tribution from the electron exchange. The large deviation observed for NO is explained
by the inapplicability of the WKtheory for transitions other than sp (rp) type. Xenon and
SF
6
also seem to be within the correlation and may have a minor contribution from the
electron exchange interaction.
Secondly, the cross section reects the nature of the dependence of the partial widths
on the intermolecular distance R. As expected from Eq. (20), C
dd
is small for large R and
increases gradually with decrease in R. On the other hand, C
ex
in its empirical form (Eq.
(17)) is extremely small at large distances and increases sharply near the repulsive wall.
Therefore, the de-excitation probability due to Penning ionization is determined by the
dipoledipole part of the decay width in collisions with large impact parameter, whereas
the probability for Penning ionization is already almost unity by the single contribution
Copyright 2004 by Taylor & Francis Group, LLC
from the dipoledipole part in collisions with a small impact parameter where the
trajectory intersects the intermolecular region in which C
ex
is comparably large relative
to C
dd
. The reason why the cross sections for a rare gas target can be explained almost
completely by the dipoledipole interaction originates from this relationship. As expected
from the results for metastable atoms in Table 4, the molecules CO
2
, O
2
, and NO could
have large C
ex
at relatively large R, and the cross sections may have contributions from the
exchange part even at large R, additional to the dipoledipole part.
The above conclusion based on the cross-section measurements, using the pulse
radiolysis method and with consideration of the Penning ionization mechanism including
the classical trajectory analysis, is further rationalized by a quantum mechanical calcu-
lation for He(2
1
P)Ar, Kr, Xe, H
2
, and N
2
collisions [148,149], where the radial wave
function for relative motion is numerically solved by taking into account the partial widths
for dipoledipole interaction and electron exchange interaction. Fitting the calculated
Penning ionization cross sections to the experimental cross sections enables to determine
the interaction potentials for these collision systems for the rst time (Fig. 14), which was
not obtained by the scattering experiments. This is also recognized as an advantage of the
pulse radiolysis method, which provides the insight of intermolecular interaction.
Figure 13 Plots of (r
M
r
WK
) vs. r
m
for an analysis of the de-excitation cross section r
M
for
He(2
1
P), where the de-excitation cross section r
m
for the metastable helium atoms is assumed to
represent the exchange part of the total de-excitation cross section. (From Ref. 142.) The r
m
values
are for He(2
3
S) (o) and He (2
1
S) (.).
Copyright 2004 by Taylor & Francis Group, LLC
It is also a fact that very few studies have been reported on simple excitation transfer,
in which Penning ionization is energetically impossible, from resonant rare gas atoms
to atoms and molecules. The temperature dependence of the de-excitation rate constants
of the resonant states of Ar,
1
P
1
, and
3
P
1
by SF
6
and N
2
has been measured in the tem-
perature range from 133 to 300 K using a pulse radiolysis method [150], thus obtaining
the collisional energy dependence of the de-excitation cross sections. The results of the
cross sections for SF
6
are compared with the WK theory and a good agreement is ob-
tained. The results for N
2
agree with predictions of the cross sections for a nonresonant
case. Even in the case of the de-excitation of the resonant state, such as Ar(
1
P
1
) or Ar(
3
P
1
),
the cross-section value and its collisional energy dependence are very similar to those for
the metastable state (i.e., Ar(
3
P
0
) or Ar(
3
P
2
)). This result is consistent with the WK theory
because of the fact that N
2
has almost no optical absorption in the energy region
Figure 14 Interaction potential and autoionization decay width employed and determined in the
calculations for He(2
1
P)M; ( ) M=Ar; ( ) M=Kr; and () M=Xe; (a) interaction
potential: position of the repulsive potentials for He(2
1
P)Kr and He(2
1
P)Xe are determined by
tting to the experimental results; (b) autoionization decay width for Penning ionization.
Copyright 2004 by Taylor & Francis Group, LLC
corresponding to the excitation energy of Ar(
1
P
1
) and Ar(
3
P
1
). It is concluded, therefore,
that a long-range dipoledipole interaction is important in the de-excitation processes of
Ar(
1
P
1
) and Ar(
3
P
1
) by SF
6
, but that a short-range interaction with curve crossing
dominates in the de-excitation of Ar(
1
P
1
) and Ar(
3
P
1
) by N
2
.
A little more complicated system is the de-excitation of He(2
1
P) by Ne, where the de-
excitation is dominated by the excitation transfer and only a minor contribution from the
Penning ionization is involved. The experimental cross section obtained by the pulse
radiolysis method, together with the numerical calculation for the coupled-channel radial
Schro dinger equation, has clearly provided the major contribution of the following
excitation transfer processes to the absolute de-excitation cross sections [151] (Fig. 15):
He2
1
P Ne !He Ne*6p
3
; 5s
2
; and6s
4
:
3.3. De-excitation of Excited Rare Gas Atoms by Molecules Containing
a Group IV Element
De-excitation cross sections of He(2
3
S, 2
1
S, and 2
1
P), Ne(
3
P
2
,
3
P
1
, and
3
P
0
), and Ar(
3
P
2
,
3
P
1
,
3
P
0
, and
1
P
1
) by molecules containing group IV elements were obtained by the pulse
Figure 15 Calculated total and state-to-state excitation transfer cross sections in the de-excitation
of He(2
1
P)Ne. (From Ref. 151.) Both electron exchange and dipoledipole interactions are included
in the coupling matrix elements. The threshold energy into each exit channel is shown on the upper
axis.
Copyright 2004 by Taylor & Francis Group, LLC
radiolysis method [152155]. The study was designed from the viewpoint of the control of
reactive plasmas. As can be seen in the above, in reactive plasmas, the excited rare gas
atoms in dierent excited states are produced simultaneously and react with atoms and
molecules with dierent probabilities. Thus, the cross-section data for dierent excited
atoms are strongly required.
Table 5 presents the de-excitation cross sections for dierent excited helium atoms.
As demonstrated in the preceding sections, the de-excitation cross sections of He(2
3
S) and
He(2
1
P) are the Penning ionization cross sections. It is natural that these cross sections are
also understood as the Penning ionization cross sections. In general, the cross sections are
extremely large compared with a simple energy transfer cross section between a neutral
atom and a neutral molecule, and they approach the gas-kinetic collision cross sections.
The values in Table 5 show an apparent dependence on the particular excited states of the
helium atoms (i.e., the cross sections for the metastable atoms, He(2
3
S) and He(2
1
S), are
much smaller than those for the resonant atom, He(2
1
P), whereas the cross sections for
He(2
1
S) are larger than those for He(2
3
S)). The behavior of the de-excitation cross
sections, dependent both on the helium excited states and the target molecules, is ascribed
to that of the Penning ionization cross sections. A systematic comparison in Table 6
between the total de-excitation cross sections and the partial cross sections for possible
fragmentation conrms that the Penning ionization is the major de-excitation mechanism.
For example, the optical emission cross sections in a UVvisible region from Si* atoms in
several excited electronic states produced in He(2
3
S)SiH
4
collisions are reported to be
0.081 A
2
. Minor emission from other excited fragments H*, SiH*, SiH
2
*
, and SiH
3
*
does
not increase the cross section signicantly. The production of nonemitting fragments H,
Si, SiH, SiH
2
, and SiH
3
has been shown to give a minor contribution of such fragments to
the formation of neutral products. The cross section of 0.081 A
2
thus represents the total
cross section for the production of emitting neutral fragments, or possibly the total cross
section for the neutral dissociation of SiH
4
in He(2
3
S)SiH
4
collisions. The de-excitation
cross section of He(2
3
S) by SiH
4
in Table 5 is, however, determined to be 18 A
2
in the present
experimenta value that is more than about 200 times larger than that of the optical
emission cross section. It is, therefore, concluded that a major part of the de-excitation
processes in He(2
3
S)SiH
4
collisions is due to processes other than the formation of neutral
dissociation fragments (i.e., due to Penning ionization). The Penning ionization here then
Table 5 De-excitation Cross Sections (r
M
) at a Mean Collisional Energy Corresponding to
Room Temperature (295 K) and De-excitation Probabilities ( P) of Excited Helium Atoms He
*
(He
*
=He(2
3
S), He(2
1
S), and He(2
1
P)) and polarizabilities (a
M
) of target molecules (From Ref. 154.)
M
2
3
S 2
1
S 2
1
P
a
M
(A
3
) r
M
(A
2
) P r
M
(A
2
) P r
M
(A
2
) P
CH
4
14 F 1 0.127 58 F 3 0.461 98 F 9 0.792 2.59
SiH
4
18 F 1 0.127 60 F 2 0.367 105 F 2 0.663 5.44
GeH
4
26 F 1 0.175 70 F 4 0.415 109 F 3 0.662 6.20
C
2
H
6
28 F 1 0.222 106 F 4 0.472 152 F 4 1.031 4.43
Si
2
H
6
41 F 1 0.228 58 F 5 0.521 179 F 7 0.904 11.1
CF
4
14 F 3 0.109 58 F 1 0.157 94 F 6 0.669 3.84
SiF
4
50 F 1 0.352 58 F 4 0.439 147 F 6 0.919 5.45
SiCl
4
59 F 3 0.309 58 F 5 0.503 172 F8 0.815 13.1
Copyright 2004 by Taylor & Francis Group, LLC
means the formation of SiH
n
+
+e
2
) (From Refs. 123 and 152.)
He(2
3
S)+CH
4
!
r
M
14
CH
n
+
(n V 4) 12
C*
CH* 0.051
CH
CH
2
, CH
3
He(2
3
S)+SiH
4
!
r
M
18
SiH
n
+
(n V 4)
Si* 0.081
SiH* 0
SiH 0
SiH
2
, SiH
3
0
He(2
3
S)+GeH
4
!
r
M
26
GeH
n
+
(n V 4)
Ge* 0.44
GeH* 0
GeH
GeH
2
, GeH
3
Copyright 2004 by Taylor & Francis Group, LLC
Figure 17 The ratios of the experimental cross sections to the theoretical ones (r
M
/r
WK
) for
He(2
1
P) as a function of photoabsorption cross section (r
abs
) of M at the corresponding energy to
the excitation energy of He(2
1
P). (From Ref. 154.)
Figure 16 Relation of de-excitation cross sections (r
M
) for metastable helium atoms with the
HotopNiehaus formula for (a): He(2
3
S) and (b) He(2
1
S). (From Ref. 154.)
Copyright 2004 by Taylor & Francis Group, LLC
A similar conclusion has been obtained for the de-excitation cross sections of Ne(
3
P
2
,
3
P
1
, and
3
P
0
) by the molecules containing the group IV elements [155]. As shown, the cross
sections are not so strongly dependent on the excited state of neon atoms. Instead, the
dependence of the cross sections on the target molecule is more prominent. A systematic
analysis of the cross sections with respect to the polarizability of target molecules has
provided the physical insight of the mobile state of the r, p, and nonbonding outermost
electrons; the marked dierence in the behavior of the cross section against the target
polarizabilities in Fig. 18 presents the relation of the localized and delocalized electron
characters of the r, p, and nonbonding outermost electrons and the probabilities of
electron exchange.
The cross sections of the de-excitation of Ar
*
(Ar
*
: Ar (
3
P
2
,
3
P
1
,
3
P
0
, and/or
1
P
1
)) by
CH
4
, SiH
4
, and GeH
4
are extremely large [153] (Fig. 19). The dierence in the magnitude
of the cross sections is ascribed to the particular de-excitation mechanisms for the
metastable and resonant excited argon atoms. However, in general, the Penning ionization
is not a major de-excitation process. This is readily seen from the following. The excitation
energy of Ar
*
is smaller than both of the vertical and adiabatic ionization potentials of
CH
4
, so that the Penning ionization of CH
4
is not accessible. In the case of Ar
*
SiH
4
and
Ar
*
GeH
4
collisions, the excitation energy of Ar* is slightly greater than the adiabatic
ionization potential of SiH
4
or GeH
4
, and less than its vertical ionization potential. It is
safely predicted that rendering the Penning ionization requires a strong structural
deformation of a target molecule, which is not accessible with the thermal collisional
energies. As a matter of fact, the ionic products such as CH
n
(0 V n V 4) or H
+
are not
Figure 18 De-excitation cross sections of Ne(
3
P
2
) by M (M=CH
4
, SiH
4
, GeH
4
, CF
4
, and SiF
4
, Ar,
Kr, Xe, H
2
, D
2
, N
2
, O
2
, CO, NO, CO
2
, N
2
O, and SF
6
) as a function of the polarizability of M.
(From Ref. 155.) The target molecules are classied by the type of their outermost molecular
orbitals: r-type (.); p-type (o); and nonbonding ( w ).
Copyright 2004 by Taylor & Francis Group, LLC
observed in mass spectrometric studies for Ar*CH
4
, Ar*SiH
4
, and Ar
*
GeH
4
, as seen in
Table 7. At the same time, the above consideration on the energetics between the excita-
tion energy of Ar
*
and the ionization potential of the target molecule leads to another
insight of the de-excitation mechanism (i.e., the excitation transfer to form the molecular
Rydberg state). The vertical excitation of a molecule with a slightly smaller excitation
energy than the rst ionization potential produces one of the molecular Rydberg states,
which exist with a fairly high density of states in the region slightly below the vertical ion-
ization potential. The density of the molecular Rydberg states increases with the increase in
the excitation energy in the region slightly below the vertical ionization potential of a target
molecule. The probability of de-excitation is enhanced with the increase in the density of
molecular Rydberg states because the number of channels accepting the energy transferred
from Ar
*
is increased.
Thus, it is concluded that the de-excitation of the metastable Ar(
3
P
2
and
3
P
0
) atoms
is ascribed to the nonadiabatic excitation transfer at large intermolecular distance by the
crossing of the intermolecular potential curves between the initial Ar
*
M channel and the
nal ArM* channel, and the de-excitation of the resonant Ar(
3
P
1
and
1
P
1
) to the resonant
excitation transfer by the dipoledipole interaction [153]. This conclusion is compatible
with the result of the above-mentioned conclusions for the de-excitation of He(2
1
P) by Ne
[135].
As a brief conclusion of this section, the cross-section measurements for the de-
excitation of excited rare gas atoms have been best performed using the pulse radiolysis
method. The pulse radiolysis method has provided not only the most reliable cross sections
Figure 19 Comparison of the de-excitation cross sections of excited argon atoms Ar
*
[Ar
*
=
3
P
1
(o) or
1
P
1
(.)] by M (M=CH
4
, SiH
4
, or GeH
4
) with the theoretical cross sections (r
WK
). (From
Ref. 153.)
Copyright 2004 by Taylor & Francis Group, LLC
both for metastable and the resonant atoms, but also several physical aspects of the colli-
sional and molecular insight, so that the research of the collisional de-excitation proceeds
into a new stage of investigation.
ACKNOWLEDGMENTS
The authors thank a number of collaborators for their excellent collaboration, particularly
Dr. N. Kouchi. They also appreciate the early contribution to and the continuous devel-
opment of the pulse radiolysis microwave conductivity/cavity technique by the late Dr. H.
Shimamori.
REFERENCES
1. Hatano, Y. In Proceedings of the 19th International Conference on Phenom. Ioniz. Gas.,
Belgrade, 1989 (Invited Papers); Zigman, V.J., Ed.; University of Belgrade Press: Belgrade,
1989; 242 pp.
2. Hatano, Y. Radiat. Phys. Chem. 1989, 34, 675.
3. Hatano, Y. In Pulse Radiolysis; CRC Press: Boca Raton, Tabata, Y., Ed.; 1991. (Chap. 9).
4. Sauer, M.C., Jr. Adv. Radiat. Chem. 1976, 5, 97.
5. Firestone, R.F.; Dorfman, L.M. In Actions Chimiques et Biologiques des Radiations; Haissinsky,
M., Ed.; Masson: Paris, 1971; Vol. 15, 7 pp.
6. Hatano, Y. New Exp. Tech. Chem. 1978, 16, 544. in Japanese.
7. Sauer, M.C., Jr. In Study of Fast Processes and Transient Species by Electron Pulse Radiolysis;
Baxendale, J.H., Busi, F., Eds; Reidel: Dordrecht, 1982; 601 pp.
8. Armstrong, D.A. In Radiation Chemistry; Principles and Applications; Farhataziz, Rodgers,
M.A.J., Eds; . VCH: Berlin, 1987 (Chap. 9).
9. Jonah, C.D.; Liu, A.; Mulac, W.A. In Radiation Research; Fielden, E.M., Fowlor, J.F.,
Hendry, J.H., Scott, D., Eds.; Taylor and Francis: London, 1987; Vol. 2, 60 pp.
Table 7 De-excitation Cross Sections of Ar(
3
P
2,0
) by CH
4
, SiH
4
, or GeH
4
in Comparison with the Respective Cross Sections for Reaction Products
(in A
2
) (From Refs. 123 and 153.)
Ar(
3
P
2,0
)+CH
4
!
r
M
80
CH
n
+
(n V 4) 0
C*
CH* (A) 0
CH(X) 4.1
CH
2
, CH
3
Ar(
3
P
2,0
)+SiH
4
!
r
M
101
SiH
n
+
(nV4)
Si* 0.27
SiH* (A) 4.0
SiH(X) 4.6
SiH
2
, SiH
3
Ar(
3
P
2,0
)+GeH
4
!
r
M
102
GeH
n
+
(nV4)
Ge* 1.7
GeH* (A) 0.03
GeH(X) 0
GeH
2
, GeH
3
Copyright 2004 by Taylor & Francis Group, LLC
10. Hatano, Y. In Electronic and Atomic Collisions; Lorents, D.C., Meyerhof, W.E., Peterson,
J.R., Eds.; Elsevier: Amsterdam, 1986; 153 pp.
11. Hatano, Y.; Shimamori, H. In Electron and Ion Swarms; Christophorou, L.G., Ed.; Pergamon:
Oxford, 1981; 103 pp.
12. Ukai, M.; Hatano, Y. In Gaseous Electronics and Its Applications; Crompton, R.W., Hayashi,
M., Boyd, D.E., Makabe, T., Eds.; KTK Scientic: Tokyo, 1991; 51 pp.
13. Shinsaka, K.; Hatano, Y. Nucl. Instrum. Methods A 1993, 327, 7.
14. Hatano, Y. In Linking the Gaseous and Condensed Phases of Matter. The Behavior of Slow
Electrons, NATO-ASI B326; Christophrou, L.G., Illenberger, E., Schmidt, W.F., Eds.;
Plenum: New York, 1994; 467 pp.
15. Hatano, Y. Aust. J. Phys. 1997, 50, 615.
16. Cooper, R.; Denison, L.S.; Zeglinski, P.; Roy, C.R.; Gillis, H. J. Appl. Phys. 1983, 54, 3053.
17. Cooper, R.; Mezyk, S.P.; Armstrong, D.A. Radiat. Phys. Chem. 1984, 24, 545.
18. Doba, T.; Arai, S. J. Chem. Phys. 1981, 75, 488.
19. Kasama, K.; Oka, T.; Arai, S.; Kurusu, H.; Hama, Y. J. Phys. Chem. 1982, 86, 2035.
20. Loeb, D.W.; Chen, M.; Firestone, R.F. J. Chem. Phys. 1981, 74, 3270.
21. Manzahares, E.R.; Firestone, R.F. J. Chem. Phys. 1982, 76, 4475.
22. Manzahares, E.R.; Firestone, R.F. J. Chem. Phys. 1983, 79, 1683.
23. Tanaka, M.; Sasaki, S.; Katayama, M. Bull. Chem. Soc. Jpn. 1985, 58, 429.
24. Sauer, M.C., Jr.; Mulac, W.A. J. Chem. Phys. 1972, 56, 4995.
25. Dreyer, J.W.; Perner, D. Chem. Phys. Lett. 1971, 12, 299.
26. Ueno, T.; Kouchi, N.; Takao, S.; Hatano, Y. J. Phys. Chem. 1978, 82, 2373.
27. Cooper, R.; Denison, L.; Sauer, M.C., Jr. J. Phys. Chem. 1982, 86, 5093.
28. Freeman, G.R. In Electron, Ion Swarms; Christophorou, L.G., Ed.; Pergamon: Oxford, 1981;
93 pp.
29. Huang, S.S.-S.; Freeman, G.R. Phys. Rev. A 1981, 24, 714.
30. Nakamura, Y.; Shinsaka, K.; Hatano, Y. J. Chem. Phys. 1983, 78, 5820.
31. Nishikawa, M.; Holroyd, R.A.; Sowada, U. J. Chem. Phys. 1980, 72, 3081.
32. Nishikawa, M.; Holroyd, R.A. J. Chem. Phys. 1982, 77, 4678.
33. Shinsaka, K.; Codama, M.; Srithanratana, T.; Yamamoto, M.; Hatano, Y. J. Chem. Phys.
1988, 88, 7529.
34. Wada, T.; Freeman, G.R. Phys. Rev., A 1981, 24, 1066.
35. Hansen, D.; Jungblut, H.; Schmidt, W.F. J. Phys. D 1983, 16, 1623.
36. Sennhauser, E.S.; Armstrong, D.A.; Wilkinson, F. J. Phys. Chem. 1980, 84, 123.
37. Warman, J.M.; Sennhauser, E.S.; Armstrong, D.A. J. Chem. Phys. 1979, 70, 995.
38. Scales, M.J.; Cooper, R.; Warman, J.M.; deHaas, M.P. Radiat. Phys. Chem. 1987, 29, 365.
39. Shizgal, B.; Hatano, Y. J. Chem. Phys. 1988, 88, 5980.
40. Suzuki, E.; Hatano, Y. J. Chem. Phys. 1986, 84, 4915.
41. Suzuki, E.; Hatano, Y. J. Chem. Phys. 1986, 85, 5341.
42. Warman, J.M.; deHaas, M.P. J. Chem. Phys. 1975, 63, 2094.
43. Warman, J.M.; deHaas, M.P. Radiat. Phys. Chem. 1988, 32, 31.
44. Warman, J.M.; Sauer, M.C., Jr. J. Chem. Phys. 1975, 62, 1971.
45. Hatano, Y.; Kimizuka, Y.; Shimamori, H. Radiat. Phys. Chem. 1982, 19, 265.
46. Fessenden, R.W.; Bansal, K.M. J. Chem. Phys. 1970, 53, 3468.
47. Matsuoka, S.; Nakamura, H.; Tamura, T. J. Chem. Phys. 1981, 75, 681.
48. Matsuoka, S.; Nakamura, H.; Tamura, T. J. Chem. Phys. 1983, 79, 825.
49. Bansal, K.M.; Fessenden, R.W. Chem. Phys. Lett. 1972, 15, 21.
50. Bansal, K.M.; Fessenden, R.W. J. Chem. Phys. 1973, 59, 1760.
51. Warman, J.M.; Fessenden, R.W.; Bakale, G. J. Chem. Phys. 1972, 57, 2702.
52. Kokaku, Y.; Hatano, Y.; Shimamori, H.; Fessenden, R.W. J. Chem. Phys. 1979, 71, 4883.
53. Kokaku, Y.; Toriumi, M.; Hatano, Y. J. Chem. Phys. 1980, 73, 6167.
54. Shimamori, H.; Fessenden, R.W. J. Chem. Phys. 1978, 68, 2757.
55. Shimamori, H.; Fessenden, R.W. J. Chem. Phys. 1978, 69, 4732.
Copyright 2004 by Taylor & Francis Group, LLC
56. Shimamori, H.; Fessenden, R.W. J. Chem. Phys. 1979, 70, 1137.
57. Shimamori, H.; Fessenden, R.W. J. Chem. Phys. 1979, 71, 3009.
58. Shimamori, H.; Fessenden, R.W. J. Chem. Phys. 1981, 74, 453.
59. Shimamori, H.; Hatano, Y. Chem. Phys. Lett. 1976, 38, 242.
60. Shimamori, H.; Hatano, Y. Chem. Phys. 1976, 12, 439.
61. Shimamori, H.; Hatano, Y. Chem. Phys. 1977, 21, 187.
62. Shimamori, H.; Hotta, H. J. Chem. Phys. 1983, 78, 1318.
63. Shimamori, H.; Hotta, H. J. Chem. Phys. 1984, 81, 1271.
64. Shimamori, H.; Hotta, H. J. Chem. Phys. 1986, 84, 3195.
65. Shimamori, H.; Hotta, H. J. Chem. Phys. 1986, 85, 887.
66. Shimamori, H.; Hotta, H. J. Chem. Phys. 1986, 85, 4480.
67. Shimamori, H.; Hotta, H. J. Chem. Phys. 1989, 90, 232.
68. Toriumi, M.; Hatano, Y. J. Chem. Phys. 1983, 79, 3749.
69. Toriumi, M.; Hatano, Y. J. Chem. Phys. 1984, 81, 3748.
70. Toriumi, M.; Hatano, Y. J. Chem. Phys. 1985, 82, 254.
71. Nagra, S.S.; Armstrong, D.A. Can. J. Chem. 1976, 54, 3580.
72. Nagra, S.S.; Armstrong, D.A. J. Phys. Chem. 1977, 81, 599.
73. Nagra, S.S.; Armstrong, D.A. Radiat. Phys. Chem. 1978, 11, 305.
74. Ma rk, T.D. In Electronic and Atomic Collisions; Gilbody, H.B., Newell, W.R., Read, F.H.,
Smith, A.C.H., Eds.; North-Holland: Amsterdam, 1988; 705 pp.
75. Ma rk, T.D.; Leiter, K.; Ritter, W.; Stamatovic, A. Phys. Rev. Lett. 1985, 55, 2559.
76. Stamatovic, A. In Electronic and Atomic Collisions; Gilbody, H.B., Newell, W.R., Read, F.H.,
Smith, A.C.H., Eds.; North-Holland: Amsterdam, 1988; 729 pp.
77. Matejcik, S.; Kiendler, A.; Stampi, P.; Stamatovic, A.; Ma rk, T.D. Phys. Rev. Lett. 1996, 77,
3771.
78. Smirnov, B.M. Sov. Phys., Usp. 1984, 27, 1.
79. Bloch, F.; Bradbury, N.E. Phys. Rev. 1935, 48, 689.
80. Herzenberg, A. J. Chem. Phys. 1969, 51, 4942.
81. Bouby, L.; Abgrall, H. In Proceedings of the 5th International Conference on Phys. Electron.
Atom. Collision, Leningrad, 1967; 584 pp.
82. Bouby, L.; Fiquet-Fayard, F.; LeCoat, Y. Int. J. Mass Spectrom. Ion Phys. 1970, 3, 439.
83. Chanin, L.M.; Phelps, A.V.; Biondi, M.A. Phys. Rev. 1962, 128, 219.
84. Crompton, R.W.; Hegerberg, R.; Skullerud, H.R. In Proceedings of the International Seminar
on Swarm Experiments in Atomic Collision Research, Tokyo; Ogawa, I., Ed.; 1979; 18 pp.
85. Hackam, R.; Lennon, J.J. Proc. Phys. Soc. 1965, 86, 123.
86. Hegerberg, R.; Crompton, R.W. Aust. J. Phys. 1983, 36, 831.
87. Hurst, G.S.; Bortner, T.E. Phys. Rev. 1959, 114, 166.
88. van Lind, V.A.J.; Wikner, E.G.; Truedblood, D.L. Bull. Am. Phys. Soc. 1960, 5, 122.
89. Nelson, D.R.; Davis, F.J. Bull. Am. Phys. Soc. 1971, 16, 217.
90. Pack, J.L.; Phelps, A.V. J. Chem. Phys. 1966, 45, 4316.
91. Stockdale, J.A.; Christophorou, L.G.; Hurst, G.S. J. Chem. Phys. 1967, 47, 3267.
92. Warman, J.M.; Bansal, K.M.; Fessenden, R.W. Chem. Phys. Lett. 1971, 12, 211.
93. Young, B.G.; Johnson, A.W.; Garruthors, J.A. Can. J. Phys. 1963, 41, 625.
94. Koike, F. J. Phys. Soc. Jpn. 1973, 35, 1166.
95. Koike, F. J. Phys. Soc. Jpn. 1975, 39, 1590.
96. Koike, F.; Watanabe, T. J. Phys. Soc. Jpn. 1973, 34, 1022.
97. Parlant, G.; Fiquet-Fayard, F. J. Phys. B 1976, 9, 1617.
98. Christophorou, L.G. J. Phys. Chem. 1972, 76, 3730.
99. Christophorou, L.G. Chem. Rev. 1976, 76, 409.
100. Christophorou, L.G. Adv. Electr. Electron Phys. 1978, 46, 55.
101. Goans, R.E.; Christophorou, L.G. J. Chem. Phys. 1974, 60, 103.
102. McCorkle, D.L.; Christophorou, L.G.; Anderson, V.E. J. Phys. B 1972, 5, 1211.
103. Stogryn, D.E.; Hirschfelder, J.O. J. Chem. Phys. 1959, 31, 1531.
Copyright 2004 by Taylor & Francis Group, LLC
104. Blaney, B.L.; Ewing, G.E. Annu. Rev. Phys. Chem. 1976, 27, 553.
105. Christophorou, L.G. Environ. Health Perspect. 1980, 36, 3.
106. Christophorou, L.G.; Stockdale, J.A. J. Chem. Phys. 1968, 48, 1956.
107. Huo, W.M.; Fessenden, R.W.; Bauschlicher, C.W., Jr. J. Chem. Phys. 1984, 81, 5811.
108. McMahon, D.R.A. In Electron and Ion Swarms; Christophorou, L.G., Ed.; Pergamon:
Oxford, 1981; 117 pp.
109. McMahon, D.R.A. Chem. Phys. 1982, 66, 67.
110. Nishikawa, M.; Holroyd, R.A. J. Chem. Phys. 1983, 79, 3754.
111. Crompton, R.W.; Hegerberg, R.; Skullerud, H.R. J. Phys. B 1980, 13, 603.
112. Koura, K. J. Chem. Phys. 1982, 76, 390.
113. Koura, K. J. Chem. Phys. 1983, 78, 604.
114. McMahon, D.R.A.; Crompton, R.W. J. Chem. Phys. 1983, 78, 603.
115. Skullerud, H.R. Aust. J. Phys. 1983, 36, 845.
116. Shimamori, H.; Tatsumi, Y.; Ogawa, Y.; Sunagawa, T. Chem. Phys. Lett. 1992, 194, 223.
117. Shimamori, H.; Tatsumi, Y.; Ogawa, Y.; Sunagawa, T. J. Chem. Phys. 1992, 97, 6335.
118. Shimamori, H.; Sunagawa, T.; Ogawa, Y.; Tatsumi, Y. Chem. Phys. Lett. 1994, 227, 609.
119. Shimamori, H.; Sunagawa, T.; Ogawa, Y.; Tatsumi, Y. Chem. Phys. Lett. 1995, 232, 115.
120. Sunagawa, T.; Shimamori, H. Int. J. Mass Spectrom. Ion Process 1995, 149/150, 123.
121. Sunagawa, T.; Shimamori, H. J. Chem. Phys. 1997, 107, 7876.
122. Sunagawa, T.; Shimamori, H. Int. J. Mass Spectrom. 2001, 205, 285.
123. Hatano, Y. Adv. At. Mol. Opt. Phys. 2000, 43, 231.
124. Yencha, A.J. In Electron Spectroscopy: Theory, Techniques and Applications; Brundle, C.R.,
Baker, A.D., Eds.; Academic Press: London, 1984; Vol. 5, 197 pp.
125. Ueno, T.; Hatano, Y. Chem. Phys. Lett. 1976, 40, 283.
126. Nakamura, H. J. Phys. Soc. Jpn. 1965, 26, 1473.
127. Nakamura, H. J. Phys. Soc. Jpn. 1971, 31, 574.
128. Nakamura, H. J. Phys. B 1975, 8, L489.
129. Miller, W.H. J. Chem. Phys. 1970, 52, 3563.
130. Niehaus, A. In Radiation Research; Nygaard, O.F., Adler, H.I., Sinclair, W.K., Eds.; Academic
Press: London, 1975; 227 pp.
131. Niehaus, A. Adv. Chem. Phys. 1981, 45, 399.
132. Niehaus, A. In Electronic and Atomic Collisions; Datz, S., Ed.; North Holland: Amsterdam,
1982; 237 pp.
133. Illenberger, E.; Niehaus, A. Z. Phys., B 1975, 20, 33.
134. Ueno, T.; Yokoyama, A.; Takao, S.; Hatano, Y. Chem. Phys. 1980, 45, 261.
135. Yokoyama, A.; Takao, S.; Ueno, T.; Hatano, Y. Chem. Phys. 1980, 45, 439.
136. Yokoyama, A.; Hatano, Y. Chem. Phys. 1981, 63, 59.
137. Koizumi, H.; Ukai, M.; Tanaka, Y.; Shinsaka, K.; Hatano, Y. J. Chem. Phys. 1986, 85, 1931.
138. Ohno, K.; Matsumoto, S.; Harada, Y. J. Chem. Phys. 1984, 81, 4447.
139. Kohmoto, M.; Watanabe, T. J. Phys. B 1977, 10, 1875.
140. Watanabe, T.; Katsuura, K. J. Chem. Phys. 1967, 47, 800.
141. Ukai, M.; Tanaka, Y.; Koizumi, H.; Shinsaka, K.; Hatano, Y. J. Chem. Phys. 1986, 84, 5575.
142. Ukai, M.; Nakazawa, H.; Shinsaka, K.; Hatano, Y. J. Chem. Phys. 1988, 88, 3623.
143. Katsuura, K. J. Chem. Phys. 1965, 42, 3771.
144. Ukai, M.; Yoshida, H.; Morishima, Y.; Nakazawa, H.; Shinsaka, K.; Hatano, Y. J. Chem.
Phys. 1989, 90, 4865.
145. Morishima, Y.; Ukai, M.; Shinsaka, K.; Kouchi, N.; Hatano, Y. J. Chem. Phys. 1991, 94,
2564.
146. Nayfeh, M.H.; Chen, C.H.; Payne, M.G. Phys. Rev., A 1976, 14, 739.
147. Penkin, N.P.; Devdariani, A.Z.; Ionih, W.Z.; Samson, A.V. In Abstract, Xth International
Conference on Phys. Electron. Atom. Collision; Paris, Watel, G., Ed; 1977; 320 pp.
148. Morishima, Y.; Ukai, M.; Kouchi, N.; Hatano, Y. J. Chem. Phys. 1992, 96, 8187.
149. Morishima, Y.; Yohshida, H.; Ukai, M.; Kouchi, N.; Hatano, Y. J. Chem. Phys. 1992, 97, 3180.
Copyright 2004 by Taylor & Francis Group, LLC
150. Ukai, M.; Koizumi, H.; Shinsaka, K.; Hatano, Y. J. Chem. Phys. 1986, 84, 3199.
151. Kitajima, M.; Hidaka, K.; Kusumori, H.; Ukai, M.; Kouchi, N.; Hatano, Y. J. Chem. Phys.
1994, 100, 8072.
152. Yohshida, H.; Kawamura, H.; Ukai, M.; Shinsaka, K.; Kouchi, N.; Hatano, Y. Chem. Phys.
Lett. 1991, 176, 173.
153. Yohshida, H.; Kawamura, H.; Ukai, M.; Kouchi, N.; Hatano, Y. J. Chem. Phys. 1992, 96, 4372.
154. Yohshida, H.; Ukai, M.; Kawamura, H.; Kouchi, N.; Hatano, Y. J. Chem. Phys. 1992, 97, 3289.
155. Yohshida, H.; Kitajima, M.; Kawamura, H.; Hidaka, K.; Ukai, M.; Kouchi, N.; Hatano, Y.
J. Chem. Phys. 1993, 98, 6190.
Copyright 2004 by Taylor & Francis Group, LLC
7
Studies of Solvation Using Electrons and
Anions in Alcohol Solutions
**
Charles D. Jonah
Argonne National Laboratory, Argonne, Illinois, U.S.A.
1. INTRODUCTION
It has long been known that the solvent plays an important part in chemical reactivity. It
can inuence reactivity by altering the energy levels of the reactants and products, by
removing energy from hot modes, by stabilizing intermediates, and by conning reactants.
For these reasons, reactions in solution may be considerably dierent than in the gas phase.
Radiation chemistry highlights the importance of the role of the solvent in chemical
reactions. When one radiolyzes water in the gas phase, the primary products are H atoms
and OH radicals, whereas in solution, the primary species are e
aq
, OH, and H
+
[1]. One
can vary the temperature and pressure of water so that it is possible to go continuously
from the liquid to the gas phase (with supercritical water as a bridge). In such experiments,
it was found that the ratio of the yield of the H atom to the hydrated electron (H/e
aq
)
does indeed go from that in the liquid phase to the gas phase [2]. Similarly, when one
photoionizes water, the threshold energy for the ejection of an electron is much lower in
the liquid phase than it is in the gas phase. One might suspect that a major dierence is
that the electron can be transferred to a trap in the solution so that the full ionization
energy is not required to transfer the electron from the molecule to the solvent.
One can consider two facets of the solvation process, the energetics and the kinetics.
Clearly, the kinetics will not matter if reactions take place on a time scale that is much
faster or much slower than the solvation process. However, if reaction and solvation occur
on the same time scale, the considerable energy changes that the solvation process can
engender will aect the reaction. In fact, exactly what solvent motions take place during
the solvation process may well be important. Thus, it is of interest to understand the ki-
netics of the solvation process.
In a paper, Jortner stated that information about solvation dynamics can be ob-
tained from (1) solvation of the electron, (2) solvation of a dipole, and (3) extracting in-
formation from the rates of electron transfer reaction [3,4]. We have added to these
*
Work performed under the auspices of the Oce of Science, Division of Chemical Science, US-
DOE, under contract number W-31-109-ENG-38.
Copyright 2004 by Taylor & Francis Group, LLC
possibilities the study of the solvation of an anion that is created very quickly, and we
show that some of the characteristics of electron solvation are specic to that species.
Solvation processes are generally studied by observing the shift of an absorption
process or an emission process as a function of time [5,6]. For example, one might excite a
molecule where the excited state is considerably more polar than the ground state. The
solvent would then rearrange the solvent around the excited state to lower the total energy
of the system. One can then observe the shift of the emission spectrum as a function of
time. This is shown pictorially in Fig. 1, with the top showing the system and the bottom
showing the potential curves against the solvent coordinate. This will then lead to a shift
of the emission spectrum towards lower energy. These techniques have been used by many
authors (see, e.g., Refs. [5,6]). The uorescence measurements can make use of the single-
photon-counting technique, and the signal-to-noise and dynamic range of those measure-
ments are much better than for optical absorption measurements. In fact, such experi-
ments are able to separate multiple-exponential processes, something that is not possible
with the absorption measurements.
An alternative technique is to create a ground state species very quickly, by creating
an anion or a cation, or possibly by decomposing a molecule and then observing the
change of its optical spectrum. This technique has the advantage over observing the
relaxation of the excited state in that one is not limited by the uorescence lifetime of
the probe molecule. In the experiments that we will discuss, we will attach an electron to a
benzophenone molecule and then watch the change of the spectrum of the resulting anion.
Fig. 2 shows a cartoon for this process. Note that the absorption will shift towards higher
energies (the blue) for these experiments.
In this paper, we will limit ourselves to the discussion of two types of anions in alcohol
solutions: the benzophenone anion [79] and the solvated electron. One can consider that
these systems are both measurements of the solvation of an anion, if we consider the electron
Figure 1 Cartoon showing solvation and how it will shift the spectrum of a dipolar excited state.
Copyright 2004 by Taylor & Francis Group, LLC
as the simplest anion. We will showthat while there are considerable similarities between the
solvation of the two species, there are also some considerable dierences. We will suggest
that these dierences arise out of the formation mechanism of the electron.
We will not discuss the solvation of the electron in water. While one might expect
that the electron solvation in alcohols should be qualitatively the same as in water, with
only quantitative changes, it turns out that the quantitative changes lead to qualitative
changes in the mechanism. It is usually assumed that the solvated electron is formed in the
excited state, which then relaxes to the ground state, and the solvent reorganizes around
the ground-state electron. This picture seems to hold reasonably well for electrons in
alcohols; however, in water, the solvent relaxation time is suciently fast that the re-
laxation of the excited state of the electron and the reorganization of the solvent occur on
the same time scale, making interpretation dicult.
2. EXPERIMENTAL
The results in this discussion will include those from our laboratory and experiments on
electron solvation from other laboratories. The experiments that were done at Argonne
made use of the stroboscopic pulse radiolysis technique, which will be discussed below.
Experiments from other laboratories have made use of pulse radiolysis and laser photolyis
techniques for the measurement of electron solvation.
Figure 2 Cartoon showing solvation and how it will shift the spectrum of an anion.
Copyright 2004 by Taylor & Francis Group, LLC
The laser photolysis techniques are discussed in detail in the appropriate references.
We will just give a short summary here. The time resolution of the laser photolysis
experiments ranged from 100s of femtoseconds to 2030 ps. An intense light pulse was
used to form the solvated electrons, either by photolyzing the solvent or by photolyzing a
solute in the solvent [1012]. The relaxation of the spectrum was then probed using a
second probe light beam. Walhout and Barbara used an alternative technique [13]. An
ionizing light pulse was used to form the electron. The electron was allowed to relax for
4 ns, and then the solvated electron was optically bleached with a second pulse and the
absorption was probed using a third light pulse.
The experiments that we have carried out make use of the Argonne subnanosecond
pulse radiolysis system, which has been described in detail elsewhere [14]. To summarize, a
30-ps electron pulse is used to create both an electron pump pulse and an optical probe
pulse. To create the probe pulse, a portion of the electron pulse intercepts a cell that is
lled with 1 atm of xenon. This pressure is sucient so that the electrons are traveling
faster than the speed of light in that medium. This generates Cerenkov light, which has the
time prole of the electron pulse and which is a continuum light source where the intensity
increases towards the blue. The portion of the electron beam that is not intercepted by the
gas cell is delayed by a magnet (a 270j magnet or alpha magnet). The Cerenkov light pulse
can be adjusted using an optical delay line so that it can arrive prior to the electron pulse
or up to approximately 3 ns after the pulse (maximum time range of approximately 3.9 ns).
For the experiments that measure the solvation of the electron in an alcohol,
dierent neat alcohols were placed in a ow cell and the kinetics at dierent wavelengths
and the evolution of the spectra were observed [1517]. Kinetics were determined to the
blue side of the solvated electron spectrum. The temperature could be lowered to provide
data over a range of temperatures. The lower-temperature experiments all utilized n-
propanol or 2-propanol because they have a long liquid range [18].
Additional experiments were done in mixtures of alcoholalkane [16,17]. The spectra
and kinetics were measured in mixtures of 1-propanoln-hexane. Some experiments were
done in cyclohexane, where the behavior was qualitatively similar; however, the exact con-
centration where spectra and kinetics changed depended on the alkane [16]. Additional
experiments observed the shift of the nal spectrum of the solvated electron in supercritical
ethanemethanol mixtures. These experiments were done using standard pulse radiolysis
techniques and thus we were unable to observe the kinetics [19].
Anion-solvation experiments were done using benzophenone as a probe molecule.
These experiments were suggested by Bernard Hickel and were based on work of Marig-
nier and Hickel and Ichikawa et al. in low-temperature alcohols [79]. The concentration
of benzophenone was in the range of 0.25 M. This concentration was shown to be suf-
cient so that it would react with all of the solvated-electron precursors; thus, virtually no
solvated electrons would be formed under these conditions [20]. The data were analyzed by
considering the time dependence of the spectra and the kinetics and evaluating a global t
for these data [20,21]. These experiments were done in a series of alcohols, as a function of
temperature, and in alcoholalkane mixtures.
The solvation processes for the anion systems were simulated using both simple ball
models and more representational models of the solvent and solute [14,22].
3. RESULTS AND DISCUSSION
The measurement of the solvation of an aromatic anion and the measurement of the
solvation of an electron in alcohols have quite a bit in common. They both observe the
Copyright 2004 by Taylor & Francis Group, LLC
response of the solvent to the sudden creation of an anion in the bulk of the solution.
Thus, one might expect that the nal states would be similar. However, there may well be
considerable dierences in the initial states, which then may well lead to changes in the
solvation process.
The electron will be solvated in a region where the solvent molecules are appropri-
ately arranged. There must be a cluster of electrons of a size of 45 to support the for-
mation of the solvated electron from the results of Gangwer et al., [23], Baxendale [24,25],
and Kenney-Wallace and Jonah [16]. This behavior does not depend on the specic
alcohol or alkane and even occurs in supercritical solutions, as has been shown in ex-
periments done using mixtures of supercritical ethanemethanol mixtures [19]. Experi-
ments have also shown that the thermodynamically lowest state might not be reached. For
example, the experiments of Baxendale that measured the conductivity of the solvated
electron in alcoholalkane mixtures showed that when there was a sucient concentration
of alcohols to form dimers, there was a sharp decrease in the mobility of the electron
[24,25]. This result showed that the electron was at least partially solvated. However, the
conductivity was not as low as one would expect for the fully solvated electron, and the
fully solvated electron was never formed on their time scale (many microseconds), a time
scale that was suciently long for the electronalcohol entity to encounter sucient
alcohols to fully solvate the electron. Similarly, the experiments of Weinstein and Fire-
stone, in mixed polar solvents, showed that the electron that was observed depended on
the initial mixture and would not relax to form the most fully solvated electron [26].
The situation for the benzophenone molecule would be expected to be considerably
dierent. The major driving force for the attachment of the electron is the electron anity
of the benzophenone molecule. The arrangement of the solvent around that molecule will
provide only a perturbation to that energy. Thus, one will observe an average of the
solvent structures around the benzophenone and not just solvent arrangements that are
optimal to trap the electron. Similarly, because the charge is more delocalized on the
benzophenone molecule, the alcohol molecules might not be so strongly aligned, so that
the alcohol molecules will be able to rearrange and reach the lowest energy conguration.
We shall discuss separately the results for electron solvation and anion solvation at
room temperature in dierent alcohols to provide a basis for the discussion of solvation
mechanisms. This will be followed by a discussion of solvation at lower temperatures and
in alcoholalkane mixtures to further highlight the similarities and dierences between
anion and electron solvation.
3.1. Electron Solvation at Room Temperature
There have been many experiments on electron solvation at very low temperatures, and
the results depend on the temperature range and alcohol. For n-propanol, which has a
very long liquid range, Baxendale and Wardman showed that the solvation behavior
changes markedly from 110 to 120 K [27]. At the higher temperatures, kinetics and spec-
tral changes are similar to what are seen at room temperature, while at lower temper-
atures, the results are similar to what are observed in glasses. The original experiments on
electron solvation in alcohols near room temperature were carried out by Chase and Hunt
[28]. In their experiments, they used their stroboscopic pulse radiolysis system to determine
the kinetics of the electron solvation. Those experiments were followed by further pulse
radiolysis experiments that obtained similar results [15,17]. With the advent of short-
pulsed lasers, several groups explored the solvation of the electron in alcohols. A summary
of the pulse radiolysis data was given by Kenney-Wallace and Jonah [16], and a summary
of the laser data has been given by Walhout and Barbara [13].
Copyright 2004 by Taylor & Francis Group, LLC
If one observes the blue of the absorption maximum, the signal appears to consist of
a fast rise followed by a slower growth. We are concerned here with the slower growth,
because it arises from the relaxation of the solvent around the electron. This behavior was
seen in both the photolytic and radiolytic formation of the electron. The data of Walhout
and Barbara are a bit more complex in this time period, because one has a bleach followed
by a recovery of the absorption of the electron (however, not all the absorption is recov-
ered, suggesting that the electron reacts with the solvent after the photoexcitation) [13]. A
summary of the measured kinetics from both laser and pulse radiolytic experiments are
given in Fig. 3. As we can see from this gure, the rate of the slow growth does not appear
to depend on the mode of formation. Early results suggested that there were dierences
between the rate of electron solvation for the dierent methods of electron formation, but
it has been suggested by Sander et al. that the dierences arose from the data analyses of
some of the early laser data [11], because the earlier analyses of the laser-ash-photolysis
data did not separate the slow growth process from the fast component, which would have
been instantaneous on their time scale.
Fig. 3 displays the solvation time of the electron (using open symbols) plotted
against the longitudinal relaxation time. The results were plotted against the longitudinal
relaxation time because that is the time scale that some continuum models would have
suggested. Note that the rates are comparable to those seen for the solvation of an anion,
and considerably faster than those measured for dipole solvation. The anion data will be
discussed below.
The shift of the spectrum has been well discussed previously. The early pulse
radiolysis data suggested that the kinetics at 1300 nm [28] matched the kinetics in the
blue [16], which led to the description of the kinetics as a two-state problem. However, the
results measured by Chase and Hunt, where the kinetics at 1050 nm were considerably
slower than the kinetics at 500 and 1300 nm, suggested that the kinetics were more
Figure 3 Solvation time in alcohols of the electron and the benzophenone anion, plotted versus the
longitudinal relaxation time s
1
. (The dipole data are from Refs. [5,6].)
Copyright 2004 by Taylor & Francis Group, LLC
complicated than a two-state model would predict [28]. Some authors have suggested the
existence of an approximate isosbestic point [12]. It seems to be clear that the response
observed well to the blue of the peak does not depend on wavelength [16].
3.2. Anion Solvation
Anion solvation has been studied by observing the shift in the absorption spectrum of the
benzophenone anion in various solvents and as a function of temperature. The benzo-
phenone anion was formed from the reaction of the benzophenone molecule and a
precursor to the solvated electron. Approximately 0.25 M benzophenone is put into the
solution so that all the presolvated electrons will react with the benzophenone and
virtually none will form the solvated electron. This process occurs much more quickly
than the solvation processes that are observed [14,20].
The rst experiments were done in a series of primary alcohols at room temperature.
In Fig. 4, we show the spectra as a function of time at room temperature at dierent times
in n-octanol. Similar data are seen in all the primary alcohols, and the nal solvated
spectrum is similar for all of the alcohols, suggesting that nal solvated anion state is
similar in all primary alcohols.
One of the models that has had considerable success for predicting solvation
processes of dipoles in non-hydrogen-bonded solvents is the dielectric continuum model
[5,14]. In this model, the amount of solvation will depend on the dipole densitythat is,
the molar concentration and strength of dipoles. While the position of the absorption
maximum is not directly related to the energy of solvation that a molecule experiences, one
would expect the two to be very strongly correlated. However, for the three dierent
Figure 4 Spectra of the benzophenone anion in n-octanol as a function of time after the generation
of the anion.
Copyright 2004 by Taylor & Francis Group, LLC
primary alcohols considered here, the dipole density changes by more than a factor of 2,
with the highest density for the butanol alcohol and the lowest density for the decanol
solvent, and the maxima of the spectra of the solvated anion do not change. We interpret
these results to mean that it is the local concentration of dipoles around an anion that
dominates the solvation and not the concentration of the dipoles in the bulk of the
solution.
We have also carried out experiments in secondary alcohols. For example, Fig. 5
shows the spectra in primary and secondary octanol (1-octanol vs. 2-octanol). These data
show that while the initial spectra are approximately the same in the two alcohols, the nal
spectra are considerably dierent. There is much less shift in the absorption spectrum in
the 2-octanol, suggesting that 2-octanol is less eective at solvating the benzophenone
molecule.
We suggest that in a secondary alcohol, there will be steric hindrance, which will
limit the number of molecules that can get near the ion. This is shown, cartoon fashion, in
Fig. 6. As one can see, the steric factors might well limit the ability of the alcohols to pack
around an anion.
To test these suggestions, simple Monte Carlo and molecular dynamics simulations
were carried out [22]. These explored the potential energy of a charge in the presence of
solvent molecules of dierent forms. The form of these approximations is shown in the
reference. These simple systems embody the primary characteristics of the present system.
The results showed that the solvation energy (the eld at the charge) does not depend on
the length of the solvent molecule but on the orientation of the dipole. If the positive end
of the dipole is at the end of the molecule, solvation will be strong. This agrees with the
experimental results that show that the nal spectra are very much the same in all primary
alcohols.
Figure 5 Spectra of the benzophenone anion in primary and secondary n-octanol.
Copyright 2004 by Taylor & Francis Group, LLC
Simulations were done where the dipole is reversed in the solvent molecule. This is
equivalent to making measurements in acetonitrile. The calculations suggested that there
would only be weak solvation in acetonitrile, despite the fact that acetonitrile is far more
polar than any of the alcohols that we have measured. This fact is indeed borne out
experimentally. The spectrum of the benzophenone anion is considerably to the red of the
spectrum of the benzophenone anion in any of the alcohols that we have measured. In
addition, there is no evidence for any shift of the spectrum in the time scale that we
observe. This lack of shift may not be surprising, because experiments in acetonitrile
suggest that the solvation time is very fast. Thus, any relaxation that is going to take place
will have occurred on the time scale of the present experiments.
Typical kinetics for the solvation of an anion in 1-octanol are seen in Fig. 7. While an
approximate isosbestic point has been seen for the solvation of the electron in an alcohol,
this gure shows there is nothing approaching an isosbestic point for this process. The
kinetics are dierent at dierent wavelengths, and even include both a growth and decay.
The kinetics were determined by tting the time proles and spectra at dierent times.
The results of these analyses are seen in Fig. 3, along with the data for the solvation of the
electron in the same solvents, as discussed above. We see that the solvation time for the
anion is longer but of the same order of magnitude as the solvation time for the electron,
and it is much shorter than would be predicted for dielectric relaxation.
3.3. Temperature Dependence
Solvation takes place through multiple processes, each with dierent energy requirements.
By making measurements at dierent temperatures, one can begin to separate the multiple
processes and possibly shift which processes are dominant. For these reasons, temper-
Figure 6 Cartoon of the eect of structure on benzophenone anion solvation.
Copyright 2004 by Taylor & Francis Group, LLC
ature-dependent measurements have always been an important technique for under-
standing fundamental reaction pathways. As has been mentioned previously, the spectral
and kinetic behavior of the solvation process alters drastically in n-propanol between 110
and 120 K [27]. There have also been studies of the activation energy of solvation for
dipolar systems, which we will compare to the present data. We will rst discuss the anion-
solvation measurements, then the electron solvation measurements, and then compare the
two dierent chemical systems.
Anion Solvation as a Function of Temperature
In Fig. 8 (top), we see the temperature dependence of the rate of solvation of the ben-
zophenone anion in primary and secondary propanol and n-butanol as a function of
temperature [21,29]. As one would expect, the solvation rate is slower in n-butanol than it
is in n-propanol; however, as we can see from the slope of the Arrhenius plot, the acti-
vation energies are about the same, both having an activation energy of approximately 22
kJ/mol. The activation energy for solvation of the anion in 2-propanol is considerably
lower, 16 kJ/mol. These values are similar to the energy required to break a hydrogen
bond in these systems, of 25F3 kJ/mol for the primary alcohols and 17F3 kJ/mol for 2-
propanol [30]. The ordering of the activation energies is reasonable in that steric hindrance
will decrease the hydrogen bonding in the secondary alcohol in comparison to the primary
alcohol. The similarity of the solvation energy to the activation energy for hydrogen-bond
breakage would suggest that hydrogen-bond breakage is important in the solvation
process; however the correlation with s
2
, the second dielectric relaxation time, which is
often attributed to the rate of rotation without breaking hydrogen bonds, is not consistent
with that picture [31].
Figure 7 Time proles of the absorption of the benzophenone anion as a function of wavelength.
Copyright 2004 by Taylor & Francis Group, LLC
Figure 8 (a) Temperature dependence of the solvation of the benzophenone anion in primary and
secondary propanol and n-butanol. (b) Temperature dependence of the solvation of the electron in
n-propanol and 2-propanol.
Copyright 2004 by Taylor & Francis Group, LLC
If one compares the log of the solvation time to the log of viscosity, the two primary
alcohols fall on the same straight line and this line has a slope of 1. The 2-propanol, while
correlated with viscosity, does not show a linear dependence [21].
Temperature Dependence of Electron Solvation
These results can be compared to the temperature dependence that has been seen in
electron solvation (see Fig. 8, bottom) [18,28]. We rst note that the dependence of the
solvation time on temperature is quite dierent for 1-propanol and 2-propanol. It appears
as if the activation energy for 2-propanol is considerably higher than 1-propanol, which is
the reverse of what was observed for the solvation of an anion or for the hydrogen-bond
energetics. The activation energy for solvation in 1-propanol is 14.4 kJ/mol, which is less
than the energy for hydrogen-bond breakage, while the activation energy for 2-propanol is
21 kJ/mol, which is considerably higher than the activation energy for anion solvation and
considerably higher than the energy of a hydrogen bond.
It has been suggested that the electron will solvate in a preexisting structure in the
liquid such as a cyclic tetramer or pentamer [16,23]. It is known that these types of
structures are quite common in alcoholalkane mixtures. Because the electron is dissolved
in the preexisting structure that has an appropriate structural arrangement, it might not be
necessary to break the hydrogen bonds to solvate the electron. In 2-propanol, exper-
imental data suggest that in dilute alcohol solutions, the alcohol tends to form polymer
chains rather than cyclic structures. This means that to create the nal electron structure,
one will need to break polymeric hydrogen bonds and allow the alcohols to rearrange to
create a structure around the electron. This then requires the breaking of hydrogen bonds
and the higher activation energy results from that fact.
3.4. Concentration Dependence of Solvation
Concentration Dependence of Anion Solvation
Figs. 9 and 10 show the shift of the spectra and the dierent kinetics that occur as a
function of alcohol concentration in the benzophenone system in propanolhexane
mixtures. These results show that there is a fast absorption, which is due to the formation
of the benzophenone anion. This absorption is then followed by a slow growth. This
growth is much slower than what is seen in the neat alcohol, and the rate of the growth
depends on the concentration of the alcohol in the propanolhexane mixture. Fig. 11
displays the rate of growth as a function of concentration at two dierent wavelengths, 750
and 800 nm. The rate at 800 nm is approximately twice as fast as the rate at 750 nm. If the
benzophenone anion is more solvated, the spectrum will be more blue-shifted. Presumably,
the more deeply solvated anions will have more alcohols near them, and it should take
longer to solvate. However, one does not expect this to be a simple function, because the
spectra will be both homogeneously and heterogeneously broadened. The dierent rates of
solvation as a function of wavelength are consistent with this result. The measured rate
constants are consistent with what one would expect for the rate of an alcohol molecule
diusing to the anion and aligning itself (reacting).
Concentration Dependence of Electron Solvation
The results for the solvation of the electron in mixed systems are greatly dierent from the
results for the anion solvation. All evidence, which we summarize below, is consistent with
Copyright 2004 by Taylor & Francis Group, LLC
Figure 9 Spectral dependence of the benzophenone anion as a function of time at 0.15 M propanol
in n-hexane.
Figure 10 Solvation kinetics of the benzophenone anion in propanolhexane mixtures measured at
800 nm.
Copyright 2004 by Taylor & Francis Group, LLC
the picture of an electron nding an appropriate location and then rearranging the solvent
in the vicinity of it and not solvating further by bringing alcohol (or other solvent) mo-
lecules into the inner shell around the electron.
In the experiments of Weinstein and Firestone [26], it was shown that the spectrum
of the electron in mixed solvents depends on the mixture of the two components, for
example, an alcohol and an ether. If complete relaxation were possible, one would expect
that the solvent molecules would interchange and the result would be the lowest energy
conguration. This does not appear to occur.
In experiments done by Baxendale [24] and Baxendale and Sharpe [25] in mixed n-
propanolhexane systems, it was found that at low concentrations of alcohol, a solvated
species was formed; however, it was not the solvated electron. The existence of the
solvated species was determined by the decrease in mobility of the electron in the solvent.
The mobility never decreased to the level that would be expected for the solvated electron
and no evidence of the solvated electron spectrum was detected.
Experiments done by Kenney-Wallace and Jonah showed a fast initial solvation of
the electron followed by a shift/increase of the spectrum towards the blue on a time scale
that is similar to (but somewhat slower than) the rate of the solvation of the electron in the
neat alcohol [16,17]. There was no long-term alteration of the spectrum on the time scale
that would be needed to allow an alcohol molecule to diuse to the solvated electron (a
nanosecond or two, as seen in the kinetics for the benzophenone anion solvation).
The results of Baxendale and Wardman, Kenney-Wallace and Jonah, and Gangwer
et al. are all consistent with the idea that there must be preformed traps in the solution
[16,2426]. Further evidence consistent with this view arises from experiments of Dimi-
trijevic et al., who measured the yield and the spectrum of the solvated electron as a
Figure 11 Rate of the solvation of the benzophenone. First-order rate constant at 800 nm is
1.1510
10
(Msec)
1
and at 750 nm the rate is 6.810
9
(Msec)
1
.
Copyright 2004 by Taylor & Francis Group, LLC
function of temperature in mixtures of supercritical ethanemethanol [19]. As one raised
the temperature, the maximum of the absorption spectrum of the solvated electron in the
mixture shifted to the red the same amount as did the spectrum of the electron in neat
alcohol. The amount of absorption decreased at higher temperatures. Above 110jC, the
absorption spectrum disappeared in the alcoholalkane mixture. In a solution, an electron
will solvate where there exist suciently large clusters of alcohols to solvate the electron or
it will react with the cations in the solution. The decrease in absorption as a function of
temperature was interpreted to mean that the concentration of alcohol-molecule clusters
capable of solvating the electron decreased at higher temperatures. At higher temper-
atures, the number of clusters that are suciently large to support the solvated electron
decreased, and at 110jC, the probability of the electron nding such a cluster decreased to
such a level that no solvated electrons were formed.
4. SUMMARY
While the solvated electron may be considered the smallest and simplest anion, its
solvation behavior in alcohols is not the same as an anion. An anion can solvate by
attaching solvent molecules from the bulk of the solution, the solvated electron does not
do this. A solvated electron is formed making use of particular structures in the solvent so
the arrangement of solvent molecules is not random. This means that we are observing a
small subclass of the solvent structure when we observe electron solvation. Electron
solvation can be an excellent probe for structure of the liquid because of this characteristic.
This fact was utilized in estimating the size and structures of alcohol molecules in
supercritical ethanemethanol mixtures [19]. However, the solvation behavior is not
necessarily characteristic of the behavior that one would expect for an anion in the
solvent. The driving force for the attachment of the electron to a large aromatic molecule
will be the electron anity of the solute molecule and not the solvent structure. This means
that the anion that is quickly produced will be initially probing the average.
While noting the dierences in behavior between electron solvation and anion
solvation, we cannot forget the similarities. The activation energies for solvation seem to
be related to the strength of the hydrogen bond in the solvent, when a solvent rearrange-
ment will take place. In addition, the solvation times are closely related to the dielectric
relaxation time s
2
, the time assigned to molecular rotation without breaking hydrogen
bonds, rather than s
1
, the time for the rotation requiring hydrogen-bond breakage. The
latter is the time that correlates with the rate of dipole solvation [5]. Whether this is a char-
acteristic of the solvent or the long-range forces that occur with anion solvation is not clear
at the present.
In this discussion, I have intentionally not discussed the solvation of the electron in
water. These results are quite confusing because of the overlap of the electronic relaxation
of the excited solvated electron and the rearrangement of the solvent. In an alcohol, these
terms are well separated and so discussion is simplied. In addition, we are not aware of
any study on the solvation of an anion in water.
There are many future roads for research that can greatly illuminate the physical
processes. It would be interesting to observe the solvation of a neutral species that is
formed from a negative species by photoionization. One might well expect solvation to be
faster for this system because the solvent molecules near the anion/neutral could rearrange
without breaking hydrogen bonds. The solvation of an anion primarily rearranges
molecules that are hydrogen bonded to other solvent molecules.
Copyright 2004 by Taylor & Francis Group, LLC
Supercritical solutions also provide a large and interesting collection of phenomena
that can be probed using these techniques. The greater uctuations in these solvents may
increase the rate of solvation because of the larger number of possible states that can exist.
Solvation is an important component of all reactivity and now it is possible to gain a
clearer insight into these processes. Hopefully, this will continue in the years ahead.
REFERENCES
1. Spinks, J.W.T.; Woods, R.J. An Introduction to Radiation Chemistry; 3rd Ed; 1990.
2. Cline, J.; Takahashi, K.; Marin, T.W.; Jonah, C.D.; Bartels, D.M. J. Phys. Chem. A 2002, 106,
12260.
3. Rips, I.; Klafter, J.; Jortner, J. J. Chem. Phys. 1988, 89, 4288.
4. Rips, I.; Klafter, J.; Jortner, J. J. Chem. Phys. 1988, 88, 3246.
5. Maroncelli, M. J. Mol. Liq. 1993, 57, 1.
6. Simon, J.D. Acc. Chem. Res. 1988, 21, 128.
7. Marignier, J.L.; Hickel, B. Chem. Phys. Lett. 1982, 86, 95.
8. Marignier, J.L.; Hickel, B. J. Phys. Chem. 1984, 88, 5375.
9. Ichikawa, T.; Ishikawa, Y.; Yoshida, H. J. Phys. Chem. 1988, 92, 508.
10. Hirata, Y.; Murata, N.; Tanioka, Y.; Mataga, N. J. Phys. Chem. 1989, 93, 4527.
11. Sander, M.; Brummund, U.; Luther, K.; Troe, J. Ber. Bunsen-Ges. 1992, 96, 1486.
12. Shi, X.; Long, F.H.; Lu, H.; Eisenthal, K.B. J. Phys. Chem. 1995, 99, 6917.
13. Walhout, P.K.; Barbara, P.F. Ultrafast Processes in Chemistry and Photobiology, 1995; 83 pp.
14. Lin, Y.; Jonah, C.D. Understanding chemical reactivity. 1994, 7, 137.
15. Kenney-Wallace, G.A.; Jonah, C.D. Chem. Phys. Lett. 1976, 39, 596.
16. Kenney-Wallace, G.A.; Jonah, C.D. J. Phys. Chem. 1982, 86, 2572.
17. Kenney-Wallace, G.A.; Jonah, C.D. Chem. Phys. Lett. 1977, 47, 362.
18. Zhang, X.; Jonah, C.D. Chem. Phys. Lett. 1996, 262, 649.
19. Dimitrijevic, N.M.; Takahashi, K.; Bartels, D.M.; Jonah, C.D. J. Phys. Chem. A 2001, 105,
7236.
20. Lin, Y.; Jonah, C.D. J. Phys. Chem. 1993, 97, 295.
21. Zhang, X.; Jonah, C.D. J. Phys. Chem. 1996, 100, 7042.
22. Lin, Y.; Jonah, C.D. Chem. Phys. Lett. 1995, 233, 138.
23. Gangwer, T.E.; Allen, A.O.; Holroyd, R.A. J. Phys. Chem. 1977, 81, 1469.
24. Baxendale, J.H. Can. J. Chem. 1977, 55, 1996.
25. Baxendale, J.H.; Sharpe, P.H.G. Chem. Phys. Lett. 1976, 41, 440.
26. Weinstein, J.B.; Firestone, R.F. J. Phys. Chem. 1975, 79, 1322.
27. Baxendale, J.H.; Wardman, P. J. Chem. Soc. Faraday Trans 1973, 1, 69, 584.
28. Chase, W.J.; Hunt, J.W. J. Phys. Chem. 1975, 79, 2835.
29. Zhang, X.; Lin, Y.; Jonah, C.D. Radiat. Phys. Chem. 1999, 54, 433.
30. Levin, B.Y. Zhur. Fiz. Khim. 1954, 28, 1399.
31. Garg, S.K.; Smyth, C.P. J. Phys. Chem. 1965, 69, 1294.
Copyright 2004 by Taylor & Francis Group, LLC
8
Electrons in Nonpolar Liquids
Richard A. Holroyd
Brookhaven National Laboratory, Upton, New York, U.S.A.
1. INTRODUCTION
Excess electrons can be introduced into liquids by absorption of high-energy radiation, by
photoionization, or by photoinjection from metal surfaces. The electrons chemical and
physical properties can then be measured, but this requires that the electrons remain free.
That is, the liquid must be suciently free of electron attaching impurities for these
studies. The drift mobility as well as other transport properties of the electron are
discussed here as well as electron reactions, free-ion yields, and energy levels.
Ionization processes typically produce electrons with excess kinetic energy. In liquids
during thermalization, where this excess energy is lost to bath molecules, the electrons
travel some distance from their geminate positive ions. In general, the electrons at this
point are still within the Coulombic eld of their geminate ions and a large fraction of the
electrons recombine. However, some electrons escape recombination and the yield that
escapes to become free electrons and ions is termed G
. Reported values of G
for molec-
ular liquids range from 0.05 to 1.1 per 100 eV of energy absorbed [1,2]. The reasons for this
20-fold range of yields are discussed here.
Electrons in nonpolar liquids are either in the conduction band, trapped in a cavity
in the liquid, or in special cases form solvent anions. The energy of the bottom of the
conduction band is termed V
0
. V
0
has been measured for many liquids and its dependence
on temperature and pressure has also been measured. New techniques have provided quite
accurate values of V
0
for the liquid rare gases. The energies of the trapped state have also
been derived for several liquids from studies of equilibrium electron reactions. A
characteristic of the trapped electron is its broad absorption spectrum in the infrared.
Electron attachment rates have been measured for numerous solutes. Many of these
studies were limited to three solvents: cyclohexane, 2,2,4-trimethylpentane, and tetrame-
thylsilane (TMS), and those rates are discussed here. What to expect in other liquids can
be inferred from these results. Considerable insight has been gained into certain reactions.
Equilibrium reactions of electrons are particularly interesting since they provide informa-
tion not only on energy levels, as mentioned above, but also on the partial molar volume
of trapped electrons. This has led to a better understanding of the mechanism of electron
transport.
This chapter presents the current understanding of electrons in nonpolar liquids.
Experimental as well as theoretical studies are discussed. For further detail than is pro-
Copyright 2004 by Taylor & Francis Group, LLC
vided here, the reader is referred to recent books on the subject [3,4] as well as references
cited herein. Some questions still remain due largely to the theoretical diculties of de-
scribing a quantum particle in a disordered environment. The future will hopefully bring
new discoveries and revelations that will answer these questions. Finally, we discuss current
applications of nonpolar liquids to indicate where future uses may develop.
2. ELECTRON ESCAPE AND RECOMBINATION
An important consideration in understanding a radiation chemical mechanism or predict-
ing the outcome of a radiation experiment is knowing the yield of free electrons or G
, the
number of ion pairs produced per 100 eV absorbed. The free ion yield is aected by the
density of ionization along the track of the ionizing particle. The highest yield is observed
for high-energy electrons, typically 12 MeV, which result in well-separated clusters of a
few ionizations each. An alpha particle creates a dense column of ionization and ion yields
are very low. X-rays show intermediate behavior. Other factors like molecular structure,
temperature, applied eld, pressure, and density also aect G
.
Making sense of free ion yield data requires rst knowing what is meant by the free
ion yield. Free electrons are those that escape initial spur or track recombination and are
therefore diusing in the bulk and can react with other species in a homogeneous fashion.
The escape process can be considered to have two steps. First, the electrons released by
ionizing the solvent molecules lose energy in scattering events and, in the process, travel
some distance, r, from their positive ions. The electrons may lose energy to vibrational
modes, but a signicant fraction of the range occurs while the electron has lower, near
thermal, energy. In general, there will be a distribution of ranges D(r) and the mean
thermalization range is designated by b. For most nonpolar liquids, the electrons after
thermalization will still be within the Coulombic eld of the positive ions, equal to e
2
/er,
where e is the dielectric constant. Consequently, the fraction escaping is low.
In the second step, the thermalized electrons will either recombine with the ions in the
track or spur or escape. The yield of free electrons is the integral of the product D(r) P(r)
times the number of electronion pairs formed initially in the spur or track, G
tot
, where
P(r) is the probability of escape. For a single ion pair, P(r) is given by [5]:
Pr exp e
2
=ek
B
T
1
The initial yield, G
tot
, is not known exactly for molecular liquids. However, the
fraction of electrons that escape geminate recombination increases with the applied electric
eld and extrapolating such results to high eld gives G
tot
=4.0 for neopentane [6], 3.1F0.3
for TMS [7], and 4.5 for Ar [8]. A value for G
tot
of 4.0 was obtained for cyclohexane by
measuring the yields of methyl radicals formed in the reaction of electrons with methyl
chloride and methyl bromide [9]. High concentrations of these solutes were used (up to 0.5
M) in order to compete with geminate recombination. An analysis by Jay-Gerin [10] of
free ion yield data led to an average G
tot
value of 3.7F0.5 for typical hydrocarbons. Others
have suggested that G
tot
may vary considerably from liquid to liquid [11].
Fast pulse radiolysis studies have shown that geminate recombination occurs on the
picosecond time scale [12,13]. Bartczak and Hummel [14] predicted that for n-dodecane,
82% of the geminate ions still remain at 5 psec for 1 MeV irradiation. Future accelerators,
with pulses of a few picoseconds length, may soon provide experimental measurements of
G
tot
directly.
Copyright 2004 by Taylor & Francis Group, LLC
2.1. G
fi
for Minimum Ionizing Radiation
For minimum ionizing radiation, where the ionization events are widely separated, the
value of G
and l
D
are
low for n-alkanes, n-alkenes, and aromatics and high for branched compounds with many
methyl groups like neopentane. For l
D
>0.1 cm
2
/Vs, the yields for alkanes follow
reasonably well a relationship suggested by Jay-Gerin [10]:
G
a l
D
x
2
Implicit in such a dependence is the recognition that scattering lengths of thermal and
epithermal electrons are similar. A least-squares t of the data for compounds containing
only hydrogen and carbon leads to the solid line shown in Fig. 1 for which a = 0.25 and
x = 0.33, for l
D
in cm
2
/Vs.
Fig. 1 includes points for ethane and propane at 298 K for which G
and l
D
for ethane
and propane are still higher [20,21]. On the other hand, at low temperatures near the
boiling points, the yields of free ions as well as the mobilities in these liquids are quite low
( G
= 0.16 for ethane for example), and points for these liquids then would be on the left
side of the gure.
The data for compounds containing a silicon or germanium atom fall on a lower line
(dashed) in Fig. 1 for which a = 0.12 and x = 0.34 in Eq. (2). That is, for any given
mobility value, compounds containing a heavier atom have a lower free ion yield than that
given by the line for alkanes [15]. A lower G
decreases [22]. However, for n-pentane and TMS, Eq. (2) predicts the
changes in G
pol
and V
0
.
Thus:
E
th
IP E
pol
V
0
4
Astudy of the photoionization of tetramethyl-p-phenylenediamine (TMPD) in solution [49]
showed the dependence of the wavelength of ionization onset on the energy V
0
of the sol-
Table 1 Energy Levels of Electrons at 298 K
Solvent V
0
(eV) DG
soln
(e)
n-Hexane 0.10 0.33
Cyclohexane 0.01 0.28
n-Pentane 0.01 0.32
3-Methylpentane 0 0.33
2-Methylbutane 0.34
2,2,4-Trimethylpentane 0.17 0.40
2,2-Dimethylbutane 0.26 0.43
2,2-Dimethylpropane 0.38 0.44
Tetramethylsilane 0.62 0.62
Bis(trimethylsilyl)methane 0.66 0.66
Hexamethyldisiloxane 0.70 0.70
Source: Refs. 2, 33, and 34.
Copyright 2004 by Taylor & Francis Group, LLC
vent. The data were used to evaluate V
0
for 18 dierent liquids. Typical results of this and
other studies [50,51] are given in Table 1. Direct single-photon ionization of solvent mole-
cules is also sensitive to the value of V
0
[5255] and can, in principle, be used to determine V
0
.
Recently, laser multiphoton ionization of solutes has been used. Dening the
threshold of ionization, E
th
, can be a problem in some of these methods. A recent
multiphoton technique, utilizing femtosecond laser pulses, appears to give quite accurate
thresholds [56]. In this work, a conductivity spectrum is measured at visible wavelengths
and a sharp drop in current occurs as the mechanism changes from n-photon excitation to
(n+1)-photon excitation, where n is typically 3 to 4. The threshold is dened by tting the
current to an analytic function that denes the midpoint of this transition. E
th
is then n
times the energy at which the midpoint occurs. The thresholds are sensitive to V
0
and
could be used for determination of this quantity.
Photodetachment from anions should also be mentioned [5759]. In this case, the
threshold, E
th
V
, given by:
E
th
V
EAV
0
E
pol
5
is well dened since the electron escapes with high probability from the neutral molecule
left behind, and the photodetachment yields as a function of photon energy follow a
known power law. Thus V
0
values can be determined from such studies if values of the
electron anity, EA, are available.
Recently, V
0
has been measured in dense rare gas uids by eld ionization of
Rydberg states of solutes lying close to the continuum. In this technique, Reininger et al.
[6063] utilized synchrotron radiation and measured photocurrent spectra in the vacuum
ultraviolet (VUV) with a resolution of 7 meV. The photocurrent spectra were measured at
several applied electric elds. A eld ionization spectrum was obtained by subtracting the
spectrum at low eld from one obtained at high eld. Such spectra typically showed one
peak at threshold energies due to those high-lying states that ionized at the high voltage
but not at the low voltage. Measurements were made in argon containing CH
3
I [60] and
H
2
S [61] as solutes, and the density was varied from dilute gas to the triple point of argon.
The position of the eld ionization peak shifts as the density is changed. These shifts are
the combined eect of changes in the ion-media polarization energy and V
0
. Accurate
calculation of the former allowed the evaluation of V
0
as a function of density, utilizing
Eq. (4) above. The resulting values of V
0
are in good agreement with earlier photoelectric
and theoretical values. Furthermore, the two solutes CH
3
I and H
2
S gave very comparable
results. For argon, V
0
reaches a minimum value of 0.294 eV at a density of 12 10
21
cm
3
, which is the density at which the electron mobility is a maximum.
The solid line in Fig. 3 is a best t of Eq. (6) to the experimental data [61]. See cited
reference for values of the parameters in Eq. (6).
V
0
N a
0
a
1
Na
2
a
3
=a
4
ln cosh a
4
Na
2
6
The density dependence of V
0
in Kr was determined by eld ionization of CH
3
I
[62] and (CH
3
)
2
S [63]. Whereas previous studies found a minimum in V
0
at a density of
12 10
21
cm
3
[66], the new study indicates that the minimum is at 14.4 10
21
cm
3
(see
Fig. 3). This is very close to the density of 14.1 10
21
cm
3
at which the electron mobility
reaches a maximum in krypton [67], a result that is consistent with the deformation poten-
tial model [68] which predicts the mobility maximum to occur at a density where V
0
is a
minimum. The use of (CH
3
)
2
S permitted similar measurements of V
0
in Xe because of its
lower ionization potential. The results for Xe are also shown in Fig. 3 by the lower line,
Copyright 2004 by Taylor & Francis Group, LLC
which represents the analytical equation (6) giving the best t to the data. Parameter values,
a
n
, can be found in the reference cited.
The earliest theoretical model of the eect of a medium on the quasi-free electron
energy was proposed by Fermi [69] to explain ionization of high Rydberg states. This
model, which predicts a linear dependence on density, works well only at low densities.
Springett et al. [37] introduced the WignerSeitz model to calculate V
0
in several liquids
employing a simple electronatom pseudopotential. This model was used by Stampi and
Bennemann to calculate V
0
for liquid Ar, Kr, and Xe, except they solved the eigenvalue
problem numerically [64]. Their results for Ar, shown in Fig. 3 by the dashed line, are in
good agreement with the experiment. Several other modications to the theoretical
treatments have been proposed [65,7073] using various techniques. Simon et al. [74,75]
used a classical percolation approach to predict V
0
for Ar, Kr, and Xe. The calculations of
Plenkiewicz et al. [65], which use an accurate electronatom pseudopotential, are shown by
the dotted lines in Fig. 3. Their calculations also agree quite well with the experimental
data and nd the minima at the right densities. These newer studies are readily applicable
to molecular liquids and have been applied to CH
4
and SiH
4
[76].
Figure 3 Energy of the quasi-free state in rare gas uids as a function of density. Solid lines
represent recent results obtained by eld ionization for Ar [61], Kr, and Xe [see Eq. (6)] [63]. Recent
theoretical calculations are shown for Ar by the dashed lines [64] and for Ar, Kr, and Xe by the
dotted lines [65].
Copyright 2004 by Taylor & Francis Group, LLC
4. SOLVATED ELECTRONS
Evidence for the existence of a trapped state of the electron in some liquids comes from
several sources. For one, pulse radiolysis studies have shown that for certain alkanes, there
is a broad absorption in the infrared that can be attributed to the electron. Second, studies
of the eect of pressure on the mobility of electrons have revealed information about
cavity sizes and the role of electrostriction in the trapping process. Finally, conductivity
studies of reversible reactions of the electron have given us the energetics of the trapping
process. These latter equilibrium studies are described in detail in Sec. 5.1. Values of
DG
soln
(e) for selected liquids are given in Table 1. For many liquids, DG
soln
(e) lies below
the V
0
state making trapping energetically favorable.
Early pulse radiolysis studies of alkanes at room temperature showed that the sol-
vated electron absorption begins around 1 Am and increases with increasing wavelength to
1.6 Am for n-hexane, cyclohexane, and 2-methylbutane [77]. More complete spectra for
three liquid alkanes are shown in Fig. 4. The spectrum for methylcyclohexane at 295 K
extends to 4 Am and shows a peak at 3.25 Am [78]. At the maximum, the extinction
coecient is 2.8 10
4
M
1
cm
1
. The spectrum for 3-methyloctane at 127 K, shown in
Fig. 4, peaks around 2 Am. The peak for methylcyclohexane is also at 2 Am at lower
temperature. Recently, the absorption spectra of solvated electrons in 2-methylpentane,
3-methylpentane, cis-decalin, and methylcyclohexane glasses have been measured accu-
rately at 77 K [80]. For these alkanes, the maxima occur at 1.8 Am, where the extinction
coecient is 2.7 10
4
M
1
cm
1
.
Figure 4 Absorption spectra of solvated electrons in alkanes vs. wavelength. Solid line is for
methylcyclohexane at 295 K [78]. Dashed line is for 3-methyloctane at 127 K [79]. Dashdot line is
for 3-methylpentane at 77 K [80]. Spectra have been normalized to unity at the peaks.
Copyright 2004 by Taylor & Francis Group, LLC
The stronger absorption at lower temperatures can be attributed to several factors.
One is that the equilibrium between quasi-free and trapped electrons shifts to favor
trapped electrons as the temperature is lowered for these liquids. Another is that homo-
geneous recombination of electrons with positive ions is slower at the lower temperatures
and therefore occurs to a lesser extent during the pulse, which, for the pulse radiolysis
studies, was typically 10 to 20 nsec. The rate constant, k
r
, for electron recombination with
positive ions, in most nonpolar liquids, is given by:
k
r
4pel
D
e
1:09 10
15
l
e
=e M
1
sec
1
for l
e
in cm
2
Vs
7
This equation is suciently valid that it is considered to be a law of electron dynamics.
Exceptions exist only for very high values of l
D
(see Chap. 10).
It is important to consider the magnitude of the recombination rate in studies of this
type. For methane, k
r
is 1.7 10
17
M
1
sec
1
at 93 K [81]. Thus if a concentration of ions
of 0.1 AM was formed in the pulse, the electrons would disappear with a rst half-life
of 50 psec. For 2,2,4-trimethylpentane, k
r
is 3.6 10
15
M
1
sec
1
and for a similar con-
centration of electrons, the recombination lifetime would be a few nanoseconds. Where the
electron mobility is lower, the recombination rate is slower. For methylcyclohexane,
where l
e
= 0.07 cm
2
/Vs [18], k
r
is 4 10
13
M
1
sec
1
and the rst half-life should be about
250 nsec.
Electrons have not been detected by optical absorption in alkanes in which the
mobility is greater than 10 cm
2
/Vs. For example, Gillis et al. [82] report seeing no infrared
absorption in pulse-irradiated liquid methane at 93 K. This is not surprising since the
electron mobility in methane is 500 cm
2
/Vs [81] and trapping does not occur. Geminately
recombining electrons have, however, been detected by IR absorption in 2,2,4-trimethyl-
pentane in a subpicosecond laser pulse experiment [83]. The drift mobility in this alkane is
6.5 cm
2
/Vs, and the quasi-free mobility, as measured by the Hall mobility, is 22 cm
2
/Vs
(see Sec. 6). Thus the electron is trapped two-thirds of the time.
The nature of the absorption spectra has been discussed by several authors
[78,80,84,85]. Since the trapped state is not far below the conduction band, it is at least
reasonable to consider that the infrared spectrum is a bound-free transition. In the study
of alkane glasses at 77 K, mentioned earlier [80], the cross sections near threshold were
found to t the Wigner [86] power law. This supports the idea that the spectra are due to a
transition from a bound S-state to a continuum P-state. The threshold binding energy at
77 K was found to be 0.48 eV. From an analysis of the spectral distribution, the authors
obtained ground state properties of the trapped electron. The experimental spectrum
matched that derived from a simple spherical well model, suggesting that the electron
resides in a cavity of radius 0.35 nm. Other studies have arrived at a similar value for the
radius. For example, Ichikawa and Yoshida [84] obtained 0.36 nm for 3-methylpentane
glass based on a model that included a short-range repulsive contribution and an
attractive, Born-like, medium polarization. An electron spin resonance study of solvated
electrons in deuterated 3-methylpentane glasses indicated that there were an average of
three molecules in the rst solvation shell. A combination of spin-echo and second-
moment analysis gave electron to proton distances from 0.35 to 0.43 nm [87]. These studies
have provided considerable insight into the nature of the trapped state and are consistent
with the transport studies discussed below.
Studies of the eect of pressure on l
D
in a series of 10 hydrocarbons revealed more
information about the trapped state [34]. The pressure data led to the conclusion that the
partial molar volume of the electron, V
e
, is small but may be either positive or negative.
Copyright 2004 by Taylor & Francis Group, LLC
Values range from +22 cm
3
/mol for m-xylene to 27 cm
3
/mol for 1-pentene. This depends
on the relative magnitude of two large volume terms that make up V
e
. One is the cavity
volume, a positive term, and the other is the electrostriction of the solvent around the
trapped electron. Whereas the electrostriction term varies considerably depending on the
compressibility of the solvent, the cavity volume does not change much and the average
value for the hydrocarbons is 96 cm
3
/mol, corresponding to a cavity radius of 0.34 nm.
Thus from pulse radiolysis, mobility measurements, and electron reaction studies, we
have information on the absorption spectra, the cavity volume, and the energy of the
trapped or solvated state. The nature of this state seems to be an electron that is localized
in a cavity in the liquid.
5. REACTIONS OF ELECTRONS
Electron attachment to solutes in nonpolar liquids has been studied by such techniques as
pulse radiolysis, pulse conductivity, microwave absorption, and ash (laser) photolysis. A
considerable amount of data is now available on how rates depend on temperature,
pressure, and other factors. Although further work is needed, some recent experimental
and theoretical studies have provided new insight into the mechanism of these reactions.
To begin, we consider those reactions that show reversible attachmentdetachment
equilibria and therefore provide both free energy and volume change information.
5.1. Electron Equilibria
Electron equilibria of the type:
e
solute
k
d
!
k
a
solute
8
have been observed for solutes that do not dissociate on attachment and have gas phase
electron anities (EA) between 1.15 eV (benzene) [88,89] and 0.3 eV (phenanthrene) [90].
Application of high pressure facilitated the studies for solutes of very negative electron
anities like toluene, benzene, and butadiene. Equilibrium constants, K
eq
, have been eval-
uated by conductivity methods from changes in electron mobility and from determination
of both the attachment rate constants, k
a
, and the detachment rate constants, k
d
. Values
of DG
r
for various solutes are shown in Fig. 5 for tetramethylsilane, 2,2,4-trimethylpentane,
n-hexane, and supercritical ethane as solvents. It has been shown that DG
r
depends on the
polarization energy of the product anion, E
pol
(S
DG
s
e 9
where DG
o
r
( g) is derived from the electron anity DG
o
r
( g) = EATDS
r
( g). For all the
reactions, DG
o
r
is lower in 2,2,4-trimethylpentane than in tetramethylpentane (TMS) partly
because DG
s
(e) is higher in the former solvent and also because the dielectric constant is
higher in 2,2,4-trimethypentane, which makes the value of the solvation energy lower. The
further lowering of DG
r
for these reactions in n-hexane is due largely to the even higher value
of DG
s
(e) in this solvent.
Equilibria of this type have been used to determine the ground state of the electron in
liquids [100]. Some of the values of DG
s
(e) given in Table 1 were evaluated this way. The
reaction of the electron with CO
2
was used to measure DG
s
(e) for hexamethyldisiloxane
Copyright 2004 by Taylor & Francis Group, LLC
and bis(trimethylsilyl)methane. The electron mobility in these liquids is unusually high,
and V
0
is therefore expected to be the same as DG
s
(e). For hexamethyldisiloxane, DG
s
(e)
was found to be 0.70 eV and that for bis(trimethylsilyl)methane, it was found to be
0.66 eV (see Table 1).
These equilibrium reactions occur with large decreases in both volume and entropy.
Volume changes range from 80 to 300 cm
3
/mol depending on the solute and pressure.
These volume changes, DV
r
, are associated with the electrostriction of the solvent around
the product anion, V
el
(ion), and, to some extent, with a contribution of the partial molar
volume of the electron, V(e). Thus:
DV
r
V
el
ion V e 10
Values of V(e) are small compared to the overall volume changes (see Sec. 4). However,
magnitudes of DV
r
were found to be much larger than could be accounted for using the
classical expression for V
el
(ion) [101]:
V
el
ion e
2
=2R
ion
1=e
2
de=dP 11
To account for the dierence between experiment and theory, Schwarz suggested that
electrostriction includes the formation of a glassy shell of 7 to 9 solvent molecules around
the ion [102]. This phase transition provides a substantial density increase, and the
observed volume changes can then be accounted for.
These attachmentdetachment equilibria [Eq. (8)] shift to the right with increasing
pressure and to the left with increasing temperature. Thus the free energies decrease with
pressure and increase with temperature. These eects are related to the solvent compres-
Figure 5 Free energies for attachment to solutes: MeStmethylstyrene, Stystyrene, Bph
biphenyl, CO
2
, Pyrpyrimidine, Tphtriphenylene, Dfbp-diuorobenzene, Toltoluene, But
1,3-butadiene, Pyzpyrazine; in TMS, 2,2,4-trimethylpentane, and n-hexane at 298 K and in
supercritical ethane at 310 K. (From Refs. 9099.)
Copyright 2004 by Taylor & Francis Group, LLC
sibility, which increases with pressure and decreases with temperature. The entropy contri-
bution is less at high pressure but increases with temperature. Because both entropy and
volume changes are due to electrostriction of the solvent, these quantities are related by:
DS
el
ion a=v
T
DV
el
ion 12
where a is the coecient of thermal expansion and v
T
is the isothermal compressibility.
Electron attachment equilibria have also been observed in supercritical ethane. The
equilibrium constants are generally smaller than in liquids and consequently DG
r
is higher.
Thus, as shown in Fig. 5, DG
r
for attachment to CO
2
is higher in ethane at all pressures
than in either TMS or 2,2,4-trimethylpentane liquids. This is mainly a consequence of the
lower density of the solvent that results in less polarization energy for the anion. However,
the continuum model of Born is no longer correct for evaluating the polarization energy
because of the signicant density augmentation around the ions in supercritical nonpolar
uids. Instead, a compressible continuum model that takes this eect into account must be
used [99,103]. In the case of a supercritical solvent like ethane, the free energy of
attachment to a solute changes rapidly with pressure, particularly near critical densities
where the compressibility of the uid changes rapidly. Thus, as the pressure increases, the
values of DG
r
for attachment to CO
2
, pyrimidine, and pyrazine decrease. The equilibria in
supercritical ethane could only be observed over the pressure ranges indicated in the gure
for these solutes.
The volume changes, DV
r
, for attachment to solutes in supercritical ethane can be
extremely large and they are always negative. For pyrazine as a solute, values of DV
r
range
from 1.0 to 45 L/mol, depending on temperature and pressure. The largest changes are
found at the densities where the compressibility is the largest at each temperature. These
changes are mostly due to electrostriction of the solvent around the ion formed and are
predicted quite well by the compressible continuum model. Generally, the density
augmentation around an ion in a supercritical uid extends to a radial distance of 1 nm.
5.2. Attachment Rates
In nonpolar liquids, bimolecular electron attachment rate constants, k
a
, are much larger
than those for conventional reactions of ions or radicals. This is, in part, related to the
high mobility of electrons in these liquids; but various other factors, like V
0
, the kinetic
energy of the electron, and dipole moment of the solute, are important as well. These and
other factors are examined below; the dependence of k
a
on the energy gap, DG
r
, in
representative liquids is also shown and discussed.
It is natural to conclude that the high rate constants for electron attachment
reactions in nonpolar liquids are associated with the high mobility of electrons. Early
studies [96,104,105] of attachment to biphenyl and SF
6
emphasized the dependence of k
a
on mobility. This relationship is apparent if the expression for the rate constant for a
diusion-controlled reaction:
k
D
4pR
e
D
e
13
is combined with the Einstein relation:
D
e
l
D
k
B
T=e 14
then:
k
D
4pR
e
l
D
k
B
T=e 15
Copyright 2004 by Taylor & Francis Group, LLC
In the solvents ethane, propane, and hexane, a reaction radius, R
e
, of 1.4 nm was found for
attachment to SF
6
. This is close to the theoretical maximum radius for electron attach-
ment in the gas phase [106].
Modern theory suggests that rates of electron transfer reactions depend on the poten-
tial energy dierence [107]. For electron attachment reactions in solution, that dierence
is given by the dierence in energy of reactants, in this case, the electron plus solute, and the
product anion [see Eq. (9)]. Since this theory works so well for electron transfer, it is inter-
esting to examine the dependence of rates on DG
r
(l ) for electron attachment reactions; such
a plot for cyclohexane as solvent is shown by Fig. 6. For one of the solutes, diuoroben-
zene, DG
r
(l ) is known fromequilibriumstudies. For the other solutes, DG
r
(l ) was calculated
using Eq. (9). Values of rate constants and electron anities are fromreferences given in the
gure legend. The continuum model of Born:
E
pol
S
e
2
2R
S
1
1
e
16
was used to calculate E
pol
(S
) to increase thus narrowing the energy gap. The eect of increasing pressure on
this reaction is of interest since increasing pressure causes the density to increase, causing
the energy levels of the electron to increase and the value of E
pol
(O
2
) to decrease. Thus
the energy gap increases with increasing pressure. Preliminary studies show that the rate
of attachment to O
2
in n-pentane as solvent decreases a factor of 2 in 1 kbar [132]. The
attachment rate to O
2
has also been studied in liquid argon and xenon [127]. The rate
decreases with increasing eld or, since the eld increases the average electron energy, the
Figure 8 Rate constants for electron attachment to solutes in tetramethylsilane at 295 K. For
solutes numbered 128, see legends to Figs. 6 and 7. Other solutes: 29duroquinone, 30CH
3
I,
31cycloC
4
F
4
, 32C
2
HCl
3
, 33phenanthrene. Dotted line is calculated diusion rate. References
for rate data: [18,19,58,90,93,122,123,125,126]; references for electron anities: [112115].
Copyright 2004 by Taylor & Francis Group, LLC
rate decreases as the kinetic energy of the electrons increases. Thus at higher elds, there
is more energy available for the reaction and the rate slows down.
There is a class of solutes, including ethylbromide, N
2
O, C
2
HCl
3
, and certain
uoroalkanes, that shows negative activation energies for attachment in liquids like
TMS and neopentane [19,123]. For ethylbromide, the rate constant of electron attachment
in TMS is 4.2 10
10
at 23jC. This is point 14 in Fig. 8. Lowering the temperature causes
the energy level of the electron, V
0
, to increase and the polarization energy of the anion
E
pol
(S
) to decrease. Thus lowering the temperature increases the energy gap between the
electron energy level and the energy of the ion in solution and the rate increases. For
neopentane as solvent, the eect is similar except that the electron energy level is higher
making the gap larger, and, at 24jC, k
a
is 3.4 10
11
M
1
sec
1
, an order of magnitude
larger than in TMS. The rate reaches a maximum in 2,2,4-trimethylpentane where k
a
is
6.3 10
12
M
1
sec
1
at 23jC (point 14 in Fig. 7) because the electron level is even higher.
Thus the rate increases as the energy gap increases. In the gas phase, this reaction shows a
maximum in the attachment rate for an electron energy of 0.76 eV [133]. This value was
used for DG
r
( g) in calculating DG
r
(l). In TMS, DG
r
(l) turns out to be +0.32 eV, or this
vertical attachment reaction is unfavorable. In 2,2,4-TMP, DG
r
(l) is close to zero or, in
terms of the electron redox level picture of Henglein [134], the occupied donor level
(electron in the solvent) matches the unoccupied acceptor level (ethyl bromide). But the
rate constant is less in cyclohexane, 2 10
12
M
1
sec
1
at 23jC, where the electron level is
even higher. This is explained by the fact that the reaction is limited by the diusion rate,
which should be somewhat less than the line in Fig. 6 because of a smaller reaction radius
for this solute.
The explanation of the negative activation energies for attachment to N
2
O, C
2
HCl
3
[19], and peruorocyclobutane [123] in TMS is similar to that for ethylbromide above.
These molecules exhibit maxima for electron attachment in the gas phase at 2.2 eV [135],
0.4 eV [136], and 0.35 eV [137], respectively. Thus rates are expected to increase as the
energy gap increases. However, in 2,2,4-TMP and n-hexane, the attachment rates show
normal Arrhenius behavior. For these solvents, the ratio k
a
/l
D
is constant over a range of
temperatures indicating that the reaction becomes diusion-limited as was the case for
ethylbromide in cyclohexane. Other peruoroalkanes like n-C
5
F
12
and n-C
6
F
14
are
reported to show maxima as is seen for cyclo-C
4
F
8
[138].
Carbon dioxide reacts in a manner similar to the solutes discussed above in that as
the energy gap increases, the rate of attachment increases. However, unlike the other
solutes, this reaction is reversible in solution and the equilibrium:
e
CO
2
VCO
2
19
has been studied in several nonpolar liquids as well as in supercritical ethane. Thus not
only the attachment rates, k
a
, have been measured, but also values of the free energy of
reaction, DG
r
, are available. The rate constants, k
a
, for this reaction, obtained at various
temperatures and pressures in several uids: TMS, 2,2,4-trimethylpentane, dimethylbu-
tane, and supercritical ethane [126,139], are plotted in Fig. 9. In liquids, the rate constant is
independent of temperature at any pressure but increases with pressure. The increase in
rate with pressure can be explained by the increase in electron energy level [39] and the
increase in stability of the ion with pressure that causes the energy gap to increase. As
shown in Fig. 9, log k
a
is linearly dependent on DG
r
and k
a
increases about 3 orders of
magnitude as the energy gap increases by 0.4 eV.
The reaction of electrons with p-benzoquinone is an unusual case. A conductivity
study showed that the reaction is slow at room temperature in TMS and neopentane and
Copyright 2004 by Taylor & Francis Group, LLC
the activation energies are negative [140]. The energy gaps are large: DG
r
(l) is 2.3 eV
in TMS and 2.44 eV in neopentane, indicating possible inverted behavior. However,
in 2,2,4-TMP, DG
r
(l) is even lower, 2.58 eV, and the rate of attachment is fast. In
cyclohexane, n-pentane, and n-hexane, where the electron levels are high, the reaction
appears to be diusion-limited. The results were explained by assuming an equilibrium
with an excited state of the anion:
e
benzoquinone Vbenzoquinone
* 20
benzoquinone
* Ybenzoquinone
21
The excited state lies 2.07 eV above the ground state; thus there is sucient energy
available in all solvents to reach this state. The kinetic analysis indicated that the
attachment rate is fast in all solvents. In TMS, neopentane, and 2,2,4,4-tetramethylpen-
tane, the energy levels of the electron are low enough that the reverse reaction, autodetach-
ment from the excited anion, occurs with small activation energy.
Recently, the excited state of benzoquinone anion was detected both by absorption
and uorescence [141]. The lifetime of the excited state in 2,2,4-trimethylpentane is c120
nsec or k
21
= 8 10
6
sec
1
. If it is assumed that the lifetime is similar in the other
nonpolar solvents, then the results of the two studies may be combined to evaluate the
detachment rate from the excited anion. At room temperature, this calculation gives
k
20
= 6 10
8
sec
1
in TMS; thus detachment readily competes with deactivation of
the excited state of benzoquinone anion. The latter process includes internal conversion as
well as uorescence. In 2,2,4,4-tetramethylpentane, where the activation energy for detach-
ment is 0.41 eV, k
20
is 4 10
6
sec
1
and some autodetachment occurs along with uo-
rescence. In solvents like n-pentane, n-hexane, and cyclohexane, the activation energies for
Figure 9 Rate constants for electron attachment to CO
2
vs. the free energy of reaction in dierent
uids: w2,2,4-trimethylpentane [126], .2,2-dimethylbutane [139], oTMS [126], nsuper-
critical ethane [99].
Copyright 2004 by Taylor & Francis Group, LLC
detachment are much higher because the electron energy levels are higher, and the reaction
proceeds directly to benzoquinone anion at the diusion rate.
For attachment to CO
2
, the rate constant clearly shows a dependence on DG
r
(l ).
However, for attachment to other solutes like aromatics, the dependence on DG
r
(l ) is not
yet clear. Questions remain like what is the role of reorganization energy, which can be
large even in nonpolar solvents [142]; it was pointed out earlier that there is considerable
density augmentation around the negative ions formed in attachment reactions. For
solvents like TMS, rates maximize for values of DG
r
(l ) near 0.7 eV; what rate is to be
expected for large values of DG
r
(l )? Is the high rate observed for C
60
typical or the low
rate for quinones more typical? Further study is needed to resolve these questions.
6. TRANSPORT PROPERTIES
6.1. Quasi-Free Mobility
The drift mobility of electrons in nonpolar liquids ranges from high values such as that for
liquid xenon of 2000 cm
2
/Vs to low values like that for tetradecane of 0.02 cm
2
/Vs. It has
often been suggested that the mobility is high for symmetrical molecules and low for
straight chain molecules like n-alkanes. Inspection of Table 2 shows that liquids with
symmetrical molecules are indeed at the top of the list. However, other less symmetrical
molecules like bis-trimethylsilylmethane and 2,2,4,4-tetramethylpentane also show high
drift mobility. A more important factor may be the existence of many methyl groups in the
molecule. In any case, for liquids for which l
D
>10 cm
2
/Vs, the electron is considered to
be quasi-free. This is supported by the Hall mobility studies, as discussed below.
The mobility of quasi-free electrons has recently been explained by the deformation
potential theory. Originally from solid-state physics, this theory was applied by Basak and
Cohen [68] to liquid argon. The theory assumes that scattering occurs when the electron
encounters a change or uctuation in the local density which results in a potential change.
The potential is assumed to be given in terms of dV
0
/dN, d
2
V
0
/dN
2
, etc. The formula they
derived for the mobility is:
l
D
2et
4
2p
p
m
e
=m*
5=2
3m
5=2
e
k
B
T
1=2
k
B
N
2
V
V2
0
Tv
T
V
W2
0
b=k
B
. . .
; 22
where v
T
is the isothermal compressibility and m* is the eective mass of the electron. A
similar expression was derived by Berlin et al. [144]. Equation (22) was later shown to
account for many of the features of the density dependence of the mobility in Ar, Kr, and
Xe [66]. Namely, it predicts a minimum in the mobility near the critical density and it
correctly predicts a maximum in the mobility at the density at which V
0
is a minimum, i.e.,
when dV
0
/dN = 0 (compare Figs. 3 and 10). Scattering in these uids is weakest at this
density. At higher densities, a decrease in mobility is predicted as is observed experimen-
tally. Although the general features of the experimental mobility are reproduced, the
predicted values of l
D
around the minimum near N
C
are much too small largely because
v
T
for the uids are very large in this region. Arguing that electronmedium interactions
are relatively short-ranged, Nishikawa [147] suggested that v
T
be replaced by the adiabatic
compressibility. This greatly improved the agreement at these densities for Ar.
Since these papers were published, accurate values of V
0
in argon have been obtained
by the eld ionization technique [60,61] as described in Sec. 3. Also, the eective mass, m*,
Copyright 2004 by Taylor & Francis Group, LLC
of the electron in argon is now available from theory [148] (see Sec. 6.5). Previous workers
had taken m* = m
e
. As a test of Eq. (22), the mobility in argon was recalculated here using
these new data. The adiabatic compressibility was used and the value of b was obtained by
making both sides of Eq. (22) equal at the density at which V
0
is a minimum. Fig. 10 shows
that this calculated mobility compares well with the experimental data of Jahnke et al.
[145]. This theory has also been shown to predict quite well the density dependence of the
mobility in xenon [149].
This theory has also been used to predict mobility for molecular liquids. Neopentane
and TMS are liquids that exhibit maxima in the electron mobility at intermediate densities
[46]. These maxima occur at the same densities at which V
0
minimizes, in accordance with
the BasakCohen theory. The drift mobility in TMS has been measured as a function of
pressure to 2500 bar [150]. The observed relative experimental changes of mobility with
pressure are predicted quite well by the BasakCohen theory; however, the predicted value
of l
D
is 2.5 times the experimental value at 1 bar and 295 K. In this calculation, the
authors used v
T
to evaluate the mobility. This is reasonable in this case since for liquids,
there is little dierence between the adiabatic and isothermal compressibilities. A similar
calculation for neopentane showed that the BasakCohen theory predicted the Hall
mobility of the electron quite well for temperatures between 295 and 400 K [151]. Itoh
Table 2 Free Ion Yields and Mobilities at 293295 K
No. Liquid G
(electrons/100 eV) l
D
(cm
2
/Vs)
1 Tetramethylsilane 0.7 100
2 Tetramethylgermane 0.63 90
3 Neopentane 1.1 69
4 Bis-trimethylsilylmethane 0.5 63
5 Bis-trimethylsilylethane 0.41 47
6 Ethane 0.94 37
7 2,2,4,4-Tetramethylpentane 0.74 26
8 Hexamethyldisiloxane 0.30 22
9 Hexamethyldisilane 0.33 20
10 2,2-Dimethylbutane 0.58 12
11 2,2,5,5-Tetramethylhexane 0.67 12
12 2,2,4-Trimethylpentane 0.33 6.6
13 2,2,3,3-Tetramethylpentane 0.42 5.2
14 Polydimethylsiloxane 0.24 4.6
15 Propane 0.19 2.6
16 Cyclopentane 0.16 1.1
17 2-Methylbutane 0.17 0.93
18 3,3-Diethylpentane 0.23 0.76
19 2-Methylpentane 0.15 0.29
20 Cyclohexane 0.15 0.24
21 3-Methylpentane 0.15 0.20
22 n-Pentane 0.15 0.14
23 n-Hexane 0.13 0.074
24 Benzene 0.054 0.13
25 Toluene 0.051 0.08
26 Methylcyclohexane 0.12 0.068
Source: Refs. 2, 1519, and 143.
Copyright 2004 by Taylor & Francis Group, LLC
et al. [152] extended the theory to mixtures taking concentration uctuations into account,
but found that the data for TMSneopentane mixtures were best t if only density
uctuations were considered.
Other theories have been proposed in recent years to account for the mobility of
quasi-free electrons. Borghesani et al. [153] studied the eld dependence of the mobility in
ArKr and ArXe mixtures and found that the results disagreed with the deformation
potential model. Instead, the results could be described by a gas-kinetic model if
concentration-dependent scattering cross sections were used. Atrazhev and Iakubov
[154] used pseudopotential theory and similarly derived density-dependent scattering
lengths, which varied from negative values at low density to positive values at high
density. Their theory qualitatively describes the density dependence of the mobility in Ar,
Kr, and Xe. Naveh and Laikhtman [155] introduced scattering from longitudinal acoustic
phonons into the deformation potential framework. Their theory gave excellent agreement
with experimental results for Ar for densities from 0.8 to 11.2 10
21
cm
3
. A dierent
approach was used by Hsu and Chandler [146] who used Feynmans polaron theory [156]
for the mobility and a pseudopotential consisting of attractive and repulsive parts. These
two terms counteract at some intermediate density and result in a peak in the mobility.
Their calculation for Ar is shown as dashed line in Fig. 10.
Figure 10 Drift mobility in argon as a function of density. .: Experimental results at 55 bar
[145]. : Calculation using modied BasakCohen (see text). - - -: Calculation by Hsu and Chandler
[146].
Copyright 2004 by Taylor & Francis Group, LLC
The deformation potential model seems to provide a suitable framework to under-
stand the quasi-free electron mobility in nonpolar liquids. Already several extensions or
modications on this theory have been proposed, and the dependence on temperature and
pressure seems to be adequately explained. However, several authors have taken dierent
approaches to the problem showing that a consensus in our understanding has not yet
been reached.
6.2. Trapped State Transport
For many nonpolar liquids, the electron drift mobility is less than 10 cm
2
/Vs, too low to be
accounted for in terms of a scattering mechanism. In these liquids, electrons are trapped as
discussed in Sec. 4. Considerable evidence now supports the idea of a two-state model in
which equilibrium exists between the trapped and quasi-free states:
e
qf
V e
tr
23
The magnitude of the mobility then depends on the value of the quasi-free mobility in such
liquids multiplied by the fraction of time the electron is quasi-free since the trapped
electron is relatively immobile. Thus:
l
D
l
qf
e
qf
e
qf
e
tr
l
qf
1
1 K
23
24
This equilibrium depends on many factors like V
0
, DG
soln
(e
ionization cross-
section spectra [172]. This procedure gave m* =0.26m
e
for argon at 87 Kand m* =0.27m
e
for TMS, 2,2-dimethylbutane, and 2,2,4-trimethylpentane at 296 K. From exciton spectra,
Reininger et al. [173] determined m* for argon to be 0.55m
e
at the triple point. In a similar
way, Reshotko et al. [174] determined m* to be 0.28m
e
for Xe at several densities near N
t
.
Copyright 2004 by Taylor & Francis Group, LLC
Recently, the eective mass of the electron has been calculated [148] within a
WignerSeitz framework [175] for Ar, Kr, and Xe. In all three liquids, m* decreases
with increasing density. At the triple point densities, m* = 0.6m
e
in argon and m* = 0.3m
e
in xenon.
6.6. Fast Negative Ions
In some liquids, the electron is trapped as a solvent negative ion, yet transport occurs
much faster than expected for an ion. A well-documented example of this type of
electronic transport is found in liquid peruorobenzene where the electron attaches to
form C
6
F
6
A V A
26
to shift to the right. The mobility observed at high pressure is 0.06 cm
2
/Vs for toluene
[179], V0.08 cm
2
/Vs for benzene [179], and 0.06 and 0.04 cm
2
/Vs for m-xylene and o-
xylene, respectively [159]. Transport at high pressure is believed to occur by hopping:
A
A !AA
27
Hopping occurs between neighboring molecules and the activation energies are small,
from 0.12 to 0.15 eV for the hydrocarbons. Fast anions have also been suggested for liquid
SF
6
[180]. Thus several examples of this type of transport for the negative charge exist
and the mobilities are comparable. More examples of fast ions are expected to be found
in the future.
6.7. Positron Mobility
It is interesting to compare the properties of positive electrons, positrons, with the
properties of electrons in nonpolar liquids. Values of the mobility of positrons, l
+
, are
now available for a few liquids. Early measurements for l
+
in n-hexane ranged from 8.5 to
100 cm
2
/Vs [181,182]. In a recent study, the Doppler shift in energy of the 511-keV
annihilation gamma ray in an electric eld was utilized to measure the drift velocity. This
method led to l
+
= 53 cm
2
/Vs in n-hexane and 69 cm
2
/Vs in 2,2,4-trimethylpentane [183].
Interestingly, these values are comparable to the mobilities of quasi-free electrons in
nonpolar liquids.
Copyright 2004 by Taylor & Francis Group, LLC
7. APPLICATIONS
This chapter would not be complete without mentioning the applications of electrons in
nonpolar liquids. Scientists who are very interested in the properties of electrons in these
liquids are the detector physicists, working in high-energy physics, gamma ray astronomy,
cosmic radiation, or positron emission tomography [184]. Ionization chambers are used in
these elds to detect particles by the current signals induced by the excess electrons
produced. Ideal liquids for such applications would have high free ion yields, high drift
velocities, and high density. Liquid purity is important as well so that the electrons live
long enough to reach the electrode.
There are a variety of detectors currently in use to detect particles from neutrinos to
weakly interacting massive particles (WIMPS). Table 3 is a partial list of some of these
that are currently in use or under construction. As noted, some utilize multiton quantities
of liquid. The rst such detector was a sampling calorimeter containing 300 L of liquid
argon [185]. Many of the detectors in use today in high-energy physics experiments are of
this type. They are calorimeters because the particles are totally stopped within the
detector and the energy of the particles can be determined by the ionization produced in
the liquid. Heavy metal plates, each a few millimeters thick, are introduced and the
ionization is sampled in the liquid between the plates. An electromagnetic calorimeter
detects electrons and gamma rays. The latter create electronpositron pairs; these in turn
produce lower-energy gammas that create more pairs, etc. This sequence of events is called
an electromagnetic shower. The D0 detector at Femilab is an example that has been in
operation since 1991 [186]. Calorimeters that record signals from hadrons (protons,
neutrons, kaons, pions, etc.) are often incorporated in such detectors. These utilize thicker
metal plates. Calorimeters have necessarily grown as the energy of the particles in high-
energy physics experiments has increased. For example, the A Torcidal Large Hadron
Collider Apparatus (ATLAS) detector at European Center for Nuclear Research (CERN)
is a 200-kt detector. Specic details about these detectors are given in the cited references
Table 3 Ionization Detectors
Name Liquid Location Type Detects Year
D0 42,000
L Ar
Fermilab EM and hadronic
sampling
calorimeter
Electrons, photons,
and hadrons
1991
Walic TMP UA1-CERN EM sampling
calorimeter
1988
210 L
TMS
Karlsruhe Calorimeter Cosmic rays 1994
KEDR 24 t Kr Novosibirsk EM calorimeter 1996
ICARUS 600 t Ar Gran Sasso
Italy
Time projection
chamber
Neutrinos p decay 2001
ATLAS 140 t Ar CERN EM and hadron
calorimeters
Electrons,
photons, hadrons,
and Higgs boson
2007
LXeGRIT 7 L Xe Balloon Compton telescope Gamma rays 1999
DAMA/Xe-2 2 L Xe Gran Sasso Scintillator WIMPS 1998
Copyright 2004 by Taylor & Francis Group, LLC
[187]. To avoid the necessity of a cryogenic container, some calorimeters have utilized
molecular liquids like TMS [188] and 2,2,4,4-tetramethylpentane [189].
Another type of liquid detector is the time projection chamber. Examples are Imag-
ing Cosmic and Rare Underground Signals (ICARUS), a large liquid Ar detector [190],
and LXeGRIT (gamma ray imaging telescope) [191]. These detectors determine both the
energy and the direction of the incoming particle or photon [192]. Liquid xenon detectors
are usually smaller due to the high cost of xenon. For the purposes of gamma-ray as-
tronomy, the scintillation of the xenon provides a trigger for an event and the direction of
the particle is obtained by measuring the time required for electrons to drift to the collecting
electrodes. Balloon ights by E. Apriles group from Columbia have demonstrated the fea-
sibility of such gamma-ray telescopes [191].
ACKNOWLEDGMENTS
The preparation of this chapter was done at Brookhaven National Laboratory and sup-
ported under contract DE-AC02-98-CH10886 with U.S. Department of Energy and sup-
ported by its Division of Chemical Sciences, Oce of Basic Energy Sciences.
REFERENCES
1. Freeman, G.R. Kinetics of Nonhomogeneous Processes; Freeman, R., Ed.;. Wiley: New York,
1987; 63 pp.
2. Nishikawa, M. Handbook of Radiation Chemistry; Tabata, Y. Ito, Y., Tagawa, S., Eds.;. CRC
Press: Boca Raton, 1991; p 395.
3. Schmidt, W.F. Liquid State Electronics of Insulating Liquids; CRC Press: Boca Raton, 1997.
4. Mozumder, A. Fundamentals of Radiation Chemistry; Academic Press: San Diego, 1999.
5. Onsager, L. Phys. Rev. 1938, 54, 554.
6. Schmidt, W.F. Radiat. Res. 1970, 42, 73.
7. Hoshi, Y.; Higuchi, M.; Iso, H.; Sakamoto, M.; Ooyama, K.; Yuta, H.; Abe, K.; Hasegawa,
K.; Suekane, F.; Kawamura, N.; Neichi, M.; Suzuki, K.; Masuda, M.; Kikuchi, R.; Miyano,
K. IEEE Trans. Nucl. Sci. 1993, 40, 532.
8. Huang, S.S.-S.; Freeman, G.R. Can. J. Chem. 1977, 55, 1838.
9. Warman, J.M.; Asmus, K.D.; Schuler, R.H. Advances in Chemistry Series 82; Gould, R.F.,
Ed.; Washington: American Chemical Society, 1968; Vol. 82, p 25.
10. Jay-Gerin, J.-P. Can. J. Chem. 1993, 71, 287.
11. Geer, S.; Holroyd, R.A. Phys. Rev. B 1992, 46, 5043.
12. LeMotais, B.C.; Jonah, C.D. Radiat. Phys. Chem. 1989, 33, 505.
13. Saeki, A.; Kozawa, T.; Yoshida, Y.; Tagawa, S. Radiat. Phys. Chem. 2001, 60, 319.
14. Bartczak, W.M.; Hummel, A. Chem. Phys. Lett. 1993, 208, 232.
15. Holroyd, R.A.; Itoh, K.; Nishikawa, M. Nucl. Instrum. Methods Phys. Res. A 1997, 390, 233.
16. Gee, N.; Senanayake, C.; Freeman, G.R. J. Chem. Phys. 1988, 89, 3710.
17. Schmidt, W.F.; Allen, A.O. J. Chem. Phys. 1970, 52, 2345.
18. Allen, A.O.; Holroyd, R.A. J. Phys. Chem. 1974, 78, 796.
19. Allen, A.O.; Gangwer, T.E.; Holroyd, R.A. J. Phys. Chem. 1975, 79, 25.
20. Nishikawa, M.; Yamaguchi, Y.; Fujita, K. J. Chem. Phys. 1974, 61, 2356.
21. Yamaguchi, Y.; Nishikawa, M. J. Chem. Phys. 1973, 59, 1298.
22. Holroyd, R.A.; Chen, P.Y.; Stradowska, E.; Itoh, K. Radiat. Phys. Chem. 1996, 48, 635.
23. Holroyd, R.A.; Sham, T.K. J. Phys. Chem. 1985, 88, 2909.
24. Holroyd, R.A.; Sham, T.K.; Yang, B.X.; Feng, X.H. J. Phys. Chem. 1992, 96, 7438.
Copyright 2004 by Taylor & Francis Group, LLC
25. Holroyd, R.A.; Sham, T.K. Radiat. Phys. Chem. 1998, 51, 37.
26. Bartczak, W.M.; Hummel, A. Radiat. Phys. Chem. 1997, 49, 675.
27. Siebbeles, L.D.A.; Bartczak, W.M.; Terrissol, M.; Hummel, A. J. Phys. Chem. 1997, 101, 1619.
28. Holroyd, R.A.; Preses, J.M. Chemical Applications of Synchrotron Radiation; Sham, T.-K.,
Ed.; River Edge, NJ: World Scientic, 2002; Vol. 12B, p 987.
29. Bartczak, W.M.; Hummel, A. J. Phys. Chem. 1993, 97, 1253.
30. Chybicki, M. Acta Phys. Pol. 1966, 30, 927.
31. Munoz, R.C.; Cumming, J.B.; Holroyd, R.A. J. Chem. Phys. 1986, 85, 1104.
32. Lindblad, T.; Bagge, L.; Engstrom, A.; Bialkowski, J.; Gruhn, C.R.; Pang, W.; Roach, M.;
Loveman, R. Nucl. Instrum. Methods 1983, 215, 183.
33. Mozumder, A. J. Phys. Chem. 1996, 100, 5964.
34. Holroyd, R.A.; Nishikawa, M. Radiat. Phys. Chem. 2002, 64, 19.
35. Balakin, A.A.; Boriev, I.A.; Yakovlev, B.S. Can. J. Chem. 1977, 55, 1985.
36. Anderson, D.F.; Charpak, G.; Holroyd, R.A.; Lamb, D.C. Nucl. Instrum. Methods Phys.
Res. 1987, A261, 445.
37. Springett, B.E.; Jortner, J.; Cohen, M.H. J. Chem. Phys. 1968, 48, 2720.
38. Holroyd, R.A.; Tames, S.; Kennedy, A. J. Phys. Chem. 1975, 79, 2857.
39. Holroyd, R.A.; Nishikawa, M.; Nakagawa, K.; Kato, N. Phys. Rev. B 1992, 45, 3215.
40. Halpern, B.; Lekner, J.; Rice, S.A.; Gomer, R. Phys. Rev. 1967, 156, 351.
41. Holroyd, R.A.; Allen, M. J. Chem. Phys. 1971, 54, 5014.
42. Holroyd, R.A.; Tauchert, W. J. Chem. Phys. 1974, 60, 3715.
43. Schiller, R.; Vass, S.; Mandics, J. J. Radiat. Phys. Chem. 1973, 5, 491.
44. Yamaguchi, Y.; Nakajima, T.; Nishikawa, M. J. Chem. Phys. 1979, 72, 550.
45. Nakagawa, K.; Ohtake, K.; Nishikawa, M. J. Electrostat. 1982, 12, 157.
46. Holroyd, R.A.; Cipollini, N.E. J. Chem. Phys. 1978, 69, 501.
47. Boriev, A.; Balakin, A.A.; Yakovlev, B.S. Khim. Vys. Energ. 1978, 12, 20.
48. Borghesani, A.F.; Carugno, G.; Santini, M. IEEE Trans. Electr. Insul. 1991, 26, 615.
49. Holroyd, R.A.; Russell, R.L. J. Phys. Chem. 1974, 78, 2128.
50. Holroyd, R.A.; Dietrich, B.K.; Schwarz, H.A. J. Phys. Chem. 1972, 76, 3794.
51. Nakato, Y.; Tsubomura, H. J. Phys. Chem. 1975, 79, 2135.
52. Bottcher, E.-H.; Schmidt, W.F. J. Chem. Phys. 1984, 80, 1353.
53. Holroyd, R.A.; Preses, J.M.; Bottcher, E.H.; Schmidt, W.F. J. Phys. Chem. 1984, 88, 744.
54. Casanovas, J.; Grob, R.; Sabattier, R.; Gruelfucci, J.P.; Blanc, D. Radiat. Phys. Chem. 1980,
15, 293.
55. Casanovas, J.; Grob, R.; Delacroix, D.; Guelfucci, J.P.; Blanc, D. J. Chem. Phys. 1981, 75,
4661.
56. Greever, J.S.; Turner, J.B.M.; Kaumann, J.F. J. Phys. Chem. A 2001, 105, 8635.
57. Sowada, U.; Holroyd, R.A. J. Phys. Chem. 1980, 84, 1150.
58. Sowada, U.; Holroyd, R.A. J. Phys. Chem. 1981, 85, 541.
59. Lukin, L.V.; Yakovlev, B.S. Chem. Phys. Lett. 1976, 42, 307.
60. Al-Omari, A.K.; Reininger, R. J. Chem. Phys. 1995, 103, 506.
61. Al-Omari, A.K.; Altmann, K.N.; Reininger, R. J. Chem. Phys. 1996, 105, 1305.
62. Al-Omari, A.K.; Reininger, R. J. Chem. Phys. 1995, 103, 4484.
63. Altmann, K.N.; Reininger, R. J. Chem. Phys. 1997, 107, 1759.
64. Stampi, P.; Bennemann, K.H. Phys. Rev. A 1991, 44, 8210.
65. Plenkiewicz, B.; Frongillo, Y.; Lopez-Castillo, J.-M.; Jay-Gerin, J.-P. J. Chem. Phys. 1996,
104, 9053.
66. Reininger, R.; Asaf, U.; Steinberger, I.T.; Basak, S. Phys. Rev. B 1983, 28, 4426.
67. Jacobsen, F.M.; Gee, N.; Freeman, G.R. Phys. Rev. A 1986, 34, 2329.
68. Basak, S.; Cohen, M.H. Phys. Rev. B 1979, 20, 3404.
69. Fermi, E. Nuovo Cim. 1934, 11, 157.
70. Lopez-Castillo, J.-M.; Frongillo, Y.; Plenkiewicz, B.; Jay-Gerin, J.-P. Chem. Phys. 1992, 96,
9092.
Copyright 2004 by Taylor & Francis Group, LLC
71. Lopez-Castillo, J.-M.; Jay-Gerin, J.-P. Phys. Rev. E 1995, 52, 4892.
72. Space, B.; Coker, D.F.; Liu, Z.H.; Berne, B.; Martyna, G. J. Chem. Phys. 1992, 97, 2002.
73. Boltjes, B.; deGraaf, C.; Leeuw, S.W. J. Chem. Phys. 1993, 98, 592.
74. Simon, S.H.; Dobrosavljevic, V.; Stratt, R.M. Phys. Rev. A 1990, 42, 6278.
75. Simon, S.H.; Dobrosavljevic, V.; Stratt, R.M. J. Chem. Phys. 1991, 94, 7360.
76. Frongillo, Y.; Plenkiewicz, B.; Jay-Gerin, J.-P.; Jain, A. Phys. Rev. E 1994, 50, 4754.
77. Baxendale, J.H.; Bell, C.; Wardman, P. J. Chem. Soc., Faraday Trans. I 1973, 69, 776.
78. Atherton, S.J.; Baxendale, J.H.; Busi, F.; Kovacs, A. Radiat. Phys. Chem. 1986, 28, 183.
79. Gillis, H.A.; Klassen, N.V.; Woods, R.J. Can. J. Chem. 1977, 55, 2022.
80. McGrane, S.D.; Lipsky, S. J. Phys. Chem. 2001, 105, 2384.
81. Nakamura, Y.; Shinsaka, K.; Hatano, Y. J. Chem. Phys. 1983, 78, 5820.
82. Gillis, H.A.; Klassen, N.V.; Teather, G.G.; Lokan, K.H. Chem. Phys. Lett. 1971, 10, 481.
83. Siebbeles, L.D.A.; Emmerichs, U.; Hummel, A.; Bakker, H.J. J. Chem. Phys. 1997, 107, 9339.
84. Ichikawa, T.; Yoshida, H. J. Chem. Phys. 1980, 73, 1540.
85. Kimura, T.; Ogawa, N.; Fueki, K. Bull. Chem. Soc. Jpn. 1981, 54, 3854.
86. Wigner, E.P. Phys. Rev. 1948, 73, 1002.
87. Kevan, L. Radiat. Phys. Chem. 1981, 17, 413.
88. Itoh, K.; Holroyd, R. J. Phys. Chem. 1990, 94, 8854.
89. Itoh, K.; Nishikawa, M.; Holroyd, R. J. Phys. Chem. 1993, 97, 503.
90. Holroyd, R.A. Ber. Bunsenges. Phys. Chem. 1977, 81, 28.
91. Sowada, U.; Holroyd, R.A. Private communication.
92. Chen, P.Y.; Holroyd, R.A. J. Phys. Chem. 1995, 99, 14528.
93. Chen, P.Y.; Holroyd, R.A. J. Phys. Chem. 1996, 100, 4491.
94. Holroyd, R.A.; Gangwer, T.E.; Allen, A.O. Chem. Phys. Lett. 1975, 31, 520.
95. Holroyd, R.A.; McCreary, R.D.; Bakale, G. J. Phys. Chem. 1979, 83, 435.
96. Warman, J.M.; deHaas, M.P.; Zador, E.; Hummel, A. Chem. Phys. Lett. 1975, 35, 383.
97. Holroyd, R.A.; Nishikawa, M.; Itoh, K. J. Phys. Chem. B 1999, 103, 9205.
98. Holroyd, R.A.; Nishikawa, M.; Itoh, K. J. Phys. Chem. B 2000, 104, 11585.
99. Nishikawa, M.; Itoh, K.; Holroyd, R.A. J. Phys. Chem. A 1999, 103, 550.
100. Holroyd, R.A.; Itoh, K.; Nishikawa, M. Chem. Phys. Lett. 1997, 266, 227.
101. Hamann, S.D. Physico-Chemical Eects of Pressure. London: Butterworths, 1957.
102. Schwarz, H.A. J. Phys. Chem. 1993, 97, 12954.
103. Itoh, K.; Holroyd, R.A.; Nishikawa, M. J. Phys. Chem. A 2001, 105, 703.
104. Beck, G.; Thomas, J.K. Chem. Phys. Lett. 1972, 13, 295.
105. Bakale, G.; Schmidt, W.F. Z. Naturforsch. 1981, 36a, 802.
106. Warman, J.M. The Study of Fast Processes and Transient Species by Electron Pulse Radiolysis;
Baxendale, J.H., Busi, F., Eds.; Reidel: Dordrecht, Holland, 1982; 433.
107. Tachiya, M. J. Phys. Chem. 1993, 97, 5911.
108. Sauer, M.C.; Schmidt, K.H. Radiat. Phys. Chem. 1994, 43, 413.
109. Bakale, G.; McCreary, R.D.; Gregg, E.C. Cancer Biochem. Biophys. 1981, 5, 103.
110. Bakale, G.; Gregg, E.C.; McCreary, R.D. J. Chem. Phys. 1977, 67, 5788.
111. Baxendale, J.H.; Keene, J.P.; Rasburn, E.J. J. Chem. Soc., Faraday Trans 1974, 70, 718.
112. Chen, E.S.; Chen, E.C.M.; Sane, N.; Talley, L.; Kozanecki, N.; Shulze, S. J. Chem. Phys.
1999, 110, 9319.
113. Christodoulides, A.A.; McCorkle, D.L.; Christophorou, L.G. Report No. DOE/EV/04703-39.
114. Gains, A.F.; Kay, J.; Page, F.M. Trans. Faraday Soc. 1966, 62, 874.
115. Kebarle, P.; Chowdhury, S. Chem. Rev. 1987, 87, 513.
116. Christophorou, L.G. Z. Phys. Chem. 1996, 195s, 195.
117. Weston, R.E., Jr; Schwarz, H.A. Chemical Kinetics. Prentice-Hall: Englewood Clis, NJ, 1972.
118. Funabashi, K.; Magee, J.L. J. Chem. Phys. 1975, 62, 4428.
119. Bakale, G.; Sowada, U.; Schmidt, W.F. J. Phys. Chem. 1975, 79, 3041.
120. Schmidt, W.F.; Bakale, G.; Sowada, U. J. Chem. Phys. 1974, 61, 5275.
121. Beck, G.; Thomas, J.K. J. Chem. Phys. 1972, 57, 3649.
Copyright 2004 by Taylor & Francis Group, LLC
122. Holroyd, R.A. Radiat. Phys. Chem., submitted.
123. Holroyd, R.A.; Gangwer, T.E. Radiat. Phys. Chem. 1980, 15, 283.
124. Kang, Y.S.; Holroyd, R.A. Radiat. Phys. Chem. 1982, 20, 237.
125. Warman, J.M.; de Haas, M.P.; Hummel, A. Conduction and Breakdown in Dielectric Liquid;
Goldschwartz, J.M., Niessen, A.K., Boone, W., Eds.; Delft Univ. Press: Delft, Netherlands,
1975; 70.
126. Nishikawa, M.; Itoh, K.; Holroyd, R. J. Phys. Chem. 1988, 92, 5262.
127. Bakale, G.; Sowada, U.; Schmidt, W.F. J. Phys. Chem. 1976, 80, 2556.
128. Bakale, G.; Sowada, U.; Schmidt, W.F. J. Phys. Chem. 1975, 79, 3041.
129. Sowada, U.; Bakale, G.; Yoshino, K.; Schmidt, W.F. Chem. Phys. Lett. 1975, 34, 466.
130. Henglein, A. Ber. Bunsenges. Phys. Chem. 1976, 79, 129.
131. Henglein, A. Can. J. Chem. 1976, 55, 2112.
132. Nakagawa, K.; Holroyd, R.A. Private communication.
133. Christophorou, L.G. Atomic and Molecular Radiation Physics; Wiley-Interscience: New York,
1971.
134. Henglein, A. Ber. Bunsenges. Phys. Chem. 1975, 79, 129.
135. Wentworth, W.E.; Chen, E.; Freeman, R. J. Chem. Phys. 1971, 55, 2075.
136. Blaunstein, R.P.; Christophorou, L.G. J. Chem. Phys. 1968, 49, 1526.
137. Christophorou, L.G.; McCorkle, D.L.; Pittman, D. J. Chem. Phys. 1974, 60, 1183.
138. Holroyd, R. The Liquid State and Its Electronic Properties; Kunhardt, E.E., Christophorou,
L.G., Luessen, L.H., Eds.; Plenum Press: New York, 1987; Vol. 193, 221.
139. Ninomiya, S.; Itoh, K.; Nishikawa, M.; Holroyd, R. J. Phys. Chem. 1993, 97, 9488.
140. Holroyd, R.A. J. Phys. Chem. 1982, 86, 3541.
141. Cook, A.R.; Curtiss, L.A.; Miller, J.R. J. Am. Chem. Soc. 1997, 119, 5729.
142. Matyushov, D.V.; Schmid, R. Mol. Phys. 1995, 84, 533.
143. Astbury, A.; Keeler, R.K.; Li, Y.; Poenberger, P.R.; Robertson, L.P. Nucl. Instrum.
Methods Phys. Res. A 1991, 305, 376.
144. Berlin, Y.A.; Nyikos, L.; Schiller, R. J. Chem. Phys. 1978, 69, 2401.
145. Jahnke, J.A.; Meyer, L.; Rice, S.A. Phys. Rev. A 1971, 3, 734.
146. Hsu, D.; Chandler, D. J. Chem. Phys. 1990, 93, 5075.
147. Nishikawa, M. Chem. Phys. Lett. 1985, 114, 271.
148. Plenkiewicz, B.; Frongillo, Y.; Plenkiewicz, P.; Jay-Gerin, J.-P. J. Chem. Phys. 1991, 94, 6132.
149. Holroyd, R.A.; Nishikawa, M.; Itoh, K. J. Chem. Phys. 2003, 118, 706.
150. Munoz, R.C.; Holroyd, R.A. J. Chem. Phys. 1986, 84, 5810.
151. Munoz, R.C.; Ascarelli, G. J. Phys. Chem. 1984, 88, 3712.
152. Itoh, K.; Nishikawa, M.; Holroyd, R.A. Phys. Rev. B 1991, 44, 680.
153. Borghesani, A.F.; Iannuzzi, D.; Carugno, G. J. Phys. 5057, 1997, 9, Condens. Matter 9.
154. Atrazhev, V.M.; Iakubov, I.T. J. Chem. Phys. 1995, 103, 9030.
155. Naveh, Y.; Laikhtman, B. Phys. Rev. B 1993, 47, 3566.
156. Feynman, R.P. Phys. Rev. 1955, 97, 660.
157. Munoz, R.C.; Holroyd, R.A.; Itoh, K.; Nakagawa, K.; Nishikawa, M.; Fueki, K. J. Phys.
Chem. 1987, 91, 4639.
158. Itoh, K.; Holroyd, R.A.; Nishikawa, M. J. Phys. Chem. B 1998, 102, 3147.
159. Itoh, K.; Nishikawa, M.; Holroyd, R.A. J. Chem. Phys. 1996, 105, 5510.
160. Drude, P.; Nernst, W. Z. Phys. Chem. 1894, 15, 79.
161. Knoesel, E.; Bonn, M.; Shan, J.; Heinz, T.F. Phys. Rev. Lett. 2001, 86, 340.
162. Mozumder, A. Chem. Phys. Lett. 1993, 207, 245.
163. Mozumder, A. J. Phys. Chem. A 2002, 106, 7062.
164. Allen, A.O. Drift Mobilities and Conduction Band Energies of Excess Electrons in Dielectric
Liquids, NSRDS-NBS 58, 1976.
165. Ascarelli, G. Phys. Rev. B 1989, 40, 1871.
166. Ascarelli, G. Phys. Rev. Lett. 1991, 66, 1906.
167. Munoz, R.C.; Holroyd, R.A. Chem. Phys. Lett. 1987, 37, 250.
Copyright 2004 by Taylor & Francis Group, LLC
168. Munoz, R.C.; Ascarelli, G. Phys Rev. Lett. 1983, 51, 215.
169. Itoh, K.; Munoz, R.C.; Holroyd, R.A. J. Chem. Phys. 1989, 90, 1128.
170. Shibamura, E.; Takahashi, T.; Kubota, S.; Doke, T. Phys. Rev. A 1979, 20, 2547.
171. Baird, J.K. J. Chem. Phys. 1983, 79, 316.
172. Sowada, U.; Holroyd, R.A. J. Chem. Phys. 1979, 70, 3586.
173. Reininger, R.; Steinberger, I.T.; Bernstor, S.; Saile, V.; Laporte, P. Chem. Phys. 1984, 86,
189.
174. Reshotko, M.; Asaf, U.; Ascarelli, G.; Reininger, R.; Reisfeld, G.; Steinberger, I.T. Phys. Rev.
B. 1991, 43, 14174.
175. Wigner, E.; Seitz, F. Phys. Rev. 1933, 43, 804.
176. Ende, C.A.M.v.d.; Nyikos, L.; Warman, J.M.; Hummel, A. Radiat. Phys. Chem. 1982, 19,
297.
177. Gee, N.; Freeman, G.R. J. Chem. Phys. 1989, 90, 5399.
178. Itoh, K.; Holroyd, R.A.; Nishikawa, M. J. Phys. Chem. A 2001, 105, 703.
179. Itoh, K.; Holroyd, R.A. J. Phys. Chem. 1990, 94, 8850.
180. Gee, N.; Ramanan, G.; Freeman, G.R. Can. J. Chem. 1990, 68, 1527.
181. Heinrich, F.; Schiltz, A. Presented at the 6th International Conference on Positron Annihi-
lation, Arlington, TX, 1982. Unpublished.
182. Linderoth, S.; MacKenzie, I.K.; Tanigawa, S. Phys. Lett. A 1985, 107, 409.
183. Wang, C.L.; Kobayashi, Y.; Hirata, K. Radiat. Phys. Chem. 2000, 58, 451.
184. Lopes, M.I.; Chepel, V.; Solovov, V.; Marques, R.F.; Policarpo, A.J.P.L. Presented at the
Proceedings of the 8th Internatl. Conf. on Calorimetry in High Energy Physics, Lisbon
Portugal, 1999. Unpublished.
185. Willis, W.J.; Radeka, V. Nucl. Instrum. Methods 1974, 120, 221.
186. Abachi, S.; Abolins, M.; Acharya, B.S.; Adam, I.; Ahn, S., et al. Nucl. Instrum. Methods
Phys. Res. A 1994, 338, 185.
187. Schinzel, D. Nucl. Instrum. Methods Phys. Res. A 1998, 419, 217.
188. Engler, J. J. Phys. G. Nucl. Part. Phys. 1996, 22, 1.
189. Albrow, M.G.; Apsimon, R.; Aubert, B.; Bacci, C.; Bezaguet, A., et al. Nucl. Instrum.
Methods A 1988, 265, 303.
190. Arneodo, F.; Badertscher, A.; Baibussinov, S.; Benetti, P.; Borio, A., et al. Nucl. Instrum.
Methods A 2000, 455, 376.
191. Aprile, E.; Curioni, A.; Egorov, V.; Giboni, K.L.; Oberlack, U.; Ventura, S.; Doke, T.; Taki-
zawa, K.; Chupp, E.L.; Dunphy, P.P. Nucl. Instrum. Methods Phys. Res. A 2001, 461, 256.
192. Lopes, M.I.; Chepel, V. Presented at the 2002 IEEE 14th Internatl. Conf. on Dielectric
Liquids, Graz, Austria, 2002. Unpublished.
Copyright 2004 by Taylor & Francis Group, LLC
9
Interactions of Low-Energy Electrons with
Atomic and Molecular Solids
Andrew D. Bass and Leon Sanche**
University of Sherbrooke, Sherbrooke, Quebec, Canada
1. INTRODUCTION
Low-energy electronmolecule scattering has been studied in the gas phase for over 80
years, beginning with experiments by Franck and Hertz [1] in 1914 (for a review of gaseous
electronmolecule collision processes, see Refs. 25). While such research has usually been
performed on isolated targets, more recent work (commencing around 1980) has extended
electron-scattering type measurements and theory into the condensed phase to deter-
mine the interactions of electrons with the atomic and molecular constituents of solids and
liquids, thus providing new challenges to our understanding of the basic physics of
electron scattering. Particular attention has been directed to the interactions of low-energy
electrons (LEEs), that is, electrons with kinetic energies lower than 30 eV, since these
species are implicated in a large number of phenomena and processes operating in the
condensed phase systems, from radiation damage [6] and the electronic aging of dielectrics
[7], friction induced damage to lubricants [8] and nanolithography [9], and to (potentially)
the production of O
2
in extraterrestrial water ices [10] and depletion of the ozone [11].
Low-energy electrons are of special relevance to radiobiology since high-energy
radiation interacting with condensed matter (i.e., liquids to solids) generates as the most
abundant intermediate species electrons with energies lower than 70 eV [6,12,13], within
nanoscopic volumes along ionization tracks. About four secondary electrons per 100 eV
deposited are created with a most probable energy lying below 10 eV. At these energies,
electrons have thermalization distances of the order of 110 nm [13], which eectively
dene the initial volume for energy deposition by high-energy radiation. In these volumes,
termed spurs, the highly excited atomic, molecular, and radical species ions and sec-
ondary electrons can induce nonthermal reactions within femtosecond times. A majority
of these reactive species, including those that initiate further chemical reactions, are
created by the secondary electrons. Thus to reach a comprehensive understanding of
radiation energy deposition processes in condensed matter at the nanoscopic level, it is
necessary to identify the interactions of these electrons.
* Canada Chair in the Radiation Sciences.
Copyright 2004 by Taylor & Francis Group, LLC
In the early 1990s, one of us (LS) reviewed the then available experimental data on
LEE interactions with atomic and molecular thin lms [14,15]. In the intervening years,
considerable progress has been made, and it is now appropriate to review this more recent
work. Thus in this contribution, we will largely restrict the discussion to research con-
ducted since 1990, although reference to older material will also be given. Moreover, it
should be noted that with the exception of materials of biological interest, discussion is
further limited to measurements and processes occurring in the multilayer atomic and
molecular solids. Particular attention will be paid to electron-driven processes that initiate
molecular fragmentation and/or other chemical change and to the formation of transient
negative ions (TNIs) which occur when an electron becomes temporarily bound to an
atom or molecule and which can greatly enhance scattering cross sections.
The article is organized as follows. In Sec. 2, we describe some of the basic physics of
the interactions of LEEs with isolated molecules, including an account of the DEA
process. In Sec. 3, we review experimental investigations of LEE impact on simple atomic
and molecular solids. Studies of electron impact on molecules of biological interest are
reviewed separately in Sec. 4. Finally, in Sec. 5, we consider with examples the relevance of
this work for application in areas other than radiobiology.
2. THEORETICAL BACKGROUND
Since electrons have a negligible mass compared to those of atoms and molecules, they can
exchange very little energy by momentum transfer with such particles at low velocities.
Hence to transfer considerable energy in a single event, the electron must be coupled to
a target by its electrodynamic interaction potential or by the constructive interference of
its wave function. The latter is associated primarily with the internal structure (i.e., orbital
congurations, internuclear distances, symmetry, etc.) of an atom or a molecule and, to a
lesser extent, with the geometrical arrangement of the atomic or molecular constituents of
a molecular solid or liquid. When intra-atomic and/or intramolecular constructive inter-
ference occurs, we refer to this phenomenon as the formation of an electron resonance.
Constructive and destructive interference of electron waves, delocalized over many atoms
or molecules of a solid or liquid, is known as electron diraction and is highly dependent
on the geometrical arrangement of the target and its band structure.
The electrodynamic interaction potential acting between a diatomic molecule and an
electron outside a molecule can be written as
V
l
e
r
2
P
1
!
r
!
R
Q
e
r
3
P
2
!
r
!
R
1
ae
2
2r
4
aVe
2
2r
4
P
2
!
r
!
R . . .
where r is the distance of the incident electron from the molecule and R is the internuclear
separation; l
e
is the electric dipole moment. The term containing l
e
is necessarily absent in
homonuclear diatomic molecules. The second term, involving Q
e
, the quadrupole moment,
characterizes the quadrupole interaction. These two terms are the electrostatic terms in
that they pertain to the interaction of the incident electron and the unperturbed molecule.
The next two terms involving a, the spherically symmetric, and aV, the nonspherical part of
the polarizability, are the dynamic terms. The latter involve the polarization of the
molecule by the incident electron. In Eq. (1),
!
r and
!
R are unit vectors in the directions r
Copyright 2004 by Taylor & Francis Group, LLC
and R, respectively, and P
n
are the Legendre polynomials. The parameters a, aV, l
e
, and Q
e
are functions of R. It can be seen from Eq. (1) that by estimating the magnitude of the
various terms, it may be possible to sort out the dominant scattering mechanism. For
example, the potential of Eq. (1) can be expanded around the equilibrium distance R
e
along an internuclear coordinate of the molecule as
V VR R
e
R R
e
BV
BR
R R
e
. . . 2
When only these two terms are considered and the molecule is assumed to be a harmonic
oscillator, solving the problem within the Born approximation (usually corresponding
to small momentum transfer) leads to the optical selection rule Dv = 1 for vibrational
transitions [16]. Thus within this most restrictive approximation, the electron behaves like
electromagnetic radiation. From this analysis, we can expect the electronmolecule
potential described by Eq. (2) to be responsible for the magnitude of the dierential
scattering cross sections which are large only for small scattering angles and the excitation
of the rst vibrational energy level (v = 1) from the ground state of the molecule.
Electron resonances occur when the scattered electron resides for a much longer time
than the usual scattering time in the neighborhood of a target atom or molecule. From an
atomic or molecular orbital perspective, a resonant state may be considered as a TNI
formed by an electron in which it occupies an orbital of the target. This concept leads to
the denition of two major types or categories of resonances or transient anions [17,18]. If
the additional electron occupies a previously unlled orbital of the target in its ground
state, the transitory state is referred to as a single particle resonance. The term shape res-
onance applies more specically when temporary trapping of the electron is due to the
shape of the electronmolecule potential. When the transitory anion is formed by two
electrons occupying previously unlled orbitals, the resonance is called core-excited or
may be referred to as a two-particle, one-hole state.
When an electron resonance occurs, the temporary capture of an electron at an
atomic or molecular site increases the interaction time of the electron at that site in
proportion to the resonance lifetime (s
a
f
10
16
10
13
sec) and the inverse of the electron
transfer rate. This local interaction causes a distortion of the atom or molecule which ac-
cepts the additional electron. When the electron leaves the molecule, nuclear motion is
initiated toward the initial internuclear distance, causing excitation of many overtones of
the molecule, due to the strong overlap between the nuclear wave function of the resonant
state and those of many vibrational states of the molecular ground state. On the other
hand, when s is much smaller than a typical vibrational period (s10
14
sec), the nuclei are
not signicantly displaced. In this case, overlap between the nuclear wave function of the
resonant state and that of the vibrational levels of the ground state occurs only between
the rst few energy levels. Thus for short resonance times, only the lower vibrational levels
become excited with considerable amplitude.
Dissociative electron attachment (DEA) occurs when the molecular transient anion
state is dissociative in the FranckCondon (FC) region, the localization time is of the
order of or larger than the time required for dissociation along a particular nuclear co-
ordinate, and one of the resulting fragments has positive electron anity. In this case, a
stable atomic or molecular anion is formed along with one or more neutral species.
Dissociative electron attachment usually occurs via the formation of core-excited reso-
nances since these possess suciently long lifetimes to allow for dissociation of the anion
before autoionization.
Copyright 2004 by Taylor & Francis Group, LLC
Within a local complex potential curve crossing model, the cross section for the
simple DEA reaction e AB
!
AB
!
A
and B upon dissociation. Hence the DEA cross section depends exponentially on the prod-
uct of the lifetime of the transient anion and the velocities of the fragments. Equation (6)
denes most of the intrinsic characteristics of the DEA process. It may be seen from this
equation that the magnitude of the DEA cross section depends on parameters which are
inuenced by the nature of the target, i.e., the attaching electron wavelength k, the
resonance lifetime s
a
, and curve crossing at R
c
.
Equation (6) denes most of the intrinsic characteristics of the DEA process. For
further information on the mechanism of transient anion formation and its eects in
isolated electronatom and electronmolecule systems, the reader is referred to the review
articles by Schulz [17] and others [7,19,20]. Information on resonance scattering from
single layer and submonolayer of molecules physisorbed or chemisorbed on conductive
surfaces can be found in the review by Palmer et al. [2123]. The present article provides
information essentially on resonances in atoms and molecules condensed onto a dielectric
surface or forming a dielectric thin lm.
3. EXPERIMENTAL STUDIES OF RARE GAS AND MOLECULAR SOLID
FILMS
3.1. Experimental Methods
A number of experimental techniques have been developed to investigate the interaction of
low-energy electrons with atomic and molecular solids. To avoid surface contamination and
Copyright 2004 by Taylor & Francis Group, LLC
to facilitate the use of low-energy electron beams, such experiments must be performed
under ultrahigh-vacuum (UHV) conditions in chambers evacuated by cryogenic, ion, and
turbo-molecular pumps to base pressures below 10
10
Torr. Molecular (and atomic) solid
targets are usually formed by vapor deposition onto a clean substrate held at temperatures of
between 15 and 100 K by a cryostat. The substrate is usually a polycrystalline metal foil, a
metallic monocrystal, or a semiconductor crystal which can be cleaned by resistive heating
and/or ion bombardment. Films of materials that exhibit a high vapor pressure are grown
using a gas-volume expansion dosing procedure [24] that can be calibrated by monitoring the
quantum size eect features observed for ultrathin lms [25]. Films of materials that under
ambient conditions might be described as low vapor pressure solids can be grown using an
oven to generate a low-density molecular ux [26]. Films of larger organic and biomolecular
materials can be prepared ex vacua as self-assembled monolayers on gold substrates [27,28].
Once formed, the sample lms may be probed with a variety of electron beamtechniques, the
most common of which are now briey described.
Low-Energy Electron Transmission Spectroscopy
In a typical low-energy electron transmission (LEET) experiment [24], a trochoidal
monochromator [29] provides an electron current of between 1 and 10 nA, with an
intrinsic resolution between 40 and 60 meV full width at half maximum (fwhm). The beam
is incident normally on the lm surface. A LEET spectrum is obtained by measuring the
current I
t
arriving at the substrate as a function of incident electron energy E
i
. In LEET
spectroscopy, and in the following other electron impact experiments, the absolute energy
scale is calibrated to within F0.15 eV of the vacuum level by measuring the onset of
electron transmission through the lms. A description of the technique is given in Ref. 24.
Charge (Electron) Trapping
The LEET technique [24] can be modied to measure cross sections for charge (electron)
trapping or stabilization by molecules condensed onto dielectric lms [30,31]. Electronic
charge trapped at the surface of a lm, following exposure to electrons of a known E
i
,
produces a retarding potential (DV), which is apparent as a displacement of any
subsequent LEET spectrum to higher incident energies. The observed rate of charging
or charging coecient A
s
can be converted into a charging cross section (r
CT
) as follows:
A
s
E
i
udDVt=dtj
t 0
LIl
o
=epr
2
r
CT
7
where L and e are the spacer layer thickness and dielectric constant, respectively, l
0
is the
surface density of molecular targets, and I and r are the total current and radius of the
incident electron beam, respectively. Absolute trapping cross sections as small as 10
19
cm
2
can be measured with an uncertainty of F50% using this technique. When combined
with the mass spectrometric measurements of the electron-stimulated desorption (ESD) of
anions, r
CT
cross sections can, under certain conditions, allow absolute cross sections for
dissociative electron attachment (DEA) to be obtained.
In the form described above, charge-trapping measurements can only be performed
on nanometer-thick lms. However, analogous measurements on dielectric materials of
macroscopic dimensions (thickness of 0.05 to 0.15 mm and area
f
2 cm
2
) have been
obtained [32] with an instrument that employs a pulsed low-current electron beam (10
14
10
12
C/pulse) to induce charging and a Kelvin probe to measure changes in the samples
surface potential DV. This latter can then be related to changes in the quantity of trapped
charge DQ
t
. Independently obtained electrometer measurements made during each electron
Copyright 2004 by Taylor & Francis Group, LLC
pulse provide an instantaneous measure of the induced charge (e.g., DQ
t
), which, with
knowledge of the incident charge per pulse Q
0
, allows a trapping probability (DQ
t
/Q
0
) to be
determined. The lm is discharged with a UV lamp, mounted outside the vacuum chamber.
High-Resolution Electron Energy-Loss Spectroscopy
The energy losses of electrons scattered near the surface of thin lms and their dependence
on E
i
can be measured with a high-resolution electron energy loss (HREEL) spectrometer
of the type illustrated in Fig. 1. In addition to a cryogenically cooled substrate, such an
instrument includes an electron monochromator and energy analyzer. Those shown in Fig.
1 employ hemispherical electrostatic deectors to achieve optimal resolution (
f
5 to 30
meV fwhm). Typically, the monochromator produces a focused electron beam that strikes
the lm surface at an angle h
0
from the lm normal. Electrons scattered in a narrow pencil
about angle h
r
relative to the surface normal are energy-analyzed with the second hemi-
spherical deector. Dependent on spectrometer design, the angles h
0
, h
r
, or both can be
varied within limits set by available space. Energy-loss spectra are recorded by sweeping
the potential of the monochromator or analyzer relative to the grounded target. The
energy dependence of the magnitude of a given energy-loss event (i.e., the excitation
function) is obtained by sweeping the energy of both deectors with the potential dierence
between them corresponding to the probed energy.
The HREEL technique can be used in several ways: for detailed spectroscopic
studies of vibrational and electronic excitation within the atomic and molecular lms
Figure 1 Schematic representation of a high-resolution electron energy loss (HREEL) spectrometer.
Copyright 2004 by Taylor & Francis Group, LLC
[33,34], for measuring eective elastic and absolute inelastic cross sections [35,36], and for
determining eective cross sections for the production of fragment species [37,38]. This
latter is achieved by measuring the amplitude of a signal associated with a particular
degradation product (for example, a particular electronic or vibrational energy-loss
feature) as a function of the integrated electron current at specic electron energies.
Similar eective cross sections for the production of fragment species may be measured
using x-ray photoelectron [39] and infrared spectroscopies [40].
Electron-Stimulated Desorption Experiments
Electron-induced dissociative processes and certain other reactions are perhaps most
eectively detected when a fragment ion or neutral species is desorbed into vacuum. Gen-
erally, ion desorption is easier to observe than that of neutral species since desorbed
charged particles can be immediately mass-analyzed. Fig. 2 illustrates schematically an
experimental system for ion desorption studies of various types of target including large
organic/biomolecular lms. Note the load-lock section for the introduction sample slides
and/or the degassing of organic molecules within an oven prior to their introduction into
the analysis chamber. In this instrument [26], an electron gun produces an electron beam
of variable energy having a current of between 5 and 300 nA with an energy resolution of
f
250 meV fwhm. The electron beam is incident on a cryogenically cooled solid lm. Ions
desorbed during electron impact enter an ion lens (containing a set of retardation grids)
which precedes a quadrupole mass spectrometer. The so-called ion yield functions are
obtained by recording a particular ion signal as a function of E
i
energy. In other ESD
instruments, the electron gun is replaced with an electrostatic [41,42] or trochoidal [43]
monochromator which produces an electron beam of a few nanoamperes at a resolution
of<80 meV. Pulsed electron sources have also been used [44] in conjunction with a time of
ight mass spectrometer.
Figure 2 Schematic overview of an apparatus suitable for measurements of the electron-stimulated
desorption of anions from vacuum deposited lms. (From Ref. 26.)
Copyright 2004 by Taylor & Francis Group, LLC
In principle, the neutral desorbed products of dissociation can be detected and mass
analyzed, if ionized prior to their introduction into the mass spectrometer. However, such
experiments are dicult due to low eective ionization eciencies for desorbed neutrals.
Nevertheless, a number of systems have been studied in the groups of Wurm et al. [45],
Kimmel et al. [46,47], and Harries et al. [48], for example. In our laboratory, studies of
neutral particle desorption have been concentrated on self-assembled monolayer targets at
room temperature [27,28]. Under certain circumstances, neutrals desorbed in electronically
excited metastable states of sucient energy can be detected by their de-excitation at the
surface of a large-area microchannel plate/detector assembly [49]. Separation of the ESD
signal of metastables from UV luminescence can be eected by time of ight analysis [49];
however, when the photonsignal is small relative tothe metastable yield, such discrimination
is unnecessary and only the total yield of neutral particles (NP) needs to be measured.
3.2. Experimental Results
Elastic and Quasi-Elastic Scattering
Elastic scattering of LEEs by multilayer rare gas and molecular solid lms has been
investigated by low-energy electron transmission (LEET) spectroscopy [24,25,5071],
photoinjection [7282], and elastic reection [35,36,66,8387] experiments. In the latter,
the spectrometer shown in Fig. 1 is adjusted to measure electron elastically scattered from
the lm at particular incident and scattered angles h
0
and h
r
. Then, the amplitude of the
elastic peak is measured as a function of E
i
. Since in these experiments both the incident
and outgoing electron momenta are specied, features due to interferences of the electron
waves are prominent in the spectra. In well-ordered lms, the diraction structure is
dominated by long-range order [66], whereas in amorphous substances, variation in the
structure factor, due to short-range order, can be detected in the energy dependence of the
elastic reectivity [83,84,87].
Both LEET and photoinjection experiments measure the electron current transmitted
through a multilayer lm deposited on a metal substrate. Usually in photoinjection experi-
ments, electrons are injected in the lmfromthe metal substrate with ill-dened energies and
momenta, although some control of electron injection energy is now possible with syn-
chrotron radiation sources [79]. In either case, the outgoing electrons which escape into the
vacuum at a given energy and momentum can be selected with an electron analyzer or
retarding electric eld. This sort of measurement [72] is particularly sensitive to electron
interactions near the substratelm interface and, for self-assembled monolayers of long-
chain organic molecules, to the direction and degree of structural order within the lm [80].
In transmission, it is only the incident beam that has a well-dened energy and
momentum since the current collected at the metal substrate has been scattered into all
angles. Furthermore, when the lm is highly disordered, electrons are scattered in all
possible directions near the surface [53], so that the penetrating momentum is also un-
specied. Electron scattering resulting in multiple energy losses to phonons (i.e., quasi-
elastic scattering) is expected to depend on the electronic conduction band density of states
(CB DOS) and on electronphonon interactions. The calculated CB DOS of solid Ar
(solid line) and that extracted from LEET spectra of multilayer (20100 monolayers) Ar
lms (dashed line) are compared in Fig. 3 [60]. There exists an obvious relationship
between the two sets of data, which indicates that it is essentially the CB DOS that governs
quasi-elastic scattering of the LEEs. In HREEL experiments on rare gas (RG) solids
[83,84], it has been shown that this correspondence arises because of multiple energy losses
Copyright 2004 by Taylor & Francis Group, LLC
to phonons by the scattered electrons. Taking again the example of an Ar multilayer lm,
this relationship is shown in Fig. 4. The curves in the gure were recorded with the
HREEL spectrometer described in Sec. 3.1.3, set to measure the excitation function at an
energy loss DE = 0.25 eV and h
r
= 45j for several incident angles between 15j and 65j.
These curves thus represent the probability that an electron, penetrating a 50-layer lm of
Ar deposited on Pt, loses 0.25 eV via multiple losses to phonons in the solid. Except for the
measurement at h
0
= 45j (i.e., in the specular direction), all features are found at essen-
tially the same energy, independent of the incident angle. The similarity between these
curves reveals an electron-scattering property of the solid that is averaged over various
directions of electron propagation (i.e., various electron states) which may consequently
reect the CB DOS. In Fig. 4(b), the CB DOS of Ar as calculated by Bacalis et al. [88] is
displayed with the bottom of the lowest conduction band xed at the measured value [25]
of 0.25 eV above the vacuum level. Clearly, the experimental curves of Fig. 4(a), especially
those for large incident angles, bare a close resemblance with the calculated CB DOS. The
progressive shift to higher energy of calculated features relative to their experimental coun-
terparts (excluding those peaks at 9 and 12 eV) likely reects a discrepancy between the
face-centered cubic lattice parameter of Ar employed in the calculation (0.526 nm) and
experimental values at 20 K (0.531 nm) [89].
Figure 3 Electron conduction band density of states (CB DOS) for solid Ar; calculated (solid line)
and determined from analysis of LEET data recorded at 20 K at energies below the rst excitonic
threshold (dashed line). The zero of energy is the vacuum level. V
0
is the energy of the bottom of the
conduction band (0.25 eV). (From Ref. 60.)
Copyright 2004 by Taylor & Francis Group, LLC
The similarity between the experimental results and the calculated CB DOS can be
explained by focusing on the electron transport properties in the bulk [83,84]. An electron
propagating in a conduction band of an RG solid suers scattering mainly from defects
and lattice waves. This can be described by introducing the scattering probability per unit
length Q(E
k0
,k
0
,E
k
,k) that a Bloch electron initially in a state jvX of momentum k
0
and
energy E
k0
is scattered into a nal state jv
k
X of momentum k and energy E
k
, while the
crystal changes from a state jiX of energy e
i
to a state j fX of energy e
f
. Then, by referring
to the golden rule and solving the Boltzmann transport equation for a plane-parallel
system in the two-stream approximation [35,36,55,90], one obtains for the electron
mean-free path k(E
i
) (i.e., the reciprocal of Q) the expression
kE
i
X
k
QE
k0
; k
0
; E
k
; k
( )
1
* +
E
i
8
In this expression, the k summation extends over the rst Brillouin zone, whereas h. . .i
Ei
stands for the average over the incident direction k
0
for a constant incident energy. If we
Figure 4 (a) Scattered electron intensity at xed energy loss DE=0.25 eV as a function of incident
electron energy E
i
at several angles of incidence h
0
on a 50-ML (monolayers) Ar lm. (b) CB DOS
for fcc structure of solid Ar as calculated by Bacalis et al. [88]. (From Ref. 83.)
Copyright 2004 by Taylor & Francis Group, LLC
replace the summations with integrations and assume for simplicity that the matrix element
for calculating Q depends only on the momentum transfer (i.e., k k
0
), Eq. (8) yields
kE
i
1
DE
i
h 8p t=XSE
i
f g 1=sE
i
f gi 9
where D(E
i
) is the CB DOS of the solid at the energy E
i
, s(E
i
) corresponds to a relaxation
time (i.e., the time between scattering events) independent of the k
0
direction, S(E
i
) is the
surface of constant energy E
i
within the rst Brillouin zone, and X is the volume of the
crystal. Within the approximations of an electron eective mass and of an electronphonon
interaction described as a deformation-potential perturbation, one has 1/s(E
i
)~k
2
0
and
S(E
0
)~k
2
0
[91]. Consequently, the expression in parentheses in Eq. (9) is independent of E
i
and the energy dependence of k(E
i
)
1
or Q (i.e., the quasi-elastic scattering probability per
unit length) becomes directly proportional to the CBDOS as shown experimentally in Fig. 4.
Inuence of Temporary Electron Trapping on Quasi-Elastic Scattering
and Phonon Creation
The formation of a TNI within an atomic or molecular solid has also been found to aect
quasi-elastic scattering of incident electrons [8587]. A TNI is formed when an electron of
a particular E
i
becomes trapped at a lattice site for a short time (10
15
to 10
13
sec). This
trapping (and detrapping) process transfers energy to the lattice, which appears in HREEL
measurements as an increase in phonon loss processes. The eects of a TNI can thus be
identied by comparing HREELS spectra obtained at the on-resonance electron energy
with those obtained o resonance. The energy and dynamics involved in the temporary
trapping of an electron in a thin solid lm can be demonstrated from electron scattering
experiments on RG solid lms. In RG, an electron resonance occurs at an energy slightly
below the threshold for creating the lowest bulk exciton. In solid Ar, for example, the
lowest transient anion state of Ar
state is a
core-excited resonance [i.e., a one-hole, two-electron state composed of a 4 s electron
bound to the electronic excited state formed from the atomic orbitals (. . . 3s
2
, 3p
5
, 4s)]. It
has the isolated (i.e., gas phase) conguration (. . . 3s
2
, 3p
5
, 4s
2
)
2
P
3/2
2
P
1/2
. In the
condensed phase, this resonance is 0.25 eV below the n = 1 bulk exciton; it is formed
by the binding of an electron to the n =1,1V (the prime denoting the j =1/2 spin-orbit
partner) excitons in the bulk. The apparent width of the resonance is greater than the spin-
orbit splitting for the neutral exciton states of Ar (
f
0.18 eV [92,93]) so that the state,
assigned as
2
P Ar
2=75
p
with
x
D
the Debye frequency [96]. In Eq. (11), the term c is referred to as the pseudo-Debye
Figure 5 HREEL spectra of a 12-ML Ar lm at three E
i
with h
0
= 15j and analysis angle h
r
= 45j.
At E
i
= 11.75 eV, an Ar
2
11
where M is the mass of the atom, a
0
is the nearest neighbor distance, c
1
is the longitudinal
sound velocity, and e is the permittivity appropriate to the frequency range s
R
.
Calculations were performed with Eqs. (10) and (11) with s
R
as the sole adjustable
parameter and experimentally determined values for a
0
, e, x
D
, and c
1
[97]. The result of the
calculation with s
R
= 2.2 10
14
sec is displayed in Fig. 5. Note that the energy of the
maximum is quite sensitive to s
R
; a variation of s
R
by 10% resulted in a 15% change for
the energy of the maximum.
Vibrational Excitation
By scattering within molecular solids and at their surfaces, LEE can excite with consid-
erable cross sections not only phonon modes of the lattice [35,36,83,84,87,90,98,99], but
also individual vibrational levels of the molecular constituents [36,90,98119] of the solid.
These modes can be excited either by nonresonant or by resonant scattering prevailing at
specic energies,* but as will be seen, resonances can enhance this energy-loss process by
orders of magnitude. We provide in the next two subsections specic examples of
vibrational excitation induced by LEE in molecular solid lms. The HREEL spectra of
solid N
2
illustrate well the enhancement of vibrational excitation due to a shape resonance.
The other example with solid O
2
and O
2
-doped Ar further shows the eect of the density
of states on vibrational excitation.
High-Resolution Electron Energy-Loss Spectra of N
2
. Vibrational excitation of
ground-state N
2
, induced by LEE impact on thin multilayer N
2
lms, is illustrated in
Fig. 6. These electron energy-loss spectra were recorded for electrons of energies E
i
= 2.9,
10.8, and 19.8 eV, h
o
= 14j and h
r
= 45j. The vertical gain in each curve or portion of a
curve is referenced to the elastic peak. Each energy-loss peak can be ascribed to vibrational
excitation of ground-state N
2
[100]. It can be seen in the gure that at certain impact
energies (E
i
= 2.9 and 19.8 eV), the intensities of vibrational energy losses are greatly
increased (i.e., up to 2 orders of magnitude for overtones). In N
2
lms, the production of
overtones at E
i
= 2.9 and 20 eV is attributable to
2
C
g
and
2
A
u
shape resonances [100].
There is, however, no electron resonance at E
i
= 10.8 eV in N
2
so that the strongest part
of the interaction leading to vibrational excitation arises mainly from induced polarization
(i.e., from direct scattering). This interaction is mainly eective to produce excitation of
one vibrational quantum which amounts to an energy loss of less than 0.4 eV.
Excitation Functions of O
2
and O
2
-Doped Ar Films. Resonances can be best
identied by the structures they produce in excitation functions of a particular energy-
loss process (i.e., the incident-electron energy dependence of the loss). Fig. 7 is reproduced
from a recent study [118] of the electron-induced vibrational and electronic excitation of
multilayer lms of O
2
condensed on the Pt(111) surface and shows the incident electron
energy dependence of major losses at the indicated lm thickness and scattering angles.
Also shown in this gure is the scattered electron intensity of the inelastic background
* For a theoretical description of the eects of a surface on resonant scattering, see Ref. 120.
Copyright 2004 by Taylor & Francis Group, LLC
located just before the v=1 peak onset at DE=0.16 eV along with that contributing to
each energy loss (dashed lines).
With the exception of that of the a
1
D
g
(v=0) loss, the vibrational excitation
functions of O
2
in the solid phase dier signicantly from their gas-phase equivalents.
For example, the well-known sharp peaks of the
2
C
g
resonance, which dominate the
vibrational cross sections up to 3 eV in the gas phase [121], are absent in the solid phase.
This absence is understood as deriving from a lowering in energy of the
2
C
g
resonance by
charge-image/charge polarization [101] of the O
2
lm. At higher energies in the gas phase,
excitation functions for the X
3
A
g
, v=1, 2, and 3 vibrational losses display a single
structure due to the
2
C
u
resonance near 6.5 eV. In contrast, in the condensed phase, the
same vibrational losses are characterized by three broad overlapping bands, attributed to
the excitation of the
2
C
u
,
2
A
g
I,
2
A
u
(I), and
2
A
g;u
(II) anion states at 6.5, 8.5, 10.5, and 13.5
eV, respectively. These contributions from A
+
transient anion states, which are forbidden
by the mirror plane symmetry selection rule in the gas phase (i.e., A
p/
!
A
+
), arise here
due to the loss of the cylindrical symmetry of O
2
in the condensed phase [122].
The vibrational excitation is further altered when O
2
molecules are embedded within
an Ar matrix [118,119]. Fig. 8 presents the results of experiments for 1% O
2
in 20 ML of Ar
and shows excitation functions for the v=1 and 2 vibrational losses of O
2
[Fig. 8(b) and (c),
Figure 6 HREEL spectra of a multilayer disordered N
2
lm recorded at the indicated E
i
. (From
Ref. 100.)
Copyright 2004 by Taylor & Francis Group, LLC
respectively] obtained after subtraction of a background of Ar phonon losses. Compared to
the gas-phase result [121] [shown in Fig. 8(a) for v = 1], and those for solid O
2
in Fig. 7, the
excitation functions for O
2
in Ar are much more structured, containing many narrow
features. There is also a signicant vibrational enhancement within the 2.55 eV incident
energy range that is absent in the pure multilayer O
2
results (Fig. 7). Furthermore, except for
the relative intensity changes, each maximum is located at the same energy independently of
the angle of incidence and azimuthal orientation of the crystal. Such a behavior strongly
suggests that the maxima arise primarily from an electron scattering property that is
averaged over various directions of electron propagation within the crystal and thus
depends on the CB DOS. In Fig. 8(d), we again reproduce the CB DOS of solid Ar [88]
Figure 7 Incident electron energy dependence of the X
3
A
g
, v = 0, 1, 2, 3 vibrational and the a
1
D
g
(v = 0) electronic loss scattered intensities from a 10-layer lm of O
2
condensed on Pt(111). h
0
was
set at 10j with h
r
at 45j and the azimuth /
0
at 10j. Also shown is the energy dependence of the
inelastic background intensity located just before the v = 1 loss peak onset at DE = 0.16 eV along
with that contributing to each energy-loss prole (dashed lines). (From Ref. 118.)
Copyright 2004 by Taylor & Francis Group, LLC
and it is easily seen that all maxima in Fig. 8(b) and (c) correlate respectively with minima in
the CB DOS as opposed to the scattered intensity arising from multiple scattering on
phonon (Fig. 4) [83,84], which was shown to follow the CB DOS.
This striking result can be qualitatively understood as related to CB DOS-inuenced
changes in the O
2
2
dE
i
E
k
i2p V
d;E
2
DE
i
; 12
Figure 8 (a) Dierential cross section for the excitation of X
3
A
g
, v = 1 vibrational level of gaseous
O
2
. (b) and (c) are the incident electron energy dependence of the net v =1, 2 energy-loss intensities for
a 20-ML lm of 1% O
2
in Ar. (d) is the CB DOS of solid Ar. (From Ref. 119.)
Copyright 2004 by Taylor & Francis Group, LLC
where V
d,k
is the coupling matrix element responsible for the decay and E
k
is the energy of
the free electron continuum. Under suciently small variations of the matrix element with
the wave vector k, the density of free electron states D(E) appears explicitly with the
summation over k [125].
For the short capture time limit, it can be shown [118] that the energy imparted to
the nuclei DT
N
can be approximated by
DT
N
i2M
red
1
F
2
r
R
i
s
2
E
r
R
i
; 13
where M
red
is the reduced mass of the molecule, s is the anion lifetime, E
r
is the energy of
the resonance, R
i
is the internuclear separation at which attachment occurs, and F
r
is the
internuclear force. It is apparent that the energy DT
N
is a combination of two factors: the
strength of the resonant vibrational coupling F
r
(R) (which depends on the shape of
internuclear potential of the TNI) and the resonance lifetime s[E
r
(R)]. In the absence of
potential energy curve crossings within the FranckCondon region, the factor F
r
(R) in Eq.
(13) varies slowly as a function of the R. Moreover, as the transient anion potential energy
curve is in a rst approximation only slightly shifted down by the polarization of the sur-
rounding medium, F
r
(R) is virtually the same in both gas and condensed phase.
On the other hand, the lifetime of the resonance s[E
r
(R)], whose inverse is explicitly
given by s
1
[E
r
(R)]i(2p/t)jV
d,Er
j
2
D[E
r
(R)] [cf., Eq. (12)], is directly linked to the CB DOS
of the host medium, i.e., D[E
r
(R)]. The remaining factor, which contains the matrix
element responsible for the decay V
d,Er
, is akin to a tunneling eect and yields essentially a
monotonic behavior in energy. Owing to its relatively short-range character, it is also
expected not to change signicantly between gas and condensed phases. Consequently, the
amount of energy transferred to the nuclei [cf., Eq. (13)] along with the corresponding
overtone scattering probabilities (i.e., cross sections) varies mostly, via the modication of
the resonance lifetime, as the opposite of the CB DOS of the host medium.
Electronic Excitation
Electron Spectroscopy. Electronic excitation induced by LEE in multilayer in
atomic and molecular solids has been studied with LEET [24,51,52,57,62,126131] and
HREELS [104,131147]. The slow-electron excitation of electronic states of atoms and
nondissociative electronic states of molecules in dielectrics can result in localized energy
deposits (e.g., Refs. 136 and 137) or into the formation of excitons [133] that propagate
through the solid with well-dened wave vector. From HREELS measurements, one notes
that electronic excitation in a molecular solids is similar and of similar magnitude to that
observed for the isolated molecule, except for the cases of Rydberg and symmetry-
forbidden transitions. It appears that a general characteristic of Rydberg states in con-
densed phases is poorly resolved since they broaden considerably into wide bands [148]
due to autoionization by fast exchange of Rydberg electrons. In contrast, the strong
exchange interaction in the low-energy regime causes singlettriplet transitions to be more
prominent in low-energy HREEL spectra than in measurements obtained with higher-
energy electrons or photon excitation. Consequently, the low-energy HREELS has
employed intensively to study the lowest triplet states of organic molecules [135146].
Electron-Stimulated Desorption of Metastable Species. Electronic excitation of
atomic and molecular solids is also demonstrated in the ESD yield of metastable atoms
and molecules (i.e., lifetimes >10
6
sec) and of UV photons from RG solids Ne and Ar
[49,149151], RG alloys [152], and from molecular solids of N
2
[153], N
2
O [154], CO
[153,155], and CO
2
[156,157], as well as N
2
- and CO-covered Xe and Kr [158160].
Copyright 2004 by Taylor & Francis Group, LLC
Desorption can proceed via several mechanisms. For solids with a negative electron
anity such as Ar [49,149151] and N
2
[153], the extended electron cloud around a
metastable center will interact repulsively with the surrounding medium and metastables
formed at the lmvacuum interface will be expelled into vacuum (the so-called cavity
expulsion mechanism [161]). Also permitted in solids with positive electron anities (e.g.,
CO) is the transfer of energy intramolecular vibration to the molecule-surface bond with
the resulting desorption of a molecule in lower vibrational level [153,155,158160].
Desorption of metastables via the excitation of dissociative molecular (or excimer)
electronic states is also possible [49,149151,154,156,157]. A concise review of the topic
can be found in Ref. 162.
Recent measurements have presented an interesting example of how metastable
desorption can provide information on TNI electronic excitation in rare gas solids. Low-
energy electron impact on solid Ar is known to induce the desorption of metastable Ar
atoms (Ar*) [49,149] via the repulsive interaction of a surface-trapped exciton with the
surrounding ground-state Ar atoms [161]. Fig. 9(b)(e) shows NP yield functions from
thin Ar lms of varying preparation [163,164]. The signal at energies above 12 eV is
correlated to the production of neutral excitons within the lm and their decay by ejection
and/or photon luminescence [151]. The maximum NP yield at
f
15 eV coincides with the
energy for optimum production of excited states in the gas [165] and condensed phases
[151]. What is of note, however, in the data of Fig. 9 is that the
2
P Ar
, shown in the
HREEL data of Fig. 9(a), also contributes to the NP signal seen from the
f
11.5 eV peak
in each of the NP yield functions of Fig. 9(b), (d), and (e). This is surprising since the
2
P
Ar
resonance lies below the energy of the rst bulk and surface Ar excitons found at 12.03
and 11.71 eV [133], so that its observation in NP (and metastable) yield functions implies
that the resonance has decayed into the higher-lying exciton energy levels [163]! In detailed
studies of the
2
P Ar
to the Pt(111) substrate at an energy of about 0.3 eV below the vacuum level.
After electron emission at the Ar/Pt(111) interface, Ar* in the form of an exciton moves to
the lmvacuum interface, where it can be ejected. Autoionization to the metal is inhibited
when a spacer layer is introduced between the Ar layer and the substrate as evidenced in
Fig. 9(c), where the resonance signal disappears for Ar condensed on a spacer of 3 ML of
crystalline n-hexane.
The remaining results in Fig. 9 demonstrate that when a small amount of molecular
oxygen is mixed in the Ar layer condensed on n-hexane [Fig. 9(e)] or deposited onto an
isolated Ar layer [Fig. 9(d)], the
2
P Ar
resonance in Fig. 9(c) and (d) was therefore interpreted [164] as due to
electron transfer to O
2
leading to the formation of O
2
in its ground-state
2
C
g
. With the
electron anity of O
2
being of the order of the binding energy of the rst electronically
excited state of Ar, the decay of Ar
2
P into lowest bulk excitons is possible by electron
transfer to O
2
.
Stable Anion Formation
The production of stable anions by low-energy electron impact has been studied by ESD
[22,41,43,44,122,147,154,156,157,166219] and charge-trapping [154,156,217,220236]
experiments. Below 20 eV, the desorption of stable anions from condensed systems is
attributed to DEA, which produces oscillatory structures in the anion yield functions, and
Copyright 2004 by Taylor & Francis Group, LLC
to dipolar dissociation (DD), which produces both anionic and cationic fragments.
Typically, DD produces a featureless signal which increases linearly with electron energy
from a threshold lying between 10 and 20 eV. Anion desorption data are often dominated
by the DEA process. The rst observations of DEA in a solid were made from the ESD of
O
from multilayer O
2
lms [166]. Subsequently, the O
2
molecule has served as a model
target to study factors aecting the physics of DEA in condensed phase systems
[41,122,172175,184,185,188191,206,209,211]. The yield of O
2
at
f
6.7 eV [237].
In contrast, the anion yield function from multilayer O
2
shows in addition to DD above
f
15 eV a much broader resonance feature between 5 and 10 eV and another structure at
13 eV. The broadening of the lower energy structure and the appearance of the 13-eV
structure are attributed to DEA via A
+
states following a relaxation of the A
===
A
+
selection rule [122].
Also shown in Fig. 10(c)(g) are the anion yield functions for submonolayer
quantities of O
2
deposited onto various multilayer atomic and molecular solids. The data
represent part of a study [41] on the environmental factors involved in the DEA process.
As can be seen, the yield of desorbed ions can vary greatly with substrate composition.
Such variations can be attributed to the so-called extrinsic factors that modify the ESD
process at times before attachment and after dissociation, for example, electron energy-
loss processes in the substrate and postdissociation interactions (PDI) of ions with the
surrounding medium [41]. These processes can be contrasted with intrinsic factors, which
Figure 10 O
2
resonances. The gain factor over each curve is relative to curve (c), and the shifted
baselines correspond to the zero anion intensity levels. Figure based on the data of Ref. 41.
Copyright 2004 by Taylor & Francis Group, LLC
modify the properties of the resonance (e.g., lifetime, energy, and decay channels). In
general, the O
yields per O
2
molecule are higher for O
2
deposited on RG solids Kr and
Xe [Fig. 10(c) and (d)see Ref. 211 for detailed comparison of O
desorption from RG
solids] than on a molecular solid where extrinsic eects such as electron energy-loss
processes and PDI are more likely. The weakest O
from O
3
[203] and CF
3
I [200]. By combining ESD
data with the results of charge-trapping experiments, DEA cross sections have also been
from H
2
O, [222,223], CF
4
[224,225], CF
3
Cl [230,231,233], HCl [236], CF
2
Cl
2
[235], CH
3
Br
[227], and the chloromethanes [226229,231].
The E
i
dependence of the surface charging cross section r
CT
for 0.1 ML of CH
3
Cl
condensed onto a 15-ML Kr lm [226,227] is shown in Fig. 11(a). The two curves Fig.
11(b) and (c) show the H
and Cl
and CH
3
along the strongly antibonding CCl
r* orbital. Since no
ESD signal is observed in Fig. 11(b) and (c) below 5 eV, the charge-trapping cross section
represents an absolute DEA cross section at these low energies. The value reported for
CH
3
Cl on a 5-ML-thick Kr lm (Fig. 12) thus represents an enhancement over gas-phase
experimental [240] and theoretical values [241] of between 4 and 6 orders of magnitude,
Copyright 2004 by Taylor & Francis Group, LLC
respectively. This remarkable enhancement has been attributed to image-charge-induced
polarization energy V
p
, which lowers the potential energy curve of the intermediate
transient anion increasing the anion lifetime [242] and decreasing the distance the anion
must stretch before dissociation is inevitable [226,227].
A quantitative understanding of the eect of V
p
on electron attachment to CH
3
Cl
adsorbed on Kr [226,226] and within its bulk [229] has been attempted with a modied
version of the R-matrix scattering model [241] which has been previously successful in
describing LEECH
3
Cl scattering in the gas phase [243]. Essentially, the model was
modied to include the eects of V
p
at short electronmolecule distances. The solid line in
Fig. 12 represents the results of the R-matrix calculation. However, to obtain this result, it
was also necessary to model in a semi-empirical fashion the electrons interaction with the
metal substrate. The discontinuity represented by the lms surface complicates consid-
erably this type of calculation, which is more straightforward for CH
3
Cl embedded within
bulk Kr. In this case [229], good agreement between theory and experiment is possible by
adjusting V
p
alone. However, it must be noted that the values of V
p
required to t the data
Figure 11 Charging cross section and anion yields induced by 010 eV electron impact on CH
3
Cl.
(a) Absolute surface charging cross section for 0.1 ML methyl chloride condensed on a 15-ML Kr
lm. (b) H
and Cl
desorption yields. (c) Total anion yield from gaseous methyl chloride. Figure
taken from Ref. 227.
Copyright 2004 by Taylor & Francis Group, LLC
for CH
3
Cl condensed on and within Kr are approximately 50% larger than those
measured elsewhere in other experiments [103]. With these latter values, the calculated
cross sections are about 1 order of magnitude too small. This disparity does not exist when
comparing similar charging data for CF
3
Cl [230] with an R-matrix calculation [231].
Comparison of the CH
3
Cl and CF
3
Cl data and calculations [231] suggest that the CH
3
Cl
on
Kr [103] due to the larger static dipole moment of this molecule.
Two further mechanisms are known to trap electronic charge in thin lms:
intermolecular and resonance stabilization. In resonance stabilization, electron attachment
to a molecular center produces an anion in a vibrationally excited state that is then de-
excited by energy exchange with neighboring molecules. When the initial anion ground
state lies below the band edge or lowest conduction level of the dielectric, then the
additional electron may become permanently trapped at the molecular site. In this case, a
permanent anion is formed (e.g., the case of O
2
[220]). Intermolecular stabilization refers
Figure 12 (a) Variation of maximum charge-trapping cross section (r
CT
) with Kr lm thickness
for low-energy electrons incident on 0.1 ML of CH
3
Cl deposited on the surface of Kr lms. The solid
line drawn from the data represents the results of an R-matrix calculation to describe electron
CH
3
Cl/Kr scattering. (b) Variation in energy E
max
of the maximum in r
CT
with Kr lm thickness.
Dashed line is a parametric t to E
max
. (From Ref. 226.)
Copyright 2004 by Taylor & Francis Group, LLC
to the trapping of a LEE by an aggregate of molecules, typically unable to do so in the
monomeric form. In this sense, the process is similar to solvation,* which usually requires
the organization of polar molecules to form a suitable trap. Intermolecular stabilization
has been observed for water clusters condensed on Kr and Xe surfaces at cryogenic
temperatures [222] and in pure water lms [223]. Both intermolecular and resonance
stabilization occur at incident electron energies below 1 eV. Since DEA can, in some
systems, occur at similar energies, it is not always easy to identify which process is
responsible for charge accumulation. Fortunately, further information can sometimes be
obtained from measurements of the ESD yield of anions from similar lms and from
comparison with gas phase results.
Electron Transfer from the Environment
In general, the properties of molecular TNI of condensed molecules are greatly aected by
their local environment. It is unsurprising then to nd that DEA can arise from electron
transfer from an electron state of the substrate to a dopant molecule. Such a transfer has
been observed between the electronexciton complex of RGS and adsorbed molecules
[179,180], between water lms and adsorbed CF
2
Cl
2
[235] and HCl [236], and between
image states and CH
3
Cl [232]. Fig. 13 presents recent charge-trapping data obtained on
10-ML-thick lms of n-hexane [232]. Fig 13(a) represents the energy variation of the
trapping coecient for pure n-hexane in the glassy amorphous state nH
g
that forms by
vacuum deposition at 25 K [245].
y
The curve shows that below E
i
= 7 eV, the threshold for
charging via DEA to n-hexane [167], charging coecients are very small. Annealing such
lms at 60 K produces crystalline n-hexane nH
c
, which is also resistant to charging
below 7 eV [Fig 13(b)]. It is this property of both n-hexane lm types (i.e., resistance to
charging) which renders them suitable substrates to study electron attachment to surface
dopant molecules at incident electron energies below 7 eV. Fig 13(c) presents A
s
(E
i
) for 0.1
ML CH
3
Cl on the surface of a 5-ML nH
g
lm. The sharp (50 meV fwhm) and intense peak
near zero energy is shown on an extended energy scale in the inset [Fig 13(d)]. The
maximum in A
s
corresponds to a charge-trapping cross section of 2 10
13
cm
2
which is
10
4
times larger than that for DEA to CH
3
Cl on the Kr lm surface [226] shown in Fig
13(e). Charging data for 0.1 ML CH
3
Cl on the nH
c
lm surface are shown in Fig 13(f) and
do not show the same intense charging at 0 eV seen on the glassy n-hexane substrate. The
DEA peak observed near 0.5 eV for CH
3
Cl on Kr [in Fig 13(e)] is hidden by the back-
ground signal in Fig 13(f).
Likewise, large charge-trapping cross sections, peaked sharply at 0 eV, were
measured for submonolayer quantities of O
2
and H
2
O deposited onto the nH
g
substrate.
Both of these molecules are known to trap near-zero electron volt electrons (as O
2
[220]
and by intermolecular stabilization within water assemblages [222]), but again, the cross
sections on nH
g
were several orders of magnitude larger than when measured on Kr or on
nH
c
lms. Signicantly, charging coecients (and the cross sections for electron trapping)
at higher energies on the nH
g
surface were not enhanced, and no 0-eV peak was observed
in charge-trapping measurements for CF
4
on nH
g
(CF
4
on Kr does not trap near 0 eV
electrons [224]). In light of these results, viz., a huge enhancement in A
s
(E
i
) at 0 eV,
independent of the mechanism of electron stabilization (but only for molecules that trap
* See, e.g., Ref. 244.
y
More recent studies (Ref. 239) have shown n-hexane lms formed at <60 K to be highly porous,
although this fact is not of great relevance to the present discussion.
Copyright 2004 by Taylor & Francis Group, LLC
electrons near 0 eV), it was argued that trapping proceeds via the initial capture of the
electron into an image state of the substrate with subsequent resonant transfer of the
surface electron to the molecular target [e.g., O
2
, CH
3
Cl, or (H
2
O)
n
]. Since in this case a
large portion of the surface would be available for electron capture, the observed large
trapping cross sections would likely result, provided that the lifetime of the electron in the
image state near the surface was suciently long. Fig. 14(a) explains this process in more
detail and shows schematically the band structure near the surface of the glassy n-hexane
as deduced from the information contained in the LEET spectra of nH
g
[55]. The nH
g
lm
has a band gap at the vacuum level with the lowest conduction band edge lying at V
0
= 0.8
eV [55]. This latter, combined with the attractive interaction of an incident electron with its
image charge, produces at the lmvacuum interface a potential well of a sucient depth
to support quasi-bound electron state(s). A realistic calculation for the nH
g
system [232]
predicts image states at 0.1 and 0.42 eV below the vacuum level with lifetimes greater than
3 nsec, assuming the decay of the image state via electron transfer through gap states.
Figure 13 Charging coecient A
s
(E
i
) as a function of incident electron energy E
i
for (a) a 10-ML
lm of glassy n-hexane nH
g
; (b) a 13-ML lm of crystalline n-hexane nH
c
; (c) 0.1 ML CH
3
Cl on a 5-
ML nH
g
surface; (d) the peak near zero energy in (c) shown on an extended scale; (e) 0.1 ML CH
3
Cl
on a 10-ML Kr surface; (f) 0.1 ML CH
3
Cl on 13 ML nH
c
surface. (From Ref. 232.)
Copyright 2004 by Taylor & Francis Group, LLC
Normally incident near-zero electron volt electrons scattered into nonspecular directions
will be captured into these states. While the image states are themselves too short-lived to
contribute to charge coecient measurements, the calculated lifetime is much longer than
phonon vibrational periods, so that the probability that the excess electron transfer to any
electron acceptor, be it O
2
, CH
3
Cl, or (H
2
O)
n
clusters, would be large. In contrast, on Kr
and nH
c
lms, the conduction band edge is lower [Fig. 14 (b)] [232] and the potential well
at the lmvacuum interface is no longer capable of supporting an image state of sucient
lifetime to transfer to the adsorbate.
Figure 14 Schematic diagram of simplied one-dimensional band structure for an nH
g
(top) and
Kr (bottom) spacer layer lm. (From Ref. 232.)
Copyright 2004 by Taylor & Francis Group, LLC
Electron-Induced Reactions
Impact of LEE on atomic and molecular solids has been observed to induce chemical
changes other than dissociation. In the next two sections, we will review examples of elec-
tron-induced reactions that proceed via formation of a TNI.
Anion Desorption Measurements. Reactive scattering by DEA fragment ions in con-
densed media was rst noted in the form of OH
and NO
2
+H
2
O!{H
3
O}*
!H
2
+OH
[46], and has thus been associated with part of the unscavengeable H
2
yield
formed in water radiolysis [247]. Another (post-DEA) ionmolecule reaction, directly
observed in the condensed phase for ion kinetic energies well below5 eV, is isotope exchange
measured in
18
O
2
/C
16
Omixed solids [248], i.e.,
18
O
+C
16
O!{C
18
O
16
O}*
18
!O
+C
18
O.
Formation of D
2
O in synchrotron-irradiated lms of N
2
O has also been attributed to
associative electron attachment reactions of DEA O
+D
2
!D
2
O*
!D
2
O+e
and Cl
desorption from
condensed CF
2
Cl
2
, have observed the appearance of new features in the Cl
yield function
at long bombardment times. These changes were attributed to the synthesis of Cl
2
in the
condensed phase by an as yet unidentied reaction pathway having DEA to CF
2
Cl
2
as its
initial step. Subsequent measurements of Cl
. (1) The OH
yield
functions resembled more closely the yield functions for O
signal froman O
2
lmincreased linearly with alkane surface coverage below1 ML.
(3) The threshold in energy for OH
desorption,
which is at least 2 eVbelow the H
withO
2
[253], the OH
C
n
H
m
! C
n
H
m
O*
! OH
C
n
H
m1
The existence of such C
n
H
m
O*
with C
2
H
4
. Further evidence for a complex anion formation and
reactive scattering in the gas phase can be seen in the studies of Parkes [255] and Lindinger
et al. [256].
Copyright 2004 by Taylor & Francis Group, LLC
More recent measurements [257] reproduced in Fig. 15(a) show the H
yield vs. E
i
.
An enhanced H
signal at low E
i
is observed from the O
2
/C
4
H
10
mixed lms. Two trends
can be observed in Fig. 15(a) with increasing O
2
concentration. Firstly, there is a decrease
in the yield of H
ESD
structure at lower energies, such that at an O
2
concentration of 20%, a substantial fraction
of the H
yield occurs at incident electron energies signicantly below those required for
DEA to C
4
H
10
. In Fig. 15(b), the H
desorption
from pure C
4
H
10
. The other (bold curve 2) is obtained from curve (1) when the H
signal
from a pure lm normalized to curve (1) at 10 eV (dotted line) is removed. This procedure
yields a second quasi-Gaussian-shaped structure centered at 7.8 eV, close to the energy of
maximum production of O
production at
incident electron energies below the threshold for DEA to C
4
H
10
was attributed to reactive
scattering of O
C
4
H
10
! C
4
H
10
O*
! H
C
4
H
9
O
Figure 15 (a) The H
production
via DEA to the alkane (dotted line), and another at 7.8 eV, associated with reactive scattering of O
production, as
evidenced from the similarity of the OH
and H
and CH
2
ESD
yield functions [259] and vibrational excitation functions [116] for condensed CH
3
OH,
Lepage et al. ascribed the shoulder at 11.5 eV and broad maximum centered at 14.5 eV in
Fig. 17 to the . . .(6aV)
1
(3saV)
2
,
2
A,. . .(1a)
1
(3saV)
2
,
2
A, and . . .(5aV)1(3saV)
2
,
2
AV core-excited
electron resonances, which decay into their parent repulsive states. The rising signal above
19 eV was attributed to direct ionization of CH
3
OH.
The same experimental techniques were applied to measure eective cross sections
for the electron-induced production of CO from condensed acetone [38], which again was
attributed to the formation of TNI and their decay to neutral dissociative states. HREELS
measurements have also been used to study electron-induced degradation of cyclopropane
[260] and CH
3
Cl on graphite [261].
4. RESULTS FOR LOW-ENERGY ELECTRON INTERACTIONS WITH DNA
AND OTHER BIOMOLECULES
4.1. LEE Damage to Plasmid DNA
The DNA molecule is composed of two strands of repeated sugar-phosphate units
hydrogen-bonded together by the bases (adenine, thymine, guanine, and cytosine) which
are covalently linked to the sugar moiety of the backbone. Results obtained from electron
bombardment of solid lms of DNA and its constituents are presented in the following
sections. These latter include H
2
O, the bases, and the backbone dioxyribose analogs such
as tetrahydrofuran.
Damage to pure dry samples of supercoiled DNA has been induced by bombard-
ment with 3 eV to 1.5 keV electrons under a hydrocarbon-free 10
9
Torr residual
atmosphere [262265]. These experiments required the adoption of sample preparation
and analytical techniques [263] unlike those already described for simple atomic and
molecular solids. They are briey discussed here. Supercoiled plasmid DNA [pGEM
3Zf(), 3199 base pairs] is rst extracted from Escherichia coli DH5a, puried, and
resuspended in nanopure water into a clean N
2
-lled glove box, where the following
manipulations are performed. An aliquot of this pure aqueous DNA solution is deposited
onto chemically clean tantalum substrates held at liquid nitrogen temperatures, lyophi-
lized with a hydrocarbon-free sorption pump at 0.005 Torr, and transferred directly in a
UHV chamber without exposure to air. After evacuation for
f
24 hr, room-temperature
DNA solid lms of 5-ML average thickness and 6-mm diameter are irradiated with a
monochromatic LEE beam of the same diameter. Each target is bombarded for a specic
time at a xed beam current density (2.210
12
electrons sec
1
cm
1
) and E
i
. After 24 hr
under UHV, the DNA still contains 2.5 H
2
O molecules/base pair as structural water [266].
After bombardment, the DNA is analyzed by agarose gel electrophoresis and quantied as
supercoiled (undamaged), nicked circle (single strand breakSSB), full-length linear
Copyright 2004 by Taylor & Francis Group, LLC
(double strand breakDSB), and short linear forms [264]. By measuring the relative
quantities of the various forms of DNA in the sample as a function exposure to 10, 30, and
50 eV electrons, it was possible to obtain the total eective cross section (
f
410
15
cm
2
)
and eective range (
f
13 nm) for destruction of supercoiled DNA at these energies [263].
Such experiments also allowed the identication of the regime in which the measured
yields were linear to electron exposure. It is within this regime that the incident electron
energy dependence of damage to DNA was recorded more continuously between 3 and
100 eV [262,263,265]. The most striking observations were that (1) these yields did not
depend on the ionization cross section and (2) below 15 eV, they varied considerably with
E
i
. Above 20 eV, the yields of SSB and DSB did not exhibit pronounced energy variations
nor did they increase considerably with increasing energy.
The incident electron energy dependence of these yields below 20 eV are displayed in
Fig. 18 and show that LEE induce SSBs and DSBs, even at electron energies below the
ionization limit of DNA (
f
8 eV) [267]. Also, the damage is highly dependent on the initial
kinetic energy of the incident electron, particularly below 14 to 15 eV, where thresholds
near 3 to 5 eV and intense peaks near 10 eV are observed. The results shown in Fig. 18 can
be understood by investigating the fragmentation induced by LEEs to the various subunits
of the DNA molecule, including its structural water.
Figure 18 Measured quantum yields, per incident electron, (a) for the induction of DSBs, (b)
SSBs, and (c) loss of the supercoiled DNA form, in DNA solids by low-energy electron irradiation as
a function of incident electron energy; the curves are guides to the eye. (From Ref. 262.)
Copyright 2004 by Taylor & Francis Group, LLC
4.2. Electron-Stimulated Desorption from Water Ice Films
The damage induced by LEE impact on amorphous ice lms has been measured by
recording H
[223,268,269], H
2
[46,270], D(
2
S), O(
3
P), and O(
1
D
2
) [47,271] desorption yield
functions in the range 530 eV. Most of these functions exhibit resonance structures below
15 eV which are characteristic of TNI formation. From anion yields, DEA to condensed
H
2
O was shown to result principally into the formation of H
via DEA
Copyright 2004 by Taylor & Francis Group, LLC
from I, II, and III has been discussed in detail by Antic et al. [272,273]. These authors
considered the possibility of H
and CH
2
OH
production, but these heavier ions were not observed to desorb. Typically, large mass
fragments do not possess sucient kinetic energy to escape induced polarization and thus
remain trapped within the lm [147]. Owing to the strong similarity of the H
desorption
proles for I, II, and III, these authors concluded that the majority of the anion yield for
all three systems arises from at least one transient anion associated with electron attach-
ment to the furan ring and located near 10 eV. Considering the largely Rydberg character
of the excited states in I near the energy range of the observed resonance, they further
suggested that this resonant state is of the core-excited type, possibly with dissociative
valence r* congurational mixing.
4.4. Anion Electron-Stimulated Desorption from Thin Films of DNA Bases
Electron impact dissociation of DNA bases has been investigated in the gas [274] and the
condensed phase [26,39,275]. In both phases, a large variety of stable anions is produced via
DEAto the bases for electron energies below30 eV. The anions H
, O
, CN
, CH
, OCN
,
and OCNH
2
, C
, O
, and/or NH
2
, CN
, OCN
,
C
4
H
5
N
3
, and/or CH
4
N
3
O
and C
3
H
3
N
2
, O
, OH
,
CN
, OCN
, and CH
2
can be induced via a simple bond cleavage fromthe bases, the production of CN
occurs under more complex ring dissociation. It shouldbe noted that the extractionof a CN
anion from any DNA base must involve more than a single bond cleavage. In other words,
during the lifetime of the negative ion, not only exocyclic single bond occurs, but also
complex endocyclic multibond cleavages are involved via most likely either stepwise [276] or
rearrangement reactions [277]. The similarity of the CN
, OH
, and O
, Cy
, or T
)R+(OCNH)*
+OH, OH
+CN, or O
+CNH.
4.5. Neutral Species Electron-Stimulated Desorption from Short Single
DNA Strands
In recent experiments, the damage produced by LEE impact on short single strands of
DNA was measured by ESD. Work in this area has been largely performed by Dugal et al.
[278,279] and Abdoul-Carime et al. [280283], who measured the yields of neutral
fragments induced by 130 eV electron impact on oligonucleotides consisting of 6 to 12
base units. The oligomers were chemisorbed on a gold surface via sulfur bounding. Their
results showed that LEE impact dissociation of DNA bases led to desorption of CN,
OCN, and/or H
2
NCN neutral species as the most intense observable yields. No sugar
moieties were detected; neither phosphorus-containing fragments nor entire bases were
observed to desorb. These results were obtained from mass spectrometric measurements of
the residual atmosphere near the target during its bombardment in UHV by a high current
10
8
A electron beam. Fig. 22 presents the electron energy dependence of neutral CN
(solid square) and OCN (and/or H
2
NCN) (open circle) fragments desorbed per incident
electron from oligomers consisting of nine bases [281]. The upper, middle, and lower
panels show the results from oligonucleotides consisting of nine cytosine bases, Cy
9
, six
cytosine bases and three thymine bases, Cy
6
T
3
, and six cytosine bases and three
bromouracil bases, Cy
6
BrU
3
, respectively. The latter base is not naturally found in
DNA but can be substituted for thymine and be incorporated into cellular DNA to
enhance the therapeutic eects of high-energy radiation [284]. Above 20 eV in Fig. 22, the
neutral fragment production increases linearly with the E
i
, which is indicative of molecular
fragmentation governed mostly by nonresonant DD and/or dissociative ionization of the
bases. Below 20 eV, base fragmentation may involve excitation to dissociative electronic
neutral states and DEA [278280]. The curves in Fig. 22 present broad peaks extending
from 7 to 15 eV for all oligomers. At these high energies, these broad maxima are likely to
reect the formation of core-excited resonances that are dissociative in the Franck
Condon region. This interpretation is supported by: (1) the electron energy losses in solid
phase DNA bases [285] in the 715 eV range, which are attributed to either p or r orbitals;
and (2) observation of resonant formation CN
, O
, and/or NH
2
and H
at 910, 79,
and 1213 eV, respectively, in electron-stimulated desorption yields from thin lms of
DNA bases [26,275]. Moreover, the threshold of neutral species production observed at 5
eV coincides with the threshold for electronic excitations, suggesting that the onset of the
nucleic acid fragmentation involves dissociative electronic transitions.
An extra peak is observed at 3 eV for BrU-substituted oligonucleotides in the inset at
the bottom of Fig. 22. Since this peak lies at an energy too low to involve electronically
Copyright 2004 by Taylor & Francis Group, LLC
excited states, it probably arises from the formation of a shape resonance consisting of the
BrU molecule in the ground state with an electron occupying a usually unlled orbital. In
XPS [39] investigations of LEE-induced damage to Bromouracil, it was found that Br
production below 5 eV involves resonant electron capture by BrU followed by dissociation
into a uracil-yl radical (U).
4.6. Discussion
In previous sections, we have seen that the dependence on E
i
of damage to elementary
constituents of DNA, probed in the form of desorbed anions, exhibits strong variations
Figure 22 Dependence on E
i
of neutral CN (solid square) and OCN (and/or H
2
NCN) (open circle)
fragment desorption yields per incident electron from Cy
6
-(Cy)
3
oligos (upper panel), Cy
6
-(Th)
3
oligos (middle panel), and C
6
-(BrU)
3
oligos (bottom panel). The spread in the data is estimated to be
20%. (From Ref. 281.)
Copyright 2004 by Taylor & Francis Group, LLC
due to electron resonances. Maxima in the anion desorption yields are due to DEA, while
those appearing in the neutral radical yields may result from either DEA or the decay of a
transient anion into an electronically excited state which is dissociative. Thus from these
data, it becomes quite obvious that the strong energy dependence of the DNA strand
breaks below 15 eV in Fig. 18 can be attributed to the initial formation of transient
anions, decaying into the DEA and/or dissociative electronic excitation channels. How-
ever, since the basic DNA components (i.e., the sugar and base units and structural H
2
O)
can all be fragmented via DEA between 5 and 13 eV, it is not possible to unambiguously
attribute SSB and DSB to the initial dissociation of a specic component. For example,
the maximum in the SSB yield lying near 8 eV is very close to the maximum in H
+OH
production from H
2
O, but the DNA bases also produce H
X H
2
O
2
B
1
;
1
A
1
! H
2
S
0
OH
2
P
1
! CF
4
! F
CF
3
CF
4
e
! CF
4
! F CF
3
as an initial step for further reactions of fragments with the Si substrate.
* For a review, see Ref. 289.
y
For example, Ref. 292.
Copyright 2004 by Taylor & Francis Group, LLC
5.3. Atmospheric Processes
In a recent review, Madey et al. [296] describe how in interplanetary space, energetic
photons, ions, and electrons from the solar wind, together with galactic and extragalactic
cosmic rays, constantly bombard surfaces of planets, planetary satellites, dust particles,
comets, and asteroids. The authors suggest that much can be learned about the tenuous
atmospheres surrounding these bodies by studying the interactions of energetic particles
producing LEE secondary electrons with the molecular constituents of their surfaces. For
example, Sieger et al. [10] have demonstrated the production of O
2
from ice lms by
electron impact at 200 eV and have posited this mechanism as a source of O
2
in the
atmospheres of Ganymede and Europa.
The presence of ionizing radiation in the upper regions of the earths atmosphere and
the realization that atmospheric chemistry can occur on the surface of ice and dust
particles have lead many authors to study on the interaction of LEEwith molecular solids of
ozone [203], HCl [236], and halogen-containing organic compounds [176,177,195197,199
202,205,214,217,224234] in aneort toshed newlight onthe problemof ozone depletion. In
a recent series of experiments, Lu and Madey [297,298] found that the F
and Cl
yields
Figure 24 Experimental cross sections for the electron-induced oxidation of the a-H:Si(111)
surface as a function of incident electron energy. The solid line is drawn to guide the eye. (From Ref.
294.)
Copyright 2004 by Taylor & Francis Group, LLC
produced by the impact of 250-eVelectrons on CF
2
Cl
2
adsorbed on a metal Ru surface were
enhanced by several orders of magnitude when co-adsorbed with polar molecules such as
H
2
O and NH
3
. Subsequent charge-trapping measurements by Lu and Sanche [235] also
display an enhancement in stable anion formation at electron impact energies near 0 eV. Fig.
25(a) shows charging coecient A
s
(E
i
) for 10 ML of Kr deposited on Pt foil (solid triangles)
and that for 0.1 ML of CF
2
Cl
2
deposited on the Kr surface (open squares). These results can
be compared to those for 5 MLof H
2
Oon 10 MLof Kr and for 0.1 MLof CF
2
Cl
2
on 5 MLof
H
2
O on Kr [solid circles and open diamonds, respectively, in Fig. 25(b)]. In contrast to pure
Kr lms, which do not trap electrons, both H
2
O- and CF
2
Cl
2
-covered lms show signicant
charging. The results for H
2
O are similar to earlier measurements [222,223]. Those for
CF
2
Cl
2
correspond to a maximum trapping cross section of 1.410
15
cm
2
[235] near 0 eV,
attributed to charge stabilization as Cl
2
r
0
r
c
r
2
0
_ _
du
dr
0
0 9
The escape probability satises the following boundary condition at innity
ur
0
!l 1 10
In the equations written so far, the reaction between the particles of the geminate pair is
not accounted for. This can be performed by introducing a boundary condition at a
distance R, which is called the reaction distance. If the reaction on encounter is very fast,
one can use the totally absorbing boundary condition
wR; t 0 11a
Equation (11a) is sometimes called the absorption boundary condition. This boundary
condition is equivalent to the following boundary condition on the escape probability [4]
uR 0 11b
By solving Eq. (9) subject to boundary conditions (10) and (11b), the escape probability
for the totally diusion-controlled geminate ion recombination is calculated as
ur
0
exp
r
c
r
0
_ _
exp
r
c
R
_ _
1 exp
r
c
R
_ _ 12
Equation (12) is derived on the assumption of an instantaneous reaction of the opposite
ions once they approach each other to the distance R. However, in many real systems, the
rate of the reaction at the reaction distance is comparable to, or even slower than, the
diusive ux of the ions toward each other. This eect can be taken into account by
Copyright 2004 by Taylor & Francis Group, LLC
modifying the boundary condition at the reaction distance R. Instead of the totally
absorbing boundary condition (11a), one can use a partially reecting boundary condition
in the form [5]
D
Bwr; t
Br
wr; t
k
B
T
BVr
Br
_ _
rR
pwR; t 13a
where the left-hand side represents the ux of particles through the reaction surface, and
the parameter p is a measure of the intrinsic rate constant at the reaction distance.
Equation (13a) is sometimes called the radiation boundary condition. This boundary
condition leads to the following boundary condition on the escape probability [3]
D
Bur
0
Br
0
_ _
r
0
R
puR 13b
In the limit p!l, Eqs. (13a) and (13b) reduce to Eqs. (11a) and (11b), respectively. The
solution of Eq. (9) subject to the boundary conditions (10) and (13b) is given by [3]
ur
0
exp
r
c
r
0
_ _
1
Dr
c
pR
2
_ _
exp
r
c
R
_ _
1 1
Dr
c
pR
2
_ _
exp
r
c
R
_ _
14
This expression characterizes the escape probability for the partially diusion-controlled
geminate ion recombination.
In the case where the reaction distance may be assumed much shorter than the
range of the Coulomb interaction (R<<r
c
), Eq. (14) reduces to the well-known Onsager
formula [2]
ur
0
exp
r
c
r
0
_ _
15
This simple expression, which gives the probability that a pair of charges initially separated
by r
0
and performing a diusive motion in the mutual Coulomb eld will escape
recombination, has found numerous applications in various elds of physics and chemistry.
Another method of accounting for the reaction between the electron and the ion,
instead of imposing a specic boundary condition, is to introduce a sink termk
r
(r)w(r,t)
on the right-hand side of Eq. (2), where k
r
(r) is a rst-order rate constant [6]. This
approach is more general than that using the totally absorbing or the partially reecting
boundary conditions, in a sense that it is applicable to the case where the reaction occurs
over a range of distances. Moreover, it does not even require introducing the reaction
distance, which is a quantity dicult to dene in a real system. If the sink term is used to
represent the reaction, the escape probability satises
D
d
2
u
dr
2
0
2
r
0
r
c
r
2
0
_ _
du
dr
0
_ _
k
r
r
0
ur
0
0 16
When the reaction is represented by a sink term, a reective boundary condition must be
imposed at a contact distance d. This boundary condition leads to the following boundary
condition on the escape probability
Bur
0
Br
0
_ _
r
0
d
0 17
Copyright 2004 by Taylor & Francis Group, LLC
The solution of Eq. (16) depends on the assumed rate constant. In many cases, an
exponential form of the rate constant k
r
(r)
f
exp(ar), where a is a parameter, provides a
realistic model of the reaction. Unfortunately, it is usually not possible to nd simple
analytical solutions of Eq. (16), and the equation has to be numerically solved [7].
We have derived the escape probability for a pair of charges initially separated by a
given distance r
0
for various cases. However, in real systems, the electron thermalization
distance is distributed. If we denote the distribution of thermalization distances by f(r), the
total averaged escape probability, u
tot
, can be calculated from
u
tot
_
l
0
urfrdr 18
The exact form of f(r) is usually unknown, and the electron thermalization in irradiated
systems still remains a subject of both theoretical and experimental scrutiny. Often, f(r) is
approximated by a simple analytical expression, e.g., the exponential distribution
fr; r
0
1
r
0
exp
r
r
0
_ _
19
or the Gaussian distribution
fr; r
0
4r
2
p
p
r
3
0
exp
r
2
r
2
0
_ _
20
where r
0
is a parameter.
One of the most important experimental methods of studying the electronion
recombination processes in irradiated systems are measurements of the external electric
eld eect on the radiation-induced conductivity. The applied electric eld is expected to
increase the escape probability of geminate ion pairs and, thus, enhance the number of free
ions in the system, which will result in an enhanced conductivity.
The escape probability of an electron in an external electric eld F can generally be
obtained by solving Eq. (7) or (16) with the potential energy V given by
Vr; #
e
2
er
eFr cos # 21
Here # is the polar angle. We assume that the eld is directed along #=p. The boundary
condition at innity is given by Eq. (10), and the reaction may be described either by a
specic boundary condition at the reaction distance [Eq. (7)], or by a sink term [Eq. (16)],
as discussed before.
The solution of Eq. (7) with the potential energy described by Eq. (21), for the
boundary condition (11b) with R=0, was found by Onsager [2]. His result is given by
ur
0
; #; F exp
_
br1 cos #
_
l
r
c
=r
J
0
2
_
br1 cos #s
1=2
_ _
e
s
ds 22
where b=eF/2k
B
T and J
0
is a zeroth-order Bessel function of the rst kind. Because in real
systems, the angular dependence of the escape probability is usually unobservable, the
quantity that is experimentally more signicant is the escape probability u(r
0
,F ) averaged
over the polar angle #. Although no closed solution for u(r
0
,F ) is available, the eld
Copyright 2004 by Taylor & Francis Group, LLC
dependence of the angle-averaged escape probability can be described by an innite series
representation [8]
ur
0
; F expa b
1
b
l
l 1
l
b
a
_ _
l=2
I
l
2
ab
p
_ _
23
where a=r
c
/r
0
, b=eFr
0
/k
B
T, and I
l
are modied Bessel functions. A couple of dierent
series representations of u(r
0
,F ) are also available [9,10].
It can be shown that the expansion of u(r
0
,F ) up to the rst order in F gives the
expression
ur
0
; F exp
r
c
r
0
_ _
1
eFr
c
2k
B
T
_ _
24
This equation shows that at low electric elds, the escape probability is a linear function of
F, and the slope-to-intercept ratio of this dependence is given by er
c
/2k
B
T. It is worth
noting that this ratio is independent of r
0
. Therefore plots of u( F )/u(0) vs. F may be used
to test the applicability of the presented theory to describe real systems, even if the dis-
tribution of electron thermalization distances is unknown.
Calculation of the electric eld dependence of the escape probability for boundary
conditions other than Eq. (11b) with R=0 poses a serious theoretical problem. For the
partially reecting boundary condition imposed at a nonzero R, some analytical treat-
ments were presented by Hong and Noolandi [11]. However, their theory was not de-
veloped to the level, where concrete results of u(r
0
,F ) for the partially diusion-controlled
geminate recombination could be obtained. Also, in the most general case, where the
reaction is represented by a sink term, the analytical treatment is very complicated, and the
only practical way to calculate the eld dependence of the escape probability is to use
numerical methods.
In Fig. 1, we plot the electron escape probability of geminate ion pairs as a function
of the external electric eld. The solid lines show the results calculated using Eq. (23) for
various initial separations r
0
. The broken lines represent the results of u(r
0
,F ) obtained by
numerically solving Eq. (16), which includes the sink term k
r
u to represent the reaction
[12]. The rate constant k
r
is assumed in the form k
r
(r) =Aexp[a(rd )], and the
parameters are chosen in such a way that the recombination is partially diusion-
controlled. We see that the electric eld dependence of the escape probability in the
partially diusion-controlled recombination is signicantly dierent from that in the
totally diusion-controlled recombination. This indicates that the eects due to the nite
rate of the reaction cannot be neglected in analyzing experimental data. We should also
bear in mind that in real systems the electron thermalization distance r
0
is distributed. To
calculate the escape probability for a given thermalization distance distribution f(r), one
has to average u(r,F ) over f(r).
In the preceding part of this section, we have concentrated on the electron escape
probability, which is an important quantity in the geminate phase of recombination, and
can be experimentally observed. However, modern experimental techniques also give us a
possibility to observe the time-resolved kinetics of geminate recombination in some
systems. Theoretically, the decay of the geminate ion pairs can be described by the pair
survival probability, W(t), dened by Eq. (4). One method of calculating W(t) is to solve
the Smoluchowski equation [Eq. (2)] for w(r,t) and, then, to integrate the solution over the
space variable. Another method [4] is to directly solve Eq. (7) under relevant conditions.
Copyright 2004 by Taylor & Francis Group, LLC
The analytical solution of the Smoluchowski equation for a Coulomb potential has
been found by Hong and Noolandi [13]. Their results of the pair survival probability, ob-
tained for the boundary condition (11a) with R = 0, are presented in Fig. 2. The solid lines
show W(t) calculated for two dierent values of r
0
. The horizontal axis has a unit of r
2
c
=D,
which characterizes the timescale of the kinetics of geminate recombination in a particular
system For example, in nonpolar liquids at room temperature r
2
c
=D
f
10
8
sec. Unfortu-
nately, the analytical treatment presented by Hong and Noolandi [13] is rather compli-
cated and inconvenient for practical use. Tabulated values of W(t) can be found in Ref. 14.
The pair survival probability of geminate ion pairs can also be calculated numerically [15].
In some cases, numerical methods may be a more convenient approach to calculate W(t),
especially when the reaction cannot be assumed as totally diusion-controlled.
It can be shown that for the totally absorbing boundary condition, and for R<<r
c
,
the long-time behavior of the pair survival probability W(t) is described by
Wt exp
r
c
r
0
_ _
1
r
c
pDt
p
_ _
25
Figure 1 Electron escape probability as a function of the applied electric eld. The solid lines
are obtained from Eq. (23) for dierent values of the initial electron cation distance r
0
. The broken
lines are calculated for r
0
= 1 nm from the numerical solution of Eq. (16) with the sink term given by
k
r
(r) = A exp[a(rd)], where a = 10 nm
1
and d = 0.6 nm. Dierent lines correspond to dierent
values of A from 10
13
(the lowest broken curve) to 10
8
, in decadic intervals. The parameter values
were assumed as e = 4, T = 298 K, and D = 510
3
cm
2
/sec.
Copyright 2004 by Taylor & Francis Group, LLC
This asymptotic form of W(t) is also plotted in Fig. 2 (as dashed lines). However, we see
that this approximation is valid only when the pair survival probability slightly exceeds the
escape probability, so its practical importance is rather limited.
When a strong electron acceptor (electron scavenger) is added in the irradiated
system, some of the electrons will react with the scavenger before they recombine with
their parent cations. Therefore the measurement of the yield of the scavenging products as
a function of the scavenger concentration can be used to monitor the geminate recombi-
nation, and the electron scavenging technique has proved to be an important tool in ex-
perimental studies.
The geminate recombination in the presence of a scavenger can be described by the
Smoluchowski equation [Eq. (2)] with an additional term representing the loss of the
geminate pairs by scavenging reactions
Bw
s
Bt
D j
2
w
s
1
k
B
T
jw
s
jV
_ _
k
s
C
s
w
s
r; t; C
s
26
Here k
s
denotes the rate constant for the scavenging reaction and C
s
is the scavenger
concentration. The probability density for the geminate pairs, w
s
(r,t,C
s
), may be related to
the corresponding probability density in the absence of scavenger, w(r,t), through
w
s
r; t; C
s
wr; texpk
s
C
s
t 27
Figure 2 Survival probability of geminate ion pairs as a function of time. The two solid lines
correspond to two dierent values of the initial electroncation distance. The broken lines show the
asymptotic kinetics calculated from Eq. (25). The value of the escape probability for r
0
= 0.5r
c
is
indicated by u. (From Ref. 13.)
Copyright 2004 by Taylor & Francis Group, LLC
which is easily veried by substitution. Experimentally important is the total scavenging
probability u
s
(k
s
C
s
), which is dened as the probability that an electron reacts with a
scavenger before recombination with a positive ion. By making use of Eq. (27), u
s
(k
s
C
s
)
was shown to satisfy the relation
u
s
k
s
C
s
k
s
C
s
_
l
0
Wtexpk
s
C
s
tdt k
s
C
s
Wk
s
C
s
28
where Ws stands for the Laplace transform of W(t). This equation shows that the total
scavenging probability is proportional to the Laplace transform, Wk
s
C
s
, of the pair
survival probability in the absence of scavenger W(t) [16]. Eq. (28) has a great practical
importance, as it allows us to calculate the lifetime distribution of the geminate pairs from
measurements of the yield of the scavenging products by inverse Laplace transformation.
This equation may also be considered as a generalization [17] of the SternVolmer relation
widely used in photochemistry to the case where the survival probability W(t) is not de-
scribed by an exponential function.
The theory of geminate recombination presented so far is based on the solutions of the
Smoluchowski equation for an isolated pair of a cation and an electron, which are assumed
to perform a diusive motion in each others Coulomb eld. The single-pair approach has
proved to be very useful in modeling early stages of the electronion recombination
processes in various systems, and was applied to interpret numerous experimental results.
There are physical situations, especially in photochemical studies, where ionization is weak
and the geminate electronion pairs are separated from one another by distances much
larger than the Onsager radius. In such systems, the single-pair model of geminate re-
combination is directly applicable. On the other hand, ionization events in radiation tracks
are known to form specic spatial patterns, in which only a fraction of the produced
electroncation pairs may be treated as independent of one another. The single-pair model is
then only an approximation, and the multipair eects due to the interactions between the
neighboring pairs of charges have to be taken into account.
Analytical treatment of the diusion-reaction problem in a many-body system
composed of Coulombically interacting particles poses a very complex problem. Except
for some approximate treatments, most theoretical treatments of the multipair eects
have been performed by computer simulations. In the most direct approach, random
trajectories and reactions of several ion pairs were followed by a Monte Carlo simulation
[18]. In another approach [19], the approximate Independent Reaction Times (IRT)
technique was used, in which an actual reaction time in a cluster of ions was assumed to
be the smallest one selected from the set of reaction times associated with each independent
ion pair.
An important result of the theoretical studies of the multipair eects is that the
recombination kinetics in a cluster of ions, in which the initial separation between
neighboring cations is
f
1 nm, is faster than the corresponding decay kinetics of a single
ion pair [18]. Furthermore, the escape probability is lower than the Onsager value [Eq.
(15)], and decreases with increasing number of ion pairs in the cluster (a relative decrease
of about 30% for two ion pairs, and about 50% for ve ion pairs). The average electron
escape probability in radiation tracks obviously depends on the distribution of ionization
events in the tracks, which is determined by the type of radiations and their energy.
Calculations of the external electric eld eect on the escape probability in multipair
clusters [18] have shown that signicant deviations from the results of the single-pair
theory are expected at high electric elds. At low elds, the escape probability is a linear
Copyright 2004 by Taylor & Francis Group, LLC
function of the eld, and the slope-to-intercept ratio of this dependence is almost the same
as that obtained in the single-pair theory [20] [cf. Eq. (24)].
A study of electron scavenging in multipair clusters [21] has shown that the total
scavenging probability decreases with increasing number of ion pairs in the cluster.
However, the Laplace transform relationship [Eq. (28)] between the scavenging probabil-
ity and the recombination kinetics was found to work reasonably well also in the multipair
case.
In this section, a theory of geminate recombination of charged species is presented by
fully taking into account the Coulomb interaction between the species. While for nonpolar
systems, the theory of geminate ion recombination in its full form has to be used, some
simple approximate theories can be used for systems of high polarity. An important ex-
ample is water, where the Onsager radius is only 0.7 nm, and is shorter than typical elec-
tron thermalization distances of a few nanometers. In this case, the Coulomb interaction
between the ionization products is not so important and, in the rst approximation, it may
even be neglected, as it was performed in numerous theoretical studies. The advantage of
this approach is that the analytical expression describing the geminate recombination of
uncharged species are much simpler than those for the geminate ion recombination.
However, theoretical modeling of the geminate recombination processes in water is still
complicated, even if one neglects the Coulomb interactions. This is because the primary
ionization products in water undergo a series of chemical reactions, which proceed on the
timescale shorter than that of the charge recombination.
Geminate Ion Recombination in High-Mobility Systems
The theory of geminate recombination presented in Sec. 10.1.2 (Diusion-Controlled
Geminate Ion Recombination) is based on the assumption that the relative motion of
electrons and cations that constitute geminate pairs is described by the diusion equation.
In a real system, the motion of a particle may be treated as ideal diusion when its scat-
tering by atoms or molecules of the system is very frequent, and its mean free path, k, is
much shorter compared with the typical length involved in the problem. In the case of the
ion recombination, it is reasonable to take as the typical length involved in the problem
the range of the Coulomb interactions in the considered system, so the applicability of the
diusion theory of geminate recombination is theoretically limited to systems where the
scattering mean free path is much shorter than the Onsager radius (k<<r
c
). In a simple
model, the mean free path of charged particles is proportional to their mobility l. There-
fore the diusion theory of geminate recombination will not be applicable, when the
mobility of charges is too high. In fact, deviations from the diusion theory were ex-
perimentally observed in systems where the electron mobility exceeds
f
100 cm V
1
sec
1
.
Limitations of the diusion approach also become apparent from the considerations
based on energy. In the conventional diusion approach, it is assumed that particles are
always in thermal equilibrium with the system, and that their velocity distribution is close
to the MaxwellBoltzmann distribution. When the mean free path is long, and the scat-
tering events not so frequent, electrons are no longer able to eectively dissipate the kinetic
energy they acquire in the Coulomb eld of the cations. The average electron energy is
then signicantly higher than the thermal level.
In this physical picture of geminate recombination in high-mobility systems, it is no
longer valid to use the space criterion of reaction expressed in terms of the distance.
Instead, the energy criterion of reaction expressed in terms of energy or, more preferably,
the energy-dependent reaction cross section should be used.
Copyright 2004 by Taylor & Francis Group, LLC
Because of the inecient dissipation of their kinetic energy, electrons in high-mobility
systems are more likely to escape geminate recombination than predicted by the diusion
theory. Such an eect has been conrmed by the results of a computer simulation study [22]
presented in Fig. 3. In this study, classical electron trajectories in the Coulomb eld of a
cation were calculated, starting at a distance r
0
from the cation, and the electron scattering
was modeled in such a way that the electron velocity was randomized at each collision to
restore the MaxwellBoltzmann distribution. The times between the collisions were selected
from an exponential distribution, and its mean value, s, was used to characterize the
strength of the dynamic interaction between the electron and the solvent. The mean free
time s may be related to the mean free path by k/r
cc
1.6s. For each electron, the simulation
was carried out until either the reaction occurred, or the electron was separated from the
cation to a distance much larger than r
c
. An energy criterion of reaction was used, namely,
the electron was assumed to have reacted once its total energy in the Coulomb eld of the
cation was reduced below 10k
B
T.
According to Fig. 3, the Onsager result [Eq. (15)], which shows the escape prob-
ability in the diusion-controlled geminate recombination, gives the lower bound for the
simulation results. The simulation results obtained for the lowest value of s (s = 0.05) are
Figure 3 Escape probability as a function of the initial electroncation distance. The lower broken
curve is calculated from the Onsager equation [Eq. (15)]. The numerical results for dierent mean
free times s were taken from Ref. [22]. The unit of s is r
c
/(k
B
T/m)
0.5
, where m is the electron mass.
The upper broken curve was calculated using the energy diusion model. (From Ref. 23.)
Copyright 2004 by Taylor & Francis Group, LLC
very close to the Onsager result. At higher values of the mean free time, the escape
probability exceeds the Onsager result and signicantly increases with increasing s.
The upper bound of the escape probability, corresponding to s!l, can also be
analytically calculated. In this limit, the electron is allowed to make many revolutions in
the same orbit around the cation before it is scattered into another orbit. Under this
condition, the electron motion may be described as diusion in energy space [23]. The
escape probability calculated by using the energy diusion model is also included in Fig. 3.
We see that the simulation results for nite s properly approach the energy diusion limit.
Computer simulation has also been used to calculate the external electric eld eect
on the geminate recombination in high-mobility systems [22]. For the mean free time s
exceeding
f
0.05, the eld dependence of the escape probability was found to signicantly
deviate from that obtained from the diusion theory. Furthermore, the slope-to-intercept
ratio of the eld dependence of the escape probability was found to decrease with
increasing s. Unlike in the diusion-controlled geminate recombination, this ratio is no
longer independent of the initial electron-ion separation [cf. Eq. (24)].
The model of electron scattering in high-mobility systems applied in the simulations
is rather simplied. Especially, the assumption that the electron velocity is randomized at
each scattering to restore the MaxwellBoltzmann distribution may be an oversimplica-
tion. If the dissipation of energy by electron collisions in a real system is less ecient than
that assumed in the simulation, the escape probability is expected to further increase.
The electronion recombination in high-mobility systems has also been analyzed in
terms of the fractal theory [24,25]. It was postulated that even when the fractal dimension
of particle trajectories is not equal to 2, the motion of particles is still described by
diusion but with a distance-dependent eective diusion coecient DV(r) =D(1+l/
r)
1
, where the parameter l is proportional to the mean free path k [24]. However, when
the fractal dimension of trajectories is not equal to 2, the motion of particles is not
described by orthodox diusion.
1.3. Bulk Ion Recombination
Diusion-Controlled Bulk Ion Recombination
When the motion of electrons and positive ions in a particular system may be described as
ideal diusion, the process of bulk recombination of these particles is described by the
diusion equation. The mathematical formalism of the bulk recombination theory is very
similar to that used in the theory of geminate electronion recombination, which was de-
scribed in Sec. 10.1.2 (Diusion-Controlled Geminate Ion Recombination). Geminate
recombination is described by the Smoluchowski equation for the probability density w(r,t)
[cf. Eq. (2)], while the bulk recombination is described by the diusion equation for the space
and time-dependent concentration of electrons around a cation (or vice versa), c(r,t),
Bcr; t
Bs
D j
2
cr; t
1
k
B
T
jcr; tjVr
_ _
29
The two equations have the same mathematical form, although they deal with dierent
physical quantities. The initial and boundary conditions are also somewhat dierent. In the
bulk recombination, the initial electron concentration is assumed to be uniformin the space
and equal to the average concentration c
0
cr; 0 c
0
30
Copyright 2004 by Taylor & Francis Group, LLC
The boundary condition at innite separation is given by
lim
r!l
cr; t c
0
31
where we assume that the average electron concentration changes little during the reaction.
The methods used in geminate recombination to account for the reaction are also
applicable to bulk ion recombination. In the simplest case of fully diusion-controlled
recombination, the reaction can be represented by the totally absorbing boundary con-
dition
cR; t 0 32
where R is the reaction distance. For partially diusion-controlled recombination, we may
use the boundary condition in the form
D
Bcr; t
Br
cr; t
k
B
T
BVr
Br
_ _
rR
pcR; t 33
where the left-hand side represents the ux through the reaction surface, and the reactivity
parameter p is a measure of the intrinsic rate constant at the reaction distance.
An example of numerical solutions of Eq. (29) for the attractive Coulomb potential,
subject to the initial and the boundary conditions (30)(32), is presented in Fig. 4 [26]. The
gure shows how the initially uniform electron concentration gradually decreases in the
vicinity of the cation until the steady state is established. The rate of reaction is determined
by the diusive ow of electrons from large distances toward the reaction sphere.
Figure 4 Time evolution of the normalized concentration of electrons around the cation c(r,t)/c
0
for D=10
4
cm
2
/sec and r
c
= 0.7 nm. Dierent curves correspond to dierent times from 10
14
to
10
7
sec, in decadic intervals. (From Ref. 26.)
Copyright 2004 by Taylor & Francis Group, LLC
The time-dependent rate coecient for the bulk ion recombination may be obtained
as the total current of particles owing through the reaction surface at r = R, divided
by c
0
,
kt 4pR
2
D
Bcr; t
Br
cr; t
k
B
T
BVr
Br
_ _
rR
_
c
0
34
The long-time limit of k(t) denes the steady state rate constant of the bulk ion recom-
bination, which for the totally diusion-controlled recombination is calculated as [27]
k
4pDr
c
1 exp
r
c
R
_ _ 35
When the reaction distance is short in comparison with the range of the Coulomb inter-
action (R<<r
c
), Eq. (35) reduces to the so-called DebyeSmoluchowski expression
k
D
4pDr
c
36a
This expression is a fundamental result of the theory of bulk ion recombination and has
been extensively used in interpreting experimental results of diusion controlled reactions.
The DebyeSmoluchowski expression can also be written in terms of the mobility,
k
D
4pe
e
l 36b
where we have used the Einstein relation between the diusion coecient and the mobility
l/D=e/k
B
T.
In the case of partially diusion-controlled recombination, which is described by Eq.
(33), the recombination rate constant is calculated as [28]
k
4pDr
c
1 1
Dr
c
pR
2
_ _
exp
r
c
R
_ _
37
The rate constants given by Eqs. (35)Eqs. (37) give the steady state rate constants of bulk
ion recombination. As seen from Fig. 4, it takes some time before the steady state is
established, and, in general, the rate coecient of the bulk ion recombination is time-
dependent. Although no simple analytical expression is available, which describes k(t) in
the whole time domain, the expression for the asymptotic behavior of k(t) at long times is
easily obtained. For fully diusion-controlled recombination, we have
kt !l
4pDr
c
1 expr
c
=R
1
expr
c
=R
1 expr
c
=R
r
c
pDt
p
_ _
38
It was shown [26] that in water (r
c
=0.7 nm), the transient eect in the bulk ion recom-
bination may be signicant on the timescale shorter than a nanosecond. For nonpolar
systems (r
c
f
30 nm), the transient eect in Eq. (38) is much smaller, and the time de-
pendence of the rate coecient is not likely to be experimentally observed.
Bulk recombination is a many-body problem. It is not so obvious whether the ap-
proach based on Eqs. (2934) correctly describes the many-body character of bulk recom-
bination. Tachiya [29] formulated the rate of bulk recombination in terms of the pair
survival probability of geminate recombination and showed that the approach described
above is exact only when the minority reactants are xed and the majority reactants are
Copyright 2004 by Taylor & Francis Group, LLC
mobile (target problem). The above approach is not exact when the majority reactants are
xed and the minority reactants are mobile (trapping problem), or when the concen-
trations of two reactants are comparable, although it gives a very good approximation in
most cases of interest.
Because of its importance for experimental studies, we will also discuss the eect of
an applied electric eld on the rate constant of bulk ion recombination. Within the frame-
work of the diusion theory, this eect can be calculated by solving the diusion equation
[Eq. (29)] with the potential energy given by Eq. (21) under relevant initial and boundary
conditions. Analytical solutions of this problem, both for the totally absorbing [Eq. (32)]
and the partially reecting [Eq. (33)] boundary conditions, are found in Ref. 30. In the case
of totally diusion-controlled recombination, the rate constant is found to increase with
increasing eld strength. The general expression for the electric eld dependence of the
recombination rate constant, k( F), obtained in Ref. 30 is quite complicated and involves
an innite series summation. An expansion of this expression in the Taylor series up to the
rst order in F gives [30]
kF
4pDr
c
1 expr
c
=R
1
expr
c
=R
1 expr
c
=R
eFr
c
2k
B
T
_ _
39
The increase of k with increasing eld strength is explained by an enhanced rate of en-
counter between positively and negatively charged particles because of their drift in the
opposite directions in the presence of an external eld. However, in nonpolar systems, this
eect is very weak and the recombination rate constant is almost independent of the elec-
tric eld.
The results obtained in Ref. 30 for partially diusion-controlled recombination show
that the eld dependence of the recombination rate constant is aected by both the
reaction radius R and the reactivity parameter p [cf. Eq. (33)]. Depending on their relative
values, the rate constant can be increased or decreased by the electric eld. The latter eect
predominates at low values of p, where the reactants staying at the encounter distance are
forced to separate by the electric eld.
Because of space limitation, we have discussed only a part of aspects of the theory of
diusion-controlled ionic processes, which we believe to be essential for understanding the
bulk phase of ion recombination in irradiated systems. Discussions of other aspects of
diusion-controlled ionic processes are found in monographs on the diusion-controlled
reactions [31]. One important aspect that was only briey touched in this section concerns
the theoretical models to account for the reaction between ions (the nal step of recombi-
nation). We have discussed the models, which introduce specic boundary conditions at the
reaction distance [Eqs. (32) and (33)] to account for the reaction. In a more general
approach, the reaction may be accounted for by introducing a sink term, as it was considered
in geminate recombination [cf. Eqs. (16) and (17) in Sec. 10.1.2 (Diusion-Controlled
Geminate Ion Recombination)]. Such an approach allows us to treat the case where the
reaction occurs not only at the contact distance, but over a range of distances, as in the case
of electron transfer reactions. Another important aspect concerns the inuence of the
solvent structure on the diusive transport of ions. This is exhibited as the potential of mean
force and hydrodynamic eects. However, these eects are usually overwhelmed by the
Coulomb attraction between the ions, and may be safely neglected, especially in nonpolar
systems. The competition eects in the diusion-controlled bulk ion recombination are also
expected to inuence the recombination rate constant at high ion concentrations, although
they may not be very important in normal experimental conditions [32].
Copyright 2004 by Taylor & Francis Group, LLC
The theory of diusion-controlled recombination presented in Secs. 10.1.2 (Dif-
fusion-Controlled Geminate Ion Recombination) and 10.1.3 (Diusion-Controlled
Bulk Ion Recombination) may not be applicable to a class of disordered solid state
systems, including amorphous semiconductors and organic glasses, where the charge
transport is strongly inuenced by the processes of trapping and detrapping of charge
carriers. The transport of charged particles in such systems is usually described as the
continuous time random walk [33,34] or using the fractional diusion approach [35].
These two approaches have also been used in the theoretical studies of both geminate and
bulk ion recombination in the disordered systems [36,37]. However, in those studies, the
Coulomb interaction between the charged particles was not taken into account.
Bulk Ion Recombination in High-Mobility Systems
The diusion theory of bulk electronion recombination presented in Sec. 10.1.3 (Dif-
fusion-Controlled Bulk Ion Recombination) is not applicable to systems where the elec-
tron mobility is high. If the mobility exceeds
f
100 cm
2
V
1
sec
1
, eects due to the nite
mean free path for electron scattering become signicant, and the electron motion can no
longer be described as ideal diusion. The general physical picture of electron transport
processes and reactions in high-mobility systems is essentially the same both in the
geminate and the bulk phases of recombination, and has already been discussed in Sec.
10.1.2 (Geminate Ion Recombination in High-Mobility Systems). In short, when the
electron mean free path increases, the character of its motion gradually changes from
diusive to ballistic. In place of the spatial diusion of electrons and cations toward one
another, the rate-determining process for their recombination becomes the dissipation of
the kinetic energy the particles acquire in the mutual Coulomb eld. In consequence, the
electronion recombination rate constants in high-mobility systems are expected to be
signicantly lower than the DebyeSmoluchowski value k
D
predicted by the diusion
theory [cf. Eq. (36)]. This was conrmed by the experimental results, which will be dis-
cussed later in this chapter.
An estimate of the electronion recombination rate constant in high-mobility systems
based on an empirical model of energy dissipation processes was provided by Warman [38].
He related the rate constant to the eld dependence of the electron mobility, and proposed
an empirical formula
k
k
D
1
55T
4
e
2
F
2
10
_ _
1
40
Here F
10
denotes the critical electric eld strength expressed in units of V/m, at which the
electron mobility deviates 10% from the thermal mobility, and is used as a measure of the
rate of the electron energy dissipation in a particular system. Despite its simplicity, Eq. (40)
is shown to give reasonable estimates of the electronion recombination rate constant for
some of the experimentally studied high-mobility systems.
The breakdown of the diusion theory of bulk ion recombination in high-mobility
systems has been clearly demonstrated by the results of the computer simulations by
Tachiya [39]. In his method, it was assumed that the electron motion may be described by
the Smoluchowski equation only at distances from the cation, which are much larger than
the electron mean free path. At shorter distances, individual trajectories of electrons were
simulated, and the probability that an electron recombines with the positive ion before
separating again to a large distance from the cation was determined. The value of the re-
combination rate constant was calculated by matching the net inward current of electrons
Copyright 2004 by Taylor & Francis Group, LLC
obtained from the simulation results with that derived from the Smoluchowski equation at
large distances. To simulate electron trajectories, he used a simplied model of electron
scattering, in which the electron velocity was randomized at each collision to restore the
MaxwellBoltzmann distribution. An energy criterion of recombination was used, namely,
an electron was assumed to have reacted once its total energy in the Coulomb eld of the
cation was reduced below a certain critical energy.
The simulation results of the electronion recombination rate constant obtained in
Ref. 39 are plotted in Fig. 5. The gure shows that the rate constant becomes lower than the
DebyeSmoluchowski value k
D
when the electron mean free path exceeds
f
0.01r
c
. At
higher values of k, the ratio k/k
D
further decreases with increasing mean free path. The
simulation results are found to be in good agreement with the experimental data on the
electronion recombination rate constant in liquid methane, which are also plotted in Fig. 5.
The dependence of the electronion recombination rate constant on the mean free
path for electron scattering has also been analyzed on the basis of the FokkerPlanck
equation [40] and in terms of the fractal theory [24,25,41]. In the fractal approach, it was
postulated that even when the fractal dimension of particle trajectories is not equal to 2,
the motion of particles is still described by diusion but with a distance-dependent
eective diusion coecient. However, when the fractal dimension of trajectories is
not equal to 2, the motion of particles is not described by orthodox diusion. For the
Figure 5 The rate constant of bulk electronion recombination, relative to the Debye
Smoluchowski value [Eq. (36)], as a function of the electron mean free path k. The solid line
represents the simulation results, and the circles show the experimental data for liquid methane [49].
(From Ref. 39.)
Copyright 2004 by Taylor & Francis Group, LLC
relation between the fractal dimensions of reactant trajectories and the transport-limited
reaction rate, see Ref. 42.
In the limit of an innitely long mean free path, the bulk electronion recombination
may be described using the energy diusion model [43,44]. This model is especially
relevant to the electronion recombination processes in the gas phase.
The theoretical models of the bulk electronion recombination presented so far in
this section use only one parameter, namely the electron mean free path, to characterize
the electron transport processes in a wide class of high-mobility systems. This is
undoubtedly an oversimplication, and more realistic models of electron scattering have
to be used, if one wants to achieve a better understanding of the bulk recombination
processes in a given system. Such models, which use electron scattering cross sections, were
applied in the simulations of the electronion recombination in liquid methane [45] and
liquid argon [46]. In the case of methane, the mean free path for elastic electron scattering
was determined from the experimental electron mobility, and the inelastic scattering was
simulated using model cross sections for vibrational and electronic excitations. In the
simulation for argon [46], the electron motion was described using a model, which takes
account of dierent rates of energy and momentum transfers in electron collisions, in
accordance with the CohenLekner theory [47] of hot electrons in liquids. In both
Figure 6 Dependence of the electronion recombination rate constant on the external electric eld
F, calculated for dierent values of the electron mean free time s. The unit of F is e=er
2
c
, and that of s
is r
c
/(k
B
T/m)
0.5
. The simulation results of Morgan [45] for liquid methane at 120 K are shown by
crosses. (From Ref. 48.)
Copyright 2004 by Taylor & Francis Group, LLC
simulations, a reasonable agreement of the calculated rate constants with the experimental
results has been obtained.
Experimental data on the electronion recombination in high-mobility systems are
usually obtained by measuring the transient conductivity of an ionized system in the
presence of an external electric eld. The applied eld increases the average kinetic energy
of charge carriers. As the dissipation of energy is believed to be the rate-determining
process for the electronion recombination in high-mobility systems, the recombination
rate constant in these systems should decrease with increasing eld strength. Such an
expectation has been conrmed by the simulations [48] in which the eld dependence of the
rate constant was calculated for systems characterized by dierent values of the electron
mean free time H (the mean free time may be related to the mean free path by k/r
cc
1.6s).
The results are presented in Fig. 6. We see that as the mean free time increases, the sup-
pression of the electronion recombination by the electric eld becomes stronger. In the
model of the electron scattering assumed in Ref. 48, the electron velocity was randomized
at each collision to restore the MaxwellBoltzmann distribution. However, in real systems,
the electron energy dissipation by collisions is usually not so ecient, and the decrease of
the recombination rate constant with increasing electric eld should be even stronger than
calculated in Ref. 48. This expectation is supported by the simulation results for liquid
methane [45], also included in Fig. 6, which were obtained by using a realistic model of
electron scattering in that system.
2. PICOSECOND AND SUBPICOSECOND PULSE-RADIOLYSIS STUDIES
OF GEMINATE ION RECOMBINATION IN LIQUID HYDROCARBONS
2.1. Introduction
The geminate ion recombination in liquid hydrocarbons has been studied both by station-
ary experiments and by kinetics data. Although the kinetics data of pulse radiolysis exper-
iments give more detailed information about the geminate recombination than stationary
data such as scavenger experiments, the very high time-resolved spectroscopy is indis-
pensable for studies on the kinetics of the geminate ion recombination. The experimental
studies on the kinetics of the geminate ion recombination have advanced with the devel-
opment of picosecond and subpicosecond pulse radiolysis development.
Here the progress in the picosecond and subpicosecond pulse radiolysis is described
rst and then the experimental studies on the kinetics of the geminate ion recombination is
explained in connection with their application to advanced technology such as the next
generation nanolithography and nanotechnology.
2.2. Progress in Picosecond and Subpicosecosecond Pulse Radiolysis
History of Picosecond and Subpicosecosecond Pulse Radiolysis
With the development of the picosecond pulse radiolysis, the kinetics data of the geminate
ion recombination have been directly obtained. The history of picosecond and subpico-
second pulse radiolysis is shown in Fig. 7. Very recently, the rst construction of the
femtosecond pulse radiolysis and the improvement of the subpicosecond pulse radiolysis
started in Osaka University.
The rst picosecond pulse radiolysis experiment was carried out in the late 1960s by
the so-called stroboscopic method (generally pomp and probe method) at University of
Copyright 2004 by Taylor & Francis Group, LLC
Toronto [50]. Picosecond electron pulse trains generated with a S-band linear accelerator
(linac) as an irradiation source and the Cerenkov light induced from the electron pulse
trains as an analyzing light were used. The system was very sophisticated, but S/N ratio
is not high. Therefore the operation of the system for obtaining small absorption and
the detailed kinetics data is very dicult. Similar system was developed in Hokkaido
University more than 10 years later [51]. The combination of picosecond electron pulse
trains with high time-resolved detection systems was applied to the picosecond pulse
radiolysis with optical [52] and conductivity [53] spectroscopy. Measurable time domain of
these picosecond pulse radiolysis systems using picosecond electron pulse trains were
limited from the time resolution of the system to 350 and 770 psec determined by the pulse
interval in the pulse trains of S-band and L-band linacs, respectively. The overlapping of
the correct signal with the signal of slow decaying intermediates species produced by the
preceding picosecond electron pulses made dicult the detailed analysis of the kinetics of
intermediates. These systems were unable to study the geminate ion recombination in
irradiated liquid hydrocarbons.
The stroboscopic pulse radiolysis with the single bunch electron pulse instead of
pulse trains started in Argonne National Laboratory in 1975 [54]. The research elds have
been extended by the stroboscopic pulse radiolysis with the picosecond single electron
bunch, although most of researches had been limited to hydrated and solvated electrons in
the aqueous and alcoholic solutions. This system was unable to study the kinetics of the
geminate ion recombination in liquid hydrocarbons until the modication of the Argonne
linac in 1983, which made possible the quality measurements of the weak absorption.
Figure 7 History of picosecond and subpicosecond pulse radiolysis and the start of construction of
the rst femtosecond pulse radiolysis.
Copyright 2004 by Taylor & Francis Group, LLC
The combination of the picosecond single electron bunch with streak cameras, in-
dependently developed in 1979 at Argonne National Laboratory [55] and at University of
Tokyo by us [56], enabled the very high time resolution for emission spectroscopy. The
research elds have been extended to organic materials such as liquid scintillators [5557],
polymer systems [58], and pure organic solvents [59]. The kinetics of the geminate ion
recombination were studied [55,57,59].
The picosecond pulse radiolysis system in the combination of very fast response pho-
todiode with the transient digitizer (the time resolution of about 800 psec) was developed
by us in 1982 [60]. This system was the rst picosecond pulse radiolysis system that could
measure the kinetics of ions and triplet states on subnanosecond timescale by using only
one picosecond single bunch electron pulse. This system has no limitation of nano- and
microsecond timescale measurement and the radiation damage of the samples of the ex-
periments performed by this system was so much less than preceding picosecond pulse
radiolysis. Most experiments of geminate recombination of solute ions were performed by
this system [6066]. The response time of this system using the sampling oscilloscope
instead of the transient digitizer was 200 psec [61], but the radiation damage of the samples
was very large.
The important picosecond pulse radiolysis study on geminate recombination of solute
ions in liquid alkanes [65] was carried out in the combination of the picosecond pulse
radiolysis system composed of very fast response photodiode and the transient digitizer [60]
with the specially designed picosecond pulse radiolysis system [67] composed of a streak
camera (the time resolution of 50 psec) developed in 1984. Our experiment [65] showed for
the rst time that the geminate recombination of solute ions could be analytically explained
by the extended Smoluchowski equation [13,68]. The S/N ratio was not high because of a
limited dynamic range of the streak camera at that time, but the radiation damage of the
samples was very low. The proof of the principle of the twin linac system, where one linac
produced picosecond electron pulses as radiation source and the other linac produces the
Cherenkov light as monitor light, was reported in 1985 [69] and the construction of the
system was carried out in 1987 [70]. Although the twin linac system was very interesting
system, the operation of the system was very dicult and it takes too much eort to obtain
the data of radiation chemistry. The radiation damage of the samples was also very large.
After the LL (LaserLinac) twin picosecond pulse radiolysis was developed in 1991 [71], the
twin linac system was not used any more. Very important results for radiation chemistry
used by the twin linac system were the measurement of very rapid formation of alkyl
radicals in irradiated liquid alkanes within the time resolution of the system and it indicates
the importance of the excited radical cations in radiolysis of cyclohexane [66].
The LL twin picosecond pulse radiolysis system, which used picosecond semiconduc-
tor laser as monitor light instead of Cherenkov light of the twin linac pulse radiolysis, was
successfully developed by us for the rst time in 1991 [71]. It is the rst picosecond pulse
radiolysis in the combination of the single electron bunch with laser monitor light, although
several trials of the picosecond pulse radiolysis with the single electron bunch and laser
monitor light had attempted, but the system was not correctly operated [72]. Because the
operation of the LL twin picosecond pulse system was much easier than that of the twin
linac, pulse radiolysis and the laser diode as monitor light enabled the absorption spectro-
scopy from visible to infrared region in picosecond time range for the rst time [71]. The
kinetics of the geminate recombination of both electrons and cation radicals were rst
measured on a picosecond timescale in the same sample and agree with each other on
picosecond timescale from 50 psec after a pulse [71]. However, it was very dicult to obtain
Copyright 2004 by Taylor & Francis Group, LLC
the complete transient absorption spectra by this system because the laser diodes are not
tunable light source.
We solved the lack of the tunability of the LLtwin pulse radiolysis by the development
of the laser synchronized picosecond pulse radiolysis in 1995 at Osaka University [73]. The
time resolution of pulse radiolysis had remained about 30 psec for 30 years since the late
1960s (Fig. 7), although many modications had been performed for picosecond pulse
radiolysis techniques. The femtosecond monitoring laser pulse is synchronized with both the
electron pulses from the electron gun and accelerating electron pulses through the accel-
erating tubes in the laser synchronized picosecond pulse radiolysis system, although the
picosecond laser pulse is synchronized with only the electron pulse from the electron gun in
the LL twin linac system. Although the time resolution of the laser synchronized pulse
radiolysis at that time was 30 psec, this system had the possibility of the development of the
femtosecond pulse radiolysis.
The time resolution of 2 psec [74] for 2-mm optical path sample cell was obtained as
shown in Fig. 8 by using laser synchronized subpicosecond pulse radiolysis with the
femtosecond single electron bunch [76] and a jitter compensation system with 200-fsec
streak camera [74]. In this case, the time dierence between electron beam and light passing
through a sample is 1.8 psec. Therefore the time resolution is mainly determined by the
optical path length. The time resolution of 800 fsec [75] for 0.5-mm optical path sample cell
was obtained as shown in Fig. 9, but the S/N ratio is very low. The S/N ratio of the laser
synchronized subpicosecond pulse radiolysis was greatly improved by the double-pulse
method as shown in Fig. 10 [77]. Now the kinetics of the geminate ion recombination of
Figure 8 The time-dependent behavior of the hydrated electron obtained in the subpicosecond
pulse radiolysis of neat water using 2-mm optical path sample cell, monitored at the wavelength of
780 nm.
Copyright 2004 by Taylor & Francis Group, LLC
electron and radical cations has been directly studied by the laser synchronized subpico-
second pulse radiolysis with the double-pulse method. The details of the double-pulse
method are described in the next section.
Now several picosecond pulse radiolysis systems are operating in Argonne National
Laboratory, Brookhaven National Laboratory and University of Tokyo and are under
construction in University of Paris South and Waseda University, although only one sub-
picopulse radiolysis is operating in Osaka University.
Figure 9 The time-dependent behavior of the hydrated electron obtained in the subpicosecond pulse
radiolysis of neat water using 0.5-mmoptical path sample cell, monitored at the wavelength of 780 nm.
Figure 10 The geminate decay of the radical cation in irradiated neat n-dodecane observed by the
subpicosecond pulse radiolysis before (a) and after (b) the improvement of the system by mainly
using the double-pulse method.
Copyright 2004 by Taylor & Francis Group, LLC
Subpicosecond Pulse Radiolysis [74,77]
The subpicosecond pulse radiolysis [74,77] detects the optical absorption of short-lived
intermediates in the time region of subpicoseconds by using a so-called stroboscopic tech-
nique as described in Sec. 10.2.2 (History of Picosecond and Subpicosecosecond Pulse
Radiolysis). The short-lived intermediates produced in a sample by an electron pulse are
detected by measuring the optical absorption using a very short probe light (a femtosecond
laser in our system). The time prole of the optical absorption can be obtained by chang-
ing the delay between the electron pulse and the probe light.
Stroboscopic pulse radiolysis can achieve high time resolution because the time res-
olution of the stroboscopic pulse radiolysis does not depend on the time resolution of the
detection systems such as photodiodes and oscilloscopes. The time resolution is principally
determined by the lengths of the electron pulse and probe light and the timing accuracy (or
jitter) between them as shown in Fig. 11. Assuming that the envelopes of both the electron
pulse and probe light and the timing uctuation have a Gaussian distribution, the response
function, g
1
(t), can be expressed as follows,
g
1
t
1
2pr
p e
t
2
2r
2
41
r
2
r
2
i
r
2
b
r
2
j
42
where r
b
is the length of the electron pulse (rms), r
i
is the length of the probe light (rms)
and r
j
is the timing uctuation (rms).
However, the time resolution is also degraded by the velocity dierence between the
light and the electron pulse in a sample. The time in which the electron pulse passes
through a sample is given by L/(bc), where L is the optical length of the sample, c is
the velocity of light in vacuum, and b is the ratio of the velocity of the electron to c. On the
other hand, in the case of light, the time in which it passes is given by Ln/c, where n is the
refractive index of the sample. Therefore the time resolution is degraded with increasing
Figure 11 The components of the timing jitter of the laser synchronized subpicosecond pulse
radiolysis. r
b
is the length of the electron pulse (rms), r
i
is the length of the probe light (rms), and r
j
is the timing uctuation (rms).
Copyright 2004 by Taylor & Francis Group, LLC
thickness of the sample. The response function with the eect of velocity dierence, g
2
(t),
can be deduced from the denition of the response function as follows,
g
2
t
c
p
p
L n
1
b
_ _
_
2ct L n
1
b
_ _
2
2rc
p
2ct L n
1
b
_ _
2
2rc
p
e
x
2
dx 43
When the optical length of the sample is much shorter than the other factors such as the
length of the electron pulse, the response function takes on a Gaussian shape. As the
optical length increases, the shape of the response function becomes trapezoidal. Fur-
thermore, a thick sample causes the prolongation of the electron pulse by electron
scattering, which leads to the degradation of time resolution. Therefore the experiment
to observe ultrafast phenomena requires the use of a thin sample.
Fig. 12 shows the subpicosecond pulse radiolysis system. The system consists of a
subpicosecond electron linac as an irradiation source, a femtosecond laser as an analyzing
light, and a jitter compensation system. A sample was irradiated by a subpicosecond
electron single pulse. The subpicosecond electron single pulse was obtained by compressing
a 30-psec electron single pulse from the ISIR L-band linac with a magnetic pulse
compressor. This system can compress the width of electron pulse to approximately 125
fsec (FWHM) [76]. The time-resolved optical absorption was detected with a femtosecond
laser. A mode locked ultrafast Ti:Sapphire laser (Tsunami, Spectra-Physics Lasers, Inc.)
was synchronized to the ISIR L-band Linac using a commercially available phase lock
loop. The width of the laser pulse was 60 fsec (FWHM). The intensity of the laser pulse was
measured by a Si photodiode. The timing between the electron pulse and the laser pulse was
controlled by radio frequency (RF) system. The time prole of the optical absorption could
be obtained by changing the phase of the RF with an electrical phase shifter.
The jitter between the laser pulse and the electron pulse was estimated from the mea-
surement using a streak camera (C1370, Hamamatsu Photonics Co. Ltd.), because the jitter
is one of important factors that decide the time resolution of the pulse radiolysis. The jitter
was several picoseconds. To avoid eects of the jitter on the time resolution, a jitter com-
pensation system was designed [74]. The time interval between the electron pulse (Cerenkov
light) and the laser pulse was measured by the streak camera at every shot. The Cerenkov
radiation was induced by the electron pulse in air at the end of the beamline. The laser pulse
was separated from the analyzing light by a half mirror. The precious time interval could be
Figure 12 The subpicosecond pulse radiolysis system.
Copyright 2004 by Taylor & Francis Group, LLC
obtained by the analysis of the streak image. All equipment described above was controlled
by a personal computer. The acquisition time was 1 sec per one shot. Fig. 8 shows the time-
dependent behavior of the absorption due to hydrated electrons at the wavelength of 780
nm was observed in distilled and deaerated water in a quartz cell with the optical length of
2.0 mm [76]. Fig. 8 shows the time-dependent behavior of the hydrated electron obtained in
the system. The solid line represents the tting curve for the experimental data. The 10
90% rise time is 2.0 psec.
The time resolution of the stroboscopic pulse radiolysis depends on the width of
electron pulse and that of laser. In this experiment, both widths are much shorter than 2.0
psec. Furthermore, the time resolution is limited by the dierence between the velocity of
the light and that of the electron pulse in the sample. The passing time of the electron pulse
through a sample is given by L/(bc), where L is an optical length of a sample, b is the ratio
of the velocity of electron to that of light in the vacuum, and c is the velocity of light in the
vacuum. On the other hand, in the case of light, the passing time is given by Ln/c, where n
is a refractive index of a sample. Therefore the time resolution limited by the thickness of
the sample is 1.8 psec (1090% rise time) at n=1.33 and 1=2 mm. Therefore the 1090%
rise time of 2.0 psec is the time resolution of the system. In the present system, the time
resolution is mainly limited by the thickness of the sample. The time resolution of 800 fsec
can be obtained by a quartz cell with the optical length of 0.5 mm as shown in Fig. 9. The
S/N ratio decreases with the optical length decreases. So the S/N ratio is not good for a
0.5-mm cell.
As discussed above, it is necessary to use a thin sample in high time resolution pulse
radiolysis. The use of a thin sample leads to the degradation of the S/N ratio because the
signal intensity is proportional to the optical length of the sample. In pulse radiolysis
experiments, the optical density is generally calculated based on the intensity of the probe
light passing through a sample with an electron beam present (I ) and that of the probe
light passing through a sample with the electron beam absent (I
0
).
O:D: log
I
0
I
44
If there is no uctuation of laser intensity, we have to measure I
0
only once. Actually,
the envelope of laser pulses changes in a relatively long time range (typically from several
minutes to a few tens of minutes) because of the change of environmental factors such as
room temperature and coolant temperature. There is also an intensity jitter caused by
factors such as the mechanical vibration of mirrors and the timing jitter of electronics.
Furthermore, in our system, the laser system is located about 15 m from the beam port to
prevent radiation damage to the laser system. (Later, it was moved into a clean room,
which was installed in the control room to keep the room temperature constant and to
keep the laser system clean. The distance is about 10 m.) Therefore it is predicted that a
slight tilt of a mirror placed upstream will cause a displacement of the laser pulse at the
downstream position where the photodetector is placed.
From Eq. (44), to achieve a good S/N ratio measurement, it is necessary and
sucient to lower the uctuation of I
0
/I. Based on this information, we devised a new
method to measure I and I
0
with a good S/N ratio. We named it the double-pulse
method. In the new method, I
0
is measured by the laser pulse passing just before the laser
pulse used to measure I. Because the frequency of the Ti:Sapphire laser is 81 MHz, the
time interval of two laser pulses is 12.3 nsec. The laser pulse for the measurement of I
passes through the sample 12.3 nsec after the laser pulse for the measurement of I
0
. An
Copyright 2004 by Taylor & Francis Group, LLC
electron pulse enters the sample between two laser pulses. The two laser pulses are
expected not to suer from the vibration of optical elements because it is thought that the
frequency of mechanical vibration mainly caused by the vibration of coolant and
ventilator systems is lower than the MHz order. The uctuation of I
0
/I is measured at
the position where the photodetector is located. The uctuation of I
0
/I is reduced from
5.34% to 0.80% using the newly developed method. The time-dependent behavior of
geminate ion recombination in neat dodecane is shown in Fig. 10 as a typical example that
demonstrates the eect of the newly developed system. The sample was deaerated by Ar
bubbling in a quartz cell having an optical length of 2.0 mm and was irradiated at room
temperature. The high S/N data obtained by the improved subpicosecond pulse radiolysis
using the double-pulse method makes possible the detailed studies on the geminate
recombination of electrons and radical cations. The details of the time-dependent behavior
of the geminate recombination shown in Fig. 10 are discussed in the next section.
2.3. Experimental Studies of the Kinetics of Geminate Ion Recombination
Kinetics of Geminate Ion Recombination
The early work of the kinetics of geminate ion recombination studied by picosecond pulse
radiolysis has been briey reviewed [78]. The studies on the kinetics of geminate ion
recombination in liquid hydrocarbons were dicult because the very high time-resolved
spectroscopy is indispensable for monitoring the time behavior of the geminate ion re-
combination. The combination of the picosecond single electron bunch with streak cameras,
independently developed in 1979 at Argonne National Laboratory [55] and at the University
of Tokyo by us [56], enabled the time behavior of recombination uorescence of the
geminate ions. The direct measurement of the recombination uorescence produced by
geminate ion pairs [55,57] stimulated studies on the kinetics of the geminate ion recombi-
nation and gave us many kinds of the problems for understanding the geminate kinetics and
the positive ions in irradiated hydrocarbons. But the recombination uorescence is the
indirect measurement of the geminate positive ions and is unable to make clear the details of
both the identication and characteristics of positive ions in irradiated hydrocarbons,
especially cyclohexane [79].
The picosecond pulse radiolysis system in the combination of very fast response
photodiode with the transient digitizer (the time resolution of about 800 psec) was devel-
oped by us in 1982 [60]. This system was the rst picosecond pulse radiolysis system that
could measure the kinetics of ions and triplet states on subnanosecond timescale by using
only one picosecond single bunch electron pulse. This system has no limitation of nano-
and microsecond timescale measurement and the radiation damage of the samples by this
system was much less than that by other preceding picosecond pulse radiolysis. The change
of the ratio of the formation rate of the solute excited triplet state to the decay rate of the
solute anion observed directly by this system showed the loss process of the spin cor-
relation of geminate ion pairs very clearly in the time range between 5 and 20 nsec [60].
The decay of the geminate ion pairs in cyclohexane [62] was analytically analyzed by the
Smoluchowski equation [13] and in various nonpolar solutions [63] on the basis of the
absorption of solute ions in time range from 800 psec to several hundred nanoseconds by
using the same system. In 1987 [64], the kinetics of the geminate pairs of solute ions was
analytically analyzed in the time range from 50 psec to several hundred nsec by the extend
Smoluchowski equation [68] on the basis of the combination of the picosecond pulse
Copyright 2004 by Taylor & Francis Group, LLC
radiolysis system composed of very fast response photodiode and the transient digitizer
[60] with the specially designed picosecond pulse radiolysis system composed of a streak
camera [67]. Then, geminate ion recombination and the formation process of solute ex-
cited states in cyclohexane solutions of biphenyl observed by the combination of two types
of picosecond pulse radiolysis with emission [56] and absorption [60] spectroscopy were
analyzed by the expansion of the extended Smoluchowski equation to the formation of
solute excited states [65]. The comparison of the theoretical treatment based on Smolu-
chowski equation with the absorption and emission data on the geminate recombination
of solute ions and the formation of solute excited states on timescale from 50 psec to
several hundred was successfully achieved [64,65].
Now we shall discuss about the geminate recombination of electrons and solvent
cation radical in liquid hydrocarbons. The original stroboscopic pulse radiolysis in
Argonne National Laboratory [54] was unable to make good quality measurements on
geminate recombination of the electron and the positive ion in liquid hydrocarbons. In
1983, the modication of the Argonne linac made possible the rst good quality measure-
ments of the weak absorption at 600 nm in neat liquid n-hexane and cyclohexane. The
transient absorptions at 600 nm are assigned to the positive ion and the excited state for
neat n-hexane and cyclohexane, respectively [79]. The dierence between broad absorption
bands of the excited states and the radical cations become larger with increasing the
number of carbon atoms of n-alkanes, although the peak wavelengths of both absorption
bands become longer with increasing the number of carbon atoms of n-alkanes [66]. The
lifetimes of alkane radical cations become shorter with decreasing the number of carbon
atoms of n-alkanes. In irradiated neat cyclohexane, very broad visible absorption bands,
mainly because of the excited state, were observed [66]. Although radiolysis of cyclohexane
and n-hexane was heavily studied, both experimentally and theoretically, the kinetics of
radical cations of both cyclohexane and n-hexane are dicult to be studied because of
their very short lifetimes [66], the absorption overlapping of radical cations and excited
states [66], and the complexity of their reactions and characters [80]. Because the
complication of the absorption spectroscopy on radical cations of smaller hydrocarbons
was clearly shown experimentally [66], the absorption spectroscopy of the geminate
recombination of electrons and cation radicals in liquid hydrocarbons has mainly been
carried out for higher n-alkanes such as n-dodecane instead of both cyclohexane and n-
hexane.
Because the LL twin picosecond pulse system with the laser diode as monitor light
enabled the absorption spectroscopy from visible to infrared region for the rst time in
picosecond pulse radiolysis, the kinetics of the geminate recombination of both electrons
and cation radicals were rst measured on picosecond timescale in the same sample and
agree with each other on picosecond timescale from 50 psec after a pulse [71]. By using the
laser synchronized picosecond pulse radiolysis [73] with the higher S/N ratio than that of
the LL twin system, the good agreement of the diusion theory with experimental data
were conrmed on a picosecond timescale from 50 psec after a pulse, but a slight dierence
between the experimental data and the simulation based on Smoluchowski equation using
the exponential function of the initial separation between electrons and radical cations
[81]. Many factors from the time resolution of the system and simulation models are
considered as the reason for the dierence. Because the time resolution and S/N ratio of
the laser synchronized subpicosecond pulse radiolysis was drastically improved by both
the jitter compensation system [74] and the double-pulse method [77] as described above,
the dierence between the experimental data and the simulation based on Smoluchowski
Copyright 2004 by Taylor & Francis Group, LLC
equation using the simple model was clearly observed on timescale within 50 psec. The
details of the kinetics of the geminate ion recombination studied by the laser synchronized
subpicosecond pulse radiolysis are described in the next section.
Subpicosecond Pulse-Radiolysis Studies of Geminate Ion Recombination
The kinetics data of the geminate ion recombination in irradiated liquid hydrocarbons
obtained by the subpicosecond pulse radiolysis was analyzed by Monte Carlo simulation
based on the diusion in an electric eld [77,81,82]. The simulation data were convoluted
by the response function and tted to the experimental data. By transforming the time-
dependent behavior of cation radicals to the distribution function of cation radicalelectron
distance, the time-dependent distribution was obtained. Subsequently, the relationship be-
tween the space resolution and the space distribution of ionic species was discussed. The
space distribution of reactive intermediates produced by radiation is very important for
advanced science and technology using ionizing radiation such as nanolithography and
nanotechnology [77,82].
The deposited energy of radiation is spent mostly for ionization of liquid hydro-
carbon molecules. During ionization, a pair of a cation radical (RH
+
) and an electron
(e
) is produced as follows,
RH
a
RH
45
After the ionization, the electrons with excess energy interact with surrounding molecules
and become thermalized. The reaction from the ionization to thermalization is estimated
to occur within about 1 psec. The initial separation length between the cation radical and
the thermalized electron is several nanometers on average.
In nonpolar materials, most of the electrons recombine with the counter cation
radicals as shown by Eq. (46). RH* is one of the main products produced through the
recombination.
RH
a
RH* 46
This occurs because the initial separation length between the cation radical and the
electron is smaller than the Onsager length (r
c
) at which the thermal energy of an electron
corresponds to the Coulomb eld. This reaction is called the geminate ion recombination
and has been investigated by many researchers globally.
The experimental method and apparatus, and a procedure of the Monte Carlo method
that simulates the geminate ion recombination are described, and the time-dependent
distribution is elucidated.
The experiments were carried out using the subpicosecond pulse radiolysis system
[77] described in Sec. 10.2.2 (Subpicosecond Pulse Radiolysis). Considering the signal
intensity and the degradation of the time resolution, a sample cell with the optical length
of 2 mm was mostly used. The sample was saturated by Ar gas to eliminate the scavenging
eect by the remaining O
2
gas.
At the initial stage of reactions, the produced intermediate species such as the cation
radical and the electron exist in a narrow space, the so-called spur. After the electron
thermalization process, a pair of a cation radical and a thermalized electron remain in a
spur. The geminate ion recombination of the cation radical and the electron occurs before
these ionic species diuse and spread uniformly in the media. Therefore the geminate ion
recombination takes place in the spur. On the condition of a so-called single pair model,
Copyright 2004 by Taylor & Francis Group, LLC
the reaction, which is described by the following equation, can be solved in spherical co-
ordinates with a radial component.
Bwr; t
Bt
D j
2
wr; t
1
k
B
T
jwr; tjVr
_ _
2
Hong and Noolandi [11,13,83] analytically solved Eq. (2) under the following initial and
boundary conditions:
wr; 0 fr initial condition 47
wR; t 0 boundary condition 11a
Equation 2 indicates that if the distance r between the cation radical and the electron
becomes less than the reaction radius R, they recombine. The typical shape of f(r,r
0
), the
initial distribution function, is expressed by the following exponential or Gaussian.
fr; r
0
1
r
0
exp
r
r
0
_ _
exponential 19
fr; r
0
4r
2
p
p
r
3
0
exp
r
2
r
2
0
_ _
gaussian 20
The solution of Eq. (2) can also be obtained by a numerical analysis similar to the calculus
of nite dierences. However, an analytical or semianalytical method based on Eq. (2) is
not suitable for discussing the time-dependent distribution function because the calcu-
lation is lengthy.
On the other hand, the Monte Carlo method enables us to simultaneously obtain the
time-dependent decay curve and the time-dependent distribution function. Therefore we
adopted Monte Carlo simulation [18,21,84,85] for the analysis. The geminate ion
recombination is also described as follows,
Dr
i
6D
i
Dtn
i
_
l
i
DtE
i
48
In Eq. (48), i represents the cation radical and electron. The D
i
and l
i
are the diusion
constant and the mobility of the ith species, respectively. The relationship between them is
described by the Einsteins relation. The calculation was carried out in the rectangular
coordinates (x, y, z). For each time step Dt, the vector of movement Dr
i
is calculated
according to Eq. (48). The rst term in Eq. (48) represents the diusion component. It was
calculated by uniform random variables n
i
ranging from 1 to 1. The second term in Eq.
(48) represents the drift in the electric eld E
i
, which is due to counter ionic species. At rst,
the cation radical and the electron were set at (0, 0, 0) and (x
0
, 0, 0) in the coordinates,
respectively. Therefore x
0
corresponds to r
0
, which is the initial separation length between
the cation radical and the electron. After that, the positions of the ith species were
calculated by Eq. (48). If the distance between the cation radical and electron became less
than the reaction radius (R), they recombine. The time taken for the recombination was
recorded. The Dt was taken to be 0.1 psec. By carrying out calculations for 4000 pairs of the
cation radical and the electron, the survival probability of the cation radicals was obtained.
The time-dependent distribution function was also obtained by recording the distance
between the cation radical and the electron. Before analyzing the experimental data, to
check the validity of the simulation code, we compared the simulation data with the
analytical solution given by Hong and Noolandi [11,13,83]. The calculation was completed
in about 15 min by a personal computer. The result is shown in Fig. 13(a). The initial
distribution function and the Onsager length (r
c
) were taken as a delta function d(rr
0
) of
Copyright 2004 by Taylor & Francis Group, LLC
r
0
=7.5 nm and r
c
=30 nm, respectively. The dots and the solid line represent the analytical
solution and the simulation curve, respectively. They perfectly agree with each other.
Therefore the validity of the code was conrmed. At the same time, the time-dependent
distribution function was obtained. It is shown in Fig.13(b). The vertical axis and the
horizontal axis represent the probability density and the distance between the cation radical
and the electron, respectively. The probability density is dened as the existing ratio of
species per unit length. We can see that the electron, which existed at r
0
=7.5 nm, diuses
and recombines with the cation radicals.
In the experiment, liquid n-dodecane was used due to the small overlap among
intermediate species. At the wavelength of the Ti:Sapphire fundamental light, a large part
of the absorption is assigned to the cation radicals in n-dodecane [66]. Thus the decay
curve of the experimental result shows a decay of the cation radicals indicated in Eq. (46).
Fig. 14(a) shows the time-dependent behavior of the cation radicals in liquid n-dodecane,
which was obtained by the subpicosecond pulse radiolysis and monitored at 790 nm [82].
The vertical axis represents optical density, which is proportional to the density of the
intermediate species. The dotted and the solid lines represent the experimental curve and
the simulation curve, respectively. It was observed that the cation radicals began to be
produced at time zero and decayed by the recombination. The rise time of the exper-
imental data corresponded to the time resolution of the measurement system. In the
simulation, the parameters of the electron diusion coecient (D
e
) =6.4 10
4
cm
2
/sec,
the cation radical diusion coecient (D
+
) =6.0 10
6
cm
2
/sec, the relative dielectric
constant e =2.01, the reaction radius R=0.5 nm and the exponential function as shown
in Eq. (19) with r
0
=6.6 nm were used. These values are the same as those used in previous
papers [81,86]. The exponential distribution function was realized by using the rejection
method with two sets of uniform random variables. The shape and mean value of the
initial distribution function are dependent on the materials used. The obtained simulation
curve was convoluted by the response function [77] and tted to the experimental curve so
that the residual between two curves was minimized in the long timescale. Although the
Figure 13 (a) Comparison of the simulation curve with the analytical solution given by Hong and
Noolandi [11,13,83]. The initial distribution function and the Onsager length (r
c
) were taken as a
delta function d(rr
0
) of r
0
=7.5 nm and r
c
=30 nm, respectively. The dots and the solid line
indicate the analytical solution and the simulation curve, respectively. (b) Time-dependent
distribution function that was obtained from the simulation of (a). r indicates the distance between
the cation radical and the electron.
Copyright 2004 by Taylor & Francis Group, LLC
time range was terminated at 350 psec in Fig. 14(b), it was conrmed that the experimental
curve is in accordance with the simulation curve in the nanosecond timescale [63,66,81,86].
As a result, the two curves corresponded with each other after a time interval of about 50
psec after irradiation. Some factors that may explain the disagreements within the time
interval of about 50 psec are discussed but was not determined [81]. Fig. 14(b) shows the
time-dependent distribution obtained by the simulation curve in Fig. 14(a). The solid line,
dashed line, and dots represent the distribution of cation radical-electrons distance at 0,
30, and 100 psec after irradiation, respectively. At 0 psec, the distribution was in the shape
of the exponential function. At 30 psec, a large part of the electrons, especially near the
cation radical, have already recombined. At 100 psec, the shape of the distribution curve is
highly similar to that of 30 psec.
The space distribution of ionic species, which move in several or a few nanometers
scale, should not be ignored in future nanolithography and nanotechnology based on beam
technology.
Further detailed kinetics of the geminate recombination of electrons and positive ions
and their application to the advanced technology will be studied by higher time resolution of
the femtosecond pulse radiolysis and both by the higher S/Nratio and the wider wavelength
monitoring light of the improved subpicosecond pulse radiolysis shown in Fig. 7.
3. EXPERIMENTAL STUDIES OF ELECTRONION BULK RECOMBINATION
IN CONDENSED MATTER
3.1. Introduction
The investigation of the dynamic behavior of low-energy electrons such as electron transport
and reactivities in nonpolar dense molecular media is of essential importance in both
Figure 14 (a) Time-dependent behavior of cation radicals in liquid n-dodecane monitored at 790
nm. The dotted and the solid lines represent the experimental curve and the simulation curve,
respectively. The parameters of the electron diusion coecient (D
e
) =6.4 10
4
cm
2
/sec, the cation
radical diusion coecient (D
+
) =6.0 10
6
cm
2
/sec, the relative dielectric constant e =2.01, the
reaction radius R=0.5 nm, and the exponential function as shown in Eq. (19) with r
0
=6.6 nm were
used. (b) Time-dependent distribution function obtained from tting curve of (a). r indicates the
distance between the cation radical and the electron. The solid line, dashed line, and dots represent the
distribution of cation radicalelectron distance at 0, 30, and 100 psec after irradiation, respectively.
Copyright 2004 by Taylor & Francis Group, LLC
fundamental and applied sciences [87100]. The values of an electron mobility l
e
have been
extensively measured in a variety of these media, and several theoretical models have been
proposed for the transport mechanism. Electron reactivities with solute molecules or
positive ions in these media have been also extensively investigated. The former reaction
is electron attachment and the latter electronion recombination. The investigation of these
electron reactivities has provided a new insight into that of the behavior of low-energy
electrons in matter.
In the following, a brief survey is given of the experimental investigation of electron
ion recombination with particular emphasis placed on the comparison between the
experimental and theoretical results as well as on that between the results in dense media
and in low-density gases or beams.
A survey of electron attachment with a similar emphasis has been given elsewhere
[92,100].
3.2. Comparative Aspects of ElectronIon Recombination Processes
in the Gas and Condensed Phases
It is of great interest to investigate the electronion recombination in dense molecular
media in the density range from dense gases to liquids and solids [96,97,99,100] because the
Coulombic interaction distance between an electron and an ion, as expressed by the
Onsager length r
c
, which is given by Eq. (1), is much bigger than the mean intermolecular
distance of the nearest neighbor molecules in the medium.
Electronion recombination processes in isolated two-body collisions in beams or in
low-pressure gases have been extensively studied mainly with a merged beam method
[101], a pulsed afterglow method [102], a owing afterglow method [103], and a recently
developed method by means of a cooler ring [104,105]. Cross sections or rate constants for
electronion recombination obtained by these experimental methods are theoretically
explained by the mechanism including molecular superexcited states as collision complex
or reaction intermediate species. Spectroscopy and dynamics studies of molecular super-
excited states have been recently in great progress [106,107], and these states are assigned
mainly to vibrationally, doubly, or inner-core excited molecular high Rydberg states
converging to each of ion states.
In the gas phase at one to several atmospheric pressures, electronion recombination
processes have been studied with a pulse-radiolysis method, and observed recombination
rate constants k
r
are expressed by,
k
r
k
2
k
3
n 49
where k
2
and k
3
are the two-body and three-body rate coecients, respectively, and n is the
density of a third-body molecule. The magnitude of observed rate constant values is ex-
plicable in terms of the collisional de-excitation of two-body formed complex or inter-
mediate species.
3.3. Experimental Methods to Measure Absolute Rate Constants
for Recombination
However, in dense media such as high-pressure gases near the critical points, liquids, and
solids, experimental studies were relatively few because of diculties in experimental
methods. But recently there has been great progress due to the development of a new
Copyright 2004 by Taylor & Francis Group, LLC
powerful method, called the decay-curve analysis method [96,97,99,100]. The following are
a brief comparative survey of the experimental methods [99].
1. In a pulse-radiolysis d.c. conductivity method, k
r
/l
e
is measured and then k
r
is
determined from the measured k
r
/l
e
and a separately known l
e
.
2. In a pulse-radiolysis charge-clearing method, a residual amount of charge that is
left unrecombined is collected and measured. Because the collected charge is
expressed as a function of k
r
, the value of k
r
/bl
e
, where b is a correction factor due
to a space charge is obtained. The kr value is thus determined by using the value of
l
e
known separately and the value of b which is approximately assumed.
3. In a pulse-radiolysis optical-absorption method, the value of k
r
/e, where e is a
molar absorption coecient, is measured by the time-resolved measurement of
the optical absorption of solvated electrons, and then the k
r
value is determined
by the observed value of k
r
/e and the value of e known separately.
4. In a pulse-radiolysis d.c. conductivity method with a long pulse width, the
resulting steady state current during the long pulse is considered by the
assumption that the rate of excess electron generation becomes equal to the rate
of electron loss because of the reaction with positive ions and impurities.
5. In a pulse-radiolysis microwave-conductivity method, the k
r
value is determined
by the time-resolved measurement of microwave conductivity.
6. In a method using a radioisotope, e.g.,
207
Bi, as an internal radiation source, the
dierence in a uorescence intensity with and without a d.c. electric eld is
assumed to be due to electronion recombination, and the k
r
value is determined
by the time-resolved measurement of the uorescence with the known number
density of ionization, which is separately determined.
7. In the decay-curve analysis method, which is one of the pulse radiolysis d.c.
conductivity methods, both k
r
and l
e
values are simultaneously determined by
analyzing observed transient current decay curves by tting them to a general
formula of conductivity both at a high pulse dose of x-rays where the
recombination is predominant and at a low pulse dose where the recombination
is negligible. This is a useful method also because of the k
r
determination without
knowing the pulse dose of x-rays or the free-ion yield. Recently, this method has
been improved by taking account of a space-charge eect to obtain more
accurate values of k
r
.
3.4. Experimental Results: The Breakdown of the Debye Theory
In the liquid phase, observed electronion recombination rate constants k
r
in a variety of
nonpolar media were, as shown in Fig. 15, in good agreement with the values of k
D
calculated from the reduced Debye equation,
k
D
4pe
e
l
e
50
viz., Eq. (36b), in the range of l
e
below about 150 cm
2
/V sec, and thus the recombination
in these media was concluded to be diusion controlled [99,100,112].
However, in liquid and high-pressure gaseous methane in which most l
e
values are
larger than 150 cm
2
/V sec, it has been found [49] that the observed k
r
values are much
lower than k
D
. This experiment has stimulated a great deal of theoretical investigations
[24,25,30,3842,48,113119] of the recombination process in dense media. Because the
Copyright 2004 by Taylor & Francis Group, LLC
magnitudes of the electron mobilities in dense Ar, Kr, and Xe media are as high as in dense
methane media, the values of k
r
and l
e
in these media have been also measured [120123]
and compared further with theoretical results. The observed k
r
values in these rare-gas
media in both liquid and dense gas phases are again found to be much smaller than those
calculated by the reduced Debye equation. However, in the solid phase, the observed
k
r
values are almost in agreement with those calculated by Eq. (50). These results are
summarized in Fig. 16. The eect of an external d.c. electric eld strength on both k
r
and l
e
values has also been measured. An example of the eld eect is shown in Fig. 17 for liquid
Ar. Recently, the eect of the addition of molecular impurities such as N
2
and CH
4
on k
r
and l
e
values has been further examined to dierentiate between the energy and spatial
criteria [39,42,114] in expectation that electrons heated up by applied external electric elds
may be cooled down by the added molecular impurities due to inelastic collisions between
electrons and molecules. Detailed comparisons have been made between these experimental
results and the theoretical ones, from which it has been concluded that the recombination
in liquid and dense gaseous Ar, Kr, and Xe as well as CH
4
is clearly dierent from a usual
diusion-controlled reaction corresponding to Eq. (50).
3.5. Comparison with Theories
Theoretical investigations trying to explain the k
r
deviation from k
D
, i.e., the breakdown
of the Debye theory, are summarized as follows:
1. A semiempirical treatment taking both the diusion controlled and electron-
energy exchange controlled recombination processes into account [38].
2. A Monte Carlo simulation with the parameter K/a, where K is the mean free path
of electrons and a is the reaction radius [39,42,114].
Figure 15 Variation of k
r
with l
e
in the l
e
region below about 150 cm
2
/V sec in condensed
nonpolar media. n-Pentane, n-hexane, TMS (5); n-hexane, cyclohexane (); neopentanen-hexane
mixtures (.); and liquid and solid neopentane (o). The straight line corresponds to k
D
(From Refs.
99,100,112.)
Copyright 2004 by Taylor & Francis Group, LLC
Figure 16 Variation of k
r
/k
D
with l
e
in condensed nonpolar media [99,100,112]. Neopentane, liquid
(D) and solid (E); neohexane (z); TMS (q); n-pentane, n-hexane (+); n-hexane, cyclohexane (5);
neopentanen-hexane mixtures (); methane, liquid (.) and solid (n); and argon, liquid (o) and solid
( w ).
Figure 17 Electric eld dependence of k
r
(.) and nl
e
(o) in liquid argon (T=87 K, n = 2.1 10
22
/
cm
3
) (From Refs. 99,100,120.)
Copyright 2004 by Taylor & Francis Group, LLC
3. A molecular dynamics simulation [115].
4. A fractal treatment [24,117,118].
5. A gas kinetic approach [116].
6. An approach based on the FokkerPlanck equation [40].
7. Theories to explain the eect of applied external electric elds on k
r
[30,48,119].
Warman [38] rst proposed a semiempirical formula of the recombination rate
constant to explain the new experimental result of k
r
<<k
D
[49]. In spite of many as-
sumptions to derive the formula, the calculated value for liquid argon is in rather good
agreement with the experimental one. However, for dense gaseous argon, the calculated
value is far from the experimental one. Tachiya [39,42,114] pointed out using a Monte
Carlo method that if K/a is not negligible the rate constant for bulk recombination
deviates from k
D
. Experimental k
r
values for liquid methane lie around the theoretical
curve of his simulation which uses K/a only as a parameter. He also explains the eect of
the eld strength on k
r
. The values of k
r
roughly increase in proportion to the eld strength
in the lower eld strength region and form a peak at around the critical eld strength.
These results are explained in terms of the eect of the eld strength on the trajectory of
electrons to be recombined. In the lower eld region, the disentangling of random electron
trajectories caused by the presence of the external electric eld should increase the k
r
value.
In the eld region higher than the critical eld strength, the internal energy of an electron
ion pair decreases the k
r
value.
Recent experiments [123] on the eects of electric eld strength and the addition of
molecular impurities on k
r
values in dense gaseous Ar and Kr at the density range near the
critical points, in which a new experimental method to measure k
r
values are developed by
including the eect of space charges on k
r
and l
e
values in the analysis of observed
conductivity signals, have clearly shown again a large deviation of k
r
values from the
Debye theory. The obtained experimental results are discussed in terms of the electron
mean free paths and electron energies based on recent theories.
It is concluded that electronion recombination processes in dense high-mobility
media such as liquid and dense gaseous Ar, Kr, and Xe as well as CH
4
are clearly dierent
from those expressed by the Debye theory, i.e., a well-established diusion-controlled
theory. Newly developed theories, instead of the Debye theory, as stimulated by the
experimental nding of the breakdown of the Debye theory, have been explicable, at least
qualitatively, in terms of the experimental results of the breakdown, i.e., k
r
<<k
D
, but still
quantitatively inexplicable not only in the magnitude of the breakdown itself but also in
the eects of external electric elds and the addition of molecular impurities on it. Further
new investigations are strongly needed both theoretically and experimentally. It is also
noted that these investigations may provide a new insight into the fundamental processes
of phenomena of ionization and excitation in condensed matter, e.g., those in liquid xenon
detectors [124].
REFERENCES
1. Smoluchowski, M. Z. Phys. Chem. 1917, 92, 129.
2. Onsager, L. Phys. Rev. 1938, 54, 554.
3. Sano, H.; Tachiya, M. J. Chem. Phys. 1979, 71, 1276.
4. Tachiya, M. J. Chem. Phys. 1978, 69, 2375.
5. Collins, F.C.; Kimball, G.E. J. Colloid Sci. 1949, 4, 425.
Copyright 2004 by Taylor & Francis Group, LLC
6. Wilemski, G.; Fixman, M. J. Chem. Phys. 1973, 58, 4009.
7. Murata, S.; Tachiya, M. J. Chim. Phys. 1996, 93, 1577.
8. Que, W.; Rowlands, J.A. Phys. Rev. B 1995, 51, 10500.
9. Abell, G.C.; Funabashi, K. J. Chem. Phys. 1973, 58, 1079.
10. Mozumder, A. J. Chem. Phys. 1974, 60, 4300.
11. Hong, K.M.; Noolandi, J. J. Chem. Phys. 1978, 69, 5026.
12. Wojcik, M.; Tachiya, M. to be published.
13. Hong, K.M.; Noolandi, J. J. Chem. Phys. 1978, 68, 5163.
14. Noolandi, J. Kinetics of Nonhomogeneous Processes; Freeman, G.R., Ed. Wiley: New York,
1987; 465 pp.
15. Hummel, A.; Infelta, P.P. Chem. Phys. Lett. 1974, 24, 559.
16. Hummel, A. J. Chem. Phys. 1968, 49, 4840; M. Tachiya, M.S. Thesis (Univ. of Tokyo, 1969).
17. Tachiya, M. Radiat. Phys. Chem. 1988, 32, 37; Green, N.J.B.; Pimblott, S.M.; Tachiya,
M. J. Phys. Chem. 1993, 97, 196.
18. Bartczak, W.M.; Hummel, A. J. Chem. Phys. 1987, 87, 5222.
19. Cliord, P.; Green, N.J.B.; Pilling, M.J.; Pimblott, S.M. J. Phys. Chem. 1987, 91, 4417; Green,
N.J.B.; Pilling, M.J.; Pimblott, S.M.; Cliord, P. J. Phys. Chem. 1989, 93, 8025.
20. Tachiya, M.; Hummel, A. Chem. Phys. Lett. 1989, 154, 497.
21. Wojcik, M.; Bartczak, W.M.; Hummel, A. J. Chem. Phys. 1992, 97, 3688.
22. Tachiya, M. J. Chem. Phys. 1988, 89, 6929.
23. Tachiya, M.; Schmidt, W.F. J. Chem. Phys. 1989, 90, 2471.
24. Lo pez-Quintela, M.A.; Buja n-Nu n ez, M.C.; Pe rez-Moure, J.C. J. Chem. Phys. 1988, 88, 7478.
25. Mozumder, A. J. Chem. Phys. 1990, 92, 1015.
26. Rice, S.A.; Butler, P.R.; Pilling, M.J.; Baird, J.K. J. Chem. Phys. 1979, 70, 4001.
27. Debye, P. J. Electrochem. Soc. 1942, 82, 265.
28. Hummel, A. Adv. Radiat. Chem. 1974, 4, 1.
29. Tachiya, M. Radiat. Phys. Chem. 1983, 21, 167.
30. Isoda, K.; Kouchi, N.; Hatano, Y.; Tachiya, M. J. Chem. Phys. 1994, 100, 5874.
31. See, for example; Rice, S.A. Diusion-Limited Reactions. Bamford, C.H. Tipper, C.F.H.,
Compton, G. eds.; Comprehensive Chemical Kinetics; Elsevier: Amsterdam, 1985; Vol. 25.
32. Traytak, S. D.; Tachiya, M. to be published.
33. Scher, H.; Montroll, E.W. Phys. Rev. B 1975, 12, 2455.
34. Blumen, A.; Klafter, J.; Zumofen, G. Optical Spectroscopy of Glasses; In Zschokke, I., Ed.
Reidel: Dordrecht, 1986; 199 pp.
35. Metzler, R.; Klafter, J. Phys. Rep. 2000, 339, 1.
36. Helman, W.P.; Funabashi, K. J. Chem. Phys. 1977, 66, 5790; 1979, 71, 2458.
37. Sung, J.; Barkai, E.; Silbey, R.J.; Lee, S. J. Chem. Phys. 2002, 116, 2338.
38. Warman, J.M. J. Phys. Chem. 1983, 87, 4353.
39. Tachiya, M. J. Chem. Phys. 1987, 87, 4108.
40. Sceats, M.G. J. Chem. Phys. 1989, 90, 2666.
41. Mozumder, A. J. Chem. Phys. 1994, 101, 10388.
42. Tachiya, M. J. Chem. Phys. 1987, 87, 4622.
43. Pitaevskii, L.P. Sov. Phys. JETP 1962, 15, 919.
44. Wojcik, M.; Tachiya, M. J. Chem. Phys. 2000, 112, 3845.
45. Morgan, W.L. J. Chem. Phys. 1986, 84, 2298.
46. Wojcik, M.; Tachiya, M. Chem. Phys. Lett. 2002, 363, 381.
47. Cohen, M.H.; Lekner, J. Phys. Rev. 1967, 158, 305.
48. Wojcik, M.; Tachiya, M. J. Chem. Phys. 1998, 109, 3999.
49. Nakamura, Y.; Shinsaka, K.; Hatano, Y. J. Chem. Phys. 1983, 78, 5820.
50. Bronskill, M.J.; Taylor, W.B.; Wol, R.K.; Hunt, J.W. Rev. Sci. Instrum. 1970, 41, 333.
51. Sumiyoshi, T.; Sawamura, S.; Koshikawa, Y.; Katayama, M. Bull. Chem. Soc. Jpn. 1982, 4,
2341.
52. Beck, G.; Thomas, J.K. J. Phys. Chem. 1972, 76, 3856.
Copyright 2004 by Taylor & Francis Group, LLC
53. Beck, G. Rev. Sci. Instrum. 1979, 50, 1147.
54. Jonah, C.D. Rev. Sci. Instrum. 1975, 46, 62.
55. Jonah, C.D.; Sauer, M.C. Chem. Phys. Lett. 1979, 63, 535.
56. Tagawa, S.; Katsumura, Y.; Tabata, Y. Chem. Phys. Lett. 1979, 64, 258.
57. Tagawa, S.; Katsumura, Y.; Tabata, Y. Radiat. Phys. Chem. 1982, 19, 125.
58. Tagawa, S.; Washio, M.; Tabata, Y. Chem. Phys. Lett. 1979, 68, 276.
59. Katsumura, Y.; Tagawa, S.; Tabata, Y. Radiat. Phys. Chem. 1982, 19, 267.
60a. Tagawa, S.; Tabata, Y.; Kobayashi, H.; Washio, M. Radiat. Phys. Chem. 1982, 19, 193.
60b. Tagawa, S.; Washio, M.; Tabata, Y.; Kobayashi, H. Radiat. Phys. Chem. 1982, 19, 277.
61. Tagawa, S.; Washio, M.; Kobayashi, H.; Katsumura, Y.; Tabata, Y. Radiat. Phys. Chem.
1983, 21, 45.
62. Yoshida, Y.; Tagawa, S.; Tabata, Y. Radiat. Phys. Chem. 1984, 23, 279.
63. Yoshida, Y.; Tagawa, S.; Tabata, Y. Radiat. Phys. Chem. 1986, 28, 201.
64. Yoshida, Y.; Tagawa, S.; Kobayashi, H.; Tabata, Y. Radiat. Phys. Chem. 1987, 30, 83.
65. Yoshida, Y.; Tagawa, S.; Washio, M.; Kobayashi, H.; Tabata, Y. Radiat. Phys. Chem. 1989,
34, 493.
66. Tagawa, S.; Hayashi, N.; Yoshida, Y.; Washio, M.; Tabata, Y. Radiat. Phys. Chem. 1989, 34,
503.
67. Kobayashi, H.; Ueda, T.; Kobayashi, T.; Tagawa, S.; Yoshida, Y.; Tabata, Y. Radiat. Phys.
Chem. 1984, 23, 393.
68. Tachiya, M. Radiat. Phys. Chem. 1987, 30, 75.
69. Kobayashi, H.; Tabata, Y. Nucl. Instrum. Methods 1985, B10/11, 1004.
70. Kobayashi, H.; Tabata, Y.; Ueda, T.; Kobayashi, T. Nucl. Instrum. Methods 1987, B24/25,
1073.
71. Yoshida, Y.; Ueda, T.; Kobayashi, T.; Tagawa, S. J. Photopolym. Sci. Tech. 1991, 4, 171.
72. Jonah, C.D.; Hart, E.J.; Matheson, M.S. J. Phys. Chem. 1973, 77, 1838.
73a. Yoshida, Y.; Tagawa, S. Proc. International Workshop Femtosecond Tech., Tsukuba, 1995;
63 pp.
73b. Tagawa, S.; Yoshida, Y.; Miki, M.; Yamamoto, T.; Ushida, K.; Izumi, Y. Proc. International
Workshop Femtosecond Tech., Tsukuba p. 31.
73c. Yoshida, Y.; Mizutani, Y.; Kozawa, T.; Saeki, A.; Seki, S.; Tagawa, S.; Ushida, K. Radiat.
Phys. Chem. 2001, 60, 313.
74. Kozawa, T.; Mizutani, Y.; Miki, M.; Yamamoto, T.; Suemine, S.; Yoshida, Y.; Tagawa, S.
Nucl. Instrum. Methods 2000, A440, 251.
75. Unpublished data measured by T. Kozawa, Y. Yoshida and S. Tagawa
76. Kozawa, T.; Mizutani, Y.; Yokoyama, K.; Okuda, S.; Yoshida, Y.; Tagawa, S. Nucl. Instrum.
Methods 1999, A429, 471.
77. Kozawa, T.; Yoshida, Y.; Tagawa, S. Jpn. J. Appl. Phys. 2002, 41, 4208.
78a. In Kinetics of Nonhomogeneous Processes; Freeman, G.R., Wiley-Interscience Pub., John
Wiley and Sons, 1987; 215 pp.
78b. Mozumder, A. Fundamentals of Radiation Chemistry; Academic Press: San Diego 1999.
79a. Jonah, C.D. Radiat. Phys. Chem. 1983, 21, 53.
79b. Sauer, M.C., Jr; Werst, D.M.; Jonah, C.D.; Trifunac, A.D. Radiat. Phys. Chem. 1991, 37,
461.
80. Saeki, A.; Kozawa, T.; Saeki, A.; Yoshida, Y.; Tagawa, S. Radiat. Phys. Chem. 2001, 60, 319.
81. Saeki, A.; Kozawa, T.; Saeki, A.; Yoshida, Y.; Tagawa, S. Jpn. J. Appl. Phys. 2002, 41, 4213.
82. Hong, Y.M.; Noolandi, J. J. Chem. Phys. 1978, 68, 5172.
83. Wojcik, M.; Hummel, A. J. Chem. Phys. 1987, 87, 5222.
84. Swallen, S.F.; Fayer, M.D. J. Chem. Phys. 1995, 103, 8864.
85. Kim, H.; Shin, S.; Lee, S.; Shin, K.J. J. Chem. Phys. 1996, 105, 7705.
86. Yoshida, Y.; Ueda, T.; Kobayashi, T.; Tagawa, S. Nucl. Instrum. Methods 1993, A327, 41.
87. Hatano, Y.; Shimamori, H. Electron and Ion Swarms; Christophorou, L.G., Ed. Pergamon
Press: New York; 1981; 103 pp.
Copyright 2004 by Taylor & Francis Group, LLC
88. Warman, J.M. The Study of Fast Processes and Transient Species by Electron Pulse Radiolysis;
Baxendale, J.H. Busi, F., Eds. NATO-ASI Series, C86, Dordrecht: D. Reidel Publ. Co, 1982;
433 pp.
89. Christophorou, L.G.; Siomos, K. Electron-Molecule Interactions and Their Applications, Vol.
2; Christophorou, L.G., Ed.; Academic Press: New York, 1984, Vol. 2, 221 pp.
90. Johnsen, R.; Lee, H.S. Swarm Studies and Inelastic Electron-Molecule Collisions; Pitchford,
L.C. Mckoy, B.V., Chutjian, A., Trajmar, S., Eds.; Springer-Verlag: New York, 1985; 23 pp.
91. Morgan, W. L. ibid. (1985) p. 43.
92. Hatano, Y. Electronic and Atomic Collisions; Lorents, D.C. Meyerhof, W.F., Peterson, J.R.,
Eds.; Elsevier: Amsterdam, 1986; 153 pp.
93. Freeman, G.R. Kinetics of Nonhomogeneous Processes; Freeman, G.R., Ed; John Wiley and
Sons Inc: New York, 1987; 19 pp.
94. Holroyd, R.A. Radiation Chemistry; Farhataziz, Rodgers, M.A.J., Ed; VCH Publ. Inc: New
York, 1987; 201 pp.
95. Holroyd, R.A.; Schmidt, W.F. Annu. Rev. Phys. Chem. 1989, 40,439.
96. Shinsaka, K.; Hatano, Y. Non-Equilibrium Eects in Ion and Electron Transport; Gallagher,
J.W., Hudson, D.F., Kunhardt, E.E., Van Brunt, R.J., Eds.; Plenum Press: New York, 1990;
275 pp.
97. Shinsaka, K.; Hatano, Y. Nucl. Instrum. Methods 1993, A327, 7.
98. Sanche, L. Excess Electrons in Dielectric Media; Ferradini, C. Jay-Gerin, J.-P., Eds.; CRC
Press: Boca Raton, 1991; 1 pp.
99. Hatano, Y. Linking the Gaseous and Condensed Phases of MatterThe Behavior of Slow
Electrons; Christophorou, L.G. Illenberger, E., Schmidt, W.F., Eds. NATO-ASI Series,
Plenum Press: New York, 1994; 467 pp.
100. Hatano, Y. Aust. J. Phys. 1997, 50, 615.
101. Mitchell, J.B.A.; Yousif, F.B. Dissociative Recombination: Theory, Experiment, and Applica-
tions; Mitchell, J.B.A., Guberman, S.L., Eds.; World Scientic: Singapore, 1989; 109 pp.
102. Shiu, Y.-J.; Biondi, M.A.; Sipler, D.P. Phys. Rev. 1977, A 15, 494.
103. Smith, D.; Adams, N.G. The Physics of Electronic and Atomic Collisions; Dalgarno, A.,
Freund, R.S., Koch, P.M., Lubell, M.S., Lucatorto, T.B., Eds.; AIP Press: New York, 1990;
325 pp.
104. Tanabe, T.; Katayama, I.; Kamegaya, H.; Chiba, K.; Watanabe, T.; Arakaki, Y.; Yoshizawa,
M.; Saito, M.; Haruyama, Y.; Hosono, K.; Hatanaka, K.; Honma, T.; Noda, K.; Ohtani, S.;
Takagi, H. The Physics of Electronic and Atomic Collisions; Dube, L.G. Mitchell, J.B.A.,
McConkey, J.W., Brion, C.E., Eds.; AIP Press: New York, 1995; 329 pp.
105. A. Muller, ibid., (1995) p. 317.
106. Hatano, Y. Phys. Rep. 1999, 313, 109.
107. Hatano, Y. J. Electron Spectrosc. Relat. Phenom. 2001, 119, 107.
108. Warman, J.M.; Senhauser, E.S.; Armstrong, D.A. J. Chem. Phys. 1979, 70, 995.
109. Senhauser, E.S.; Armstrong, D.A.; Warman, J.M. Radiat. Phys. Chem. 1980, 15, 479.
110. Van Sonsbeek, R.J.; Cooper, R.; Bhave, R.N. J. Chem. Phys. 1992, 97, 1800.
111. Cooper, R.; Burgers, M.; Bhave, R.; Van Sonsbeek, R.; Cauleld, K.; Lowke, J. Gaseous
Electronics and Its Applications; Crompton, R.W. Hayashi, M., Boyd, D.E., Makabe, T., Eds.;
KTK Sci. Publ.: Tokyo, 1991; 73 pp.
112. Tezuka, T.; Namba, H.; Nakamura, Y.; Chiba, M.; Shinsaka, K.; Hatano, Y. Radiat. Phys.
Chem. 1983, 21, 197.
113. Morgan, W.L.; Bardsley, J.N. Chem. Phys. Lett. 1983, 96, 93.
114. Tachiya, M. J. Chem. Phys. 1986, 84, 6178.
115. Morgan, W.L. J. Chem. Phys. 1990, 92, 1015.
116. Kaneko, K.; Usami, Y.; Kitahara, K. J. Chem. Phys. 1988, 89, 6420; Kaneko, K.; Tanimoto,
J.; Usami, Y.; Kitahara, K. J. Phys. Soc. Jpn. 1990, 59, 56.
117. Lopez-Quintela, M.A.; Bujan-Nunes, M.C. Chem. Phys. 1991, 157, 307.
118. Mozumder, A. J. Chem. Phys. 1993, 98, 8347; Chem. Phys. Lett. 1993, 245, 359.
Copyright 2004 by Taylor & Francis Group, LLC
119. Fedorenko, S.G.; Burshtein, A.I. J. Chem. Phys. 1997, 107, 6659.
120. Shinsaka, K.; Codama, M.; Srithanratana, T.; Yamamoto, M.; Hatano, Y. J. Chem. Phys.
1988, 88, 7529; Shinsaka, K.; Codama, M.; Nakamura, Y.; Serizawa, K.; Hatano, Y. Radiat.
Phys. Chem. 1989, 34, 519.
121. Honda, K.; Endou, K.; Yamada, H.; Shinsaka, K.; Ukai, M.; Kouchi, N.; Hatano, Y. J.
Chem. Phys. 1992, 97, 2386; Honda, K.; Endou, K.; Yamada, H.; Isoda, K.; Shinsaka, K.;
Ukai, M.; Kouchi, N.; Hatano, Y. ibid, 1992, 98, 8348.
122. Ukai, M.; Odaka, T.; Yamada, H.; Isoda, K.; Shinsaka, K.; Kouchi, N.; Hatano, Y. Int. J.
Mass Spectrom. Ion Process. 1995, 149/150, 451.
123. Takeda, K.; Kato, R.; Hayashida, M.; Odaka, T.; Shinsaka, K.; Kameta, K.; Odagiri, T.;
Kouchi, N.; Hatano, Y. J. Chem. Phys. 2001, 114, 3554.
124. Hatano, Y. Proceedings of Xe-01 Symposium, December 34, 2001, Kashiwa; Suzuki, Y., Ed.;
World Scientic: Singapore, 2000. In press.
Copyright 2004 by Taylor & Francis Group, LLC
11
Radical Ions in Liquids
Ilya A. Shkrob and Myran C. Sauer, Jr.
Argonne National Laboratory, Argonne, Illinois, U.S.A.
1. INTRODUCTION
The interaction of ionizing radiationfast electrons, a-particles, x- and g-rays, and
ultraviolet (UV) and vacuumultraviolet (VUV) photonswith molecular solids and liquids
causes the formation of short-lived electron-hole pairs that, in such media, thermalize and,
eventually, localize yielding radical ions and/or trapped (solvated) electrons and holes (see
Chaps. 15). The distinction between the radical anions and the solvated electrons is
arbitrary. For the time being, it will be assumed that radical ions have an excess electron
or electron deciency in the valence orbitals of a single solvent molecule (molecular ions
or monomer ions) or a small group of such molecules (dimer ions or multimer ions)
that do not share charge with neutral solvent molecules that solvate them. Naturally, the
excess electron in a radical anion is indistinguishable from other valence electrons in this
anion. By contrast, in the solvated electron (also known as cavity electron, see Chap.
7), the electron density resides mainly in interstitial sites between the solvent molecules
(solvation cavity) that are polarized by the negative charge at its center (thereby forming
the outer shell of a negative polaron). The underlying assumption of this visualization is
that the solvated electron is a single-electron state whose properties can be given by a band
model in which the valence electrons in the solvent and the excess electron in the cavity are
wholly separately treated [1]in the exact opposite way to how the electronic structure of
the solvent radical ion is viewed. An additional assumption is usually made that the excess
electron interacts with (rigid, exible, or polarizable) solvent molecules by means of an
empirical classical potential. Both of these simplifying assumptions nd little support in
structural studies of trapped electrons in vitreous molecular solids using magnetic res-
onance spectroscopy [2].
The primary speciessolvated/trapped electrons/holes and solvent radical ions
are ecient donors and acceptors of electrons and protons; they readily react with the
solvent, with the solute and dopant molecules, with each other, and with the short-lived
species (radicals, molecular fragments, and excited states) generated in the ionization and
excitation events along with these primary charges. In most radiation chemistry studies,
the species of interest are the resulting secondary ions, radicals, and excited neutrals.
For example, radiolysis of liquid solutions and doped solid matrices is widely used for
producing high yields of secondary radical ions (via ionmolecule reactions of the primary
ions) for detailed structural studies, e.g., by using magnetic resonance and UVVis and/or
Copyright 2004 by Taylor & Francis Group, LLC
infrared (IR)-Raman spectroscopies [3]. The general approach in these structural studies is
to generate a primary ion that rapidly reacts with a dopant (or solute) of interest yielding a
secondary ion that is relatively stable in the matrix (on the time scale of observation). Such
stabilization usually requires reduced mobility (low sample temperature), low reactivity,
and high ionization potential (IP) or electron anity (EA) of the solvent/matrix relative to
the solute. A classical example of the matrix stabilization is radiolytic generation of
hydrocarbon radical cations in frozen halocarbon (freon) solutions [3,4]. Rapid dissocia-
tive electron attachment to the halocarbon solvent yields relatively immobile halide anions
and C-centered radicals that recombine slowly (if at all) in these low-temperature matrices
[3]. The solute radical cations are rapidly formed (on the time scale of the observations)
because of the oxidation of the dopant by a mobile solvent hole [3]. As a result, the ionization
energy is eciently converted into the product (radical cation), enabling many types of
structural studies. A similar approach is used for stabilization of radical cations in rare gas
matrices doped by SF
6
(a matrix favored by IR-Raman spectroscopists [5]), zeolites, and
microporous solids (favored by electron paramagnetic resonance (EPR) spectroscopists [6]),
etc. A review of the literature shows that almost a third of radiolytic studies on organic
radical ions are done in this fashion.
Another popular approach is to react the intermediates of liquid water radiolysis
hydroxyl radicals and hydrated electronswith the parent molecules of interest (see
several representative reviews in Ref. 7). The resulting solute radical ions are observed in a
time-resolved fashion, using pulse radiolysistransient absorption spectroscopy with
nanosecond-to-millisecond time resolution; other detection techniques may involve uo-
rescence, time-resolved EPR, and Raman spectroscopy. The emphasis in such studies is to
observe structural rearrangements and reaction patterns of short-lived solute radical ions.
The reaction behavior is usually complex (ring opening, protonation/deprotonation,
rearrangements, etc.), especially for radical ions whose parent molecules are multifunc-
tional organic compounds, such as biomolecules and biomimetic systems (see, e.g., Chap.
15). Whether the solute is simple (inorganic ions) or complex (biomolecules), the general
approach stays the same. A variation of this common technique is pulse radiolysis of
solutes in aprotic, high-IP liquids, such as acetone, CC
l4
, CH
2
Cl
2
, 1,2-dichloroethane, and,
most often, n-butyl chloride [8]. Unlike the OH radicals in water, solvent holes in such
liquids oxidize almost any organic solute; the resulting secondary cations are usually more
stable than the same species in water, for lack of ecient donors and acceptors of the
proton. There are many examples of using this approach for studies of radical cations of
hydrocarbons, heterocyclic compounds, phenols, etc.
Finally, solute radical ions can be generated by light-induced, one-photon or multi-
photon ionization of their parent compounds (Chaps. 5 and 16). This approach is par-
ticularly useful in the ultrafast studies of short-lived, unstable radical ions that aim to
unravel their solvation, recombination, reaction, and vibrational relaxation dynamics of
the primary charges (see, e.g., Chap. 10). Whereas the time scale of radiolytic production
of secondary ions is always limited by the rate with which the primary species reacts with
the dispersed parent molecules, light-induced charge separation can occur in <100 fsec.
There are many studies on photoionization of solute molecules in liquid solutions; we do
not intend to review these works.
There are also many photoreactions, in which rapid, ecient charge transfer between
electron donor and acceptor moieties occurs, yielding a radical ion pair that subsequently
dissociates; because the solvent is not oxidized or reduced in these light-induced reactions,
these reactions belong to photo- rather than radiation chemistry. Of particular interest, in
the context of radiation chemistry studies, are photoreactions in which the solvent is
Copyright 2004 by Taylor & Francis Group, LLC
oxidized and the resulting hole reacts with the solutes yielding secondary radical ions, just
like in radiolysis. This occurs, for example, in one-photon excitation of certain salts (such
as N-methyl quinolinium pentauoride) in low-IP nonpolar uids (benzene and toluene)
[9,10,11]. Upon photoexcitation, the cation moiety accepts a valence electron from the
solvent, and the resulting dimer radical cation of the solvent reacts with solute molecules.
This reaction is so ecient that it has been used for preparative organic synthesis [9].
Similar photooxidation reactions of aliphatic alcohols, by excited transition metal (e.g.,
Fe
III
) [10] or stable organic cations (e.g., methylviologen
2+
) [11], have been studied.
Another class of photooxidation reactions mimicking radiolytic reactions involves elec-
tronically excited radical cations [12,13]. Radical cations are more ecient oxidizers than
stable cations because they are closer in energy to the valence band of the solvent.
Photoexcited aromatic radical cations (IR to UV) readily oxidize most solvents, including
saturated hydrocarbons [12], water, and alcohols [13]. These charge-transfer-from-solvent
(CTFS) photoreactions mirror the better-known charge-transfer-to-solvent (CTTS) reac-
tions in which photoexcited anion (e.g., halide) injects an electron into a polar liquid [14].
The photoelectrons released in the CTTS or photoionization reactions can be used to
generate secondary radical anions, although this approach is seldom used in practice.
Understandably, most workers who use radiolysis, photoionization, CTFS, or CTTS
as the means for generation of (secondary) radical ions pay little attention to the nature of
short-lived precursors of these ions. After all, the subject of interest is a secondary rather
than a primary ion. This ad hoc approach is justiable because radiolytic production is
just another means of obtaining a sucient yield of the radical ion. Quite often in such
studies, the radiolysis is complemented by other techniques for radical ion generation,
such as plasma oxidation, electron bombardmentmatrix deposition, and chemical and
electrochemical reduction or oxidation. While the data obtained in these studies are useful,
there is little radiation chemistry in suchnominally, radiation chemistrystudies.
There are many excellent books and reviews on the structure and reactions of
secondary radical ions generated in radiolytic and photolytic reactions. Common topics
include the means and kinetics of radical ion production, techniques for matrix stabiliza-
tion, electronic and atomic structure, ionmolecule reactions, structural rearrangements,
etc. On the other hand, the studies of primary radical ions, viz. solvent radical ions, have
not been reviewed in a systematic fashion. In this chapter, we attempt to close this gap. To
this end, we will concentrate on a few better-characterized systems. (There have been many
scattered pulse radiolysis studies of organic solvents; most of these studies are inconclusive
as to the nature of the primary species.)
Before we review specic systems, note that the primary species should be considered
on a dierent footing than the secondary radical ions. The latter ions are well isolated
from their parent molecules by the matrix or the solvent. By contrast, in the primary
species, the charge is residing on a molecule(s) that is surrounded by like solvent
molecules. This often results in unusual properties because the barriers for charge hopping
and charge delocalization are lowered. We will examine several examples of the reaction
and migration dynamics of such primary radical ions; the multiplicity of examples and
commonality of the observed behavior suggest a general pattern.
As demonstrated below, a primary charge viewed as a solvated electron or the mo-
lecular ion residing in an inert liquid does not account for experimental observations in
many, if not in most, of the systems. While we cannot oer a specic, general model of
these exceptional ions, we provide a general introduction to the known properties of
such species. Furthermore, we argue that these species comprise the rule rather than the
exception. The reader is invited to reach his or her own conclusions.
Copyright 2004 by Taylor & Francis Group, LLC
2. ELECTRONS AND SOLVENT ANIONS IN SUPERCRITICAL CO
2
The rst exceptional system that we review is carbon dioxide [1521]. Supercritical (sc)
CO
2
nds numerous industrial applications as a green solvent, and this practical con-
sideration stimulates interest in its radiation chemistry. Though the studies of sc CO
2
are
recent, this system is particularly interesting because of the simplicity of the solvent
molecule and extensive gas phase and matrix isolation studies of the corresponding ions
(see below).
Like other sc uids, sc CO
2
is a collection of nanoscale molecular clusters that
rapidly associate, dissociate, and exchange molecules among each other [22]. Prethermal-
ized electrons in sc CO
2
attach to these (CO
2
)
n
clusters. In the gas phase, the attachment
can be dissociative [reaction (2)] and or nondissociative [reaction (1)],
e
CO
2
n
! CO
2
n
1
!COO
CO
2
n1
!COCO
3
CO
2
n2
2
depending on the electron energy [for n=2, reaction (2) is 3.6 eV more endothermic] [23].
Although in radiolysis, a large fraction of electrons have initial excess energies >10 eV, the
yield of CO
3
,
shown in Fig. 1a (with lifetime>2 msec), and further increases to 3.4 eV for n = 6 clusters.
In larger clusters, the VDE rst decreases to 2.49 eV (n = 7), then monotonically increases
to 3.14 eV (n = 13); then, for nz14, the VDE suddenly increases to 4.55 eV. The onset of
the photoelectron spectrum increases from 1.5 eV for n = 27 clusters to 1.82 eV for n =
813 clusters to 3 eV for n = 1516 clusters. Tsukuda et al. [24] argue that while the core
of small (nV6) and large (nz14) clusters is a D
2d
dimer anion, the core of 6<n<14 clusters
is a monomer anion weakly coupled to several CO
2
molecules (with binding energy ca.
0.22 eV per molecule). Both of these forms coexist in n = 6 and n = 13 clusters. It appears
that the two isomers are close in energy, and core switching readily occurs when the cluster
geometry and size change. Because in sc uid there is a wide distribution of the cluster sizes
and shapes [22], the solvent radical anion does not have a well-dened geometry.
In CO
2
gas, the density-normalized electron mobility l
e
s
e
is independent of temper-
ature (2 10
22
molecule/cm V sec [25]), although the apparent mobility steadily decreases
with the pressure: free electrons are trapped by neutral (CO
2
)
n
clusters (n
c
6) with nearly
collisional rates, and the electron anity of these clusters >0.9 eV. When extrapolated to
solvent densities of (215) 10
21
cm
3
typical for sc CO
2
, these estimates suggest that the
free electron mobility l
e
is ca. 1 cm
2
/V sec and its collision-limited lifetime s
e
<30 fsec [18].
If the lifetime were this short, the electrons would negligibly contribute either to the
conductivity or the product formation. However, this extrapolation is not supported by
experiment [18,20].
In low-temperature solid matrices (e.g., Ne and Ar at 511 K), the CO
2
monomer,
(CO
2
)(CO
2
)
n
multimer anion (n = 1, 2), and the C
2
O
4
69
clusters
observed by Tsukuda et al. [24] in the gas phase; it is distinctly dierent from the electron
photodetachment spectra of CO
2
qf
(with mobility >10 cm
2
/V sec) is trapped by
dense sc CO
2
solvent, a high-mobility radical anion, (CO
2
)
n
, withl() of 10
2
cm
2
/Vsec is formed. The
binding energy of the electron is 1.61.8 eV; the electron can be detrapped by absorption of a photon
with energy >1.76 eV. (b) Electron photodetachment (empty symbols) and photoabsorption (lled
circles) spectra of solvent radical anion in sc CO
2
. The arrow points to the onset of the
photodetachment band. (c) Photoinduced electron detachment in sc CO
2
observed by dc photo-
conductivity. The initial (clipped) narrow peak is the prompt conductivity signal from free electrons;
the time prole of this signal follows the shape of the 248-nmexcitation laser pulse. The arrows indicate
the delay times at which a second (532 nm) laser pulse was red. The 532-nm photoexcitation detraps
the electrons that subsequently trap within the duration of the 532-nm laser pulse (sharp spikes).
Copyright 2004 by Taylor & Francis Group, LLC
Both the electrons and the solvent anions react with nonpolar electron acceptor
solutes with EA
g
>0.4 eV [18]. For oxygen (EA
g
of 0.4 eV), the electron transfer from the
solvent anion to the solute (which yields a stable CO
4
anion rapidly forms OO bound complex with a solvent molecule, yielding a stable
radical anion, CO
4
. (a) The 532-nm photoexcitation can detach an electron from the solvent anion
(Fig. 1a). CO
4
is a 0.44 eV deeper trap and the electron cannot be detached from it. By determining
the magnitude of the spike from quasifree electrons (generated by 532-nm laser excitation of the
pholysate) as a function of time, one can obtain the decay kinetics of the solvent anion. (b) Ex-
perimental realization of this concept for a sc CO
2
solution containing 120 AM of O
2
. The conduc-
tivity signal (r) shown on the double-logarithmic scale shows a gradual transformation of the solvent
anion to CO
4
. The ratio Dr/r of the 532-nm laser induced conductivity signal to the conductivity
tracks the concentration of the solvent anion in a reaction mixture that contains both (CO
2
)
n
and
CO
4
. After the rst 10 Asec, this fraction persists at a small value; this represents the settling of the
equilibrium between the solvent anion and CO
4
.
Copyright 2004 by Taylor & Francis Group, LLC
from the complexes to the CB is much less ecient than that of free solvated electrons
(Chap. 8). No changes in the absorption band of the solvated electron in the NIR are
observed upon the complexation of these electrons with the monomers and dimers of polar
molecules [32]. Higher alcohol multimers, such as tetramers, provide very deep traps for
these electrons; absorption spectra of such cluster-trapped electrons are almost identical to
the spectra of solvated electrons in neat aliphatic alcohols [33].
That the solvent anion in sc CO
2
demonstrates behavior similar to that of solvated
electrons in saturated hydrocarbons suggests that the mechanism of binding to the polar
molecules must be similar. Apparently, increasing the trapping energy by ca. 0.15 eV
because of the electrostatic binding to a monomer or dimer molecule with a dipole
moment of (24) Debye is sucient to halt the charge hopping completely [29]. The same
eect of polar solutes on the hopping rate of high-mobility solvent radical cations (with
fairly similar thermochemistry) was observed in neat cis- and trans-decalin (Sec. 5).
sc CO
2
is not the only liquid for which high-mobility solvent anions are observed (see
below), but it is the simplest one. In monoatomic liquids (liqueed Ar, Kr, and Xe),
ammonia, and simple hydrocarbons (CH
4
, C
2
H
6
), solvent anions are not formed (quasi-
free or solvated electrons are formed instead; see Chap. 8). In diatomic liquids, such as N
2
and O
2
, solvent anions are generated, but the formation of strongly bound dimer anions
precludes rapid electron hopping. Even for CO
2
, high-mobility anions are not observed in
the low-temperature liquid: Considerable thermal activation is needed to break the dimer
anion prior to every hop of the electron. In liquids whose solute molecules show less
tendency to form strongly bound dimer anions, the electron hopping is faster and requires
much less activation energy. As argued below, such a situation occurs in carbon disulde,
hexauorobenzene, benzene, and toluene.
3. SOLVENT ANIONS IN LIQUID CS
2
, C
6
F
6
, AND AROMATIC
HYDROCARBONS
Carbon disulde is isovalent to carbon dioxide and it also has a bent monomer anion.
While gas-phase CO
2
has negative EA
g
of 0.6 eV [24], for CS
2
, EA
g
is +0.8 eV [34].
Despite this very dierent electron anity, Gee and Freeman [34] observed long-lived
electrons in CS
2
(with lifetime>500 Asec) with mobility ca. 8 times greater than that of
solvent cations. Over time, these electrons converted to secondary anions whose
mobility was within 30% of the cation mobility. Between 163 and 500 K, the two ion
mobilities scaled linearly with the solvent viscosity, as would be expected for regular ions.
Of course, Gee and Freemans identication of the long-lived high-mobility solvent anions
as electron is just a manner of speech: Obviously, quasifree or solvated electrons cannot
survive for over a millisecond in a positive-EA
g
liquid.
To the best of our knowledge, pulse radiolysistransient absorption studies of neat
CS
2
have not been reported. CS
2
anions, the CS
2
anions, CS
2
and C
2
S
4
(CS
2
)
n2
anions
results in the fragmentation of the dimer core to C
2
S
2
and S
2
[37]. The dimer anion was
also observed in the EPR spectra of g-irradiated CS
2
-doped frozen alcohols; the g-tensor
parameters suggested SS bonding [39]. Interestingly, while C
2
O
4
with C
6
F
6
results in diusion-controlled degenerate electron
transfer rather than anion dimerization [44]. Thus, while the data clearly point to charge
sharing and charge hopping, there seems to be no evidence that a metastable dimer is
formed. Once more, one needs to postulate a exible-structure multimer anion in which
fractional negative charge freely exchanges between the solvent molecules.
Importantly, both for liquid C
6
F
6
and CS
2
, there is a relatively high yield of ortho-
positronium (o-Ps) observed in e
+
irradiation of these uids [34]. The o-Ps is formed in the
e
+
e
recombinations that occur in the end-of-track spurs. The higher the negative charge
mobility, the higher the probability that these e
+
e
,
has not been observed in the gas phase. In the neutral trimer, one of the monomers couples
sideways to the antiparallel pair; this molecule binds the electron in the same way as the
monomer; the adiabatic electron anity of this trimer (1420 meV) is higher than that of
the monomer [47]. Higher multimer anions, {CH
3
CN}
n
cluster was
modeled using a density functional (B3LYP) method [30]. A 631+G** basis set that
included polarized (d,f) and diuse functions was used and the C
3h
symmetry was imposed.
A ghost hydrogen atom with zero charge was placed at the center of the cluster to
provide s-functions for the solvated electron. The polarizable (overlapping spheres)
continuum model was implemented in the integral equation formalism. The lowest energy
state was a propeller-like 2AV state shown in Fig. 6. The CCN angle in the acetonitrile
subunits is 178j in vacuum and 168j in solution (vs. 180j in the neutral molecule). This
bending is considerably smaller than in the monomer and dimer anions (ca. 130j). The
solvated cluster anion is compact: The closest methyl hydrogens are 0.171 nm away from
the symmetry center. The CC bond in the acetonitrile subunits is elongated from 0.144 to
0.148 nm, while the CN bond is very slightly changed. In this structure, the singly
occupied molecular orbital (SOMO) envelopes the whole cluster anion. The main negative
nodes are on methyl carbons, while the main positive nodes are at the center of symmetry,
on the in-plane hydrogens, and on carbonyl carbons. This structure may be viewed both
as a multimer anion and a solvated electron: The SOMO has a noticeable s-character at
the symmetry center (ca. 0.34), although the main spin density is on the methyl carbons.
The latter atoms exhibit large hyperne coupling constants (hfcc) for
13
C: the isotropic
hfcc is 6.9 mT; the anisotropy is negligible. Isotropic hfcc for methyl protons are relatively
small: 0.19 mT for in-plain hydrogens (the principal values of the dipole tensor are 0.29,
0.15, and 0.44 mT) and 0.086 mT for out-of-plane hydrogens (0.29, 0.16, and
+0.45 mT, respectively). The isotropic hfcc for cyanide
13
C and
14
N nuclei are 0.4 and
0.36 mT, respectively.
Copyright 2004 by Taylor & Francis Group, LLC
The structure bears strong resemblance to the trapped electron in saturated hydro-
carbons studied by Kevan et al. [2]. The electron is solvated by methyl groups; the
positive charge on these groups is increased because of the considerable elongation of CC
bonds. This elongation, as demonstrated by our discrete Fourier transform (DFT)
calculations, is the consequence of large electron density on the skeletal carbon atoms.
In the semicontinuum model of Kevan et al. [2], this (multielectron) interaction is treated
in terms of a polarizable CC bond; our calculation justies their ad hoc approach. The
Figure 6 Singly occupied molecular orbital (SOMO) of a propeller-like trimer radical anion of
acetonitrile obtained using density functional theory. The structure was immersed in a polarizable
dielectric continuum with the properties of liquid acetonitrile. Several surfaces (on the right) and
midplane cuts (on the left) are shown. The SOMO has a diuse halo that envelops the whole cluster;
within this halo, there is a more compact kernel that has nodes at the cavity center and on the
molecules.
Copyright 2004 by Taylor & Francis Group, LLC
size of the solvation cage, the juxtaposition of methyl groups, and the hfcc tensors for
methyl protons favorably compare with those experimentally obtained by Kevan et al. for
the trapped electron in frozen 3-methylpentane [2]. Therefore it is reasonable that the
multimer {CH
3
CN}
n
aq
, H
.
,
.
OH, H
2
, and
H
2
O
2
resulting from the spur reactions and how they are measured using scavengers, or
by direct observation using pulse radiolysis. Next, the physical and chemical properties of
e
aq
, H
.
, and
.
OH are presented, followed by methods of converting these primary radicals
to secondary ones of a single kind, either oxidizing or reducing, and covering a wide range
of reduction potentials. From this information, one can determine what G-values to use
for the radicals and can tune the reduction potential for the chemical system of interest.
Finally, reference is made to a recently published textbook [Radiation Chemistry. Present
Status and Future Prospects, Jonah, C. D., Rao, B. S. M., Eds.; Elsevier: Amsterdam,
2001] that gives details of the wide range of chemical studies that have been made using
water radiolysis as a tool.
1. HISTORICAL PERSPECTIVE
Systematic studies of the eects of ionizing radiation on liquid water began in earnest in the
1940s, at the University of Chicago [1], as part of the work on the atomic bomb project. This
project involved building a nuclear reactor to produce plutoniumand, because of its neutron
absorption cross-section, fast-owing water was deemed to be the best method for cooling
the reactor. Fast neutrons knock out protons from water molecules and these energetic
charged particles ionize and excite further molecules to produce a multitude of free electrons
Copyright 2004 by Taylor & Francis Group, LLC
and positively charged species. The ejected electrons themselves can also have sucient
energy to cause ionization and excitation of water molecules. Qualitatively, whether the
energetic particle is a fast neutron, charged particle, or photon (i.e., x-ray or g-ray), the result
is that a single interaction of such a particle causes several water molecules to be ionized or
excited. These initial events are classed as radiation physics. The radiation chemistry
comprises the sequence of events that lead to the nal products of water decomposition
resulting from the ionization and excitation processes. The yields of these nal products do,
however, depend on the nature of the original energetic particle. As will be seen later, this
sequence of events is divided into a physicochemical stage, during which the initial products
reach thermal equilibrium with the liquid, followed by a kinetically nonhomogeneous
reaction stage.
The idea that ionizing radiation decomposes water into hydrogen atoms and hydroxyl
radicals according to reaction (1) was suggested rst by Risse [2] in 1929 and later by Weiss
[3] in 1944:
H
2
OQH
.
.
OH 1
In 1938, Fricke et al. [4] had shown that although x-rays seemed not to decompose pure
water, they caused oxidation and reduction to take place in aqueous solutions. On the other
hand, it was also known from the earliest days of radiochemistry that a-rays decompose
liquid water to H
2
, O
2
, and H
2
O
2
. This dependence of the decomposition of water on the
linear energy transfer (LET) of the ionizing radiation was rst explained by Allen [5], who
put forward the concept that the molecular products are formed in tiny regions of high
radical concentrations, so-called spurs, along the track of the ionizing particle. Essen-
tially, there is a competition between the back reactions of the radicals in the spurs to form
the molecules H
2
, H
2
O
2
, and H
2
O, and their escape by diusion into the bulk liquid. This led
to the development of the spurdiusion model (see Section 2 below) and was a major
contribution to the understanding of the radiation chemistry of water, which at that time
was represented by reaction (2):
H
2
OQH
.
;
.
OH; H
2
; H
2
O
2
2
The yields of these so-called primary species, present at the time when radical combina-
tion in, and diusive escape from, the spurs is complete, were obtained by adding solutes to
the water to capture the radicals and by measuring the stable identifying products. It was
from a number of these studies that it became clear that the reducing radical must exist in
two forms, which turned out to be the hydrogen atom and the hydrated electron (e
aq
). For
example, Hayon and Weiss [6] found that the yields of H
2
and Cl
produced by irradiating
solutions of chloroacetic acid varied with pH in a manner that was consistent with the
following reactions:
e
aq
H
! H
.
3
e
aq
ClCH
2
CO
2
H ! Cl
.
CH
2
CO
2
H 4
H
.
ClCH
2
CO
2
H ! H
2
.
CHClCO
2
H 5
It is now generally agreed that for low LET radiation, such as x-rays, g-rays, and fast elec-
trons, the radiolysis of water can be represented to a good approximation by reaction (6):
4:36H
2
OQ2:8e
aq
; 0:62H; 2:8OH; 0:47H
2
; 0:73H
2
O
2
; 2:8H
6
Copyright 2004 by Taylor & Francis Group, LLC
The numbers in reaction (6) are the so-called primary yields (expressed here in units of
10
7
mol J
1
) of the radical and molecular products that are present when the spur processes
are complete at ca. 10
7
sec. The original units for these yields, which are still in use, are
molecules (100 eV)
1
, and the relationship between the two sets is 1 mol J
1
u9.6510
6
molecules (100 eV)
1
. Thus, reaction (6) may also be written as reaction (7):
4:2H
2
OQ2:7e
aq
; 0:6H; 2:7OH; 0:45H
2
; 0:7H
2
O
2
; 2:7H
7
In this chapter, the various radiation chemical yields, known as G-values, are dened as
follows: g(X) is the yield of the species in reactions (6) and (7); G
o
(X) is the yield of the initial
products of water radiolysis at the end of the physicochemical stage; and G(X) is an
experimentally measured yield. In some publications G
X
, rather than g(X), is used to
represent the primary yield of the species X. The yields in reactions (6) and (7) are
numerically very similar, but one must be sure of the units that are being used to express
them. In the older literature, it was common practice to quote G-values without units when
the units were in fact molecules (100 eV)
1
.
Shortly after the discovery of the hydrated electron, Hart and Boag [7] developed the
method of pulse radiolysis, which enabled them to make the rst direct observation of this
species by optical spectroscopy. In the 1960s, pulse radiolysis facilities became quite widely
available and attention was focussed on the measurement of the rate constants of reactions
that were expected to take place in the spurs. Armed with this information, Schwarz [8]
reported in 1969 the rst detailed spurdiusion model for water to make the link between
the yields of the products in reaction (7) at ca. 10
7
sec and those present initially in the spurs
at ca. 10
12
sec. This time scale was then only partially accessible experimentally, down to ca.
10
10
sec, by using high concentrations of scavengers (up to ca. 1 mol dm
3
) to capture the
radicals in the spurs. Fromthen on, advancements were made in the time resolution of pulse
radiolysis equipment from microseconds (10
6
sec) to picoseconds (10
12
sec), which
permitted spur processes to be measured by direct observation. Simultaneously, the increase
in computational power has enabled more sophisticated models of the radiation chemistry
of water to be developed and tested against the experimental data.
More generally, water radiolysis provides a very versatile method of generating other
free radicals and unstable intermediates such as hyperreduced metal ions. This, when
combined with the technique of pulse radiolysis, has, without doubt, made an enormous
contribution to our knowledge of one-electron redox chemistry in aqueous inorganic,
organic, and biochemical systems that cannot be studied readily by thermal or photo-
chemical methods. Most commonly, this means using radiation of low LET such as
60
Co g-
rays or beams of high-energy electrons. The eects of increasing LETand/or temperature on
the chemical stage of water radiolysis are covered in Chaps. 14 and 23, respectively.
In this chapter, an outline of the radiation chemistry of water is given in Section 2.
Section 3 describes how the yields of the radical and molecular products evolve with time,
and indicates the bounds of useful experimental conditions for the use of water radiolysis as
a tool in general chemistry. In these two sections, emphasis is placed on recent developments
in modelling experimental data that shed newlight on our understanding of the fundamental
aspects of water radiolysis. Section 4 contains a summary of the chemical and physical
properties of the primary radicals. In Section 5, examples are given to show how these
radicals can be converted into secondary radicals, often of a single kind. This is an important
consideration for the study of the chemistry of free radicals and metal ions in unusual
oxidation states. Finally, reference is given in Section 6 to examples of applications that have
resulted in signicant advances being made in general chemistry.
Copyright 2004 by Taylor & Francis Group, LLC
2. SCHEME FOR THE RADIATION CHEMISTRY OF WATER
The approximate time scale of events initiated by the absorption of energy by water fromthe
incident ionizing radiation is shown in the following scheme:
Initially, a water molecule is ionized or electronically excited in about 10
16
sec or less,
depending on the energy of the particle causing the ionization. The ejected electron generally
has sucient energy to ionize or excite further water molecules. This process continues until
the energy of the ejected electron is no longer sucient to raise a water molecule above its
ground electronic state. This so-called subexcitation electron loses the rest of its excess
energy by exciting vibrational and rotational modes of the solvent molecules.
At the end of the physical stage, which is within about 10
16
sec of the passage of the
ionizing particle through the liquid, the track made by the particle contains H
2
O
+
,
subexcitation electrons e
se
, and electronically excited water molecules H
2
O* in small clusters
called spurs. From about 10
16
to 10
12
sec, the following processes are thought to occur
and comprise the physicochemical stage [9,10]:
H
2
O
se
! H
2
O
vib
*
I
Copyright 2004 by Taylor & Francis Group, LLC
H
2
O
vib
*
H
.
.
OH
H
2
O
.
1
D ! H
2
H
2
O
2
or 2
.
OH
2H
.
O
.
3
P
Ia
Ib
Ic
H
2
O
H
2
O ! H
aq
.
OH II
e
se
H
2
O !
.
OH H
!
.
OH H
2
OH
III
e
se
H
2
O ! e
aq
IV
H
2
O
elec
*
.
OH H
.
H
2
O
.
1
D ! H
2
H
2
O
2
or 2
.
OH
2H O
.
3
P
Va
Vb
Vc
Processes (I) and (II) account for H
2
O
+
, whereas processes (I), (III), and (IV) describe the
fate of e
se
. According to Kaplan et al. [11], process (I) produces water molecules in high
vibrational levels of their electronic ground state. The remaining H
2
O
+
reacts with water to
form H
aq
and
.
OH in process (II). This ionmolecule reaction is known to occur in the gas
phase with a rate constant of 810
12
dm
3
mol
1
sec
1
[12], which, when extrapolated to
liquid water, sets the lifetime of H
2
O
+
in this medium at less than 10
14
sec. However,
Hamill [13] pointed out that H
2
O
+
initially has the structure of a neutral water molecule so
that it may migrate rapidly over distances of a fewmolecular diameters by resonant electron
transfer with a succession of neighboring water molecules.
The subexcitation electrons that escape process (I) either undergo a dissociative
attachment to H
2
O as in process (III), or thermalize and become solvated (process (IV)).
Process (III), which is important in the gas phase, was suggested by Platzman [14] and
Faraggi and Desalos [15] to be responsible for an apparently unscavengeable yield of
molecular hydrogen in the radiolysis of liquid water. That a precursor of the hydrated
electron is the source of at least part of G(H
2
) has been established by the nding that
scavengers such as selenate ion, dichromate ion, and nitrate ion, which are known to react
eciently with a precursor of the hydrated electron (see Section 3.1.1), lower G(H
2
) to near
zero at suciently high concentrations, for both low[16] and high [17] LET radiation. It has
also been shown that when water is photoionized using a subpicosecond laser pulse [18],
process (IV) occurs in less than 10
12
sec.
Very little is known about the decay channels for excited water molecules in the liquid
phase. Consequently, they are generally assumed [9,10] to be essentially the same as those in
the gas phase.
2.1. Modelling
In recent years, two dierent approaches, deterministic [9,19] and stochastic [10,20], have
been used with a good level of success to model the radiation chemistry of water. Each
approach leads to reasonable agreement between calculated results and experimental data
obtained for a wide range of LET fromroomtemperature up to ca. 300jC[9,10]. There are,
however, fundamental dierences between the two models. The deterministic model is based
on the concept of an average spur [8,9,19,23] at the end of the physicochemical stage (ca.
10
12
sec), which contains the products of processes (I), (II), (III), (IV), and (V) in certain
yields and spatial distributions, and in thermal equilibrium with the liquid. For low LET
Copyright 2004 by Taylor & Francis Group, LLC
radiation, the spurs are treated as isolated spheres with usually a Gaussian distribution of
each species, and for high LETas continuous cylindrical tracks with a Gaussian distribution
having pseudo-two-dimensional symmetry. The initial yields ( G
o
) of the species and the
width of their spatial distributions are chosen to t with the experiment, but the two sets of
parameters are not independent [24]. The nonhomogeneous kinetics of the spur processes of
reaction and diusion is treated as a set of coupled dierential equations, which are solved
numerically.
The stochastic model [10,20] uses Monte Carlo simulations of the events that take
place during the physical stage up to ca. 10
15
sec, which, in principle, lead naturally to the
spatial distributions at the end of the physicochemical stage (ca. 10
12
sec). However,
assumptions have to be made about the energy loss processes of the incident ionizing
radiation and the secondary electrons that are thereby generated because these processes are
not well known quantitatively for liquid water. As in the deterministic approach, the
parameters are adjusted to give a t with experiment. Thereafter, the nonhomogeneous
reaction kinetics between ca. 10
12
and 10
6
sec is treated by the method of Independent
Reaction Times (IRT) [25]. This method assumes that the reaction time of each pair of
reactants is independent of the presence of other reactants in the system. The important spur
reactions are listed in Table 1.
The major dierence between the deterministic and stochastic methods lies in the
spatial distributions, and their dependence on temperature, of the species at the end of the
physicochemical stage (ca. 10
12
sec). As mentioned above, the spatial distribution is
assumed to be Gaussian in the deterministic approach. For G
o
(e
aq
) = 4.78 molecules
(100 eV)
1
, which is the value derived by Schwarz [8], the best t to the experimental data
was obtained with the width parameter r
o
of the Gaussian distributions at 25jC ca. equal
to 2.3 nm [9] or 2.23 nm [19] for e
aq
and the products of processes (Ia)(Ic), and 0.85 nm for
the other species [9,19]. It was later shown [24] that for G
o
(e
aq
) = 4.1 molecules (100 eV)
1
(see Section 3.2.1), a good t could be obtained by increasing r
o
for e
aq
to 3.8 nm. In contrast
to this, a recent version of the stochastic model [10] generates a very asymmetrical
distribution of hydrated electrons with most probable and average thermalization dis-
tances (equivalent to r
o
) of
f
2.5 and 9.2 nm, respectively at 25jC, and G
o
(e
aq
) = 5.3
Table 1 Reaction Scheme of the Chemical Stage in Water Radiolysis at 298 K for Low LET
Radiation
Symbol Reaction k/10
10
dm
3
mol
1
sec
1 a
DG/molecule (100 eV)
1 b
R1 e
aq
+H
+
!H
.
2.3 0.55
R2 e
aq
+
.
OH!OH
3.0 1.15
R3 H
+
+OH
!H
2
O 14.0
c
0.65
R4 e
aq
+e
aq
!H
2
+2OH
2k=1.1 0.27
R5 e
aq
+H
.
!H
2
+OH
2.5 0.10
R6 H
.
+H
.
!H
2
2k=1.5 0.007
R7 H
.
+
.
OH!H
2
O 1.5
d
0.16
R8
.
OH+
.
OH!H
2
O
2
2k=1.1 1.53
R9 e
aq
+H
2
O
2
!
.
OH+OH
1.1 0.07
a
Source: Ref. 59.
b
DG is the change in yield of e
aq
, H
.
, or
.
OH for each reaction between 10
12
and 10
6
sec predicted by the
deterministic model [9].
c
Source: Laidler, K.J. Chemical Kinetics; McGraw-Hill, New York, 1965.
d
Source: Ref. 63.
Copyright 2004 by Taylor & Francis Group, LLC
molecules (100 eV)
1
at 10
12
sec. The model predicts that G(e
aq
) has decreased by 11.6%to
4.7 molecules (100 eV)
1
at 10
10
sec, which is about the shortest time at which experimental
measurements have been made. In an earlier version [20], it was recognized that the initial
distribution would be asymmetrical but would rapidly become Gaussian through Brownian
motion. The initial distribution was therefore assumed to be Gaussian for convenience.
The best t with experiment was obtained with r
o
= 4 nm and G
o
(e
aq
) = 4.93 molecules
(100 eV)
1
. This version [20] also predicts G(e
aq
) =4.9 and 4.7 molecules (100 eV)
1
at 10
12
and 10
10
sec, respectively, a fall of only 2.1%.
In their deterministic modelling, Swiatla-Wojcik and Buxton [9] scaled the change in
the initial spur radii r
o
with temperature t using Eq. (8):
r
o
t r
o
25
j
Cq25
j
C=qt
1=3
8
where q is the density of water. The use of Eq. (8) results in r
o
increasing by no more than
11.5%up to 300jC. In the stochastic model [10], the temperature dependence of the electron
thermalization distance r
th
was deduced from considerations of g(H
2
), which was found to
be the yield most sensitive to r
th
. The best t with the experiment was obtained using Eq. (9):
r
th
300
j
C r
th
25
j
C=2 r
th
t decreasing linearly with t 9
Herve du Penboat et al. [10c] suggested that g-values at 300jC calculated using the
deterministic approach [9] would not be sensitive to a 50% decrease in r
th
, whereas when
this condition is imposed, it is found [26] that g(e
aq
) decreases from 3.410
7
to 2.810
7
mol J
1
, which is a signicant change. On the other hand, the g-values of the other products
are changed only slightly, the change being within the error of the measured values.
As would be expected, these results indicate that the thermalization distances and
spatial distribution of the hydrated electron are key parameters in modelling the radiation
chemistry of water. Although the stochastic approach is the more logical one to adopt, its
present status does not appear to outweigh the advantages of using the simpler deterministic
model to represent the essential features of water radiolysis over a wide range of conditions.
Before leaving the topic of modelling, it is pertinent to note that, at roomtemperature,
the deterministic model [9] and the stochastic model [10b] both predict that ca. 50%of g(H
2
)
comes fromthe sumof reactions (R4) and (R5), and negligible amounts fromreaction (R6).
The values for the deterministic model are given in Table 1. The corresponding values for the
stochastic model are 0.24 and 0.15 molecules (100 eV)
1
, respectively [10,b,], and 0.28 and
0.14 molecules (100 eV)
1
, respectively [20]. On the other hand, g(H
2
O
2
) is predicted to be
almost entirely accounted for by reactions (R8) and (R9) in both types of model.
3. YIELDS FOR LOW LET RADIATION
Water radiolysis has proven to be a very useful method of generating free radicals whose
reactions can be studied in a wide range of aqueous solutions. Indeed, a wealth of kinetic [27]
and mechanistic data has been obtained in this way, almost invariably using low LET
radiation, such as
60
Co g-rays for continuous radiolysis and high-energy electron beams
for pulse radiolysis. An important quantity in this kind of work is the yield of the radical
or molecule under investigation because it is dependent on the experimental conditions. The
g-values in Eqs. (6) and (7) have been derived from numerous scavenging studies where
the scavenging power, which is dened as k
(X+S)
[S], is typically ca. 10
7
sec
1
, is generally
appropriate for studies of the radiolysis of dilute aqueous solutions on the order of
Copyright 2004 by Taylor & Francis Group, LLC
10
3
mol dm
3
. In more concentrated solutions, one needs to consider scavenging by the
solute of e
aq
, H
.
, and
.
OH within the spurs, whereas in very dilute solutions, possible
depletion of the solute has to be taken into account. As will become clear in the following
section, scavenging studies have played an important role in our current understanding of
the chemical stage of water radiolysis.
Ageneralized reaction scheme for intraspur scavenging is represented by the following
reactions, using the principal radicals e
aq
and
.
OH in neutral water as examples:
e
aq
S
1
! P
1
10
.
OH S
2
! P
2
11
e
aq
P
2
! S
2
12
.
OH P
1
! S
1
OH
13
where S and P represent scavenger and product, respectively. As noted above, the
eectiveness of a given scavenging solute S is measured by its scavenging power, which,
for reactions (10) and (11), is k
10
[S
1
] and k
11
[S
2
] sec
1
, respectively. In principle, therefore,
data for dierent S can be normalized in terms of their scavenging power; in practice,
however, this procedure is not always valid. One reason for this is that when a single
scavenger is present, the product of the scavenging reaction (e.g., reaction (10) for e
aq
) may
react to some extent with the sibling radical
.
OHin reaction (13). In such a case, G(P
1
) will be
smaller than the scavenged yield of e
aq
. A second reason is that some scavengers may react
with a precursor of e
aq
, but at relative rates dierent from those for their reactions with e
aq
itself. Examples of such behavior are described in the following sections.
Although the time dependence of G(e
aq
) and G(
.
OH) can be obtained by direct
observation using pulse radiolysis, as well as by adding scavengers, the evolution of the
yields of H
.
, H
2
, and H
2
O
2
can only be derived from scavenger studies.
3.1. The Molecular Yields G(H
2
) and G(H
2
O
2
)
G(H
2
)
In dilute aqueous solution, g(H
2
) = 0.45 molecule (100 eV)
1
is well established. In more
concentrated solutions of certain solutes, G(H
2
) decreases as the concentration of solute
increases, showing that precursors of H
2
are being scavenged. The question to be answered
is: Which are the relevant precursors? According to the scheme presented in Section 2, the
possibilities are e
sec
, H
2
O*, H
, e
aq
, and H
.
. Several studies have been made in attempts to
answer this question; the main ideas are summarized below.
Faraggi and Desalos [15] showed that G(H
2
) is independent of [S
1
] for S
1
=Ni
2+
,
Co
2+
, and Zn
2+
although these cations all react eciently with e
aq
. They concluded,
therefore, that H
, rather than e
aq
or H
.
, is the main precursor of H
2
. On the other hand,
Peled and Czapski [28] investigated the eects on G(H
2
) of high concentrations of reactive
anions and neutral molecules as well as cations, and concluded that e
aq
is the main precursor
through reactions (R4) and (R5) in Table 1. The apparent conict with the results of Faraggi
and Desalos [15] for S
1
= Ni
2+
, Co
2+
, and Zn
2+
was eectively resolved by adding 10
3
mol dm
3
NO
3
to the solutions and allowing for the eect of ionic strength on k
10
. The role
of NO
3
is to oxidize the M
+
formed in reaction (10) back to M
2+
to prevent the possibility
of H
2
being produced in further reactions of M
+
. Peled and Czapski [28] also showed that
the results for anions, neutral molecules, and cations as scavengers for e
aq
could be brought
Copyright 2004 by Taylor & Francis Group, LLC
into near coincidence on a single curve of G(H
2
) vs. log(k
10
[S
1
]) when ionic strength eects
were properly taken into account.
The idea that e
aq
is the main precursor of H
2
was supported by results obtained by
Draganic and Draganic [29]. These are exemplied in Fig. 1, which shows that measure-
ments made of the ratios G(H
2
)/g(H
2
) and G(H
2
O
2
)/g(H
2
O
2
) as a function of scavenging
power for a number of dierent scavengers fall very close to a single curve. Draganic and
Draganic [29] also found that when S
1
and S
2
are both present, raising [S
1
] at xed [S
2
] causes
G(H
2
O
2
) to increase; the corresponding eect is also observed for G(H
2
). These results
seemed to provide clear evidence that scavenging in the spurs of one of the primary radicals
promotes the self-reaction of the other one.
Recent studies by Pastina et al. [16] of solutes having a wider range of reactivity toward
the hydrated electron demonstrate that the single scavenging curve shown in Fig. 1 is not
always obtained. Some of the results reported [16], which illustrate this point, are shown in
Fig. 2. Here, the left-hand set of points is plotted against the scavenging power of S
1
for e
aq
,
whereas the right-hand set is plotted against the scavenging power of S
1
for the precursor(s)
of e
aq
, designated here as e
pre
. There is much less scatter in the right-hand set of data, which
led Pastina et al. [16] to conclude that the yield of molecular hydrogen is better para-
meterized by the scavenging capacity for the precursors to the hydrated electron than by the
scavenging capacity for the hydrated electron. In determining the rate coecients for the
reactions of the precursor of e
aq
, Pastina et al. [16] used a procedure outlined previously by
Pimblott and LaVerne [30]. However, for the same scavengers, Pastina et al. [16] quote
values for the rate coecients that are quite dierent from those reported by Pimblott
and LaVerne [30]. The dierence is greatest for H
2
O
2
, with values of zero [30] and 6.33 10
13
dm
3
mol
1
sec
1
[16], but there are signicant dierences between the values for other
solutes. Pastina et al. [16] concluded that process (III) could make a signicant contribution
to g(H
2
) in the radiolysis of water (reaction (7)). Similarly, a good agreement among SeO
2
4
,
Cr
2
O
2
7
, and H
2
O
2
has been obtained with high LET radiation [17] where g(H
2
), as is well
Figure 1 Decrease of G(H
2
)/g(H
2
) or G(H
2
O
2
)/g(H
2
O
2
) with increasing reactivity toward e
aq
or
.
OH. For H
2
: S
1
=H
2
O
2
+10
3
mol dm
3
Br
(o); NO
3
+510
4
mol dm
3
I
(5); Cu
2+
+10
3
mol
dm
3
Br
(D).For H
2
O
2
: S
2
jC
2
H
5
OH+2.510
3
mol dm
3
NO
3
(
.
). (From Ref. 29.)
Copyright 2004 by Taylor & Francis Group, LLC
known, is larger. However, process (III) is kinetically rst-order, whereas the increase in
g(H
2
) with increasing LET is expected to involve second-order processes because the local
concentration of radiolysis products also increases with LET. On the basis that the lifetime
of H
2
O
+
is on the order of 10
13
sec, rather than 10
14
sec (see Section 2), LaVerne and
Pimblott [17] concluded that process (I), followed by process (Ib), is even more signicant
than process (III) as a source of the yield of molecular hydrogen. These results imply that the
precursor of e
aq
is scavenged while it still has excess energy [17].
The sources of g(H
2
) according to deterministic [9] and stochastic [10] modelling are
processes (I), (III), and (V) and also the spur reactions (R4) (R5) (R6). As shown in Table 1,
the deterministic model indicates that approximately 50% of g(H
2
) is produced in reactions
(R4) and (R5), and similar results are predicted by the stochastic model [10b,20]. The models
also predict that the extent of reactions (R4) and (R5) increases with LET. For an increase in
LET from 0.2 to 60 eV nm
1
, g(H
2
) increases to 0.27 and 0.42 molecules (100 eV)
1
due to
reactions (R4) and (R5), respectively [21]. These data, together with numerous experimental
results, are not consistent with the conclusion [17] that H
2
is generated mainly through
reaction of e
se
with H
2
O
+
. There will, of course, be no e
aq
produced under conditions where
e
se
is scavenged eciently in suciently concentrated solution. However, it does not follow
that reactions (R4) and (R5) do not contribute to g(H
2
) in dilute solution or pure water.
Reactions of e
pre
. It is convenient to consider these reactions here. That the electron
can react with solutes before it becomes solvated is clearly demonstrated by the data in
Fig. 3, which were obtained by picosecond pulse radiolysis [31]. The data show that there is
no good correlation between the C
37
value of a solute and its rate constant k for reaction
with e
aq
measured at the same high concentrations of solute used to determine C
37
. The
parameter C
37
is the concentration of solute that lowers the earliest measurable value of
G(e
aq
) to 37% of the value obtained in the absence of solute. A linear correlation between
C
37
and k would be expected if the solute merely reacted with e
aq
before this earliest time of
Figure 2 Decrease of G(H
2
) with increasing reactivity toward e
aq
(left-hand data set) and its
precursor, e
pre
of e
aq
exist [33]. Recent studies [33] show that, using a static scavenging model, the
eciencies with which these precursors are scavenged are very similar for NO
3
, Cd
2+
, and
SeO
2
4
, whether the energized state is reached by multiphoton ionization of water,
photoionization of FeCN
4
6
, optical excitation of e
aq
, or ionization of H
2
O by radiolysis.
These results are considered to be consistent with there being no energy barrier to the
transfer of the electron due to the high exoergicity of these reactions and the availability
of open electronic and vibrational channels involving excited states of the scavenger [33].
Kee et al. [33] showed that the rate constant for reaction of the precursors of e
aq
with NO
3
,
Cd
2+
, and SeO
2
4
is inversely proportional to the volume of e
pre
. On this basis, they obtained
a best-t value of 9.0210
13
sec
1
for the static scavenging rate constant for e
aq
(i.e., the rate
of the scavenging reaction when e
aq
is within the reaction distance of a scavenger molecule).
An anomaly in the radiolysis results was noted by Kee et al. [33]. This disappears when rate
coecients for e
pre
from Ref. 30, which were used by Kee et al. [33], are replaced by those
from Ref. 16.
G(H
2
O
2
)
There seems to be little doubt from modelling [9,10,19,20] and experimental results (see
Fig. 1) that the major precursor of the molecular yield of H
2
O
2
is
.
OHthrough reaction (R8)
in Table 1. That reaction (R8) is a more important spur reaction than reactions (R4) and
Figure 3 Dependence of C
37
on k(e
aq
+S
1
) obtained at [S
1
]=C
37
mol dm
3
. Ref. 32: open points;
Ref. 31: solid points.
Copyright 2004 by Taylor & Francis Group, LLC
(R5) is consistent with
.
OHhaving a narrower spatial distribution than e
aq
. The unied curve
for G(H
2
)/g(H
2
) and G(H
2
O
2
)/g(H
2
O
2
) in Fig. 1 may therefore be a somewhat fortuitous
result of the spatial distributions of the radicals and the rate constants of their reactions with
one another. If, as proposed by LaVerne and Pimblott [17], process I(b) is an important
contributor to g(H
2
), then there should also be a signicant initial yield of H
2
O
2
resulting
fromthe reaction of O
.
(
1
D) with H
2
O, but there is no experimental evidence for such a yield.
3.2. G(e
aq
)
Numerous measurements have been made of g(e
aq
) for lowLET radiation and most of these
have been carried out under steady-state conditions using dilute (ca. 10
3
mol dm
3
) solu-
tions of a suitable scavenger S
1
to form an identiable stable product P
1
, as in reaction (10).
As the scavenging power k
10
[S
1
] increases, G(P
1
) also increases because reaction (10) com-
petes with other reactions of e
aq
in the spur (see Table 1). Fromtheir extensive results on spur
scavenging, Draganic and Draganic [34] obtained unifying curves that describe the depend-
ence of the fractional changes in the radical and molecular yields G(X)/g(X), on scavenging
power. These curves are reproduced in Figs. 4 and 5. Draganic and Draganic [34] concluded
on the basis of these results that secondary spur reactions such as reactions (12) and (13) are
not signicant for the scavengers they studied. Moreover, their data showed [34] that
ecient scavenging of e
aq
not only increases G(P
1
) in reaction (10) but also increases G(
.
OH)
by reducing the extent of water reformation (see Table 1) and releasing more
.
OH for other
reactions such as reaction (11) when the solution contains S
2
as well as S
1
. This condition
is known as cooperative scavenging. Similarly, the scavenging of
.
OH in reaction (11) leads
not only to an increase in G(P
2
) but also in G(e
aq
). The results obtained by Draganic and
Draganic [34] are presented in Figs. 4 and 5.
Other data for a number of dierent scavengers showing the eect on G(P
1
) of
increasing the concentration of the scavenger S
1
are reproduced in Fig. 6 for nitrous oxide
Figure 4 Dependence of the fractional yields G(X)/g(X) of the primary products of water
radiolysis (see reaction (7)) on the scavenging power of S
1
. (From Ref. 34.)
Copyright 2004 by Taylor & Francis Group, LLC
Figure 6 Comparison of experimental and predicted values of G(P
1
) from electron scavenging as a
function of the scavenger power k
10
[S
1
]. S
1
= N
2
O (n); CH
3
Cl (E); C(NO
2
)
4
(x); glycylglycine (.);
NO
3
(o). The broken line is predicted from direct observation of the time dependence of G(e
aq
) in
pulse radiolysis experiments. (From Ref. 49.) The solid line is the t obtained by Pimblott and
LaVerne [43] with the restriction that G
o
(e
aq
) = 4.80 molecules (100 eV)
1
.
Figure 5 Dependence of the fractional yields G(X)/g(X) of the primary products of water
radiolysis (see reaction (7)) on the scavenging power of S
2
. (From Ref. 34.)
Copyright 2004 by Taylor & Francis Group, LLC
[3537], methyl chloride [38,39], nitrate ion [29,40], glycylglycine [41], tetranitromethane
[37], and methyl viologen [40]. Contrary to the unifying curve obtained [34] for e
aq
in Fig. 4,
the data in Fig. 6 show considerable scatter, which is probably at least partially due to
scavenging of electrons in their presolvated state as discussed in Section 3.1. Note that G(P
1
)
continues to fall as k
10
[S
1
] is decreased below 10
6
sec
1
.
The rst attempt at providing a mathematical formulation of the eect of scavengers
on G-values in aqueous solution was made by Balkas et al. [38] based on an equation
developed by Warman et al. [42] for scavenging of ion pairs in nonpolar liquids. Balkas et al.
[38] showed that the yield of chloride ion from reaction (14) can be represented by Eq. (15):
e
aq
CH
3
Cl !
.
CH
3
Cl
14
GCl
2:55
2:23kS=k
1=2
1 kSk
1=2
15
In Eq. (15), g(e
aq
) =2.55 molecules (100 eV)
1
, G
o
(e
aq
) =4.78 molecules (100 eV)
1
, and k is
a constant estimated [38] to be 810
8
sec
1
. The term (k[S]/k)
1/2
/(1+(k[S]/k)
1/2
) is referred
to as the scavenger function F(S). The value of G
o
(e
aq
) was chosen to agree with the value of
Schwarz [8]. The value of g(e
aq
) =2.55 molecules (100 eV)
1
is a t parameter that refers to
very low values of scavenger power k[S] (i.e., F(S)!0). This g-value is a little lower than the
generally accepted value in Eq. (7), which is appropriate to dilute solutions where, typically,
k[S] = 10
7
sec
1
and F(S) c0.1.
The form of Eq. (15) assumes that G(e
aq
) = 2.55 molecules (100 eV)
1
is the yield of
e
aq
that escapes from the track into the bulk solution at innite time. This is a reasonable
assumption under most experimental conditions where k
10
[S
1
]z10
6
sec
1
. On the other
hand, the data in Fig. 6 for lower scavenger power do suggest that the true yield that would
escape from a low LET track in pure water is less than 2.55 molecules (100 eV)
1
.
La Verne and Pimblott [19] and Pimblott and La Verne [43] rened Eq. (15) and
developed an analytical description of these eects of scavengers using the deterministic
diusion kinetic model outlined in Section 2. For a single scavenger, they showed that the
dependence of the amount of scavenging reaction on the concentration of S could be better
described by [22]:
GX
s
gX G
o
X gXFS 16
where X is the species of interest (e.g., e
aq
, H
2
, etc.) and F(S) is given by:
FS aS
1=2
aS=2=1 aS
1=2
aS=2 17
Here a =ak
s
where a, which is equal to 1/k in Eq. (15), is equivalent to a phenomenological
decay or formation time of the species X in the absence of scavenger and k
s
is the rate
constant of the scavenging reaction. Thus, the eects of dierent scavengers for a given
species X can be normalized using the relative values of k
s
. Applying this model to
scavenging data for e
aq
, LaVerne and Pimblott [19] and Pimblott and LaVerne [43] obtained
a good t to experimental data for the scavengers N
2
O, NO
3
, and CH
3
Cl with a = 0.905
nsec, g(e
aq
) =2.55 molecules (100 eV)
1
, and choosing G
o
(e
aq
) =4.80 molecules (100 eV)
1
.
The result is shown as the solid line in Fig. 6. They noted [19] that when no restriction was
placed on the value of G
o
(e
aq
), a best-t value of 4.3 molecules (100 eV)
1
was obtained for
this parameter.
Copyright 2004 by Taylor & Francis Group, LLC
Pimblott and LaVerne [43] then extended their model to cooperative scavenging and
obtained the following expression for cooperative scavenging of e
aq
:
Ge
aq
S
1
; S
2
2:55 2:25F9:1 10
10
s
1
1:2F5:5 10
10
fs
1
s
2
g
F5:5 10
10
s
1
18
where s
1
and s
2
are the scavenging powers of S
1
and S
2
, respectively, and the value of a for
reaction (R2) in Table 1 is 5.5 10
10
sec. The rst two terms on the right-hand side of
Eq. (18) represent the scavenging of e
aq
by the primary scavenger S
1
; the last two terms
represent the eect of the secondary scavenger S
2
for
.
OH through its interference with
the spur reaction e
aq
+
.
OH. The maximum yield of this reaction was estimated [43] to be
1.2 molecules (100 eV)
1
; a very similar value was obtained by Swiatla-Wojcik and Buxton
[9] in their deterministic modelling (see Table 1).
Time Dependence of G(e
aq
)
The time dependence of the yield of hydrated electrons during the chemical stage of water
radiolysis (ca. 10
12
to 10
6
sec) can be obtained from scavenging studies and by direct
observation using pulse radiolysis. In the former case, the scavenger dependence G(S) is
converted into the time dependence using the inverse Laplace transform of F(S) to F(t),
which is given by Eqs. (19) and (20):
Ft expkterfckt
1=2
for Eq: 15 19
Ft 2F
f
4:0kt=p
1=2
for Eq: 17 20
In Eq. (20), F
f
is the auxiliary function for the Fresnel integrals [44]. In practice, it is usual to
choose empirical scavenging functions F(S) that have analytical inverse Laplace transforms.
The rst pulse radiolysis experiments to measure G
o
(e
aq
) directly were made in the
1970s, with reported values of 4.0 F 0.2 molecules (100 eV)
1
at 30 psec [45] and 4.1 F 0.1
molecules (100 eV)
1
at >200 psec [46]. The latter value was subsequently revised to 4.6 F
0.2 molecules (100 eV)
1
at 100 psec [47], and later a yield of 4.8 F0.3 molecules (100 eV)
1
at 30 psec was reported by Sumiyoshi et al. [48]. The method of evaluating G(e
aq
) at these
short times is either to use dosimetry [45,46] and the molar absorption coecient of e
aq
, or to
compare the optical absorbance at short times with that observed at 10
7
10
6
sec and take
G(e
aq
) = 2.7 molecules (100 eV)
1
at this time [48]. The causes of the discrepancies between
these pulse radiolysis values have been reviewed recently by Bartels et al. [49], who have also
made new measurements of the spur decay of e
aq
.
In their pulse radiolysis experiments, Bartels et al. [49] measured directly the decay
of e
aq
from 100 psec out to microseconds. By using the four exponential sum in Eq. (21) to
t the observed decay, they obtained a ratio G
o
(0)/G
inf
of 1.59, eectively equivalent to
G
o
(e
aq
)/g(e
aq
):
G
o
t=G
inf
1 0:090expt=139nsec 0:128expt=24:4nsec
0:255expt=3:51nsec 0:118expt=0:480nsec 21
Fromthe time dependence of the optical absorbance due to e
aq
, Bartels et al. [49] found that
if they assumed G
inf
=2.5 molecules (100 eV)
1
, then they calculated G(e
aq
) =2.7 molecules
(100 eV)
1
for a scavenging power of 10
7
sec
1
(see Fig. 6). Thus, these data are in accord
with G(e
aq
) in reaction (7) and with the rst term on the right-hand side of Eq. (15). On the
other hand, Bartels et al. [49] obtained G
o
(0) (i.e., G
o
(e
aq
) =4.0 F0.2 molecules (100 eV)
1
),
some 20% lower than the value of Schwarz [8]. Their predicted scavenging curve is
Copyright 2004 by Taylor & Francis Group, LLC
compared with that of Pimblott and LaVerne [43] in Fig. 6. Because of the scatter in the
experimental data, the t parameters of the calculated curves are not tightly dened. This is
especially true for G
o
(e
aq
) because the maximum scavenging power that can be used
experimentally is limited to ca. 10
10
sec
1
, corresponding to
f
100 psec in real time. A
reason put forward by Bartels et al. [49] to account for G(P
1
)>G(e
aq
) is that the electron may
react with the scavenger in its presolvation state e
pre
aq
) =4.0 F0.2 molecules (100 eV)
1
[49]
has implications for the stochastic and deterministic models of the radiolysis of water
because, as noted in Section 2, G
o
(e
aq
) at 1 psec is calculated to be 5.3 molecules (100 eV)
1
[10] and 4.9 molecules (100 eV)
1
[20]. On the other hand, for the deterministic model [9],
it has been shown [24] that G
o
(e
aq
) = 4.1 molecules (100 eV)
1
is accommodated by
increasing the initial width of the assumed Gaussian distribution of e
aq
from 2.3 to 3.8 nm.
Another factor to be considered is the extent to which e
aq
decays in the time interval 1100
psec, which at present is largely inaccessible experimentally. The ratio calculated from Eq.
(21) at 100 psec (i.e., G(100 psec)/G
inf
) is 1.56, indicating that e
aq
decays by only 1.8%
between 10
12
and 10
10
sec. This is quite close to the decay of 2.1% predicted by the
stochastic simulations of Pimblott and LaVerne, assuming a Gaussian distribution of
thermalized electrons with r
o
=4 nm [20], but much smaller than the 11.6% decay obtained
by Frongillo et al. [10b] (see Section 2). The deterministic model calculations predict decays
of 8.0% and 3.1% for Gaussian distributions with r
o
= 2.3 nm [9] and 3.8 nm [24],
respectively. In the deterministic model, G
o
(e
aq
) and r
o
are adjustable parameters, whereas
the stochastic treatment automatically generates the value of G
o
(e
aq
) and the distribution of
thermalized electrons. If the results obtained by Bartels et al. [49] are correct, then the
stochastic calculations [10,20] need to be revisited.
To make a complete interpretation of the scavenging data in Fig. 6, one has to consider
the reactivity of the electron in the various states in which it can exist before it reaches its
fully solvated state [33] as discussed in Section 3.1. Static scavenging must also be taken into
account when e
aq
is formed within the reaction distance of the solute so that it reacts before it
can be observed [50]. Hence, it is not surprising that the data in Fig. 6 do not conformto the
unied curve for G(e
aq
) in Fig. 4. Therefore, in chemical applications involving concentrated
solutions, one should measure G(P
1
) in the system of interest. If this is not feasible, then it
should be recognized that the value calculated from scavenging functions based only on the
reactivity of e
aq
could be in error by as much as 10%.
3.3. G(
.
OH)
By comparison with G(e
aq
), relatively few independent measurements of G(
.
OH) have been
made. In contrast to e
aq
, only the relative change in G(
.
OH) with time has been reliably
measured by pulse radiolysis [51]. In practice, absolute values of G(
.
OH) have been obtained
from scavenger studies or by material balance (reaction (7)). Fig. 7 shows data for aerated
solutions of formate ion [52] and hexacyanoferrate(II) [53] taken fromFig. 1 of Ref. 54. The
data for formic acid, which were included by LaVerne and Pimblott [54], have been omitted
here because they were obtained at low pH where the primary yields are dierent (see Sec-
tion 3.4). The solid line shows the best t obtained using Eqs. (16) and (17) and the broken
line is the best t when the term a[S]/2 is omitted from Eq. (17). The respective sets of
parameters are a = 1.64 and 1.69 nsec, g(
.
OH) = 2.53 and 2.50 molecules (100 eV)
1
, and
G
o
(
.
OH) = 4.48 and 4.86 molecules (100 eV)
1
. These values dier signicantly from those
obtained by LaVerne and Pimblott [54], which were a = 0.258 nsec, g(
.
OH) = 2.66
molecules (100 eV)
1
, and G
o
(
.
OH) = 5.50 molecules (100 eV)
1
. The reason for the
dierence is that LaVerne and Pimblott [54] chose G
o
(
.
OH) = 5.50 molecules (100 eV)
1
.
Copyright 2004 by Taylor & Francis Group, LLC
When no constraints were placed on the tting parameters, G
o
(
.
OH) was found to be 4.4
molecules (100 eV)
1
[54], which is closer to those given above for the data in Fig. 7. Sub-
sequently, Pimblott and LaVerne [20] compared the results of their stochastic simulations
with the scavenging data. However, their simulations predicted G
o
(e
aq
) c 4.9 molecules
(100 eV)
1
and G
o
(
.
OH) c5.5 molecules (100 eV)
1
, respectively, which nowseemto be too
large when compared with the results and conclusions of Bartels et al. [49] (i.e., G
o
(e
aq
) c4.0
molecules (100 eV)
1
and G
o
(
.
OH) c5.1 molecules (100 eV)
1
).
The question of the magnitude of G
o
(
.
OH) has been addressed recently by Jay-Gerin
and Ferradini [55] who concluded that a value of 4.6F0.25 molecules (100 eV)
1
at 100 psec
is consistent with the scavenger data in Fig. 7 and with stochastic simulation results, which
give G
o
(
.
OH) equal to 5.8 molecules (100 eV)
1
at 1 psec and G(
.
OH) = 4.8 molecules (100
eV)
1
at 100 psec [10]. This represents a 17.2% decrease in G(
.
OH) over this period. The
eect of using G
o
(e
aq
) =4.1 instead of 4.78 molecules (100 eV)
1
in the deterministic model
[9,24] is to make G
o
(
.
OH) 4.9 molecules (100 eV)
1
instead of 5.5 molecules (100 eV)
1
. The
corresponding decreases in G(
.
OH) between 10
12
and 10
10
sec are 8.3% and 14.1%,
respectively.
In contrast to the data in Fig. 7, Sutton et al. [56], using pulse radiolysis, measured
G(Br
.
2
)e(Br
.
2
), where e is the molar absorption coecient, and found it to be independent
of the concentration of Br
! HOBr
.
22
HOBr
.
WOH
Br
.
23
Br
.
Br
WBr
.
2
24
Figure 7 Comparison of experimental and predicted values of G(P
2
) from hydroxyl radical
scavenging as a function of the scavenger power k
11
[S
2
]. S
2
= HCO
2
(n); FeCN
4
6
(E). The full
and broken lines are the best ts of Eqs. (16) and (17) without any restriction on the parameters (see
text). (From Ref. 54.)
Copyright 2004 by Taylor & Francis Group, LLC
To explain their results, Sutton et al. [56] suggested that spur reactions (R2) and (R8) in
Table 1 be replaced by reactions (25) and (26), which are equally ecient:
e
aq
Br
.
=Br
.
2
! Br
=2Br
25
Br
.
Br
.
! Br
2
26
Similarly, it was found that G(P
2
) is independent of [S
2
] for the
.
OH scavengers thiocyanate
ion [57] and iodide ion [58]. Clearly, there is good evidence that back reactions such as
reaction (25) can occur in the spur and thus modify G(P
2
).
G(
.
OH) in N
2
O-Saturated Solution
In applications of water radiolysis to free radical chemistry, it is particularly convenient to
use aqueous solutions saturated with N
2
O, where e
aq
is converted to
.
OH via reactions (27)
and (28) [59]:
e
aq
N
2
O ! O
.
N
2
; k
27
9:1 10
9
dm
3
mol
1
sec
1
27
O
.
H
2
O W
.
OH OH
; k
28
1:7 10
6
dm
3
mol
1
sec
1
28
k
28
1:2 10
10
dm
3
mol
1
sec
1
Under normal conditions of room temperature and atmospherical pressure, the saturation
concentration of N
2
O is ca. 2.5 10
2
mol dm
3
and its scavenging power is ca. 2.3 10
8
sec
1
. Thus, one has to consider that reactions (27) and (28) generate extra
.
OH in the spurs
because k
28
[H
2
O] =9.4 10
7
sec
1
. This problemwas addressed by Schuler et al. [60], who
showed that the scavengeable yield of
.
OH is described by the empirical equation, Eq. (29),
which is analogous to Eq. (15):
G
.
OH 5:2
3:0kS=k
1=2
1 kSk
1=2
molecule 100eV
1
29
where k = 4.7 10
8
sec
1
and k[S] is the scavenging power of the solute reacting with
.
OH
(i.e., k
11
[S
2
]). Thus, the
.
OHthat replaces e
aq
in the spurs through reactions (27) and (28) also
participates in spur reactions. Values of G(
.
OH) calculated from Eq. (29) are plotted vs.
scavenging power in Fig. 8.
3.4. pH Dependence of the Primary Yields g(X)
This is illustrated in Fig. 9 for low LET radiation. At pH >3, the g-values are eectively
constant; at pH<3, the extent of reaction (R1) increases through scavenging of e
aq
in the
spurs. The observed increase in g(e
aq
) + g(H
.
) is probably due to the diusion coecient
of H
.
(D
H
= (7.0 F 1.5) 10
9
m
2
sec
1
[61]) being larger than that of e
aq
(D
e
= (4.90 F
0.02) 10
9
m
2
sec
1
[62]), and reaction (R7) (k
R7
=1.5 10
10
dm
3
mol
1
sec
1
[63]) being
slower than reaction (R2) (k
R2
=3.010
10
dm
3
mol
1
sec
1
[59]), so that conversion of e
aq
to
H
.
in the spurs favors diusive escape into the bulk liquid. The change in the individual
values of g(e
aq
) and g(H
.
) with pH has been measured [6] with chloroacetic as the
scavenger, as described in Section 1. The data are shown in Fig. 10:
e
aq
ClCH
2
CO
2
H ! Cl
.
CH
2
CO
2
H 4
H
.
ClCH
2
CO
2
H ! H
2
.
CHClCO
2
H 5
The values of g(
.
OH) and g(H
2
O
2
) also increase at pH<3, as shown in Fig. 9. An
increase in the yield of oxidizing equivalents is expected, of course, to maintain the balance
Copyright 2004 by Taylor & Francis Group, LLC
with the increase in g(e
aq
)+g(H
.
), but the increase in g(H
2
O
2
) does suggest that reaction (R8)
becomes more important when reaction (R7) replaces reaction (R2) at low pH.
The conversion of H
.
to e
aq
! e
aq
; k
30
2:2 10
7
dm
3
mol
1
sec
1
59 30
Although reaction (30) is too slowto interfere signicantly with spur processes up to pH14,
it may compete with the reaction of H
.
with other solutes in the bulk solution under these
conditions and thus increase G(e
aq
).
Figure 8 Values calculated from Eq. (29) of G(
.
OH) vs. scavenging power k
11
[S
2
] in N
2
O-saturated
solution.
Figure 9 Dependence of the primary yields g(X) on pH. (From Ref. 52.)
Copyright 2004 by Taylor & Francis Group, LLC
4. PROPERTIES OF THE PRIMARY RADICALS
The properties of the primary radicals produced by the radiolysis of water are collected in
Table 2.
4.1. The Hydrated Electron
Although its precise structure has not yet been settled, the hydrated electron may be
visualized as an excess electron surrounded by a small number of oriented water molecules
and behaving in some ways like a singly charged anion of about the same size as the iodide
ion. Its intense absorption band in the visible region of the spectrummakes it a simple matter
to measure its reaction rate constants using pulse radiolysis combined with kinetic
spectrophotometry. Rate constants for several hundred dierent reactions have been
obtained in this way, making e
aq
kinetically one of the most studied chemical entities.
As expected from its standard reduction potential of 2.9 V (Table 2), e
aq
reacts
rapidly with many species having more positive reduction potentials. Its mode of reaction
can be represented as the one-electron transfer process (reaction (31):
e
aq
S
n
! S
n1
31
where n is the charge on the solute. In some cases, the electron adduct S
n1
immediately
dissociates (e.g., as in reactions (4), (14), and (27) shown above). Values of k
31
range fromca.
10
1
dm
3
mol
1
sec
1
up to the diusion-controlled limit of ca. 10
10
dm
3
mol
1
sec
1
.
However, the activation energies are invariably small and generally fall in the range 630 kJ
mol
1
, with the majority around 15 kJ mol
1
. The latter observation led Hart and Anbar
[67] to suggest that reaction (31) has an activation energy associated with reorientation of
the solvent shell to facilitate transfer of the electron, and that this reorientation energy is the
same as that required for e
aq
to diuse in water. The corollary to this argument is that the
Figure 10 Dependence of g(e
aq
) and g(H
.
) on pH. (From Ref. 6.)
Copyright 2004 by Taylor & Francis Group, LLC
entropy of activation energy is the dominant kinetic parameter, and this can be understood
in terms of the availability of a suitable vacant orbital on the acceptor molecule S
n
into which
the electron can transfer from its solvent shell. Molecules such as water, simple alcohols,
ethers, and amines have no low-lying vacant orbitals, and this explains why solvated
electrons can be observed in these liquids.
The kinetics of the reactions of e
aq
can be rationalized as follows. The general
expression for a time-independent rate constant in solution is [68]:
1=k
obs
1=k
diff
1=k
read
32
where k
obs
, k
di
, and k
react
are the observed, diusion-controlled, and activation-controlled
rate constants, respectively. When the rate of a reaction is diusion-controlled (k
react
k
di
),
it is impossible to obtain details of the chemical step in the overall process of encounter and
reaction, but such information is revealed by raising the temperature so that the diusion
step becomes much faster than the chemical step. In this way, it has been shown [69] that
some reactions of e
aq
(e.g., with NO
2
, NO
3
, SeO
2
4
, and phenol) are reversible in the chemical
step, which accounts for the abnormally low activation energy that was reported for NO
2
[70]. In fact, the activation energy becomes negative as the temperature is raised above
100jC, as shown in Fig. 11.
These reactions can be described by the following general scheme for transfer of an
electron from a donor D to an acceptor A [71]:
D AWD=AWD=A* WD
=A
* ! D
=A
! product 33
where [D/A]* and [D
+
/A
, H
.
,
.
OH, and O
.
in aqueous solution
Property e
aq
H
. .
OH O
.
Absorption maximum (nm) 720 <200
f
225 240
Molar absorption
coecient (m
2
mol
1
)
200 (720 nm)
[65]
95 (200 nm)
[66]
50 (250 nm)
[63]
24 (240 nm)
[64]
Diusion coecient
(10
9
m
2
sec
1
)
4.9 [62] 7 [61] 2.2
Equivalent conductivity
(S cm
2
mol
1
)
190
Mobility (10
3
cm
2
V
1
sec
1
)
1.9
Reduction potential (V) 2.9
a
2.3
b
2.7
c
, 1.9
d
1.7
e
pK
a
9.6
f
11.9
g
The data are taken from Table 1 in Ref. 59 unless stated otherwise.
a
(H
2
O+e
! e
aq
).
b
(H
2
O+e
+H
+
!
.
H
aq
).
c
(
.
OH+e
+H
+
! H
2
O).
d
(
.
OH+e
!OH
).
e
(O
.
+e
+H
+
! OH
).
f
(H
.
+H
2
O!H
3
O
+
+e
aq
).
g
(
.
OH+H
2
O!H
3
O
+
+O
.
).
Copyright 2004 by Taylor & Francis Group, LLC
case of NO
2
, the calculated line in Fig. 11 was obtained with values of E
act
=24.9 and 7.2 kJ
mol
1
, respectively, for these two steps [69].
The hydrated electron behaves as a nucleophile in its reactions with organic molecules
and its reactivity is greatly enhanced by electron-withdrawing substituents attached to
aromatic rings or adjacent to double bonds. Some of the features of the reactivity of e
aq
are
illustrated by the data in Table 3.
4.2. The Hydrogen Atom
The hydrogen atom is the conjugate acid of e
aq
and can be thought of as a weak acid with a
formal pK
a
of 9.7, obtained from the rate constants of reactions (30) and (34) [59]:
e
aq
H
2
O ! H
.
OH
; k
34
19dm
3
mol
1
sec
1
34
It is not an important species in the radiation chemistry of neutral and alkaline solution
(reaction (7)), but it is the major reducing species in acidic solution (see Fig. 10). Having
a reduction potential of 2.3 V (Table 2), H
.
is a slightly less powerful reducing agent than
e
aq
, but its chemistry is often quite dierent. Agood example of this is provided by reactions
(4) and (5). H
.
readily reduces inorganic ions having more positive reduction potentials than
itself, but often at slower rates than e
aq
. In some cases, in strongly acidic solution, it
eectively acts as an oxidant, forming a hydride intermediate, which reacts with H
+
to
eliminate H
2
and leave the oxidized form of the solute as in reactions (35) and (36):
H
.
Fe
2
! Fe
3
H
!
H
Fe
3
H
2
35
H
.
I
! HI
!
I
HI
2
2
!
H
I
2
.
H
2
36
Figure 11 Arrhenius plot of k(e
aq
NO
2
). The solid circles show the values of k
react
obtained from
Eq. (32). The solid line shows the t of reaction (33). Note the inuence of k
di
on k
obs
. (From Ref.
69.)
Copyright 2004 by Taylor & Francis Group, LLC
H
.
generally reacts with organic compounds by abstracting a hydrogen atom from
saturated molecules and by adding to centers of unsaturation, for example:
H
.
CH
3
OH !
.
CH
2
OH H
2
37
H
.
CH
2
jCH
2
!
.
CH
2
CH
3
38
Because H
.
absorbs only weakly in the ultraviolet region of the spectrum(Table 2) and
g(
.
H) is relatively small in neutral solution, its optical signal cannot generally be used for
making direct measurements of its rate of reaction with solutes, but in suitable cases, the
required data can be obtained by observing the rate of growth of an absorbing product. Rate
constants have also been determined directly by observation in pulse radiolysis experiments
of the ESRsignal of the hydrogen atom, which is possible because of its enhancement by spin
polarization eects [72,73].
More generally, the method of competition kinetics is used to determine H-atom rate
constants. The hexacyanoferrate(III) ion is a suitable solute because reaction (39) can be
followed fromthe decrease in absorbance at 420 nmdue to FeCN
3
6
(e
420
=104 m
2
mol
1
).
When a second solute is present so that reaction (40) competes with reaction (39), G(Fe
CN
3
6
) is given by Eq. (41):
H
.
FeCN
3
6
! FeCN
4
6
H
39
H
.
S ! P 40
1
GFeCN
3
6
1
GH
.
1
k
40
S
k
39
FeCN
3
6
!
41
Table 3 Rate Constants for Selected Reactions of e
aq
at Room Temperature
Solute (inorganic) k (dm
3
mol
1
sec
1
) Solute (organic) k (dm
3
mol
1
sec
1
)
Ag
+
3.910
10
C
6
H
6
1.110
7
H
3
O
+
2.310
10
C
6
H
5
Cl 6.410
8
NH
4
1.510
6
C
6
H
5
I 1.210
10
Cd
2+
5.310
10
C
6
H
5
NO
2
3.710
10
Co(NH
3
3
8.710
10
C
6
H
5
NH
2
2.810
7
Fe
3
6.010
10
C
5
H
5
N (pyridine) 1.010
9
ClO
4
<110
6
MV
2+ a
7.210
10
MnO
4
2.610
10
CH
4
<110
7
CrO
2
4
1.810
10
CH
3
CN 3.710
7
FeCN
3
6
3.110
9
CH
3
I 1.610
10
FeCN
4
6
<710
4
CH
3
OH <110
4
CO
2
7.710
9
CH
2
I
2
3.410
10
H
2
O 19 C(NO
2
)
4
5.310
10
H
2
O
2
1.110
10
C
2
H
2
2.010
7
N
2
O 9.110
9
CH
2
jCH
2
<310
5
O
2
1.910
10
CH
2
jCCl
2
2.310
10
a
MV
2+
=1,1V-dimethyl-4,4V-bipyridinium ion.
Source: Ref. 59.
Copyright 2004 by Taylor & Francis Group, LLC
4.3. The Hydroxyl Radical
The hydroxyl radical is a powerful oxidant, having a reduction potential of 2.7 V in acidic
solution. In neutral solution, where the free energy of neutralization of OH
by H
3
O
+
is not
available, the reduction potential decreases to 1.9 V (Table 2). Several inorganic anions and
low-valency transition metal ions readily undergo one-electron oxidation by reaction with
.
OH, which is often represented as a simple electron transfer:
.
OH S
n
! S
n1
OH
42
where n is the ionic charge. However, although the stoichiometry is described by reaction
(42), there are many examples which show that the reaction actually proceeds via the
formation of an adduct, followed by an inner-sphere electron transfer. In the case of halide
and pseudo-halide ions, X
43
for M
n+
= Tl
+
, Ag
+
, Cu
2+
, Sn
2+
, Fe
2+
, Mn
2+
, and Cr
3+
. It has been proposed [74] that
.
OH is generally unlikely to react by accepting an electron fromthe donor by a simple outer-
sphere transfer because of the large solvent reorganization energy involved in the formation
of the hydroxide ion. This idea is certainly supported by the experimental results relating to
reactions (42) and (43).
Although
.
OH reacts at near-diusion-controlled rates with inorganic anions [59],
there seems to be an upper limit of ca. 3 10
8
dm
3
mol
1
sec
1
in the case of simple hydrated
metal ions, irrespective of the reduction potential of M
n+
. Also, there is no correlation
between the measured values of k
43
and the rates of exchange of water molecules in the rst
hydration shell of M
n+
, which rules out direct substitution of
.
OH for H
2
O as a general
mechanism. Other mechanisms that have been proposed are (i) abstraction of H from a
coordinated H
2
O [75,76], and (ii)
.
OH entering the rst hydration shell to increase the
coordination number by one, followed by inner-sphere electron transfer [77,78]. Data
reported [78] for M
n+
= Cr
3+
, for which the half-life for water exchange is of the order of
days, are consistent with mechanism (ii):
.
OH Cr
III
WCr
III
.
OH ! Cr
IV
OH
44
In alkaline solution at pH >10,
.
OH reacts rapidly with hydroxide ion to form O
.
,
with k
45
= 1.2 10
10
dm
3
mol
1
sec
1
and k
45
= 9.3 10
7
sec
1
[59] so that equilibrium
(reaction (45)) is established when k
11
[S
2
]<10
7
sec
1
for
.
OH reactions:
.
OH OH
WO
.
H
2
O 45
It is important to note that, as expressed here, K
45
= 130 dm
3
mol
1
is a concentration
equilibrium constant. Thus, when equilibrium is established,
.
OH makes up 8% of the total
concentration of hydroxyl radicals in 0.1 mol dm
3
OH
1
K
45
OH
48
so that by varying [OH
, CO
2
3
, and FeCN
4
6
, although these ions all react rapidly with
.
OH.
This dierence in reactivity was used to establish the pK
a
of
.
OHas 11.9 (Table 2). O
.
reacts
rapidly with O
2
as in reaction (49), whereas
.
OH is unreactive:
O
.
O
2
WO
.
3
49
For this reaction, k
49
=3.5 10
9
dm
3
mol
1
sec
1
and K
49
=1.8 10
6
dm
3
mol
1
at 20jC
[79]. These kinetic properties and the characteristic absorption spectrumof O
.
3
(k
max
=430
nm, e
430
= 190 m
2
mol
1
[64]) make O
2
a suitable reference solute for the measurement of
rate constants of O
.
reactions by the method of competition kinetics [79].
In their reactions with organic compounds,
.
OH is electrophilic and O
.
is nucleo-
philic. Thus,
.
OH behaves like H
.
, readily adding to double bonds and aromatic rings,
whereas O
.
does not. Both forms of the radical abstract an H atom from CUH bonds and
this can result in dierent products in neutral and alkaline solutions. This is exemplied by
toluene, which reacts with
.
OH and O
.
as in reactions (50) and (51), respectively:
.
OH C
6
H
5
CH
3
! HOC
6
H
5
CH
3
.
50
O
.
C
6
H
5
CH
3
!
.
CH
2
C
6
H
5
OH
51
with k
50
= 5.1 10
9
dm
3
mol
1
sec
1
[80] and k
51
= 2.110
9
dm
3
mol
1
sec
1
[81].
Although
.
OHbehaves mechanistically like H
.
in its reactions with organic molecules,
it is less selective and more reactive than H
.
in H-abstraction reactions because the formation
of the HUOH bond is 57 kJ mol
1
more exothermic than that of HUH. Examples of the
reactivities of H
.
,
.
OH, and O
.
are listed in Table 4.
Measurements of the rates of reactions such as reaction (50) over an extended tem-
perature range reveal that k
obs
increases by less than threefold up to 150jC and then de-
Table 4 Rate Constants for Selected Reactions of
.
OH, H
.
, and O
.
at Room Temperature
k (10
7
dm
3
mol
1
sec
1
)
Solute Reaction type
.
OH H
.
O
. a
CH
2
=CHCONH
2
Addition 590 3100 65
C
6
H
5
CH
3
Addition 510 260 210
C
6
H
5
CO
2
Addition 590 85
f
4
CH
3
OH Abstraction 97 0.26 75
CH
3
CH
2
CH
2
OH Abstraction 280 2.4 150
(CH
3
)
2
CHOH Abstraction 190 7.4 120
HCO
2
Abstraction 320 21 140
a
Abstraction reactions.
Source: Ref. 59.
Copyright 2004 by Taylor & Francis Group, LLC
creases slightly up to 200jC as shown in Fig. 12 for C
6
H
6
, C
6
H
5
Cl, C
6
H
5
NO
2
, C
6
H
5
CO
2
H,
and C
6
H
5
CO
2
[82].
The variations of k
obs
with temperature are all very similar, but the absolute values of
k
obs
depend on the substituent on the aromatic ring. To explain these results, Ashton et al.
[82] proposed the following mechanismwhereby
.
OHadds reversibly to forma p-complex as
a precursor of the nal r-bonded hydroxycyclohexadienyl radical:
.
OH C
6
H
5
X WHOC
6
H
5
.
p
52
HOC
6
H
5
p
.
! HOC
6
H
5
.
r
53
Application of the steady-state approximation to HOC
6
H
5
.
r
in this reaction scheme
leads to:
k
obs
k
52
k
53
=k
52
k
53
54
On the assumption that reaction (52) is diusion-controlled, Ashton et al. [82] found that
rearrangement of HOC
6
H
5
.
p
to HOC
6
H
5
.
r
requires virtually no activation energy, whereas
dissociation of HOC
6
H
5
.
r
in reaction (52) has an activation energy of ca. 20 kJ mol
1
. The
dierences in the absolute values of k
obs
are reected largely in the values of k
53
, suggesting
that the electronic eects of the substituents govern the rate of rearrangement. These
parameters are summarized in Table 5.
4.4. The Perhydroxyl Radical HO
2
.
Although HO
.
2
and its conjugate base O
.
2
are only signicant primary radicals for high LET
radiation, they are important secondary radicals in oxygenated solution where they are
formed in reactions (55)(58):
e
aq
O
2
! O
.
2
55
H
.
O
2
! HO
.
2
56
Figure 12 Arrhenius plots for the reaction of
.
OH with some aromatic compounds. (From Ref.
82.)
Copyright 2004 by Taylor & Francis Group, LLC
HO
.
2
W O
.
2
H
57
The pK
a
of HO
.
2
is 4.8 [83], and both O
.
2
and HO
.
2
have reducing and oxidizing properties.
The reduction potentials are 0.33 Vfor the O
2
/O
2
.
couple with respect to a standard state
of 1 atm pressure of O
2
[83] and 0.05 V for the O
2
, H
+
/HO
.
2
couple [84] so that O
.
2
is the
better reductant. Conversely, HO
.
2
is the better oxidant, the reduction potentials being 1.45
V for the (H
+
, HO
.
2
/H
2
O
2
) couple and 1.03 V for the (O
.
2
, H
+
/HO
2
) couple [84].
HO
.
2
and O
.
2
have characteristic absorption spectra with e
max
=140 m
2
mol
1
at 225
nm [83] and e
max
= 189 m
2
mol
1
at 245 nm [85], respectively, which are suciently intense
to permit their reactions to be followed by direct observation in pulse radiolysis experiments.
Both radicals are relatively unreactive with organic molecules [83], abstracting only weakly
bonded hydrogen atoms in, for example, ascorbic acid, cysteine, and hydroquinone. O
.
2
undergoes reversible electron transfer in its reaction with quinones (Q), which was used to
establish its reduction potential [86]:
O
.
2
Q WO
2
Q
.
58
Acareful study of the bimolecular decay of HO
.
2
/O
.
2
in the absence of metal ions has shown
[84] that the reactions involved are reactions (59) and (60):
HO
.
2
HO
.
2
! H
2
O
2
O
2
59
HO
.
2
O
.
2
!
H
H
2
O
2
O
2
60
There is no evidence that reaction (61) takes place at a measurable rate under ambient
conditions, but it is rapid in the presence of aquo-complexes and some organo-complexes of
copper ions via reactions (62) and (63):
O
.
2
O
.
2
2H
! H
2
O
2
O
2
61
O
.
2
Cu
II
! Cu
I
O
2
62
O
.
2
Cu
I
2H
! H
2
O
2
Cu
II
63
Reaction (63) is an example of O
.
2
acting as an oxidant and it probably proceeds via an
inner-sphere electron transfer mechanismin which incompletely coordinated Cu
I
binds O
.
2
prior to electron transfer [87]. HO
.
2
and O
.
2
also react readily with a number of other
transition metal ions, either by electron transfer or through the formation of a complex [83],
for example:
O
2
.
Mn
2
! MnO
2
64
HO
.
2
Fe
2
! Fe
3
.
HO
2
!
H
Fe
3
H
2
O
2
65
Table 5 Rate Constants and Activation Energies for Reactions (52) and (53)
k (10
9
dm
3
mol
1
sec
1
) E
act
(kJ mol
1
)
X Reaction (52) Reaction (53) Reaction (52) Reaction (53)
H 6.5 14 18 0.5
Cl 4.0 12 22 0.5
NO
2
4.0 4 20 1.3
CO
2
3.8 13 21 1.0
CO
2
H 4.0 8 23 1.0
Source: Ref. 82.
Copyright 2004 by Taylor & Francis Group, LLC
Reactions (66) and (67) provide a very convenient way of converting
.
OH to O
.
2
:
.
OH HCO
2
! CO
.
2
H
2
O 66
CO
.
2
O
2
! O
.
2
CO
2
67
Thus, all the primary water radicals can be converted into HO
.
2
/O
.
2
in reactions (55), (56),
(66), and (67).
5. GENERATION OF SECONDARY RADICALS
As we have seen in Sections 1 and 4, the principal primary products of the radiolysis of water
are powerful oxidizing and reducing radicals in approximately equal yields. For water
radiolysis to be a useful tool in general chemistry, it is desirable to convert the primary
radicals to a single kind of secondary radical to achieve either totally oxidizing or reducing
conditions. Moreover, there is the possibility of designing the system to have the required
redox properties by suitable selection of the secondary radicals. Some useful systems that
meet these requirements are described below.
5.1. Oxidizing Conditions
As noted in Section 3.3, reactions (27) and (28) provide a very convenient way of converting
e
aq
to
.
OH. Because reaction (68) is relatively slow, approximately 10% of the radicals in
N
2
O-saturated neutral solution remain as reducing radicals in the formof H
.
, but this small
fraction can be allowed for or even be neglected in chemical applications:
H
.
N
2
O !
.
OH N
2
; k
68
2:1 10
6
dm
3
mol
1
sec
1
68
At pH>11, where reaction (30) becomes increasingly important, G(
.
OH) can increase by up
to 0.6 molecule (100 eV)
1
(i.e., g(H
.
), whereas at pH<3, G(
.
OH) becomes smaller because
reaction (3) competes with reaction (27). One should not forget that the product of reaction
(27) is O
.
rather than
.
OH[88,89]. Thus, if a solute that reacts with O
.
in competition with
its protonation in reaction (28) is present, then the reaction products may not be the same in
N
2
O-saturated solutions containing hydroxyl radical scavengers in high and low concen-
trations because O
.
can react dierently from
.
OH (see Section 4.3).
More selective oxidizing conditions can be achieved by converting
.
OH into another
inorganic radical (e.g., Br
.
2
, I
.
2
, SCN
.
2
, CO
.
3
, N
.
3
, etc.), thereby tuning the reduction
potential (see Table 6). One of the most powerful oxidizing radicals is SO
.
4
, which is
produced in reaction (69):
e
aq
S
2
O
2
8
! SO
.
4
SO
2
4
69
Because of its high reduction potential of 2.43 V (see Table 6), it can be used to generate
other strongly oxidizing radicals such as C1
.
and NO
.
3
through electron transfer reactions
with C1
and NO
3
, respectively [27].
Secondary inorganic radicals are particularly useful for studying redox changes in
metalloproteins and organometallic complexes because these radicals are more likely to
react at the metal center by electron transfer, whereas
.
OH will also attack the organic
moiety and, by abstracting H, create a reducing radical there.
Copyright 2004 by Taylor & Francis Group, LLC
5.2. Reducing Conditions
Although it is possible to convert
.
OHto e
aq
through reactions (70) and (30), the slowness of
these reactions means that conditions of 100 atm of hydrogen and high pH are required.
Nevertheless, these have been realized in experiments crucial to the measurement of spur
reactions (R5) and (R6) in Table 1 [91,92]:
.
OH H
2
! H
.
H
2
O; k
70
4:2 10
7
dm
3
mol
1
sec
1
59 70
A much more convenient method of obtaining totally reducing conditions is to convert all
the primary radicals to the same secondary reducing radical by adding the appropriate
organic solute to N
2
O-saturated water. A prime example of this is the generation of CO
.
2
through reactions (66) and (71):
H
.
HCO
2
! CO
.
2
H
2
; k
71
2:1 10
8
dm
3
mol
1
sec
1
59 71
Alternatively, N
2
O can be replaced by CO
2
so that CO
.
2
is produced in reaction (72):
e
aq
CO
2
! CO
.
2
; k
72
7:7 10
9
dm
3
mol
1
sec
1
59 72
Simple alcohols are often used as the source of the secondary radicals because they also react
rapidly with both
.
OH and H
.
(e.g., 2-propanol in reactions (73) and (74)):
.
OH CH
3
2
CHOH ! CH
3
2
.
COH H
2
O 73
H
.
CH
3
2
CHOH ! CH
3
2
.
COH H
2
74
It is the hydrogen atom on the a-carbon that is most readily abstracted, as indicated in
reactions (73) and (74). However,
.
OH does abstract H from other positions so that a
mixture of radicals is produced where only one is desired. The distributions of radicals
generated from a number of simple alcohols are given in Table 7.
Radicals derived from simple alcohols have a dissociable proton on the hydroxyl
group, for example:
CH
3
2
.
COH WCH
3
2
.
CO
; pK
a
12:03 94 75
Thus, one should bear in mind that the basic form of the radical will be generated when
pH>pK
a
and it is usually more strongly reducing than the acidic form. The values of pK
a
Table 6 Reduction Potentials of Some Inorganic Radicals
Redox couple E
o
(V) Redox couple E
o
(V)
Cl
.
/Cl
2.41 CO
.
3
/CO
2
3
1.5
Cl
.
2
/2 Cl
2.09 N
.
3
/N
3
1.33
Br
.
/Br
1.92 NO
.
2
/NO
2
1.04
Br
.
2
/2 Br
1.62 NO
.
3
/NO
3
2.5
I
.
/I
1.33 SO
.
3
/SO
2
3
0.63
I
.
2
/2 I
1.03 SO
.
4
/SO
2
4
2.43
SCN
.
/SCN
1.63 CN
.
/CN
2.59
SCN
.
2
/2SCN
1.32 OCN
.
/OCN
2.66
Source: Ref. 90.
Copyright 2004 by Taylor & Francis Group, LLC
and reduction potential for the more commonly used organic radicals are summarized in
Table 8.
CO
.
2
is a useful reducing radical because of its low pK
a
, which means that it can be
used in acidic solutions where e
aq
would react predominantly with H
+
. However, where it
is required to use e
aq
as the reductant, it is common practice to add 2-methyl-2-propanol
(tert-butanol) as the scavenger for
.
OH to convert it into the relatively unreactive radical
.
CH
2
(CH
3
)
2
COH:
.
OH CH
3
3
COH !
.
CH
2
CH
3
2
COH H
2
;
76
k
76
6:0 10
8
dm
3
mol
1
sec
1
59
In this way, the powerful reducing properties of e
aq
can be exploited to generate hyper-
reduced states of metal ions, for example:
e
aq
Cd
2
! Cd
77
The monovalent ions Cd
+
, Co
+
, Ni
+
, and Zn
+
are themselves quite strong reductants
and, being positively charged, can be used to reduce metalloproteins at sites where there
is a local negative charge that makes them less accessible to e
aq
[98].
Table 7 Percentage Abstraction of H-Atom by
.
OH from Various Positions
in Simple Alcohols
Alcohol a-CUH Other CUH OH
CH
3
OH 93.0 7.0
CH
3
CH
2
OH 84.3 13.2 2.5
CH
3
(CH
2
)
2
OH 53.4 46.0 <0.5
(CH
3
)
2
CHOH 85.5 13.3 1.2
CH
3
(CH
2
)
3
OH 41.0 58.5 <0.5
(CH
3
)
3
COH 95.7 4.3
(CH
2
OH)
2
100 <0.1
CH
3
CH(OH)CH
2
OH 79.2 20.7 <0.1
CH
3
CH(OH)CH(OH)CH
3
71.0 29.0 <0.1
Source: Ref. 93.
Table 8 Values of pK
a
and E
o
for Commonly Used Reducing Radicals
Obtained by Reaction of
.
OH with Simple Alcohols (see Table 7) and
Formate Ion
Radical pK
a
[94]
E
o
(RO, H
+
/ROH
.
)
(V) [95]
E
o
(RO/RO
.
)
(V) [95]
.
CH
2
OH 10.71 1.18 1.81
CH
3
.
CHOH 11.51 1.25 1.93
(CH
3
)
2
.
COH 12.03 1.39 2.10
.
CH
2
C(CH
3
)
2
OH 0.1 [96]
CO
.
2
0.2 [97] 1.90
Copyright 2004 by Taylor & Francis Group, LLC
6. APPLICATIONS IN GENERAL CHEMISTRY
The great advantage of radiolysis over other methods of producing free radicals is that only
the solvent absorbs the ionizing radiation in most chemical applications. Moreover, it is
clear from the foregoing information that the radiolysis of water provides a powerful
method of generating one-electron redox agents that can be nely tuned in terms of
reduction potential and electric charge. The use of pulse radiolysis in particular has resulted
in a great wealth of information being obtained on the kinetics and mechanisms of the
reactions of free radicals, and of metal ions in unusual oxidation states, in aqueous solution.
The chapters, each written by experts, in Radiation ChemistryPresent Status and Future
Trends [99] provide excellent descriptions of the diverse nature and applications of this eld
of study.
REFERENCES
1. Allen, A.O. The story of the radiation chemistry of water. In Early Developments in Radiation
Chemistry; Kroh, J., Ed.; Royal Society of Chemistry: London, 1989; p 1.
2. Risse, O. Strahlenchemie 1929, 34, 578.
3. Weiss, J. Nature 1944, 153, 7481.
4. Fricke, H.; Hart, E.J.; Smith, H.P. J. Chem. Phys. 1938, 6, 229.
5. Allen, A.O. J. Phys. Colloid Chem. 1948, 52, 479.
6. Hayon, E.; Weiss, J.J. Proceedings of the 2nd International Conference on Peaceful Uses Of
Atomic Energy; Geneva, 1958, 29, 80.
7. Hart, E.J.; Boag, J.W. J. Am. Chem. Soc. 1962, 84, 4090.
8. Schwarz, H.A. J. Phys. Chem. 1969, 73, 1928.
9. Swiatla-Wojcik, D.; Buxton, G.V. J. Phys. Chem. 1995, 99, 11464.
10a. Cobut, V.; Frongillo, Y.; Patau, J.P.; Goulet, T.; Fraser, M.-J.; Jay-Gerin, J.-P. Radiat. Phys.
Chem. 1998, 51, 229.
10b. Frongillo, Y.; Goulet, T.; Fraser, M.-J.; Cobut, V.; Patau, J.P.; Jay-Gerin, J.-P. Radiat. Phys.
Chem. 1998, 51, 245.
10c. Herve du Penboat, M.-A.; Goulet, T.; Frongillo, Y.; Fraser, M.-J.; Bernat, P.; Jay-Gerin, J.-P.
J. Phys. Chem. A 2000, 104, 11757.
11. Kaplan, I.G.; Miterev, A.M.; Sukhonosov, A.M. Radiat. Phys. Chem. 1990, 36, 493.
12. Lampe, F.W.; Field, F.H.; Franklin, J.L. J. Am. Chem. Soc. 1957, 79, 6132.
13. Hamill, W.H. J. Phys. Chem. 1969, 73, 1341.
14. Platzman, R.L. In Abstracts of Papers, Second International Congress of Radiation Research,
Harrogate, August, 511, 1962; Ebert, M., Howard, A., Eds.; North-Holland Publishing:
Amsterdam, 1963; p 128.
15. Faraggi, M.; Desalos, J. Int. J. Radiat. Phys. Chem. 1969, 1, 335.
16. Pastina, B.; LaVerne, J.A.; Pimblott, S.M. J. Phys. Chem. A 1999, 103, 5841.
17. LaVerne, J.A.; Pimblott, S.M. J. Phys. Chem. A 2000, 104, 9820.
18. Wiesenfeld, J.M.; Ippen, E.P. Chem. Phys. Lett. 1980, 73, 47.
19. LaVerne, J.A.; Pimblott, S.M. J. Phys. Chem. 1991, 95, 3196.
20. Pimblott, S.M.; LaVerne, J.A. J. Phys. Chem. A 1997, 101, 5828.
21. Swiatla-Wojcik, D.; Buxton, G.V. J. Chem. Soc. Faraday Trans. 1998, 94, 2135.
22. Herve du Penboat, M.-A.; Meesungnoen, J.; Goulet, T.; Filali-Mouhim, A.; Manketkorn, S.;
Jay-Gerin, J.-P. Chem. Phys. Lett. 2001, 341, 135.
23. Magee, J.L.; Chatterjee, A. J. Phys. Chem. 1980, 84, 1937.
24. Swiatla-Wojcik, D.; Buxton, G.V. Phys. Chem. Chem. Phys. 2000, 2, 5113.
25. Cliord, P.; Green, N.J.B.; Oldeld, M.J.; Pilling, M.J.; Pimblott, S.M. J. Chem. Soc., Faraday
Trans. I 1986, 82, 2673.
Copyright 2004 by Taylor & Francis Group, LLC
26. Swiatla-Wojcik, D. personal communication.
27. www.rcdc.nd.edu.
28. Peled, E.; Czapski, G. J. Phys. Chem. 1970, 74, 2903.
29. Draganic, Z.D.; Draganic, I.G. J. Phys. Chem. 1971, 75, 3950.
30. Pimblott, S.M.; LaVerne, J.A. J. Phys. Chem. A 1998, 102, 2967.
31. Jonah, C.D.; Miller, J.R.; Matheson, M.S. J. Phys. Chem. 1977, 81, 1618.
32. Lam, K.Y.; Hunt, J.W. Int. J. Radiat. Phys. Chem. 1975, 7, 317.
33. Kee, T.W.; Son, D.H.; Kambhampati, P.; Barbara, P.F. J. Phys. Chem. A2001, 105, 8434 (and
references therein).
34. Draganic, Z.D.; Draganic, I.G. J. Phys. Chem. 1973, 77, 765.
35. Head, D.A.; Walker, D.C. Nature 1965, 207, 517.
36. Dainton, F.S.; Logan, S.R. Trans. Faraday Soc. 1965, 61, 715.
37. Buxton, G.V.; Lynch, D.A.; Stuart, C.R. J. Chem. Soc. Faraday Trans. 1998, 94, 2379.
38. Balkas, T.I.; Fendler, J.H.; Schuler, R.H. J. Phys. Chem. 1970, 74, 4497.
39. Schmidt, K.H.; Han, P.; Bartels, D.M. J. Phys. Chem. 1995, 99, 10530.
40. Elliot, A.J.; Chenier, M.P.; Oullette, D.C. J. Chem. Soc. Faraday Trans. 1993, 94, 1193.
41. Yoshida, H. Radiat. Res. 1994, 137, 145.
42. Warman, J.M.; Asmus, K.-D.; Schuler, R.H. J. Phys. Chem. 1969, 73, 931.
43. Pimblott, S.M.; LaVerne, J.A. J. Phys. Chem. 1992, 96, 8904.
44. Ambramwitz, M.; Stegun, I.A. Handbook of Mathematical Functions; Dover: New York, 1970.
45. Wol, R.K.; Bronskill, M.J.; Aldrich, J.E.; Hunt, J.W. J. Phys. Chem. 1973, 77, 1350.
46. Jonah, C.D.; Hart, E.J.; Matheson, M.S. J. Phys. Chem. 1973, 77, 1838.
47. Jonah, C.D.; Matheson, M.S.; Miller, J.R.; Hart, E.J. J. Phys. Chem. 1976, 80, 1276.
48. Sumiyoshi, T.; Tsugaru, K.; Yamada, T.; Katayama, M. Bull. Chem. Soc. Jpn. 1985, 58, 3073.
49. Bartels, D.M.; Cook, A.R.; Mudaliar, M.; Jonah, C.D. J. Phys. Chem. A 2000, 104, 1686.
50. Czapski, G.; Peled, E. J. Phys. Chem. 1973, 77, 893.
51. Chernovitz, A.C.; Jonah, C.D. J. Phys. Chem. 1988, 92, 5946.
52. Draganic, I.G.; Nenadovic, M.T.; Draganic, Z.D. J. Phys. Chem. 1969, 73, 2564.
53. Schuler, R.H.; Behar, B. In Proceedings of the 5th Tihany Symposium on Radiation Chemistry;
Dobo, J. Hedvig, P., Schiller, R., Eds.; Akademai Kiado: Budapest, 1983; Vol. 1, p 183.
54. LaVerne, J.A.; Pimblott, S.M. J. Chem. Soc. Faraday Trans. 1993, 89, 3527.
55. Jay-Gerin, J.-P.; Ferradini, C. Chem. Phys. Lett. 2000, 317, 388.
56. Sutton, H.C.; Adams, G.E.; Boag, J.W.; Michael, B.D. In Pulse Radiolysis; Ebert, M., Keene,
J.P., Swallow, A.J., Baxendale, J.H., Eds.; Academic Press: London, 1965; p 61.
57. Adams, G.E.; Boag, J.W.; Currant, J.; Michael, B.D. In Pulse Rdaiolysis; Ebert, M., Keene,
J.P., Swallow, A.J., Baxendale, J.H., Eds.; Academic Press: London; 1965; p 117.
58. Buxton, G.V.; Dainton, F.S. Proc. R. Soc. A 1965, 287, 427.
59. Buxton, G.V.; Greenstock, C.L.; Helman, W.P.; Ross, A.B. J. Phys. Chem. Ref. Data 1988, 17,
513.
60. Schuler, R.H.; Hartzell, A.L.; Behar, B. J. Phys. Chem. 1981, 85, 192.
61. Benderskii, V.A.; Krivenko, A.G.; Rukin, A.N. High Energy Chem. 1980, 14, 303.
62. Schmidt, K.H.; Han, P.; Bartels, D.M. J. Phys. Chem. 1992, 96, 199.
63. Buxton, G.V.; Elliot, A.J. J. Chem. Soc., Faraday Trans. 1993, 89, 485.
64. Hug, G.L. Optical Spectra of Nonmetallic Inorganic Transient Species in Aqueous Solution, Nat.
Stand. Ref. Data Ser. Nat. Bur. Stand. (U.S.A.). Washington, 1981; p 69.
65. Elliot, A.J.; Oullette, D.C. J. Chem. Soc., Faraday Trans. 1994, 90, 837.
66. Sehested, K.; Christensen, H. Radiat. Phys. Chem. 1990, 36, 499.
67. Hart, E.J.; Anbar, M. The Hydrated Electron, Wiley-Interscience: New York, 1970.
68. Noyes, R.M. In Progress in Reaction Kinetics; Porter, G., Ed.; Pergamon: London, 1961; Vol. 1,
p 129.
69. Buxton, G.V.; Mackenzie, S.R. J. Chem. Soc. Faraday Trans. 1992, 88, 2833.
70. Cercek, B. Nature 1969, 223, 491.
71. Newton, M.D.; Sutin, N. Annu. Rev. Phys. Chem. 1984, 35, 437.
Copyright 2004 by Taylor & Francis Group, LLC
72. Fessenden, R.W.; Verma, N.C. Faraday Discuss. Chem. Soc. 1977, 63, 104.
73. Han, P.; Bartels, D.M. Chem. Phys. Lett. 1989, 159, 538.
74. Meyerstein, D. Faraday Discuss. Chem. Soc 1977, 63, 203.
75. Collinson, E.; Dainton, F.S.; Tazuke, S.; Smith, D.R. Nature 1963, 206, 198.
76. Berdnikov, V.M. Russ. J. Phys. Chem. 1973, 47, 1547.
77. Meyerstein, D. Acc. Chem. Res. 1978, 11, 43.
78. Buxton, G.V.; Djouider, F.; Lynch, D.A.; Malone, T.N. J. Chem. Soc. Faraday Trans. 1997, 93,
4265.
79. Elliot, A.J.; McCracken, D.R. Radiat. Phys. Chem. 1989, 33, 69.
80. Boder, M.; Wojnarovits, L.; Foldiak, G. Radiat. Phys. Chem 1990, 36, 175.
81. Christensen, H.C.; Sehested, K.; Hart, E.J. J. Phys. Chem. 1973, 77, 983.
82. Ashton, L.; Buxton, G.V.; Stuart, C.R. J. Chem. Soc., Faraday Trans. 1995, 91, 1631.
83. Bielski, B.H.J.; Cabelli, D.E.; Arudi, R.L.; Ross, A.B. J. Phys. Chem. Ref. Data 1985, 14, 1041.
84. Allen, A.O.; Bielski, B.H.J. In Superoxide Dismutase; Oberley, L.W., Ed.; CRC Press: Boca
Raton, 1982; Vol. 1, p 125.
85. Elliot, A.J.; Buxton, G.V. J. Chem. Soc., Faraday Trans. 1992, 88, 2465.
86. Meisel, D.; Czapski, G. J. Phys. Chem. 1975, 79, 1503.
87. Goldstein, S.; Czapski, G. J. Am. Chem. Soc. 1983, 105, 7276.
88. Buxton, G.V. Trans. Faraday Soc. 1970, 66, 1656.
89. Zehavi, D.; Rabani, J. J. Phys. Chem. 1971, 75, 1738.
90. Stanbury, D.M. Adv. Inorg. Chem. 1989, 33, 69.
91. Sehested, K.; Christensen, H. Radiat. Phys. Chem. 1990, 36, 499.
92. Chrisrensen, H.; Sehested, K.; Logager, T. Radiat. Phys. Chem. 1994, 43, 527.
93. Asmus, K.-D.; Mo ckel, H.; Henglein, A. J. Phys. Chem. 1973, 77, 1218.
94. Laro, G.P.; Fessenden, R.W. J. Phys. Chem. 1973, 77, 1283.
95. Schwarz, H.A.; Dodson, R.W. J. Phys. Chem. 1989, 93, 409.
96. Endicott, J.P. In Concepts of Inorganic Photochemistry; Adamson, A.W., Fleishauer, P.D., Eds.;
Wiley: New York, 1975; p 88.
97. Jeevarajan, A.S.; Carmichael, I.; Fessenden, R.W. J. Phys. Chem. 1990, 94, 1372.
98. Govindaraju, K.; Christensen, H.E.M.; Lloyd, E.; Olsen, M.; Salmon, G.A.; Tomkinson, N.P.;
Sykes, A.G. Inorg. Chem. 1993, 32, 40.
99. Radiation Chemistry. Present Status and Future Prospects; Jonah, C.D., Rao, B.S.M., Eds.;
Elsevier: Amsterdam, 2001.
Copyright 2004 by Taylor & Francis Group, LLC
13
Photochemistry and Radiation Chemistry
of Liquid Alkanes: Formation and Decay
of Low-Energy Excited States
L. Wojnarovits
Hungarian Academy of Sciences, Budapest, Hungary
1. INTRODUCTION
The radiation chemistry of alkanes was reviewed several times in the past. The books of
Topchiev [1] and Foldiak (editor) [2] deal mainly with the yield and distribution of nal
products. In the book edited by Gaumann and Hoigne [3] the yields of radical intermediates,
LET eects (linear energy transfer, energy lost when the particle passes unit length) reaction
mechanisms, etc. are discussed. Hydrocarbon chemistry dominates the book edited by
Ausloos [4]. Freeman [5] and, more recently, Hummel [6] explained ionizing-radiation-in-
duced chemical changes in alkanes based on the radiation chemical theories available at the
time of writing of their review papers. The most recent review published by Shkrob et al. [7]
concentrates on the liquid phase ion chemistry detailing the dierent views on the high-
mobility cations observed inseveral liquidcycloalkanes; they also discuss some aspects of the
excited-state chemistry. The formation and decay of excited alkane molecules was reviewed
in a recent book chapter of Mayer and Szadkowska-Nicze [8]. In addition to these special
books, book chapters or review articles on hydrocarbon radiolysis, or more specically on
alkane radiolysis, general radiation chemistry books also provide great attention to the
radiolysis of alkanes [913].
This chapter is concerned mainly with the chemistry of excited molecules in photon
and high-energy ionizing-radiation-induced processes of alkanes. Our knowledge on the
higher-energy excited states is rather limited, so the results on the reactions of lower energy
excited molecules are reviewed. We will compare the processes taking place in radiolysis
and photolysis and try to give more chemistry than reported in the previous reviews. Most
of the work done in this eld was published in the 1980s; however, as can be seen from the
reference list, more recent papers also appeared, which are incorporated in the review.
2. PHOTOCHEMISTRY OF LIQUID ALKANES
The electron excitation of molecules in nearly all classes of organic compounds can be
performed by visible or ultraviolet light (k
ex
>190 nm,<6.5 eV). There are a few ex-
Copyright 2004 by Taylor & Francis Group, LLC
ceptions, e.g., the alkanes, where excitation requires higher energies in the so-called
vacuum UV (VUV) region. The photon irradiation of alkanes is generally done in the
100- to 180-nm wavelength range; this range corresponds to a photon energy range of 712
eV. From the energy dependence and by comparing the data obtained in radiolysis and
photolysis, valuable information can be obtained concerning the activated intermediates
leading to the formation of stable end products in radiolysis.
Most publications dealing with the photodecomposition of alkanes discuss the
processes in the gas phase; several comprehensive works have already been published in
this eld [1417]. In the present work, we summarize the results of liquid phase photolytic
studies and compare them with those obtained in radiolysis. An early review on liquid
alkane photochemistry was published in Ref. 18, a brief overview of the eld was given in
Ref. 19.
2.1. Absorption and Emission Spectra of Alkanes, Ionization
in the Liquid Phase
The VUV absorption spectra of alkanes are rather featureless, usually without marked
peaks (as shown in Fig. 1 for n-hexane) in contrast to the spectra of aromatics or to the
spectra of compounds containing heteroatoms. Earlier works usually assumed that at low
energies in alkanes the excitation results in j!j*-type valence-electron transitions, but
later works concluded that the excitations, at least partly, result in Rydberg states [2023].
Figure 1 Absorption [20,21] and emission [25] spectra of n-hexane.
Copyright 2004 by Taylor & Francis Group, LLC
Rydberg-type orbitals are extended in space and resemble atomic orbitals, whereas the
valence excitations do not lead to signicant spatial extension. Consequently, the Rydberg
states are, in contrast to the valence-electron excitations, sensitive to external perturba-
tions coming from changes of pressure or phase (liquidgas). The probability of formation
of Rydberg orbitals in condensed phases is less than in gases, although many experimental
results indicate their existence, e.g., energy transfer from higher excited states (Sec. 3.2).
During irradiation of many alkanes with VUV photons, with visible or infrared
photons causing double or multiphoton excitation, or ionizing radiation, weak uores-
cence was rst observed at the end of the sixties with quantum yields of U
f
c10
5
to 10
2
[2426] (Table 1). The uorescence spectrum is a broad, structureless band around 170
250 nm; the uorescence intensity for a given alkane has approximately a Gaussian
dependence on the uorescence photon energy. The uorescence spectrum has been
characterized by the wavelength of maximum k
f
and the full width at the half height r
f
of the emission spectrum. k
f
lies by
f
1 eV below the adsorption onset. This large energy
gap between the absorbed and emitted energy in alkanes (D) is presumably a consequence
of the large dierence between the nuclear distances in the ground and excited states.
Thus, D has a similar meaning as the Stokes shift [25]. For aromatics, where a conjugated
electron system is excited, the Stokes shift is only 0.1 eV. For most compounds studied,
the uorescence quantum yields decline monotonically with the energy of photons used for
excitation. U
f
attens out or reaches a minimum a few tenths of an electron volt above the
liquid phase ionization threshold. The minimum is sometimes followed by a slow increase
[26,27]. The minimum and the slow increase are attributed to the contribution of
recombination uorescence as will be discussed later in this section. Fluorescence was
not detected (U
f
V 10
5
) in the irradiation of geminally substituted branched alkanes, such
as isooctane, and the C
5
, C
7
C
10
cycloalkanes.
The energy of the relaxed S
1
state of alkanes in the liquid phase is certainly
somewhere between the energy of the absorption onset and the energy of the uorescence
maximum. In energy-transfer experiments, the energy of absorption threshold is usually
Table 1 Photon Absorption and Fluorescence Properties of Some Selected Alkanes
Alkanes k
a
, nm
a
k
f
, nm
b
D, eV
c
r
f
, eV
d
U
f,165 nm
e
I
g
, eV
f
I
l
, eV
g
n-Hexane 171 206 1.24 1.02 0.0006 10.18 8.7
n-Decane 173 207 1.19 0.95 0.0042 9.65 8.4
3-Methylpentane 173 231 1.8 1.24 0.0002 10.08 8.5
3-Methylhexane 174 225 1.6 1.14 0.002 10.0 8.65
2,3-Dimethylbutane 175 242 1.95 1.08 0.0061 10.02 8.6
2,3,4-Trimethylpentane 179 244 1.10 0.0016 9.6 8.4
Cyclohexane 177 201 0.83 0.95 0.0088 9.86 8.4
Methylcyclohexane 179 213 1.08 1.1 0.011 9.64 8.3
trans-Decalin 186 217 0.94 1.1 0.023 9.24 8.0
a
Wavelength at which the decadic extinction coecient of the neat liquids equals 5 mol
1
dm
3
cm
1
.
b
Wavelength of uorescence maximum.
c
The dierence between the energies of the absorption onset and the uorescence maximum.
d
Full width at the half height of the uorescence spectrum expressed in energy units.
e
Quantum yield of uorescence measured at 165-nm excitation wavelength.
f
Energy of the gas phase ionization potential.
g
Energy of the liquid phase ionization potential.
Source: Refs. 2527, and 38.
Copyright 2004 by Taylor & Francis Group, LLC
regarded as the energy of the uorescing excited state (S
1
). On the contrary, Kimura and
Hormes, when they constructed term schemes for excited cyclohexane and n-hexane
molecules, used the energy of k
f
as the energy of the lowest lying singlet excited state
[28]. Because of the very low yield of uorescence, other processes, as will be discussed
later, chemical decomposition dominates the decay of alkane excited molecules.
As it is obvious from this review, during the last three decades much information has
accumulated on the properties of singlet excited alkane molecules. At the same time, little
has been known about the properties of triplet excited molecules because there are no
direct methods for studying them [2931]. Our knowledge originates mainly from indirect
sources. Because the S
0
!T
n
optical transitions are spin forbidden, in contrast to S
0
!S
n
transitions, the optical absorption spectra give little information on the triplet energies.
The S
0
!T
n
transitions do appear in the electron-energy-loss spectra, but the resolution of
most of these spectra is poor. The energy of the lowest triplet state of the liquid alkanes
can also be estimated from biphotonic sensitization experiments performed with aromatic
hydrocarbons in alkanes usually in the solid state [3234]. In this process, by consecutive
absorption of two photons, high-energy triplet states of the aromatic molecules are
produced, which by energy transfer may produce the triplet state of the alkane molecule.
The transfer is recognized by the uorescence and phosphorescence properties of the
aromatic molecule [33] or by the decomposition of the alkane molecule [32,34]. The T
1
energies estimated from these experiments, e.g., for the cyclohexanes, are all around 5.8
6.0 eV [29,33,34]. Using another technique, i.e., bombardment of thin lms of alkanes by
low-energy electrons, Leclerc et al. [35] estimated a value of 5.8 eV for the T
1
of n-hexane.
Therefore, the energy separation between S
1
and T
1
is
f
1 eV. For Rydberg-type
excitations, the separation is, as a rule,
f
0.1 eV; for valence-type excitations, it is 12 eV
[22]. The separation of
f
1 eV is in favor of valence excitations in agreement with the
results of those quenching experiments in which the kinetic distance was found to be
comparable with the molecule diameter (Sec. 3.1).
Based on energy-transfer experiments in some works, low-energy (and possibly long-
living) alkane triplet state is also indicated; however, there is no direct proof for its
existence [7,36,37].
According to precise photoconductivity measurements, the ionization onset, which is
usually taken as the ionization potential, is ca. 1.5 eV lower in liquid alkanes than in the
gas phase [38]. The ionization potentials in liquid and gas phases (I
l
and I
g
, respectively)
are related by the equation:
I
1
I
g
P
V
o
1
where V
o
is the lowest energy of the electron-conducting band and P
+
denotes the
polarization energy of the positive ion. P
+
can be represented by Borns equation:
P
c
e
2
2R
o
1
1
e
2
where e is the elementary charge, R
o
is the ion radius, and e denotes the dielectric constant
of the liquid. For liquid alkanes, P
+
has a value between 1.6 and 1.3 eV and V
o
is
between 0.6 and +0.2 eV, so that Eq. (1) predicts a dierence between I
1
and I
g
of
f
1.5
eV in agreement with the experimental values (Table 1).
Passing through the threshold energy for ionization of liquid alkanes, the quantum
yield of ionization, U
i
, increases very slowly with the photon energy [23,26,3840].
Copyright 2004 by Taylor & Francis Group, LLC
According to conductivity and uorescence measurements 1.5 eV above the threshold, U
i
c 0.10.4. Ausloos et al. used chemical decomposition measurements to estimate the
ionization quantum yield of liquid cyclopentane: they reported U
i
= 0.26 F0.05 and 0.4 F
0.1 at energies of 1.3 and 2.9 eV, respectively, above the liquid phase ionization onset [40].
Thus, 13 eV above the ionization threshold the photon absorption mostly produces
neutral excited molecules, so-called superexcited molecules. The photoionization is nearly
complete at 16- to 20-eV photon energies. Ostan and Lipsky suggested that the initially
produced e
+ RH
+
geminate ion pair rst rapidly converts internally to an ion-pair state
ca. 1.6 eV below the liquid phase ionization threshold, then collapses to generate the
uorescent state of RHwith an energy close to its absorption threshold value [26]. Therefore,
in (singlet) recombination uorescing excited molecules form with a high probability.
2.2. Excitation Sources
As mentioned earlier, the light absorption of alkanes causing electron excitation is in the
far-UV range, where the emission of the so-called continuous light sources is little, e.g.,
that of high-pressure mercury or xenon lamps. For photodecomposition measurement,
mostly line-emitting sources are applied. They give a light intensity of 10
15
to 10
16
photon
sec
1
. The photon energies of a nitrogen lamp emitting at 174 nm (7.1 eV) [41], a bromine
lamp emitting at 163 nm (7.6 eV) [42], and a xenon lamp emitting at 147 nm (8.4 eV)
[43,44] are just slightly above the VUV absorption onset of alkanes, but they are lower by
23 eV than the gas phase ionization potential and by 01.5 eV than the liquid phase
ionization onset. The energies of the krypton and argon lamps emitting at 124 nm and at
105107 nm (10 eV and 11.611.8 eV) are the ranges of gas phase ionization potentials
[43]. In liquids, the absorption takes place in an extremely thin layer (
f
1 Am); thus, to
avoid high local conversion and rise of temperature, vigorous stirring is necessary. Some
types of lasers were also utilized for exciting alkanes, e.g., following ArF* excimer laser
excitation with 193 nm (6.34 eV) the nal products were analyzed [45]. Other lasers (e.g.,
N
2
laser, Nd
+
YAG laser, and some dye lasers) in double or multiphoton absorption
mode because of the weak intensities were used only for uorescence measurements or
transient absorption measurements [4650]. In uorescence lifetime measurements apply-
ing the single-photon counting technique, the VUV component of the radiation induced
by electric discharges in medium-pressure (110 bar) gases (H
2
, N
2
, Xe, or air) was also
used [51,52]. In photon absorption and emission studies, synchrotron radiation is often
applied [5356].
Finally, ionizing radiations should be mentioned. A disadvantage of applying
ionizing radiation is that the excited molecule formation is a delayed process; most excited
molecules form probably in electroncation neutralization [7]. Another disadvantage is
that in the irradiated liquid dierently excited molecules and ions are present, including
fragments, which may cause complications, e.g., in energy-transfer measurements.
2.3. Lifetime of the Excited Molecules
For the determination of the excited-state lifetime, s, energy-transfer measurements in
steady-state experiments, orescence decay measurements, or transient photon absorption
measurements were applied.
In steady-state quenching experiments [57,58], usually the ratio of quantum yields of
products or uorescence photons, respectively, measured in the absence (U
0
) and presence
(U) of the quencher showed a linear dependence on the quencher concentration c. From
Copyright 2004 by Taylor & Francis Group, LLC
the linear dependencies, U
0
/U = 1 + Kc, the K quenching parameters were calculated:
K = k
d
s. When energy-transfer experiments were used to deduce the s values, the quench-
ing was assumed to proceed with a diusion-controlled rate, and the transfer was sup-
posed to take place when the acceptor and donor molecules were in contact (no static
quenching). The k
d
transfer rate coecients were calculated using the usual diusional
equations [24,25,59] (see Sec. 3.1).
In Table 2, we show the room temperature ss for some selected alkanes, and among
them there are lifetime data that were determined in quenching experiments. (Data for
larger groups of alkanes are discussed in Refs. 48, 49, 51, 5961, 72, and 73.) For the
Table 2 S
1
Lifetimes (nsec) of Selected Liquid Alkanes at Room Temperature Determined in
Steady-State Excited State Quenching or in Transient Fluorescence and Absorption Studies
Alkanes Quenching Fluorescence Absorption
n-Hexane 0.3 [77];
f
0.32 [48];
V0.7 [59]; 0.7 [73]
0.25 [47];
0.28 [49]
n-Heptane 0.73 [48,60]; 0.9 [51];
1.05 [59]; 1.2 [73]
0.65 [49]
n-Octane 1.08 [48]; 1.35 [81];
1.47 [51]; 1.5 [59]; 1.6 [73]
>1.0 [49]
n-Decane 1.76 [48]; 2.4 [73,81];
2.5 [51,59]; 2.6 [60]; 3.0 [54]
>0.6 [49]
n-Dodecane 2.5 [48]; 3.1 [60]; 3.7 [81];
3.9 [51,59]; 4.0 [73]; 4.2 [54]
>2.8 [49]
n-Tetradecane 3.01 [48]; 4.3 [50,59];
4.4 [73]
n-Hexadecane 3.7 [48]; 5.2 [59] >3.0 [49]
2-Methylbutane 0.75 [59]
2,3-Dimethylbutane 1.0 [81]; 1.4 [59]
Cyclohexane(CH) 1.2 [59] 0.3 [60]; 0.68 [51]; 0.9 [46];
1.1 [59,69,78]; 1.2[72]
>0.3 [49];
1.0 [47]
MethylCH 0.65 [51]; 0.69 [79] 0.81 [48];
1.2 [72]; 1.25 [59]
EthylCH 1.25 [59]
cis-1.2-DimethylCH 1.02 [79]; 1.15 [59]
trans-1,2-Dime.CH 0.9 [59]; 1.11 [79]
cis-Decalin 2.1 [61,63]; 2.18 [51];
2.45 [54]
trans-Decalin 2.24 [48]; 2.7 [59]; 2.82 [51];
2.86 [79]; 3.0 [54]
>1.4 [49];
3.0 [47]
Bicyclohexyl 1.58 [51]; 1.6 [60]; 1.62 [48];
1.8 [54]; 1.9 [72]
n-Dodecane-d
26
6.5 [83]
Cyclohexane-d
12
2.1 [83]; 2.5 [78] 2.7 [47]
Cyclopentane V0.1 [59] 0.1 [49]
Cycloheptane 0.1 [59]
Cyclooctane 0.3 [59]
2,2,4-Trimethylpentane V0.2 [59] <0.01 [47];
0.04 [49]
Copyright 2004 by Taylor & Francis Group, LLC
uorescence lifetime of cyclohexane, the quenching experiments gave 1.2 nsec, a value that
agrees with the results of the newer uorescence and absorbance decay measurements.
The lifetimes of the alkane excited states were generally measured by detecting the
decay of uorescence. The s values obtained by dierent excitations, by x-ray pulses [60,62],
high-energy electron pulses [61,6374], pulses of photons [51,52,54,55,7577], laser pulses
causing double or multiphoton excitations [46,48,50,7883], are generally in good agree-
ment indicating that uorescence originates in each case from the same excited state. These
uorescence lifetimes also agree with the lifetimes that were measured by detecting the
absorbance decay of excited molecules. As regards s, cyclohexane is a special case. For its
excited-state lifetime, Henry and Helman in 1972 suggested 0.3 nsec, based on the very rst
direct measurements [60]. Two years later, Ware and Lyke found 0.68 nsec [51]. However,
over 10 later measurements yielded lifetimes in the 0.8- to 1.2-nsec range: the most probable
room-temperature value is 1.0 nsec.
As shown by Tagawa et al. [74], the alkane excited molecules have a broad absorption
band in the visible region with maxima increasing from
f
430 to
f
680 nm between C
5
and
C
20
for the n-alkanes. This spectrum strongly overlaps with the absorption spectrum of the
radical cations: with lowcarbon atomnumber alkanes the two maxima practically coincide.
With increasing carbon atom number, the red shift in the radical cation absorbance is
stronger than in the S
1
molecule absorbance [47,49,8486]. The decay of the excited
molecule absorbance was composed of two components with 0.1- and 1.0-nsec decay times
in cyclohexane, and 0.17 and 2.7 nsec in perdeuterocyclohexane [47]. The nature of the
faster-decaying component is as yet unclear.
As mentioned before, the excited-state lifetimes determined by photon and (low-
LET) ionizing radiation excitations practically coincide. However, when 5- to 14-keV soft
synchrotron x-rays were used for the irradiation in the experiments of Holroyd et al. [87],
instead of the usual fast electrons with energies in the MeV range, shorter ss were
measured than in VUV photon irradiation. The lifetimes and singlet yields decreased with
decreasing x-ray energy, thus decreased with increasing LET value. This decrease between
14 and 5 keV for cis- and trans-decalin and n-dodecane is ca. 15%. The decrease was
attributed to the quenching of alkane excited molecules by alkyl radicals whose concen-
trations may be as high as 10 mmol dm
3
in the high ionizing density tracks. Shibata et al.
[88] explained similarly the decrease of n-dodecane uorescence lifetime with increasing
LET in heavy-ion radiolysis.
In contrast to the liquid phase lifetimes, those in the gas phase showed marked depen-
dence on the energy of excitation [52,55,75]. They decreased with increasing excitation en-
ergy, and for some alkanes the ss at the red edge in the vapor phase were longer than the
liquid phase lifetimes. For instance, in the 170-nmphotoirradiation of cyclohexane at 1 Torr
pressure, Ware and OConnor found s = 1.21.3 nsec [52], whereas Wickramaaratchi et al.
measured 1.5 F0.2 nsec [75]. The liquid phase value is around 1 nsec. In the literature, there
is no satisfactory explanation for this phase eect. In the lower pressure range, the increase
of the vapor pressure, or added inert gases increased the lifetime because of vibrational
cooling of excited alkane molecules. Most of the energy dependence in the gas phase is
probably because of a competition between relaxation and the chemical decomposition.
In photoirradiated solid cyclohexane (freezing point 6.5jC), much higher uores-
cence quantum yields and longer uorescence lifetimes were observed than in the liquid
phase [89]. In solid Ar matrices, the uorescence characteristics, energy dependence of the
lifetime and intensity, were found to be very similar to these characteristics in the gas
phase. This points to the importance of cyclohexanecyclohexane interactions to deter-
mine the excited-state characteristics in the liquid phase [76].
Copyright 2004 by Taylor & Francis Group, LLC
The lifetime of the triplet excited molecule with energy
f
1 eV below the S
1
energy is
probably extremely short: the triplet excitations are suggested to take place to dissociative
states [90]. Such conclusion can be obtained from the biphotonic sensitization experiments
mentioned in connection with the energy of the triplet states (Sec. 2.1): if the decom-
position had not been fast enough (s V 10
12
sec) due to the back transfer of energy, the
sensitization could not have been observed [29].
2.4. The Radiative Rate Coefficient at
f
25jC
By using the techniques mentioned before, room-temperature ss for about 50 alkanes were
determined. In Fig. 2, we show the uorescence quantum yields as a function of lifetimes.
The U
f
values were generally taken from the work of Rothman et al. [25]; most of the
uorescence quantum yields were measured using 165-nm photons for excitation. This
wavelength is close to the absorption onset of most alkanes and (with the exception of the
smaller molecules) the measured quantumyield is close to the uorescence quantumyield of
the relaxed S
1
molecules [26]. The plot in Fig. 2 is similar to the plot we published in Ref. 59
using only our measurements. Here we use practically all the data that are available in the
literature. For most of the alkanes, several lifetime measurements were published. When
Figure 2 Relationship between lifetime (s) and uorescence intensity (U
f
) for a large group of
alkanes.
Copyright 2004 by Taylor & Francis Group, LLC
these measurements do not dier considerably we use the average. When there are larger
dierences between the results of dierent laboratories we try to select the most probable
value. In Fig. 2, the points representing compounds with similar chemical structure lie
around the same straight line. The slope of the lines in the gure is the ratio of the
uorescence intensity and lifetime (or in other words the product of the uorescence
intensity and the decay rate coecient). Therefore, the slope gives the so-called radiative
rate coecient, R, which is one of the most important photophysical characteristics of
excited states. R is a rate coecient with which the excited molecules would disappear from
the solution if there was no chemical decomposition and only uorescence existed. Five
groups can be distinguished; as the sixth group, we mention the nonuorescing alkanes
whose excited-state lifetimes are extremely short (Table 2):
n-Alkanes, R c 1.6 10
6
sec
1
Methylalkanes, R c 2 10
5
sec
1
Vicinally branched dimethylalkanes, R c 5.3 10
6
sec
1
Cyclohexane and alkylcyclohexanes including deuterated cyclohexane and decalins,
R c 8 10
6
sec
1
Di- and trimethylcyclohexanes, R c 1.2 10
7
sec
1
Nonuorescing alkanes, R < 3 10
3
sec
1
The similarity of R within a class of alkanes (the fact that within a class the points
are on the same straight line) indicates that the properties of excited states are similar. This
is also supported by Lipskys measurements on the uorescence spectra. Within a class the
value of the Stokes shift (D) and the half width of the uorescence peak (r
f
) are nearly
constant (see also Table 1).
For n-alkanes the lifetime increases with the number of carbon atoms in the molecule
together with the increase of the uorescence intensity, whereas the radiative rate
coecient remains constant with a value of
f
1.6 10
6
sec
1
; that is, in the absence of
other competing reactions the excited molecules would decay with uorescence exhibiting
a lifetime of about 600 nsec. Addition of a branch to the carbon skeleton greatly inuences
the photophysical and photochemical properties. In comparison with n-alkanes, r
f
is
larger, the absorption onset is slightly red-shifted, and the emission maximum is strongly
red-shifted (Table 1). These properties point to large nuclear distortions between ground
state and the rst excited state (large Stokes shift, D). The large nuclear distortions lead to
rapid decay, low U
f
, and low R. The high D values of the vicinally substituted 2,3-
dimethylalkanes and other optical properties indicate larger nuclear distortions than those
of the singly substituted ones. These compounds exhibit large uorescence quantum yields
and moderate lifetimes. Their radiative rate coecients are the highest among the aliphatic
alkanes: R c 5.3 10
6
sec
1
.
The S
1
excited-state lifetimes of cyclohexane and alkyl cyclohexanes are relatively
short, all around 12 nsec; however, their U
f
values are relatively high with the result of
relatively high radiative rate coecient:
f
8 10
6
sec
1
. R is the largest for the di- and
trimethylcyclohexane group,
f
1.2 10
7
sec
1
.
In the group of nonuorescing alkanes the lifetimes are very short, s V 0.3 nsec; an
upper limit of R < 3 10
3
sec
1
is estimated. The absence of uorescence for these
compounds may have two causes: the low R, i.e., low rate coecient of the S
1
!S
0
radiative transition and the short lifetime, i.e., the very fast chemical decomposition. In the
C
5
and C
7
C
10
cycloalkanes the ring strain, which is mainly caused by the repulsive
interaction of their unfavorably displaced H atoms, may enhance the rate coecient of the
chemical decay by CH decompositions. In the excited sates of the geminally branched
Copyright 2004 by Taylor & Francis Group, LLC
alkanes the excess energy strongly localized to the branch points can cause rapid local
decomposition.
2.5. Temperature Dependence of the S
1
Lifetimes, Depopulation
of S
1
Excited Molecules
In the early eighties, several papers [7981] reported that the uorescence lifetime exhibits
characteristic temperature dependence: it increases with the decreasing temperature and
tends to level o at low temperature. The Arrhenius plots, log s
1
versus T
1
, did not give
straight lines. In photophysics, such behavior is usually interpreted in terms of several
competing decay channels, and the description is usually made by the sum of several
Arrhenius-type decays. Because at higher temperatures the k decay rate coecient (= s
1
)
can be well described by a single Arrhenius-type equation, the following description was
introduced by Orlandi et al.:
s
1
k k
0
A expE
a
=RT 3
According to this equation the decay is composed of a temperature-independent and a
thermally activated part [55,67,7577,7983]. For cis- and trans-decalins, however, the
temperature dependence was found to be very small [62,67,79,80]. Because of the small
temperature dependence, the activated and nonactivated processes were not separated.
The k
0
, A, and E
a
parameters obtained for a few alkanes are collected in Table 3. k
0
is around 10
8
sec
1
, A c10
11
to 10
12
sec
1
, and E
a
c10 to 20 kJ mol
1
. In principle, the
decay of excited states may involve S
1
!S
x
-type internal conversion transitions [IC, where
S
x
is some singlet state that gives the product(s) of chemical decomposition] and S
1
!T
n
-
type intersystem crossing processes (ISC). The temperature-independent decay was
attributed, on the basis of the size of the rate parameter (k
0
c 10
8
sec
1
), to S
1
!T
n
-
type intersystem crossing. At the same time the temperature-activated decay with a fre-
quency factor of A c 10
11
to 10
12
sec
1
was attributed to an internal conversion process
that takes place by overcoming a barrier of E
a
c1020 kJ mol
1
and leads nally to some
Table 3 Temperature-Independent Term (k
0
), Preexponential Factor (A), and Activation Energy
(E
a
) Obtained by Using Eq. (3) for Selected Groups of Alkanes
Alkanes
k
0
, sec
1
A, sec
1
E
a
, kJ
mol
1
U
nonact.
U
act.
U
radical
U
molec.
n-Octane 2.9 10
8
1.1 10
11
14 0.4 0.6 0.2 0.8
n-Decane 1.5 10
8
5.3 10
11
18 0.3 0.7 0.2 0.8
2,3-Dimethylbutane 5.2 10
8
2.0 10
11
15 0.5 0.5 0.5 0.5
Cyclohexane (CH) 3.6 10
8
1.6 10
11
13 0.3 0.7 0.2 0.8
MethylCH 3.4 10
8
7.5 10
11
16 0.2 0.8 0.3 0.7
trans-1,4-DimethylCH 2.9 10
8
6.8 10
11
18 0.4 0.6 0.4 0.6
cis-1,3-DimethylCH 3.6 10
8
2.1 10
12
20 0.4 0.6 0.5 0.5
IsopropylCH 7.1 10
8
7.5 10
12
18 0.6 0.4 0.6 0.4
The table also shows the quantum yields of nonactivated and activated processes and the yields of the radical-
forming and unimolecular reactions.
Source: Refs. 29, 67, 79, and 81.
Copyright 2004 by Taylor & Francis Group, LLC
singlet state of the products. We summarized the reaction possibilities and their identi-
cation in Fig. 3. (The chemical reactions shown in the gure will be discussed later.)
The identication of the temperature-dependent decay with S
1
!S
x
IC and the tem-
perature-independent decay with S
1
!T
n
ISC was conrmed by investigations on xenon
heavy-atom eect and on deuterium eect as well. Heavy atoms are known to increase the
rates of spin-forbidden processes, whereas the rates of spin-allowed processes remain
essentially unchanged. In agreement with the expectations, the A preexponential factor
and the E
a
activation energy were found to be unaected by the presence of Xe added to the
samples at a concentration of 0.10.4 mol dm
3
[67,82]. At the same time k
0
was found to
increase. For deuteration, the eect was just the reverse. The IC process is connected with
chemical decomposition; for the investigated compounds the chemical decomposition was
mainly H
2
elimination, and therefore its rate should be inuenced by deuteration. At the
same time, the ISC process is connected with multiplicity changes only; therefore the eect
of deuteration on that process should be negligible, in agreement with the experimental
ndings [83].
2.6. Characteristic Decomposition Modes of the Low-Energy Excited
Alkane Molecules
The photodecomposition of n-alkanes at excitation energies slightly above the absorption
onset involves both CH and CC bond decompositions [18]. The dominant process is the
CH scission, U(H
2
) = 0.80.9, and the contribution of CC decomposition is small. In
the photolysis of cyclohexane, cycloheptane, cyclooctane, and cyclodecane, however, only
hydrogen evolution was observed [U(H
2
) c 1]. In the photolysis of cyclopentane, CC
decomposition also takes place with low yield. In the reactions of cyclopropane and
Figure 3 Schematic representation of formation and decay of alkane S
1
excited molecules.
Copyright 2004 by Taylor & Francis Group, LLC
cyclobutane, the ring decomposition is the decisive process; the yield of CH decom-
position is very low. The branchings on the carbon skeleton in isoalkanes and alkyl
cycloalkanes increase the yield of CC decompositions: the highly branched alkanes such
as 2,2-dimethylbutane or 2,2,4-trimethylpentane decompose mainly by CC bond scissions
at the branch points.
H
2
and H Elimination
The mechanism of H
2
formation involves both unimolecular H
2
elimination, when both of
the H atoms of a hydrogen molecule originate from the same alkane molecule, and H atom
elimination with subsequent H atom abstraction reaction. We show the mechanism on the
example of cyclohexane photolysis (hm = 7.6 eV) [91]:
c C
6
H
12
!
hm
c C
6
H
12
* 4
c C
6
H
12
!c C
6
H
10
H
2
UH
2
u
0:85 5
!c C
6
H
.
11
H UH 0:14 6
!c C
6
H
12
hmV U
f
0:01 7
H c C
6
H
12
!c C
6
H
11
.
H
2
8
2c C
6
H
11
.
!
k
d
c C
6
H
12
c C
6
H
10
9a
!
k
c
c C
6
H
11
2
9b
The H atom detachment in Eq. (6) is followed by H atom abstraction from another
cyclohexane molecule in Eq. (8) and the cyclohexyl radicals disappear in disproportiona-
tion and combination reactions giving in nearly equal amounts cyclohexene and dicyclo-
hexyl end products. In the liquid phase, the secondary decomposition of the primarily
formed energy-rich reaction products is of minor importance because of the eective
collisional deactivation. This is in contrast to the gas phase reaction, where the primarily
formed products readily undergo decomposition in the absence of deactivation (at low
pressures), e.g.:
c C
6
H
12
* !
H
2
c C
6
H
10
!C
2
H
4
C
4
H
6
5a
The yield of unimolecularly formed cyclohexene, and by that the yield of unim-
olecular H
2
elimination (quantum yield in photolysis or G value in radiolysis), X(H
2
), may
be calculated from the yields of alkene and dimer dehydrogenation products by the
relation [91,92]:
XH
2
u
Xalkene k
d
=k
c
Xdimer 10
where k
d
/k
c
stands for the disproportionation-to-combination ratio of the relevant radicals.
Using a value of 1.1 for this ratio and the yields of cyclohexene and bicyclohexyl U(H
2
)
u
=
0.85 and G(H
2
)
u
=1.31.5 has been calculated. The molecular and atomic channels can also
be distinguished by applying the isotope dilution technique, introduced by Dyne [93] and
Dyne and Denhartog [94].
Copyright 2004 by Taylor & Francis Group, LLC
The quantum yields and G values of H
2
elimination for a larger group of alkanes are
collected in Table 4; the values were mostly determined by using Eq. (10). Because in the
photolysis and radiolysis of the n-alkanes shown in the table various kinds of radicals are
produced simultaneously (e.g., n-propyl and sec-propyl from propane), the weighted
averages of several k
d
/k
c
values were used in Eq. (10). The ratios can be determined by
suppressing the radiolytic alkene and dimer yields in the presence of radical scavenger
(e.g., I
2
):
k
d
k
c
DGalkene
DGdimer
11
As shown in the table, during photolysis of n-alkanes and C
5
C
10
cycloalkanes,
U(H
2
)
u
is 0.70.9, slightly exceeding the values of 0.60.7 measured at low photon energies
in the gas phase [18,19]. Cyclopropane and cyclobutane undergo H
2
elimination to a
negligible extent: U(H
2
)
u
c0.02. The value of U(H
2
)
u
in branched alkanes is much smaller
than in normal and cyclic alkanes. The results obtained in the gas phase show that the
yields decrease with the increasing branching [18]: for instance, in the series of C
5
H
12
isomers, the isomer with maximum branching, neopentane, essentially does not eliminate
hydrogen [16].
In connection with the gas-phase studies, it was rst suggested at the beginning of the
1960s [95,96] that the H
2
elimination from the lower-molecular-mass alkanes takes place
mainly from a single carbon atom (1,1 elimination) and not from two neighboring ones
Table 4 Quantum Yields and G Values of Unimolecular H
2
Elimination and the G Values of S
1
Molecule Yields Estimated from the Yields of Unimolecular Hydrogen Elimination
Alkanes U(H
2
)
u
Energy, eV G(H
2
)
u
G(S
1
)
Propane 0.76 8.4 1.4 1.8
n-Pentane 0.82 8.4 1.4 1.7
n-Hexane 0.9 7.6 1.48 1.6
n-Heptane 0.9 7.6 1.31 1.5
n-Octane 0.92 7.6 1.4 1.5
Cyclopentane 0.88 7.6 1.6 1.8
Cyclohexane (CH) 0.85 7.6 1.3 1.5
Cycloheptane 0.89 7.6 1.3 1.6
Cyclooctane 0.87 7.6 1.9 2.2
Cyclodecane 0.91 7.6 2.0 2.3
2,2,4-Trimethylpentane 0.08 8.4
Methylcyclopentane 0.46 7.6
Methylcyclohexane 0.7 7.6
Ethylcyclohexane 0.63 7.6
cis-1,2-DimethylCH 0.28 7.6
trans-1,2-DimethylCH 0.21 7.6
cis-1,3-DimethylCH 0.5 7.6 11.5 23
trans-1,3-DimethylCH 0.53 7.6 0.51 12
cis-1,4-DimethylCH 0.45 7.6
trans-1,4-DimethylCH 0.45 7.6
Source: Refs. 18, 29, and 92.
Copyright 2004 by Taylor & Francis Group, LLC
(1,2 elimination). In liquid-phase radiolysis experiments, 1,1 elimination was also dem-
onstrated by deuterium-labeling experiments [97,98]:
-CH
2
-CH
2
-*!
H
:::::::
H
j j
CH
::::
CH !
H
2
1;2 elimination
CHjCH 12
-CH
2
-CH
2
-* !
H
2
C CH
2
!
1;1 elimination
CHjCH 13
As a result of 1,1 elimination, a carbene intermediate remains that quickly rearranges to
alkene. During the photolysis and radiolysis of cycloalkanes with 710 carbon atoms,
apart from the formation of the usual cycloalkene products, dehydrogenation into
cycloalkanes with interesting bridged rings was also observed [99102]. Similar products
were also found in catalytic reactions [103]. The cross-bridged products are formed by
stabilization of the carbene in transannular insertion. During photolysis of C
7
C
10
cycloalkanes, the distribution of products remaining after H
2
elimination in several
solvents [99] agreed with the distribution of products formed during stabilization of
carbenes produced in an independent way, by the UV photolysis of cycloalkanone p-
tosylhydrazones under aprotic conditions, e.g., for the C
8
ring:
Photolysis of
diazo compound (%)
54 43 3
Radiolysis (%) c60 c37 3
7.6-eV
photolysis (%)
57 41 2
Copyright 2004 by Taylor & Francis Group, LLC
It means that the H
2
elimination from these compounds takes place exclusively through
1,1 elimination [99,100].
Alkane and Alkyl Radical Elimination
In the alkane elimination, a smaller alkane and an alkene molecule are formed in the decay
of excited molecule without intermediate radicals. A few examples of this reaction are
shown in Table 5. Homolytic split of alkane molecules to two alkyl radicals and the
subsequent reactions of radicals may give the same products. Radical-scavenging techni-
ques or deuterium-labeling studies may be used to distinguish the two processes. Although
liquid phase elimination has rarely been investigated, a great amount of data are available
in the literature concerning the reaction in the gas phase. For example, Ausloos et al. [16]
using partially deuterated compounds, showed that in the photolysis of isopentane during
methane elimination (from position 1) the H atom is taken o with comparable
probability from positions 2 and 3:
5 CH
3
j
CH
3
---
1
CH
2
- CH
2
3
- CH
3
4
Alkane elimination has a low yield during the photolysis of liquid n-alkanes (e.g., n-
pentane [104,106]). This reaction takes place with high yield only for branched alkanes
where it is likely to be a main primary-decomposition step [105,107].
Isomerization of Cycloalkanes into Aliphatic Alkenes
Alkylcyclopentanes and cyclohexanes rearrange to aliphatic alkenes during both photol-
ysis and radiolysis [108113]. The double bond in the product aliphatic alkenes can be
found connected to one of the carbon atoms taking part in ring opening. The derivation of
Table 5 Quantum Yield of Alkane Elimination in the Photolysis
of Some Liquid Alkanes
Propane, 8.4 eV [105] U
C
3
H
8
* ! C
2
H
4
+ CH
4
0.1
n-Pentane, 8.4 eV [104]
C
5
H
12
* ! C
2
H
4
+ C
3
H
8
0.06
! C
3
H
6
+ C
2
H
6
0.05
! C
4
H
8
+ CH
4
0.1
Isobutane, 8.4 eV [105]
C
4
H
10
* ! C
4
H
8
+ CH
4
f
0.5
2,2-Dimethylpropane, 7.6 eV [107]
C
5
H
12
* ! iso-C
4
H
8
+ CH
4
0.57 F 0.14
2,2,4-Trimethylpentane, 7.6 eV, 8.4 eV [104,106]
C
8
H
18
* ! iso-C
4
H
8
+ iso-C
4
H
10
0.36
! C
3
H
6
+ neo-C
5
H
12
0.05
! C
7
H
14
+ CH
4
0.17
Copyright 2004 by Taylor & Francis Group, LLC
the alkenes formed is shown schematically on the example of ethylcylohexane:
From this compound as well as from other alkylcyclohexanes the yield of ring-opening
products is relatively small, about U = 0.10.4, and G = 0.31.6 [108,110] (Table 6), while
usually the main decomposition process is the hydrogen formation, which leaves the cyclic
structure intact. Here, and with the other alkylcyclohexanes and alkylcyclopentanes, the
scission of the ring to smaller molecular mass alkenes and cyclopropane derivates was
detected with very low yield.
It was an important observation fromthe point of viewof the opening mechanismthat
in dimethylcyclohexane isomers the yield of transformation from cis into trans and from
trans into cis also takes place. In the 7.6-eV photolysis of 1,2-dimethylcyclohexanes the cis
formisomerizes to trans with a yield of U(c!t) =0.12, while the reversed reaction has a yield
of 0.07. In the photolysis of 1,3-dimethylcyclohexanes U(c!t) = 0.07 and U(t!c) = 0.15,
Table 6 Quantum Yields and G Values of C
n
H
2n
Aliphatic Alkenes in the 7.6-eV Photolysis
and Radiolysis of C
n
H
2n
Cycloalkanes
C
n
H
2n
cycloalkane U (aliphatic alkenes) G (aliphatic alkenes)
Cyclopentane 0.04 0.88
Cyclohexane No product 0.34
Methylcyclohexane 0.10 0.62
cis-1,2-Dimethylcyclohexane 0.37 1.50
trans-1,2-Dimethylcyclohexane 0.39 1.60
cis-1,3-Dimethylcyclohexane 0.10 0.71
trans-1,3-Dimethylcylohexane 0.10 0.60
cis-1,4-Dimethylcyclohexane 0.16 0.68
trans-1,4-Dimethylcyclohexane 0.17 0.73
In radiolysis a fraction of C
n
H
2n
alkenes forms in bimolecular radical reactions.
Source: Refs. 18, 29, 108, 110, and 111.
Copyright 2004 by Taylor & Francis Group, LLC
while for the 1,4-dimethylcyclohexanes U(c!t) = 0.13 and U(t!c) = 0.06 [108]. Similar
phenomena were observed during the radiolytic processes as well [110,112]. In dimethylcy-
clohexanes the methyl groups as substituents can be of equatorial position, falling into the
plane of the ring, or of axial position, roughly perpendicular to this. If the chair form of
cyclohexane rearranges intoanother chair form, the axial and equatorial substituents change
their positions, i.e., the axial position turns into the equatorial one and vice versa in the
dimethyl-cyclohexanes. Both (e,e)X (e,a) and (a,a)X (e,a) transformations can take place
through the previous dissociation of at least one bond of the molecule.
In all cases when photolytic isomerization into aliphatic alkanes was observed it was
found that the quantum yield of primary decomposition calculated from the measured end
products is slightly smaller than unity.
On the basis of these results it is assumed that isomerization to aliphatic alkenes
proceeds through alkyl biradicals whichapart from alkene formationcan reclose again
into the initial cycloalkanes giving in this way products not measured by the usual
techniques. In 1,2-, 1,3-, and 1,4-dimethylcycloalkanes, reclosing can also result in the
opposite isomer, thereby providing a possibility for the observation of the process. The
biradical may also fragment yielding ethylene and a smaller biradical. The cyclopropanes
observed among the products may form in stabilization of the 1,3-biradicals.
Correlation Between Photophysical and Photochemical Data
Because uorescence plays a very small role in the depopulation of alkane excited
molecules the sum of the quantum yields of the chemical decompositions in the thermally
activated and nonactivated channels is practically unity: U(S
l
!S
x
) + U(S
l
!T
n
) is 1.
Using Eq. (3), the temperature dependencies of the product yields formed in the activated
and nonactivated channels have the following forms:
US
1
! S
x
A expE
a
=RT
k
0
A expE
a
=RT
15
US
1
! T
n
1 US
1
! S
x
16
According to these equations, the product formation in the S
1
!S
x
activated channel is
increasing with the increasing temperature, whereas the product formation in the S
1
!T
n
nonactivated channel is decreasing. Xenon is expected to have an opposite eect on the
yields. The temperature and Xe eect studies showed that the H
2
elimination is related to
the activated channel, and the H atom elimination to the nonactivated [67,79,81,113]. In
the 7.6-eV photolysis of n-decane between 20 and 45jC, U(H
2
)
u
increased from 0.62 to
0.82, whereas U(H) decreased from 0.4 to 0.08. Xe in a concentration of 0.2 mol dm
3
decreased U(H
2
)
u
and increased U(H) by
f
20% [67]. In cycloalkane photolysis, the yields
of ring-opening products showed a denite nonactivated character: in the temperature
range mentioned, the yield of linear octenes from ethylcyclohexane decreased from 0.32 to
0.13; Xe increased the yield by
f
25% [113]. From the point of view of activated and
nonactivated character the CC decompositions in aliphatic alkane photolysis were not
investigated. However, based on the values of the yields, the alkane elimination was
tentatively identied as a singlet decomposition, and the homolytic split to two alkyl
radicals as a triplet decomposition [81]. In Table 3 we show for some selected alkanes the
quantum yields of the decompositions in the activated and nonactivated channels at 25jC
calculated using Eqs. (15) and (16). The table also contains the yields of unimolecular
(hydrogen and alkane) eliminations and the yields of radical-forming (H atom elimination,
CC scission to radicals or biradicals) reactions. The correlation between the yields of
Copyright 2004 by Taylor & Francis Group, LLC
activated decay channel and the unimolecular decompositions and between the yield of the
nonactivated decay and the yield of radical/biradical forming reactions supports the
identications of the individual decomposition types.
The very small temperature dependence in the decay of excited cis- and trans-decalin
molecules may be interpreted in terms of absence of thermally activated decay. The
chemical decomposition of excited decalin molecules yields a large number of products
making the identication of the individual decomposition channels and the determination
of their yields rather complicated [81,114]. However, from the distribution of products it is
certain that the yield of H
2
elimination is low, and the majority of H
2
formation, U(H
2
) =
0.42 and 0.50 for cis- and trans-decalin, respectively, originates via H atom formation and
H abstraction. Another important reaction is the CC bond decomposition yielding
probably a biradical. The formation of C
4
cyclo-C
6
products, cyclodecenes, and partly
also cis!trans/trans!cis isomerizations observed in 7.6-eV photolysis is probably due to
the stabilization of these biradicals. Both the H-atom- and the biradical-forming reactions
occur with a high probability at the tertiary carbon atoms: previously these reactions were
attributed to nonactivated (triplet) decompositions. Therefore, the absence of temperature
dependence in the excited molecule decay is probably because of a strong localization of
the excitation energy between the two tertiary carbons where there is no possibility for
molecular elimination reaction.
2.7. Energy Dependence of the Primary Decomposition
Similarly to the uorescence quantum yields, the yields of individual primary decom-
position steps generally show considerable excitation energy dependence: the yields of the
unimolecular H
2
and alkane eliminations and also those of the radical-type decomposi-
tions show a continuous variation with photon energy [27,39,42,107,115]. In cyclohexane
photolysis the sum of the quantum yields of the two primary decompositions described by
Reactions (5) and (6) is practically unity between photon energies 7.6 and 11.6 eV: U(H
2
)
u
+ U(H) c 1. Close to the absorption threshold the H
2
elimination predominates: at 7.6
eV U(H
2
)
u
= 0.85. The yield decreases with the energy, U(H
2
)
u
= 0.78 and 0.64 at 8.4 and
10.0 eV, and it is only 0.4 at 11.6 eV. Thus, at the highest energy the yields of H
2
and H
elimination are comparable.
In the photolysis of cyclopentane [39,116] the energy dependencies of the H
2
elimination and H atom detachment are very similar to the energy dependencies of these
processes in cyclohexane photolysis. However, in cyclopentane photolysis ring decom-
position is also observed producing 1-pentene and ethylene + C
3
H
6
U(1-pentene) = 0.05,
U(C
2
+ C
3
) = 0.05 at 7.6 eV]. The yields of both ring decomposition reactions were found
to increase with the increasing photon energy at the expense of hydrogen formation.
The product yields in the photolysis of n-pentane, n-hexane, n-octane, and n-decane
also show decreasing importance of H
2
elimination and increasing importance of H atom
detachment with the increasing photon energy [27,117].
In contrast to the results obtained with n-alkanes and cycloalkanes, H
2
elimination is
found to be an unimportant process in the photolysis of neopentane [107]. At 7.6 eV the
methane elimination and direct CC bond cleavage to radicals are the predominant
processes:
neo-C
5
H
12
*!CH
4
iso-C
4
H
8
U 0:57 F 0:14 17
neo-C
5
H
12
*!CH
.
3
t-C
4
H
.
9
U 0:38 F 0:14 18
Copyright 2004 by Taylor & Francis Group, LLC
With an increase in photon energy, the importance of the radical-forming reaction increases
at the expense of the methane elimination process.
The energy dependence of the photodecomposition is attributed to an energy-
dependent relaxation of the higher excited molecules to the relaxed S
1
state. When the
photon energy is higher than the energy of the ionization onset ionic processes may also
contribute to the energy dependence. Although there are not enough measured data to
make a denite conclusion, from the published data it seems that when the energy is
increased above the ionization threshold the yield of the molecular-elimination-type
reactions continues to decrease and that of the radical-type reactions continues to increase.
Thus the chemical decomposition does not parallel the uorescence where sometimes an
increase in the uorescence yield was observed (Sec. 2.1). The energy dependence is higher
for the smaller molecules than for the larger ones.
The ionization, electroncation recombination, and the time scale of formation,
relaxation, and decay of alkane excited molecules were studied several times in pump-and-
probe-type subpicosecond laser experiments [49,118120]. Sander et al. [49] used 0.4-, psec
248.5-nm laser excitation in the double-photon mode. At their probe wavelength of 497
nm, both the excited molecules and the radical cations contribute to the light absorption.
These experiments may supply direct data on the time scale of relaxation. After the fast
rise of absorbance during the laser pulse there is a 0.4- to 1.4-psec ultrafast component of
absorption decay that they attribute to fragmentation from higher singlet excited states,
which competes with internal conversion to the S
1
state of the parent molecules. The fast
component is followed by a 5- to 15-psec slower component in which for some alkanes there
is a re-rise of absorbance. They attribute the second component to geminate recombination.
A third component is due to the S
1
molecule decay. It should be noted, however, that other
pump-and-probe experiments, e.g., the fsec transient measurements of Long et al. [119],
suggest much faster geminate recombination than those proposed by Sander et al. Thereby,
these experiments contradict the view of Sander et al.
It would be elegant to nish the part on photophysics and photochemistry of liquid
alkanes by giving a picture that unies the temperature- and energy-dependence results
obtained in uorescence and photodecomposition studies. However, the spectroscopic
information available for alkane molecules is not sucient to identify the exact excited
states involved in the radiative and nonradiative processes [55]. Because of the lack of
information, there are dierent views on the positions and identities of excited states
involved [52,55,83,121,122].
In Fig. 4 we try to explain and, where it is possible, to unify the dierent views using
the usual simplied two-dimensional potential energy diagram for CH bonds. The
potential energy curve of S
1
along the length of a CH bond is displaced to higher R
CH
nuclear distances (reaction coordinate) to explain the strong nuclear distortions and large
Stokes shift reected by uorescence. This is in contrast to the views of Plotnikov [122]
and Flamigni and Orlandi [83]: they assumed that the potential energy surfaces of the S
0
and S
1
states are essentially identical, except that S
1
is shifted to higher values by
f
7 eV,
because, in their opinion, the excitation energy of S
1
is delocalized over the entire
molecule. Other authors, however, suggest strongly localized alkane excited states. From
the S
1
potential energy curve, by passing through a small E
a
energy barrier S
x
can be
reached, which, according to the suggestion of Orlandis group, may give the carbene +
H
2
decomposition products. Wichramaaratchi et al. suggest a dissociative state denoted by
S
2
(this is shown in the gure), which crosses S
1
, and the crossing point lies near a very low
vibrational level of S
1
, thereby promoting S
1
!S
2
IC. S
2
is assumed to be responsible for
two reaction channels: one leading to molecular H
2
and other leading to atomic H [55]. If
Copyright 2004 by Taylor & Francis Group, LLC
H atom elimination is of lower importance from this state, S
1
!S
2
IC in the mechanism
suggested by Wichramaaratchi et al. is practically identical with S
1
!S
x
used in the
terminology of Orlandis group. The authors of Refs. 55, 83, 121, and 122 all agree that
there is a dissociative triplet state, T
n
, which can easily be populated by ISC from S
1
. We
show the potential energy curve of this state (T
n
) as it has been suggested by Orlandi et al.
[83,121]. This state gives only radical decomposition products. In order to explain the
results of the biphotonic sensitization experiments (Sec. 2.1) another triplet state
f
1 eV
below S
1
(T
1
) is also needed. Most probably, this triplet is also dissociative.
The proposed mechanism may explain such excited-state characteristics as the
temperature dependence of lifetime or k
ex
dependence of the uorescence intensity at
low excitation energies. However, in order to explain the energy dependence of the
photodecomposition at high energies at least one more dissociative state should be
included in the mechanism, which decompose to radicals.
3. ENERGY TRANSFER
Energy transfer from an electronic excited state of an alkane molecule to a solute molecule
was rst suggested 50 years ago [123]. Since then over 100 papers discussed the various
Figure 4 Two-dimensional potential energy curve for CH bond decomposition.
Copyright 2004 by Taylor & Francis Group, LLC
aspects of the transfer, e.g., the properties of excited states involved in the transfer
(reaction diameter, lifetime) or the nature of the transfer reaction (diusional type with
short-range contact interaction, Forster type). An important practical aspect of the subject
is the protecting eect exerted by dierent additives on radiation induced decomposition
of saturated hydrocarbon systems (lubricants, insulators, polyethylene). Most frequently,
the transfer in liquid cyclohexane was studied, but a few papers discussed the transfer
reactions also in decalins, 2,2,4-trimethylpentane, and n-alkanes. Mostly the low solute
concentration region was studied; however, by photoexcitation or radiolysis in some
alkanealkane mixtures the product formation, and by that the energy transfer, was
investigated in the entire concentration range [124126]. The techniques used to obtain
transfer data include stationary photolysis or radiolysis with detection of nal products or
uorescence intensity, pulsed photon or electron irradiation, and detection of transient
uorescence or absorbance.
3.1. Energy Transfer from the S
1
Excited State
Based on the magnitude of the transfer rate coecients and other transfer characteristics it
is obvious that in the energy transfer from alkane excited states to solutes long-range
interactions play a minor role (Forster-type energy transfer) [70,127]. In the Forster-type
energy transfer, the overlap of the donor emission and the acceptor absorption spectra
determines the transfer rate coecients. However, in the quenching of alkane excited states
no correlation was found between the overlap integrals and the transfer rate coecients:
the transfer is suggested to take place because of short-range contact interactions.
Therefore, in the transfer mutual diusion of the reactants to each others proximity
has a determining role. The diusional character is also supported by the temperature
dependence of the transfer rate coecient; this dependence practically agrees with the
temperature dependence of the diusion [42,62,64,128]. However, the simple Stokes
EinsteinSmoluchowski equation (SES), k
d,SES
= 8RT/3000g, where g is the solvent
viscosity at temperature T, fails to describe properly the values of transfer rate coecients
[42,129131]. In alkane excited state quenching, where the lifetimes are short (s c1 nsec),
and the quencher concentrations are high, c = 0.010.2 mol dm
3
, perhaps static
quenching eects and also transient quenching eects play a role in the transfer in
addition to the usual diusional quenching [131].
When the reaction radius is larger than the sum of the molecular radii of the reacting
molecules, there is an instantaneous quenching by the quencher molecules that reside
inside the reaction radius of the excited molecule, rV (also called kinetic distance), in the
moment of excitation. The term that takes into account instantaneous quenching is called
static quenching term and usually has the form of exp(NV
1
c), where c is the quencher
concentration and N is Avogadros number. V
1
is dened as the volume around an excited
molecule, where instantaneous quenching occurs:
V
1
4p
3
rV
3
r
3
19
r (= r + r
s
) is the collision distance, i.e., the sum of the molecular radii of the solute and
solvent molecules.
The quencher molecules that are near the excited molecule in the moment of
excitation have a much higher probability for depleting the excited molecules than the
other quencher molecules. For that reason the rate coecients used to describe the decay
of excited molecules often have a so-called transient term, k
0
r
1
(pDt)
1/2
: the transient term
Copyright 2004 by Taylor & Francis Group, LLC
has some importance only at short times (t), much shorter than the natural lifetime of the
excited molecule.
The third term k
0
is the usual diusion controlled rate coecient: k
0
= 4pNrVD, where
D = D
S
+ D
s
is the sum of the diusion coecients of the solute and solvent molecules.
In the literature there are several, mostly just slightly dierent, equations that describe
the rate coecient of the diusion controlled reactions: these equations are usually based on
the solutions of Fick II diusion law assuming that the reaction probability at contact
distance is 1. Andre et al. [131] used the following equation to describe the time dependence
of excited molecule concentration [RH*] produced by an innite excitation pulse:
RH* RH*
0
expNV
1
cexpk k
0
c1 2rVpDt
1=2
t 20
In Eq. (20), k is the rate coecient of the excited molecule decay without quencher (k =s
1
).
Using this equation in Fig. 5, we show the time dependence of the relative excited
cyclohexane molecule concentration in a solution containing 0.05 mol dm
3
CCl
4
. In order
to show the eect of static quenching we chose a large reaction radius of rV = 1.3 nm. It is
Figure 5 Decay of excited cyclohexane molecules in the pure liquid and in solution containing 0.05
mol dm
3
CCl
4
. Parameters used for the calculations: k =1.05 10
9
sec
1
[67], rV =1.3 nm[64], r =
0.7 nm, and D = 3 10
9
m
2
sec
1
[59].
Copyright 2004 by Taylor & Francis Group, LLC
obvious from Fig. 5 that static quenching is important in stationary experiments; however,
when the rate coecient is calculated from decay measurements, the static quenching,
because of its instantaneous nature, is unobserved. In pulsed experiments, the quenching
rate coecient has the form of Eq. (21):
k
d
k
0
1 rVpDt
1=2
21
For stationary experiments, Andre et al. [131] suggest the form:
k
d
NV
1
s
1
k
0
1 rVDs
1=2
22
In the usual treatments there is only one parameter to be tted, i.e., the reaction
radius. In practice three cases should be distinguished:
1. The kinetic distance exceeds the collision distance, r > r, which is typical for
Forster-type energy transfer, but as mentioned before there is evidence against this
type of transfer in alkane systems. The reaction radius may also be high if the
transfer takes place froma large-orbit Rydberg state. In early studies using steady-
state quenching or uorescence decay experiments, high kinetic distances in the
order of 1.5 nmwere suggested for cyclohexane and the decalins [42,64,129]. Most
of the later studies agree, however, that the largest kinetic distance values
practically agree with the contact distances (
f
0.70.8 nm) or they are just slightly
larger. This result suggests that the relaxed excited states of these alkanes do not
have large orbit character and that static quenching has a minor role in the
transfer.
2. When rV cr (=0.8 F0.2 nm) the reaction probability is 1 at the contact distance.
This condition is fullled for good quenchers such as CCl
4
, SF
6
, uorinated
hydrocarbons, and in some experiments alkenes or benzene [70,128,130,132].
When rV V r, the term NV
1
= 0.
3. When rV < r not all of the encounters lead to quenching. The p reaction
probability may be calculated as the ratio of the measured and calculated rate
coecient using rV = r: p = k
q
/k
d
. Reaction probability values determined for
larger groups of quenchers in cyclohexane or n-hexadecane have been reported
in Refs. 70 and 130.
For the description of partially diusion-controlled energy-transfer reactions often
the reaction scheme and theory introduced by Rehm and Weller [133] as well as by Balzani
et al. [134] are applied:
RH* Q X
k
d
k
d
RH*
: : : : :
Q X
k
e
k
e
RH
: : : : :
Q* X
k
d
k
d
RH
# 1=s
Q*
# 1=s
q
23
where k
d
and k
d
are the rate coecients of the formation (diusion-controlled rate
coecient) and dissociation of the encounter pair, respectively, and k
e
and k
e
represent
the rate coecients of the energy transfer step and its reverse, respectively. For the energy-
transfer step one can write:
k
e
k
e
expDG=RT 24
k
e
k
0
e
expDG
=RT 25
DG is the free energy change of the forward energy transfer and k
e
0
and DG
+
are the
preexponential factor and the free energy of activation of its rate coecient, respectively.
Copyright 2004 by Taylor & Francis Group, LLC
DG can be approximated by the energy dierences of excited states of donor and acceptor
molecules:
DGcDE ERH*S
1
; RHS
0
EQ*S
1
; QS
0
26
DG
+
is given by the RehmWeller theory:
DG
DG
2
DG
2
2
k
4
2
" #
1=2
27
k is often an adjustable parameter that is connected with reorganization of the inner
coordinates of the reactants and reorganization of solvent molecules. For a given donor
molecule and chemically similar acceptor molecules, k is assumed to be constant. When k
is small and in Eq. (28), which is developed to describe the transfer rate coecient
assuming 1/s
q
>>k
d
c, the second term in brackets, (k
d
/k
e
0
) exp(DG
+
/RT), can be
neglected, we get the so-called Sandros equation (29):
k
q
k
d
1 exp
DG
RT
k
d
k
0
e
exp
DG
RT
1
28
k
q
k
d
1 expDG=RT
29
For excited alkane donors k is assumed to be high. Eq. (28) was often used to evaluate the
molecular structure dependence of the transfer coecients [69,70,127,135,136]. In Fig. 6,
we show the rate coecients of the energy transfer from excited cyclohexane molecules to
alkanes, alkenes, and benzene as a function of the DE energy dierence based on the
experiments of Hermann et al. [135]. The solid lines were calculated with k
d
= 2 10
10
mol
1
dm
3
sec
1
and k = 0.5 or 1 eV. Using a similar treatment to describe the energy
transfer from excited cyclohexane molecules, Busi and Casalbore [127] obtained the best t
with k
d
= 3.5 10
10
mol
1
dm
3
sec
1
and k = 0.55 eV. For n-hexadecane solutions a
lower value of k
d
was applied,
f
1 10
10
mol
1
dm
3
sec
1
[70]. The calculations again
strongly support that the transfer occurs with a collision mechanism and the free energy
dierence determines the rate coecient of the energy-transfer step.
Seemingly, a dierent mechanism determines the rate of energy transfer to such
inorganic molecules as N
2
O, CCl
4
, SF
6
, and organic molecules with halogen atoms. For
these additives, k
q
may be high while DE is low or negative [130,135]. Those molecules
were found to be good quenchers that were good electron scavengers in radiolysis [42].
This correlation suggests that electron transfer and energy transfer may have similar
mechanism, e.g.:
RH* B ! RH B* ! neutral products 30
RH* B ! RH
! neutral products 31
Reactions (30) and (31) may give the same products. In (31) the polarization energy
decreases the energy demand for temporal charge separation and it can be exothermic
when B has a considerable electron anity. For aromatic hydrocarbon quenchers (e.g.,
anthracene) such mechanism leads to dissipation of the excitation energy on the vibra-
tional levels. When the quencher molecules contain Cl or Br atom in the intermediate step,
Cl
or Br
! C
6
H
5
CH
.
2
Cl
32
Copyright 2004 by Taylor & Francis Group, LLC
As nal products of the transfer HCl, C
6
H
5
CH
3
, C
6
H
5
CH
2
-c-C
6
H
11
and (C
6
H
5
CH
2
)
2
were
found. The same products are expected in the electron-scavenging reaction [130].
3.2. Energy Transfer from Higher-Energy Excited States
The energy transfer from excited alkane molecules to 2,5-diphenyloxazole (PPO) was
frequently investigated, often with unusual results. For example, Kimura and Hormes
investigated the transfer using for excitation 3.310.5 eV photons from a synchrotron
source. They reported extremely eective energy transfer at energies 7.81 and 9.67 eV in
cyclohexane. The transfer at 7.81 eV was tentatively ascribed to energy transfer from the
eighth peak of the Rydberg series with a diameter of
f
3 nm (static quenching). They did
not observe similar ecient transfer in n-heptane [28]. Lipsky et al. also observed ecient
transfer in several saturated hydrocarbons at photon energy 7.7 eV. According to Wang
Figure 6 Energy transfer rate coecients from excited cyclohexane molecules to solutes as a
function of the DE energy dierence: (1) isooctane, (2) n-heptadecane, (3) cyclooctane, (4) trans-1,2-
dimethylcyclohexane, (5) isopropylcyclohexane, (6) cis-1,2-dimethylcyclohexane, (7) decalin, (8) 1-
decene, (9) 2-hexene, (10) tetramethylethylene, (11) benzene. (From Ref. 135.)
Copyright 2004 by Taylor & Francis Group, LLC
et al. [137] and Krishna and Lipsky [138], the transfer takes place from some unidentied
ion-pair state of alkanes.
4. FORMATION AND YIELD OF ALKANE EXCITED STATES IN RADIOLYSIS
The yield determined in a certain type of experiment usually strongly depends on the
assumptions made about the formation mechanism. In the older literature, the excited
molecules were often assumed to be produced solely in neutral excitations [127,139143]
and energy-transfer experiments with SternVolmer-type extrapolation (linear concen-
tration dependence) were used to derive G(S
1
). For instance, by sensitization of benzene
uorescence, Baxendale and Mayer established G(S
1
) = 0.3 for cyclohexane [141]. Later
Busi [140] corrected this value to G(S
1
) = 0.51 on the basis that in the transfer, in addition
to the uorescing benzene state S
1
, the S
2
and S
3
states also form and the S
2
!S
1
and
S
3
!S
1
conversion eciencies are smaller than 1. Johnson and Lipsky [144] reported an
eciency factor of 0.26 F 0.02 per encounter for sensitization of benzene uorescence via
energy transfer from cyclohexane. Using this eciency factor the corrected yield is G(S
1
)
= 1.15. Based on energy-transfer measurements Beck and Thomas estimated G(S
1
) = 1
for cyclohexane [145]. Relatively small G(S
1
) values were determined in energy-transfer
experiments for some other alkanes as well: n-hexane 1.4, n-heptane 1.1 [140], cyclo-
pentane 0.07 [142] and 0.12 [140], cyclooctane 0.07 [142] and 1.46 [140], methylcyclohexane
0.95, cis-decalin 0.26 [140], and cis/trans-decalin mixture 0.15 [142].
A basic problem of the energy-transfer experiments is that the quenchers used may
also react with electron or cationic species, in addition to the excited molecules. If the
excited molecules also form after charge recombination (which is now unambiguously
established), the quenchers may considerably hinder the formation of excited molecules
[8]. Comparison of the results obtained in this manner with those obtained via other
techniques shows that the solute technique strongly underestimates the G(S
1
) value. It
should be mentioned that in most of the sensitization experiments unrealistically high K
SternVolmer constants were obtained [141,142].
If the excited molecule formation is mainly because of electroncation recombina-
tion the concentration dependence of quenching of the precursor charged species may
follow some form of the modied WarmanAsmusSchuler (WAS) semiempirical equa-
tion [8], e.g.:
GScavenged S
1
precursor f
s
G
fi
G
gi
ac
n
1 ac
n
33
where n is usually taken as
1
2
(square-root concentration dependence). In the equation G
and G
gi
are the yields of free and geminate ions, respectively, and a is an adjustable
parameter characteristic to the scavenger. The fraction of S
1
excited molecules formed in
recombination, f
s
, can be smaller than 1 mainly because of three reasons:
1. Some of the cations may fragment or undergo ionmolecule reactions before
neutralization. As shown in the older literature, the fragmentation has higher
yield in the radiolysis of the smaller and/or highly branched alkanes such as
neopentane or isooctane. Ionmolecule reactions, such as the H transfer, may
also reduce the S
1
yield:
c-C
6
H
12
.
c-C
6
H
12
!c-C
6
H
13
c-C
6
H
11
.
34
Copyright 2004 by Taylor & Francis Group, LLC
The ionmolecule reactions and the spin eects (see below) were detailed in the
recent review of Shkrob et al. [7]; therefore, we only briey mention these
reactions.
2. As discussed earlier, the recombination is nearly completed within a few pico-
seconds; during this short time the spin correlation is maintained and from the
originally singlet geminate ion pair singlet excited molecule forms. However, if
several ion pairs are in each others vicinity (in the spur) there is a high probability
for cross-recombination. In cross-recombination in 3:1 ratio triplet and singlet
molecules form. In a very large spur in the limit statistically
1
4
of the recom-
binations give singlet product. The singlet fraction was calculated several times by
using experimentally derived spur-size distribution [8,146148]. Numerous
computer simulations were also made on the singlet/triplet formation probability:
the theoretical calculations predict singlet formation probabilities in the 0.50.7
range.
3. The singlet excited molecules formed in charge recombination may not convert
with 100% eciency to S
1
as discussed in Sec. 2.1. This eect may be important
for smaller molecules; for larger molecules such as the decalins or n-dodecane the
eciency is probably very close to unity.
The experimentally obtained f
s
values show larger divergence: based on alkane
uorescence measurement. Walter et al. [148] and Luthjens et al. [65,128] published 0.8
0.9. Later the latter authors modied their value to 0.65 [149]. In solutions of isooctane,
cyclohexane, or n-hexane with naphthalene, Sauer and Jonah established this value as 0.5
F 0.1; this singlet formation probability is approximately constant during the decay of
charged species up to 70 nsec [150]. In nal product experiments f
s
= 0.34 [151], 0.53, and
0.47 [84] was estimated for cyclohexane, cis-decalin, and trans-decalin, respectively. The
low value for cyclohexane is due to the ionic reactions before recombination.
Most papers published during the last 20 years on the formation mechanism of S
1
alkane molecules agree that they form mostly or nearly exclusively in charge recombina-
tion [65,72,73,128,148153]. In pulse radiolysis done with
f
50-psec pulses, denite after-
formation of uorescence was observed for the larger alkanes such as the decalins. In the
radiolysis of cyclohexane, Sauer et al. used electron scavenger in subnanosecond uo-
rescence measurements to determine the formation route. Because the charge scavenging
occurs on the time scale of geminate recombination (a few picoseconds), while the excited
molecule quenching takes place during the excited molecule decay (
f
1 nsec), they
separated these eects [153]. Based on the eective initial inhibition of excited molecule
formation they concluded that less than 10% of the S
1
excited cyclohexane molecules were
formed in direct excitation. Low direct excitation contributions were suggested also for
other alkanes [7,128,148].
The uncertainty about the formation mechanism is eliminated when such method is
used for the determination of the G(S
1
) value, which is invariant to the formation route.
Such method is the comparison of the yields of those products, which are formed only in
one way in the reaction of the S
1
molecules. Such products come from the singlet decay
channel: uorescence photons, products of H
2
or alkane elimination. The triplet channel
due to the large triplet yield originating from other processes than the S
1
!T
n
ISC (e.g.,
direct triplet excitation, or charge recombination with triplet product) cannot be used for
the calculation of the S
1
yield. The calculations are based on the equation:
GX UX GS
1
35
Copyright 2004 by Taylor & Francis Group, LLC
G(X) and U(X) are the Gvalue and the quantumyield of the product used for the calculation,
respectively. Two possible sources of errors should be mentioned. One is the energy
dependence of the photodecomposition. In order to obtain intrinsic yield U(X) should
be measured at photon energies as close as possible to the absorption onset, or appropriate
corrections should be used [154,155]. Both photon emission and photodecomposition
studies show that the energy dependence is more severe for the smaller molecules than for
the larger ones. Therefore, G(S
1
) for the larger molecules can be estimated with higher
accuracy. The second source of error is that X may not solely form in S
1
molecule trans-
formation.
The uorescence of liquid alkanes is supposed to originate entirely from the relaxed
S
1
state. Walter and Lipsky [154], by measuring the uorescence yields of alkane solutions
irradiated with 165 nm photons or
65
Kr beta particles (E
max
= 0.67 MeV) relative to
benzene uorescence, determined the following yields: 2.3-dimethylbutane G(S
1
) < 1.3,
cyclohexane 1.41.7, methylcyclohexane 1.92.2, dodecane 3.33.9, hexadecane 3.33.9,
cis-decalin 3.4, and bicyclohexyl 3.5. After reinvestigating the intrinsic quantum yield of
cyclohexane uorescence, Choi et al. published G(S
1
) = 1.45 for this alkane in Ref. 155.
For trans-decalin a G(S
1
) value of 2.83.1 has been accepted [65,128,132]. The uncertain-
ties in the values reect the uncertainties in the intrinsic uorescence quantum yields.
As discussed in Sec. 2.6, the characteristic singlet photodecomposition mode of n-
alkanes and cycloalkanes is the H
2
-molecule elimination (Table 4) with U(H
2
)
u
= 0.70.9.
This process has a high yield also in radiolysis [9193,111]. Based on experiments with
deuterated hydrocarbons, Fujisaki et al. [156] and Shida et al. [157] reported values of
G(H
2
)
u
= 1.4 and 1.6 for the yield of unimolecular H
2
detachment occurring in the
radiolysis of propane and n-butane liquids. In cyclohexane experiments with c-C
6
H
12
/c-
C
6
D
12
mixtures, Dyne and Jenkinson suggested that 1020% of the hydrogen yield of
G(H
2
) = 5.6 forms in unimolecular elimination [158]. Using the same technique Nevitt
and Remsberg [159] published G(H
2
)
u
V 1.4 for cyclohexane. The unimolecular H
2
elimination yields for a large group of alkanes were determined based on the alkene
and dimer yields using Eq. (10) [91,92,111,160]. Although this method of G(H
2
)
u
determination, because of the comparable alkene and dimer yields, is less accurate than
the calculation of U(H
2
)
u
, we think that the precision of the G(H
2
)
u
values in Table 4 are
better than F 15%. The accuracy of the G(S
1
) values calculated by using the U(H
2
)
u
and
G(H
2
)
u
yields and Eq. (35) is around F 20%. In some works H
2
elimination from the
alkane radical cations were also suggested [7,161]:
c-C
6
H
12
.
!c-C
6
H
10
.
H
2
36
However, the yields are probably very low, and for most compounds this reaction does not
change considerably the determination of the G(S
1
) values by the H
2
-elimination method
[7]. When interpreting the G(S
1
) values obtained by this method the possibility should also
be considered that besides S
1
states, to some extent, higher-energy excited states might also
contribute to the unimolecular H
2
elimination. This would cause the G(S
1
) data calculated
using this method to be higher than the real G(S
1
) values. Such an assumption would,
however, contradict the general observation that with increasing excitation energy and,
consequently, decreasing lifetime of the excited state, molecular processes lose their
importance and atomic and molecular processes become predominant [160]. The yield
determined by this method for cyclohexane, G(S
1
) = 1.41.7 [91,92,111,160], agrees with
the yield obtained in uorescence experiments.
Copyright 2004 by Taylor & Francis Group, LLC
In the case of alkanes with tertiary or quaternary carbon atoms the characteristic
decomposition mode is usually the fragment alkane elimination (Sec. 2.6) and its yield can be
used to estimate G(S
1
) values. Based on the fragment alkane elimination yield Pitchuozhkin
et al. [162] calculated a singlet G value of 3.3 F 1 for 2,2,4-trimethylpentane.
We collected the singlet yields determined by comparing the uorescence or product
yields measured in photolytic and radiolytic experiments in Fig. 7. The G value of low-
energy excited states is increasing with the size of the molecules. The trend is by no means
unexpected because the density of excited states is known to increase with the size of
molecules, giving rise to a more ecient internal conversion from the higher excited states
to the rst excited level, S
1
. The f
s
singlet fraction of the charge recombination may have
size dependence also due to the decreasing yield of ion decomposition with the molecular
weight of ions [160].
Figure 7 G(S
1
) value as a function of the carbon atom numbers in the molecules. When more than
one measured value was published, we tried to select the most probable value. Alkanes: (1) propane,
(2) n-butane, (3) n-pentane, (4) cyclopentane, (5) n-hexane, (6) cyclohexane, (7) n-heptane, (8)
cycloheptane, (9) methylcyclohexane, (10) n-octane, (11) cyclooctane, (12) isooctane, (13) n-decane,
(14) cyclodecane, (15) cis-decalin, (16) trans-decalin, (17) n-dodecane, (18) dicyclohexyl, (19) n-
hexadecane. (From Refs. 18, 29, 65, 92, 148, and 155.)
Copyright 2004 by Taylor & Francis Group, LLC
Copyright 2004 by Taylor & Francis Group, LLC
Under high-LET irradiation conditions by using uorescence detection [87,88] or
nal product measurements [102,163165], a continuous decrease of the S
1
alkane excited
molecule yields was detected with increasing LET. Holroyd et al. [87], in their soft x-ray
experiments, attributed this decrease to the fast quenching of excited molecules by radicals
in the spur. LaVerne and coworkers [102,163165] investigated the distribution of nal
product of cyclopentane, cyclohexane, and cyclooctane radiolysis in the presence and
absence of I
2
radical scavenger in a very wide LET range. They found that the proportion
of H-atom-forming reaction is increasing with increasing LET at the expense of other
hydrogen-forming reactions, e.g., H
2
molecule elimination. The shift was attributed to
changes in charge recombination from basically geminate recombination in low-LET
radiolysis to predominantly cross-recombination in high-LET heavy-ion radiolysis. In the
high-LET limit the S
1
yields were estimated as G c1.2, which is about one fourth of the G
= 45 ion pair yield.
5. THE ORIGIN OF STRUCTURE EFFECT IN ALKANE RADIOLYSIS
The products that form in photolysis and radiolysis of liquid alkanes at room temperature
can be divided more or less arbitrarily into two groups: products reecting primary CH
or primary CC ruptures. On this ground, one is able to calculate the U value or the G
value of molecules activated by the photon or ionizing radiation absorption and under-
going primary CH or CC bond scission [2,160]. There are several examples for the
calculation of the yields in Ref. 2. By investigating the low-LET radiolysis of about 30
alkanes the overall yield of these two processes was found to be approximately constant: G
= 6.5 F 1. The eect of chemical structure can be observed particularly in the ratio of the
competing primary CH and CC bond rupture reactions. This is obvious from Fig. 8a,
where the G(H
2
) values are displayed as a function of the G values of the primary CC
bond rupture reactions. In contrast to the decomposition of n-alkanes and cycloalkanes,
which yield mostly hydrogen and dehydrogenation products, some branched alkanes, e.g.,
neopentane and 2,2,4-trimethylpentane, decompose mainly by CC bond rupture.
In photolysis of liquid alkanes (Fig. 8b) the competition and structure eect are more
apparent than during radiolysis: the highly branched alkanes decompose nearly exclusively
by CC bond rupture. Since, during radiolysis of many alkanes, the decompositions
originating from low-energy excited states can make a signicant contribution (2555%)
to the nal product formation, and the reactions from this state show higher selectivity, a
considerable part of the structure eect and selectivity in radiolysis is due to the
decomposition of low-energy excited molecules. Another important source of the structure
Figure 8 Connection between primary CH and CC bond ruptures during radiolysis and
photolysis. Alkanes: (1) propane, (2) n-butane, (3) n-pentane, (4) n-hexane, (5) n-heptane, (6) n-
octane, (7) n-decane, (8) isobutane, (9) neopentane, (10) 3-methylpentane, (11) 2,2-dimethylbutane,
(12) isooctane, (13) cyclopentane, (14) cyclohexane, (15) cycloheptane, (16) cyclooctane, (17)
cyclodecane, (18) methylcyclopentane, (19) methylcyclohexane, (20) ethylcyclohexane, (21) 1,1-
dimethylcyclohexane, (22) cis-1,2-dimethylcyclohexane, (23) trans-1,2-dimethylcyclohexane, (24) cis-
1,3-dimethylcyclohexane, (25) trans-1,3-dimethylcyclohexane, (26) cis-1,4-dimethylcyclohexane, (27)
trans-1,4-dimethylcyclohexane. (From Refs. 18, 29, 91, 92, 99, 100, 108, 110, 111, 113, 114, and
160.)
Copyright 2004 by Taylor & Francis Group, LLC
eect is the ionic decomposition, which is eective for highly branched and especially
smaller molecular mass alkane radical cations. With most molecules these two types of
structure eects can act in the same direction: e.g., in the reaction of highly branched
alkanes both the ionic decomposition and S
1
decomposition result in high yield of CC
bond ruptures at the branched carbons. There is no agreement regarding the yields of ionic
decompositions, the yields for some molecules may be as high as G = 13.
It is obvious from this analysis that while the decompositions from the S
1
ex-
cited molecules or the radical cations tend to take place selectively, the bond ruptures
from other species, e.g., from higher-energy excited molecules are closer to a random
distribution.
6. CONCLUDING REMARKS
Because of such diculties as the featureless absorption and emission spectra in the
vacuum ultraviolet region, very weak and energy-dependent uorescence intensity, short
excited-state lifetime, etc. the photophysics and photochemistry of alkanes is much less
known than those of other organic molecules, for instance, aromatic hydrocarbons. In this
chapter, the present status was reviewed.
At present there are dierent views about the nature of the alkane excited states. The
excited states are probably admixtures of Rydberg- and valence-type states. There are
some views that the initial energy absorption takes place in a basically Rydberg state and
than it relaxes to a valence-type excitation. The Rydberg-type excitations of alkanes have
valence type counterparts with the same symmetry [22]. According to some works the
lower-energy states are delocalized all over the molecules, while other works mention
strongly localized excited orbitals. Some excited-state properties may point to the localized
nature of the excited states. These include the characteristics of uorescence, the large
disparity between the absorption onset and the uorescence maximum (Stokes shift, D c
0.81.6 eV, width of the uorescence peak (r
f
c 1 eV), and also the molecular structure
dependence of these characteristics, e.g., the very strong and characteristic inuence of the
alkane chain branching and the high probability of decompositions at the tertiary and
quaternary carbons. In the photoexcitation of many alkanes the uorescence quantum
yields decrease with increasing photon energy; this dependence is attributed to an energy-
dependent relaxation of the higher excited states to the relaxed S
1
state from where the
uorescence occurs.
The excited-state lifetimes (s c 1 nsec) determined by uorescence decay, absorb-
ance decay, or steady-state quenching experiments after single-photon, multiphoton, or
ionizing radiation excitation generally agree indicating that the same excited state (the
relaxed S
1
) is observed in all experiments. Within a homologous series, e.g., for the n-
alkanes, s is increasing with the size of the molecules, similarly to the uorescence in-
tensity. The ratio of the two quantities, R = U
f
/s, the so-called radiative rate coecient,
within a group of compounds remains constant: R c1.6 10
6
sec
1
for n-alkanes,
f
2
10
5
sec
1
for linear methylalkanes,
f
5.3 10
6
sec
1
for vicinally branched dimethylal-
kanes,
f
8 10
6
sec
1
for cyclohexanes, and
f
1.2 10
7
sec
1
for dimethylcyclohexanes.
The lifetimes in the group of the so-called nonuorescing alkanes (geminally substituted
alkanes, C
5
, C
7
C
10
cycloalkanes) are short (s V 0.3 nsec): the absence of uorescence may
have two causes, low radiative rate coecient and fast chemical decay.
The radiationless S
1
!S
0
transition has a negligible role in the depopulation of the
excited molecules; the process, which competes with the weak uorescence (
f
1%) is the
Copyright 2004 by Taylor & Francis Group, LLC
chemical decomposition (
f
99%). Two basic chemical decomposition types are distin-
guished: hydrogen molecule and alkane molecule elimination (in these reactions no
intermediate radicals are observed) and CH or CC scission yielding two radicals or
CC scission yielding a biradical (in the reaction of cycloalkanes).
The decomposition of a C
n
H
2n+2
alkane molecule to H
2
molecule and C
n
H
2n
alkene
molecule belongs to the rst class of reactions. This is the typical decomposition mode of
the S
1
excited n-alkanes and C
5
C
10
cycloalkanes at room temperature. The quantum yield
of the reaction decreases with decreasing temperature. Thus, it needs some thermal
activation. The reaction is attributed to an S
1
!S
x
-type IC process; S
x
denotes some
singlet state that gives the products. The elimination takes place mostly or probably
entirely from a single carbon atom yielding an H
2
molecule and a carbene intermediate.
The carbene, by intramolecular 1,2-H atom shift, rearranges to an olen. In the cases of
C
7
C
10
cycloalkanes the carbene may also stabilize in the form of cross-bridged bicy-
cloalkanes by 1,3-, 1,5-, or 1,6-H atom shifts.
Alkane elimination is a basic photodecomposition mode of highly branched alkanes,
e.g., most of the excited neopentane molecules split directly to methane and isobutene.
The radical-forming reactions are suggested to take place mostly after an S
1
!T
n
type ISC: the reactions have nonactivated character. The homolytic split to H atom and
alkyl radical has a considerable yield in the photolysis of n-alkanes and cycloalkanes, while
the scission to two radicals is characteristic of the decay of excited branched alkane
molecules.
With the increase of the photon energy used for excitation, the yield of the radical-
forming reactions increases, whereas that of the H
2
elimination decreases.
By now it is unambiguously established that the energy transfer from the S
1
excited
alkane molecules takes place with a diusional mechanism at the contact distance: the
reaction radii deduced from such experiments (0.8 F0.2 nm) practically agree with the sum
of the radii of the reacting molecules. In some special cases, however, very fast energy
transfer was also observed (e.g., to PPOmolecules): the fast transfer was attributed to energy
transfer from high-orbit Rydberg excited states or from some unspecied ion-pair state.
The properties of the alkane triplet states are much less known than the properties of
the singlet states. It seems there is a dissociative, and thereby short-living triplet state
f
1
eV below the lowest singlet excited state [29]. The other triplet states are probably also
dissociative. The low-lying and long-living triplet states, supposed in some works based on
unusual energy-transfer results, still needs real experimental verication [7].
In radiolysis most of the S
1
excited molecules (up to 90%) form in charge
recombination. The singlet fraction, f
s
, based on the newer experimental and theoretical
studies is around 0.50.7. Both theoretical and experimental work indicate that f
s
decreases with increasing LET value of radiation, because of preferred cross-recombina-
tion and by preferred triplet formation in the high ionizing density tracks.
The frequency of CH and CC bond ruptures in the radiolysis of liquid alkanes is
far from a statistical distribution, and there is a correlation between the dissociation
reactions and the chemical structure. In contrast to the decomposition of n-alkanes and
C
5
C
10
cycloalkanes, which yield mostly hydrogen and dehydrogenation products, some
branched alkanes, e.g., neopentane and 2,2,4-trimethylpentane, decompose mainly by C
C bond rupture. The considerable structure eect can to some extent be traced back to the
fact that with many alkanes 25% to 55% of the total primary decomposition of G = 6.5 F
1 occurs from low-energy singlet excited states. Chemical reactions from these states tend
to take place selectively, whereas the bond ruptures from higher-energy states are closer to
a random distribution. Fragmentation of the highly branched and especially smaller
Copyright 2004 by Taylor & Francis Group, LLC
molecular mass alkane radical cations can also contribute signicantly to observed
selectivity of bond ruptures in radiolysis.
REFERENCES
1. Topchiev, A.V., In: Radiation Chemistry of Hydrocarbons, English Ed.; Holroyd, R.A.;
Elsevier; Amsterdam, 1964.
2. Foldiak, G., Ed.; Radiation Chemistry of Hydrocarbons. In Studies in Physical and Theoretical
Chemistry 14; Elsevier: Amsterdam, 1981.
3. Gaumann, T., Hoigne, J., Eds.; In Aspects of Hydrocarbon Radiolysis. Academic Press: London,
1968.
4. Ausloos, P., Ed; In: Fundamental Process in Radiation Chemistry. Interscience: New York,
1968.
5. Freeman, G.R. Radiat. Res. Rev. 1968, 1, 1.
6. Hummel, A. Radiation Chemistry of Alkanes and Cycloalkanes. In The Chemistry of Alkanes
and Cycloalkanes; Patai, S.; Rappoport, Z., Eds.; John Wiley and Sons: Chichster, 1992; 743 pp.
7. Shkrob, I.A., Sauer, M.C., Jr., Trifunac, A.D. Radiation chemistry of organic liquids:
saturated hydrocarbons. In Radiation Chemistry: Present Status and Future Trends, Studies in
Physical and Theoretical Chemistry 87; Jonah, C.D., Rao, B.S.M., Eds.; Elsevier: Amsterdam,
2001; 175 pp.
8. Mayer, J., Szadkowska-Nicze, M. Excited states in liquid alkanes and related polymers. In
Properties and Reactions of Radiation Induced Transient Species, Selected Topics; Mayer, J.,
Ed.; Polish Scientic Publishers: Warsawa, 1999; 77 pp.
9. Spinks, J.W.T.; Woods, R.J. An Introduction to Radiation Chemistry; Wiley-Interscience: New
York, First ed. (1964), Second ed. (1976), Third ed. (1990).
10. Swallow, A.J. Radiation Chemistry. An Introduction. Longman: London, 1973.
11. Farhataziz, Rodgers, M.A.J., Eds.; In Radiation Chemistry, Principles and Applications. VCH
Publishers: New York, 1987.
12. CRC Handbook of Radiation Chemistry, Tabata, Y., Ed.; CRC Press: Boca Raton, 1991;
Pulse Radiolysis, Tabata, Y., Ed.; CRC Press: Boca Raton, 1991.
13. Mozumder, A. Fundamentals of Radiation Chemistry; Academic Press: San Diego, 1999.
14. Ausloos, P.; Lias, S.G. Radiat. Res. Rev. 1968, 1, 75; Rev. Phys. Chem. 1971, 22, 85.
15. Ausloos, P. Mol. Photochem. 1972, 4, 39.
16. Ausloos, P.; Lias, S.G. Far ultraviolet photochemistry of organic compunds. In Chemical
Spectroscopy and Photochemistry in the Vacuum Ultraviolet; Sandorfy, C.; Ausloos, P.; Robin,
M.B., Eds.; Chemical Spectrospy and Photochemistry in Vacuum Ultraviolet; Reidel:
Dordecht, 1974; Vol. 8, Ser. 6, 465 pp.
17. a) Dorofeev, Y.I.; Skurat, V.E. Russ. Chem. Rev. 1982, 51, 527.
b) Dorofeev, Y.I.; Skurat, V.E. Radiation Chemistry, Photochemistry, Vol 3. Photochemical
Processes in the Presence of Far Ultraviolet Radiation (Alkanes, Alkenes, Halohydrocarbons,
Organic Polymers). VINITI, Moscow (in Russian), 1983.
18. Foldiak, G.; Wojnarovits, L. Photochemistry of liquid alkanes, in Proceedingd of Baxendale
memorial symposium, Centro StampaLo Scarabeo: Bologna, 1983; 21 pp.
19. Oppenlander, T.; Zang, G.; Adam, W. Rearrangements and Photochemical Reactions
Involving Alkanes and Cycloalkanes. In The Chemistry of Alkanes and Cycloalkanes; Patai, S.,
Rappoport, Z., Eds.; John Wiley and Sons, Chichester, 1992, 681 pp.
20. Raymonda, J.W.; Simoson, W.T. J. Chem. Phys. 1967, 47, 430.
21. Sowers, B.L.; Arakawa, M.W.; Hamm, R.N.; Arakawa, E.T. J. Chem. Phys. 1972, 57, 167.
22. Robin, M.B. Higher Excited States of Polyatomic Molecules; Academic Press: New York; Vol.
I (1974), Vol. II (1975), Vol. III (1985).
23. Hatano, Y. Radiat. Phys. Chem. 2003, 67, 187.
Copyright 2004 by Taylor & Francis Group, LLC
24. Hirayama, F.; Lipsky, S. J. Chem. Phys. 1969, 51, 3616.
25. Rothman, W.; Hirayama, F.; Lipsky, S. J. Chem. Phys. 1973, 58, 1300.
26. Ostan, A.E.; Lipsky, S. J. Chem. Phys. 1993, 98, 5408.
27. Schwarz, F.P.; Smith, D.; Lias, S.G.; Ausloos, P. J. Chem. Phys. 1981, 75, 3800.
28. Kimura, K.; Hormes, J. J. Chem. Phys. 1983, 79, 2756.
29. Wojnarovits, L., Foldiak, G. Triplets states of alkanes. In Fifth Working Meet. Radiat.
Interact; Mai, H., Brede, O., Mehnert, R., Eds.; ZFI: Leipzig, 1991; 64 pp.
30. Klein, G.; Voltz, R. Int. J. Radiat. Phys. Chem. 1975, 7, 155.
31. Tramer, A.; Voltz, R. Time-resolved studies of excited molecules. Excited States. Lim, E.C.,
Ed.; Academic Press: New York, 1979, 4, 281 pp.
32. Lamotte, M.; Pereyre, J.; Joussot-Dubien, J.; Lapouyade, R. J. Photochem. 1987, 38, 177.
33. Smirnov, V.A.; Nazarov, V.B.; Gerko, V.I.; Almov, M.V. Chem. Phys. Lett. 1975, 34, 500.
34. Lamotte, M. J. Phys. Chem. 1981, 85, 2632.
35. Leclerc, G.; Cui, Z.; Sanche, L. J. Phys. Chem. 1987, 91, 6461.
36. Hirayama, F.; Lipsky, S. Saturated Hydrocarbons as Donors in Electronic Energy Transfer
Processes New York. In Organic Scintillators and Liquid Scintillation Counting; Horrocks,
D.L., Peng, C.-T., Eds.; Academic Press, 1971; 205 pp.
37. Lipsky, S., Excited states in hydrocarbons. Proceedings 8th International Congress of
Radiation Research; Fielden, E.M.J., Hendry, H., Scott, D., Eds.; Taylor and Francis:
London, 1987; Vol 2., 72 pp.
38a. Casanovas, J., Grob, R., Delacroix, D., Guelfucci, J.P., Blanc, D. J. Chem. Phys. 1981, 75, 4661.
38b. Casanovas, J., Guelfucci, J.P., Terrissol, M. Radiat. Phys. Chem. 1988, 32, 361.
38c. Casanovas, J., Guelfucci, J.P., Caselles, O. IEEE Trans. Electr. Insul. 1991, 26, 603.
38d. Guelfucci, J.P., Casanovas, J., Huertas, M.L., Salon, J. J. Phys. Chem. 1993, 97, 10352.
39. Ausloos, P.; Lias, S.G.; Rebbert, R.E. J. Phys. Chem. 1981, 85, 2322.
40. Ausloos, P.; Rebbert, R.E.; Schwarz, F.P.; Lias, S.G. Radiat. Phys. Chem. 1983, 21, 27.
41. Wieckowski, A.; Collin, G.J. J. Phys. Chem. 1977, 81, 2592.
42. Nasi-Movaghar, J., Hatano, Y. J. Phys. Chem. 1974, 78, 1899; Wada, T., Hatano, Y. J.
Phys. Chem. 1975, 79, 2210; Wada, T., Hatano, Y. J. Phys. Chem. 1977, 81, 1057.
43. Gorden, R., Jr.; Rebbert, R.E.; Ausloos, P. Rare Gas Resonance Lamps. NBS Technical Note
496, 1969.
44. McNesby, J.R.; Braun, W.; Ball, J. Vacuum ultraviolet techniques in photochemistry. In
Creation and Detection of Excited State; Lamola, A.A., Ed.; Marcel Dekker Inc.: New York;
1971; 503 pp.
45. Ouchi, A.; Yabe, A.; Inoue, Y.; Daino, Y.; Hakushi, T. J. Chem. Soc. Chem. Commun., 1989,
1669.
46. Dellonte, S.; Gardini, E.; Barigelletti, F.; Orlandi, G. Chem. Phys. Lett. 1977, 49, 596.
47. Miyasaka, H.; Mataga, N. Chem. Phys. Lett. 1986, 126, 219.
48. Volkmer, A.; Wynne, K.; Birch, D.J.S. Chem. Phys. Lett. 1999, 299, 395.
49. Sander, M.U.; Brummund, U.; Luther, K.; Troe, J. J. Phys. Chem. 1993, 97, 8378.
50. Lakowicz, J.R.; Gryczynski, I. Biospectroscopy 1993, 1, 3.
51. Ware, W.R.; Lyke, R.L. Chem. Phys. Lett. 1974, 24, 195.
52. Ware, W.R.; OConnor, D.V. Chem. Phys. Lett. 1979, 62, 595.
53. Hatano, Y. Phys. Rep. 1999, 313, 110.
54. Shinsaka, K.; Koizumi, A.; Yoshimi, T.; Kouchi, N.; Nakamura, Y.; Toriumi, M.; Morita,
M.; Hatano, Y.; Asaoka, S.; Nishimura, H. J. Chem. Phys. 1985, 83, 4405.
55. Wickramaaratchi, M.A.; Preses, J.M.; Holroyd, R.A.; Weston, R.E., Jr. J. Chem. Phys. 1985,
82, 4745.
56. Preses, J.M.; Grover, J.R.; Kvick, A.; While, M.G. Am. Sci. 1990, 78, 424.
57. Orlandi, G.; Dellonte, S.; Flamigni, L.; Barigelletti, F. J. Chem. Soc. Faraday Trans. I 1982,
78, 1465.
58. Choi, H.T.; Hirayama, F.; Lipsky, S. J. Phys. Chem. 1984, 88, 4246.
59. Hermann, R.; Mehnert, R.; Wojna rovits, L. J. Lumin. 1985, 33, 69.
Copyright 2004 by Taylor & Francis Group, LLC
60. Henry, M.S.; Helman, W.P. J. Chem. Phys. 1972, 56, 5734.
61. Codee, H.D.K. Excited States of a Number of Saturated and Cyclic Hydrocarbons due to
Irradiation. INIS-mf-5335, 7.Jun.1979.
62. Helman, W.P. Chem. Phys. Lett. 1972, 17, 306.
63. Luthjens, L.H.; de Haas, M.P.; de Leng, H.C.; Hummel, A.; Beck, G. Radiat. Phys. Chem.
1982, 19, 121.
64. Luthjens, L.H.; Codee, H.D.K.; de Leng, H.C.; Hummel, A. Chem. Phys. Lett. 1981, 79,
444.
65. Luthjens, L.H.; Codee, H.D.K.; de Leng, H.C.; Hummel, A.; Beck, G. Radiat. Phys. Chem.
1983, 21, 21.
66. Luthjens, L.H.; de Leng, H.C.; Wojnarovits, L.; Hummel, A. Radiat. Phys. Chem. 1985, 26, 509.
67. Wojnarovits, L.; Luthjens, L.H.; de Leng, H.C.; Hummel, A. J. Radioanal. Nucl. Chem.
Articles, 1986, 101, 509.
68. Grigoryants, V.M.; Lazovoi, V.V. High Energy Chem. 1996, 30, 38.
69. Mehnert, R.; Brede, O.; Naumann, W.; Hermann, R. Ber. Bunsenges. Phys. Chem. 1983, 87,
387.
70. Hermann, R.; Mehnert, R.; Brede, O.; Neumann, W. Radiat. Phys. Chem. 1985, 26, 513.
71. Mehnert, R. ZFI-Mitt. (Leipzig) 1984, 87, 1101.
72. Katsumura, Y.; Tabata, Y.; Tagawa, S. Radiat. Phys. Chem. 1982, 19, 267.
73. Katsumura, Y.; Yoshida, Y.; Tagawa, S.; Tabata, Y. Radiat. Phys. Chem. 1983, 21, 103.
74. Tagawa, S.; Washio, M.; Kobayashi, H.; Katsumura, Y.; Tabata, Y. Radiat. Phys. Chem.
1983, 21, 45.
75. Wichramaaratchi, M.A.; Preses, J.M.; Weston, R.E., Jr. J. Chem. Phys. 1986, 85, 2445.
76. Preses, J.M. J. Chem. Phys. 1988, 89, 1251.
77. Preses, J.M.; Holroyd, R.A. J. Chem. Phys. 1990, 92, 2938.
78. Barigelletti, F.; Dellonte, S.; Mancini, G.; Orlandi, G. Chem. Phys. Lett. 1979, 65, 176.
79. Flamigni, L.; Barigelletti, F.; Dellonte, S.; Orlandi, G. Chem. Phys. Lett. 1982, 89, 13.
80. Orlandi, G.; Flamigni, L.; Barigelletti, F.; Dellonte, S. Radiat. Phys. Chem. 1983, 21, 113.
81. Dellonte, S.; Flamigni, L.; Barigelletti, F.; Wojnarovits, L.; Orlandi, G. J. Phys. Chem. 1984,
88, 58.
82. Orlandi, G.; Barigelleti, F.; Flamigni, L.; Dellonte, S. J. Photochem. 1985, 31, 49.
83. Flamigni, L.; Orlandi, G. J. Photochem. Photobiol. A 1988, 42, 241.
84. Mehnert, R. Radicals cations in pulse radiolysis. In Radical Ionic Systems. Properties in
Condensed Phases; Lund, A., Shiotani, M., Eds.; Kluwer: Dordrecht, 1991, 231 pp.
85. Tagawa, S.; Hayashi, N.; Yoshida, Y.; Washio, M.; Tabata, Y. Radiat. Phys. Chem. 1989, 34,
503.
86. Mehnert, R.; Brede, O.; Naumann, W. J. Radioanal. Nucl. Chem., Articles 1986, 101, 307.
87. Holroyd, R.A.; Preses, J.M.; Hanson, J.C. Radiat. Res. 1993, 135, 312; J. Phys. Chem. A.
1997, 101, 6931.
88. Shibata, H.; Yoshida, Y.; Tagawa, S.; Aoki, Y.; Namba, H. Nucl. Instrum. Methods Phys.
Res. A 1993, 327, 53.
89. Andre, J.C.; Lyke, R.; Ware, W.R. React. Kinet. Catal. Lett. 1984, 26, 3.
90. Marconi, G.C.; Orlandi, G.; Poggi, G. Chem. Phys. Lett. 1976, 40, 88.
91. Wojnarovits, L.; Foldiak, G. Acta Chim. Hung. (Budapest) 1980, 105, 27.
92. Wojnarovits, L.; Foldiak, G. ZFI-Mitt. (Leipzig) 1981, 43a, 243.
93. Dyne, P.J. Can. J. Chem. 1965, 43, 1080.
94. Dyne, P.J.; Denhartog, J. Can. J. Chem. 1962, 40, 1616.
95. Okabe, H.; McNesby, J.R. J. Chem. Phys. 1961, 34, 668.
96. McNesby, J.R.; Okabe, H. Adv. Photochem. 1964, 3, 157.
97. Gaumann, T. Liquid hexanes. In Aspects of Hydrocarbon Radiolysis; Gaumann, T.; Hoigne, J.
Academic Press: London, 1968; 213 pp.
98. Ballenegger, M.; Ruf, A.; Gaumann, T. Helv. Chim. Acta 1971, 54, 1373.
99. Wojnarovits, L.; Szondy, T.; Szekeres-Bursics, E.; Foldiak, G. J. Photochem. 1982, 18, 273.
Copyright 2004 by Taylor & Francis Group, LLC
100. Wojnarovits, L. J. Chem. Soc. Perkin Trans. 2, 1984, 1449.
101. Foldiak, G.; Wojnarovits, L. Acta Chim. Hung. (Budapest) 1974, 82, 269.
102. Wojnarovits, L.; LaVerne, J.A. J. Phys. Chem. 1994, 98, 8014.
103. Paal, Z. Metalcatalysed cyclization reactions of hydrocarbons. In Advances in Catalysis;
Eley, D.D.; Pines, H.; Weisz, P.B., Eds.; Academic Press: New York, 1980; Vol. 29, 273 pp.
104. Holroyd, R.A. J. Am. Chem. Soc. 1969, 91, 2208.
105. Horvath, Zs.; Ausloos, P; Foldiak, G. Comparison between liquid phase photolysis and
radiolysis of C
3
C
4
hydrocarbons mixtures. In Proceedings Fouth Tihany Symposium on
Radiation Chemistry; Hedvig, P.; Schiller, R., Eds.; Akademiai Kiado: Budapest, 1977; 57 pp.
106. Antonova, E.A.; Pichuzhkin, V.I. High Energy Chem. 1977, 11, 201.
107. Rebbert, R.E.; Lias, S.G.; Ausloos, P. Picosecond ultraviloet multiphoton laser photolysis
related to radiolysis. J . Photochem. 1975, 4, 121.
108. Wojnarovits, L.; Kozari, L.; Keszei, Cs.; Foldiak, G. J. Photochem. 1982, 19, 79.
109. Wojnarovits, L.; Foldiak, G. Acta Chim. Hung. (Budapest) 1977, 93, 1.
110. Kozari, L.; Wojnarovits, L.; Foldiak, G. Acta Chim. Hung. (Budapest) 1982, 109, 249.
111. Wojnarovits, L.; LaVerne, J.A. J. Phys. Chem. 1995, 99, 3168.
112. Eberhardt, M.K. J. Phys. Chem. 1968, 72, 4509.
113. Foldiak, G.; Wojnarovits, L. Radiat. Phys. Chem. 1988, 32, 335.
114. Hummel, A.; de Leng, H.C.; Luthjens, L.H.; Wojnarovits, L. J. Chem. Soc. Faraday Trans.
1994, 90, 2459.
115. Yang, J.Y.; Servedio, F.M.; Holroyd, R.A. J. Chem. Phys. 1968, 48, 1331.
116. Wojnarovits, L. Radiochem. Radioanal. Lett. 1978, 32, 267.
117. Pitchoozhkin, V.I.; Yamazaki, H.; Shida, S. Bull. Chem. Soc. Jpn. 1973, 46, 67.
118. Miyasaka, H.; Mataga, N. Picosecond Ultraviolet Multiphoton Laser Phtotlysis Related to
Radiolysis. In Pulse Radiolysis; Tabata, Y., Ed,; CRC Press: Boca Raton, 1991; 173 pp.
119. Long, F.H.; Lu, H.; Eisenthal, K.B. J. Phys. Chem. 1995, 99, 7436.
120. Siebbeles, L.D.A.; Emmerichs, U.; Hummel, A.; Bakker, H.J. J. Chem. Phys. 1997, 107, 9339.
121. Dellonte, S.; Barigelletti, F.; Orlandi, G.; Flamigni, L. Intramolecular Deactivation Processes
and Energy Transfer Mecahanism in Liquid Alkanes Studied by N
2
Laser Two Photon
Excitation. Proceedings 5th Tihany Symposium on Radiation Chemistry. Dobo, J.; Hedvig,
P.; Schiller, R., Eds.; Akademiai Kiado: Budapest, 1983; 437 pp.
122. Plotnikov, V.G. Radiat. Phys. Chem. 1985, 26, 519.
123. Manion, J.P.; Burton, M. J. Phys. Chem. 1952, 56, 560.
124. Szondy, T.; Wojnarovits, L.; Foldiak, G. Anomalous Energy Transfer in Radiolysis of
Alkanes. A possible Theoretical Explanation. Proceedings 5th Tihany Symposium on
Radiation Chemistry. Dobo, J.; Hedvig, P.; Shciller, R., Eds.; Akademiai Kiado, Budapest,
1983; 119 pp.
125. Wojnarovits, L.; Szondy, T.; Foldiak, G. Radiochem. Radioanal. Lett. 1981, 48, 175.
126. Kudo, T.; Shida, S. J. Phys. Chem. 1967, 71, 1971.
127. Busi, F.; Casalbore, G. Gazz. Chim. Ital. 1981, 111, 443.
128. Luthjens, L.H.; de Leng, H.C.; Appleton, W.R.S.; Hummel, A. Radiat. Phys. Chem. 1990, 36,
213.
129. Choi, H.T.; Lipsky, S. J. Phys. Chem. 1981, 85, 4089.
130. Wojnarovits, L. J. Photochem. 1984, 24, 341.
131. Andre, J.C.; Niclause, M.; Ware, W.R. Chem. Phys. 1978, 28, 371.
132. Johnston, D.B.; Wang, Y-M.; Lipsky, S. J. Phys. Chem. 1991, 95, 5524.
133. Rehm, D.; Weller, A. Ber. Bunsenges. Phys. Chem. 1969, 73, 834.
134. Balzani, V.; Bolletta, F.; Scandola, F. J. Am. Chem. Soc. 1980, 102, 2152.
135. Hermann, R.; Brede, O.; Mehnert, R. Energy Transfer from Alkane Singlets and Quenching
of AlkaneFluorescences by Alkyl Chlorides. A. Pulse Radiolysis Study. Proceedings 5th
Tihany Symposium on Radiation Chemistry; Dobo, J.; Hedvig, P.; Schiller, R., Eds.;
Akademiai Kiado: Budapest, 1983; 503 pp.
136. Mehnert, R. .Pulse Radiolysis Investigation of Charge and Energy Transfer Reactions in
Copyright 2004 by Taylor & Francis Group, LLC
Alkanes and Alkyl Chlorides. Proceedings 5th Tihany Symposium on Radiation Chemistry;
Dobo, J.; Hedvig, P.; Schiller, R., Eds.; Akademiai Kiado: Budapest, 1983; 489 pp.
137. Wang, Y.-M.; Johnston, D.B.; Lipsky, S. J. Phys. Chem. 1993, 97, 403.
138. Krishna, T.S.R.; Lipsky, S. J. Phys. Chem. A 1998, 102, 496.
139. Busi, F.; Casalbone, G. Gazz. Chim. Ital. 1983, 113, 83.
140. Busi, F. Labile Species and Fast Processes in Liquid Alkanes. In The Study of Fast Processes
and Transient Species by Electron Pulse Radiolysis; Baxendale, J.F.; Busi, F., Eds.; Reidel;
Dordrecht, 1982; 417 pp.
141. Baxendale, J.H.; Mayer, J. Chem. Phys. Lett. 1972, 17, 458.
142. ONeill, P.; Salmon, G.A.; May, R. Proc. R. Soc. Lond. A 1975, 347, 61.
143. Busi, F.; Flamigni, L.; Orlandi, G. Radiat. Phys. Chem. 1979, 13, 165.
144. Johnston, D.B.; Lipsky, S. J. Phys. Chem. 1991, 95, 1896.
145. Beck, G.; Thomas, J.K. J. Phys. Chem. 1972, 76, 3856.
146. Brocklehurst, B. J. Chem. Soc., Faraday Trans. 1992, 88, 167.
147. Bartczak, W.M. Computer Simulation of Early Physico-Chemical Processes in Irradiated
Media. In Properties and Reactions of Radiation Induced Transient Species, Selected Topics;
Mayer, J., Ed.; Polish Scientic Publishers: Warsawa, 1999; 101 pp.
148. Walter, L.; Hirayama, F.; Lipsky, S. Int. J. Radiat. Phys. Chem. 1976, 8, 237.
149. Luthjens, L.H.; Dorenbos, P.; De Haas, J.T.M.; Hummel, A.; van Eijk, C.W.E. Radiat. Phys.
Chem. 1999, 55, 255.
150. Sauer, M.C., Jr.; Jonah, C.D. Radiat. Phys. Chem. 1994, 44, 281.
151. LaVerne, J.A.; Pimblott, S.M.; Wojnarovits, L. J. Phys. Chem. A 1997, 101, 1628.
152. Yoshida, Y.; Tagawa, S.; Tabata, Y. Radiat. Phys. Chem. 1984, 23, 279.
153. Sauer, M.C., Jr.; Jonah, C.D.; LeMotais, B.C.; Chernovitz, A.C. J. Phys. Chem. 1988, 92, 4099.
154. Walter, L.; Lipsky, S. Int. J. Radiat. Phys. Chem. 1975, 7, 175.
155. Choi, H.T.; Askew, D.; Lipsky, S. Radiat. Phys. Chem. 1982, 19, 373.
156. Fujisaki, N.; Shida, S.; Hatano, Y. J. Chem. Phys. 1970, 52, 556.
157. Shida, S.; Fujisaki, N.; Hatano, Y. J. Chem. Phys. 1968, 49, 4571.
158. Dyne, P.J.; Jenkinson, W.M. Can. J. Chem. 1960, 38, 539.
159. Nevitt, T.D.; Remsberg, L.P. J. Phys. Chem. 1960, 64, 969.
160. Wojnarovits, L.; Foldiak, G. Radiat. Res. 1982, 91, 638.
161. Werst, D.W.; Trifunac, A.D. J. Phys. Chem. 1991, 95, 3466.
162. Pichuzhkin, V.I.; Antonova, E.A.; Chudakov, V.M.; Bakh, N.A. High Energy Chem. 1978,
12, 275.
163. LaVerne, J.A.; Schuler, R.H.; Foldiak, G. J. Phys. Chem. 1992, 96, 2588.
164. Wojnarovits, L.; LaVerne, J.A. J. Phys. Chem. 1995, 99, 11292.
165. Wojnarovits, L.; LaVerne, J.A. J. Radioanal. Nucl. Chem. 1998, 232, 19.
Copyright 2004 by Taylor & Francis Group, LLC
14
Radiation Chemical Effects of Heavy Ions
Jay A. LaVerne
University of Notre Dame, Notre Dame, Indiana, U.S.A.
1. INTRODUCTION
The chemical eects induced by the passage of heavy ions in matter have been actively
examined for more than 100 years. Curie and Debierne in 1901 performed the rst radia-
tion chemistry study on water by any type of radiation using a-particles from solutions
of radium salts [1]. All of the early studies examined the production of gases, mainly
hydrogen, from radium solutions [29]. Radium was used because of its availability, but
equally important is the reality that products could not be observed using other types
of radiation because of impurities and the crude techniques of the time. In contrast to the
observable transitions in radium solutions, Fricke could detect no decomposition of pure,
air-free water by x-rays as late as 1933 [10]. Many of the transient species discussed in the
preceding chapters of this book were just being discovered at about this time and a com-
plete description of the observed chemical eects was understandably dicult. However,
experimental techniques quickly advanced and many early monographs on radiation eects
note the dierent eects observed with heavy ions compared to that found with fast elec-
trons, x-rays or g-rays [1116].
A wide variety of early experimental [17] and theoretical studies [1822] have formed
much of the basis for understanding the chemical eects of heavy ions. The general con-
cepts of heavy ion radiolysis have not changed since these studies. However, more informa-
tion on specic systems is constantly being accumulated, especially in complicated mixtures
and in heterogeneous systems. The details of heavy ion radiolysis will not be discussed in
depth, but rather an overview of the major eects and the reason for their occurrence will be
given here. An examination of rare events specic to heavy ion radiolysis can be found in a
recent review [23]. Descriptions of typical experimental techniques used in heavy ion
radiolysis are published elsewhere [24,25]. The discussion in this chapter will be limited to
the radiolysis of water because most of the work has been performed in this medium and
because of the importance of water in a variety of applications. Many of the fundamental
ideas discussed here for liquid water are applicable to other liquids. However, the nature of
the processes occurring in many solids and almost all gases may lead to very dierent eects
than discussed here.
In addition to the fundamental scientic aspects, many studies on the chemical eects
of heavy ion radiolysis have signicant practical applications. These applications range
from the nuclear power industry [26,27], space radiation eects [28], medical therapy [29],
Copyright 2004 by Taylor & Francis Group, LLC
and the environmental management of radioactive waste materials [30]. Much of the future
eort on the examination of heavy ion radiolysis will driven by these need-based demands.
The complexity of such systems can be overwhelming and good fundamental research on
simple systems is required.
The term heavy ions usually refers to any ionizing radiation other than fast elec-
trons. In this chapter, the term will apply to charged atomic nuclei of all the elements from
protons to uranium ions. The concepts presented may apply to other particles such as
muons; however, a number of collision parameters are dierent between these ions and
atomic nuclei and one should refer to specic monographs for details. High velocity molec-
ular ions fragment in the rst few collision processes and one gets atomic nuclei of the
same velocity, which can then be treated as individual heavy ions. Neutrons, being neutral,
are not classied as heavy ions either. They produce recoil ions in passing through matter.
The recoil ions, oxygen or hydrogen in water, are usually formed with a wide energy dis-
tribution and each ion can be examined independently.
2. ENERGY LOSS PROPERTIES OF HEAVY IONS
Most radiation chemical studies involve ions with initial energy of a few MeV or more. At
these energies, the main interaction responsible for energy loss is the Coulombic force
between the charge of the incident heavy ion and that of the electrons of the medium. Energy
transferred from one of these collisions leads to excitation or ionization of the medium
molecule and the rate at which this energy is lost per unit path length, dE/dx, is called the
electronic stopping power of the medium. Rutherford was the rst to successfully explain
the scattering of charged particles using Coulombic forces and classical mechanics [31].
Bohr applied this cross section using an impact parameter approach to determine the stop-
ping power and ranges of ions in matter [32]. A quantum mechanical theory of stopping
powers was later developed by Bethe based on energy and momentum transfer [33]. Both
the Bohr and Bethe formalisms are equally valid depending on the charge and velocity of
the incident heavy ion [34]. Almost all other methods for determining the energy loss by
charged particles through electronic processes are modications of these two early theories
and a complete discussion is given in Chapter 2 of this book.
The details of the stopping power equations have been very well summarized else-
where [22,3438]. Most applications involve ions of low charge and high velocity in which
the Bethe formalism is valid. The nonrelativistic stopping power equation of Bethe for
heavy ions is given approximately by,
dE=dx
4pe
4
Z
2
N
mV
2
ln
2mV
2
I
1
where e and m are the charge and mass of the electron, Z and V are the charge and velocity
of the incident particle, and N and I are the electron density and mean excitation potential
of the medium. Relativistic terms, shell eects, the Fermi density correction, and other
modications to the stopping power are usually added to the logarithmic term as needed.
Sometimes stopping powers are further subdivided into hard and soft [32] or glancing and
knock-on [22] collisions. One can also encounter restricted stopping powers in which an
upper limit is placed on the energy loss in a given collision [39]. These distinctions are
made according to the energy loss in the collision and aided in the early development of
Copyright 2004 by Taylor & Francis Group, LLC
track structure as described below. However, such distinctions are rarely necessary for
understanding basic radiation chemistry. They can also lead to erroneous assumptions
because the energy cuto is often arbitrary.
Equation (1) is rarely used in its given form to determine stopping powers or ranges
of ions. There are too many diculties in estimating the eective charge of the incident
ion, screening eects, eective mean excitation potentials, and other factors that must be
taken into account for accurate stopping powers. Most investigators use standard com-
pilations that employ semi-empirical ts to experimental data [4042]. The compilations
typically compare the stopping of dierent ions in the same target medium and scale to
other media or standardize the stopping of protons in dierent media and scale to other
ions. The accuracy of the compilations is thereby increased considerably. However, the
predictions of these compilations should be veried for the specic radiolytic condition to
be examined as some compilations are more accurate than others for a particular ion in a
given energy range. Ref. 42 is developing into the standard compilation for most heavy ion
studies, while Ref. 41 is especially suitable for very high Z particles.
Low-energy, high Z particles may also have a sizable nuclear stopping power due to
elastic nuclearnuclear interactions [4042]. These processes can be thought of as billiard-
ball type collisions between the nucleus of the incident ion and the nuclei of the atoms
making up the medium molecules. These collisions occur at very low energies and can
usually be ignored, e.g., nuclear stopping equals electronic stopping for carbon ions in water
at energy of about 20 keV. On the other hand, the nuclear component can be considerable
with high Z particles and should be included in range or stopping power determinations,
e.g., nuclear stopping equals electronic stopping for uranium ions in water at energy of
about 3.5 MeV. However, no specic chemistry of water radiolysis has been identied un-
equivocally to the nuclear stopping, so it will not be treated separately. The nuclear stopping
power is not to be confused with nuclear fragmentation. The latter process occurs at very
high energies and leads to fragmentation of the incident heavy ion.
General trends of the stopping power can be easily deduced from Eq. (1). The veloc-
ity term in the numerator of the logarithmic term and in the denominator of the pre-
logarithmic term gives rise to the familiar Bragg peak, i.e., with decreasing velocity of the
incident particle the stopping power increases to a maximum and then decreases at lower
velocities. Specically, the Bragg peak refers to the increase in ionization toward the end
of the particle track since he rst noticed this eect in the ionization of gases [43]. How-
ever, the expression is also commonly used in reference to the stopping power. The dis-
tinction is negligible in water because ionization is the dominant outcome due to energy
loss by the incident heavy ion. The stopping powers of water for ve dierent ions are shown
in Fig. 1 as a function of incident energy [42]. Also shown in this gure is the stopping power
for fast electrons [44]. The important aspect of the stopping powers is the huge dierence
between heavy ions and electrons. Heavy ion stopping powers are typically several orders of
magnitude higher than that for fast electrons. There are two major consequences due to the
high rate of energy loss for heavy ions that aects the radiation chemistry: the very short
range and the track structure.
The dotted lines in Fig. 1 show the stopping powers for the dierent ions at a
constant velocity in units of MeV/amu. This unit of energy is very often used in heavy ion
radiolysis and it is based on the classical formula for kinetic energy, E =
1
2
MV
2
, where M
is the heavy ion mass. As seen in Eq. (1), the ion velocity is a dominant parameter in
energy loss processes and the MeV/amu energy unit is more convenient to use than con-
verting to absolute velocity units. Remember that MeV/amu is actually proportional to
the square of the velocity.
Copyright 2004 by Taylor & Francis Group, LLC
There are several methods used to describe the range of an energetic heavy ion. The
penetration, projected range, or depth refers to the mean displacement in the initial di-
rection of motion and is the most commonly used parameter in radiation chemistry.
Scattering will always add a radial component to the trajectory, but the magnitude of this
component is usually small for ions of energies typically used in radiation chemistry ex-
periments. Therefore the range of heavy ions often can be determined from the continuous
slowing down approximation, csda, by the following equation.
R
csda
Z
E
o
0
dE= dE=dx 2
where E
o
is the initial energy, and the stopping power, dE/dx, is obtained from one of the
compilations. Fig. 2 shows the ranges corresponding to the ions in Fig. 1. The extremely
short ranges make experiments with heavy ions dicult. A 1-MeV electron has a range of
about 4 mm in water and experiments are relatively easy to conduct. On the other hand, a
1-MeV proton has a range of only 24 Am. Window materials and sample thickness become
major concerns at these ranges. A proton would need about 20 MeV of energy to have the
same range as a 1-MeV electron, and other heavier ions would need considerably higher
energy. The ranges given in Fig. 2 are csda ranges and somewhat over estimate the true
penetration. Applications requiring greater accuracy should use Monte Carlo simulations
of heavy ion trajectories. However, these simulations require detailed knowledge of the
elastic cross sections, which are not known for heavy ions in many materials. In practice,
the csda ranges are usually sucient for the determination of penetration, but the true
range should always be checked experimentally.
The electronic stopping power involves a collision between the heavy ion nucleus and
an electron of the medium. The reduced mass of the encounter can always be assumed to
Figure 1 Stopping power, dE/dx, of some heavy ions [42] and electrons [44] in water as a function
of energy. The dotted lines show the stopping power for heavy ions of equal velocity.
Copyright 2004 by Taylor & Francis Group, LLC
be equivalent to that of the electron, even with incident protons. Therefore energy loss
depends on the velocity and not the mass of the incident ion. In other words, there are
really no isotope eects for radiolysis with ions of the same element, provided they have
the same velocity. For example, protons and deuterons of equal velocity give similar
radiation chemical yields [45]. Subtle mass eects on the stopping power of heavy ions
are predicted in more detailed theories [35,46], but their eects are small and have not yet
been observed in normal water radiolysis.
The charge and velocity of the incident particle are important parameters in the
characterization of energy loss processes as predicted by Eq. (1). The velocity can usually
be obtained by energy or time of ight analysis, but the charge is more dicult to deter-
mine [4749]. It is obvious from Eq. (1) that for two ions of the same velocity but with
dierent charges, e.g., He
+1
and He
+2
, the ion with twice the charge will have four times
the stopping power. Experiments with gas targets show the expected dependence of the
cross section on the charge of the incident particle [50]. However, in gases one is dealing
with a single isolated collision. The rates of capture and loss of electrons in condensed
media are very rapid and the heavy ion attains an equilibrium charge state depending on
its velocity [5153]. Strictly, each individual heavy ion has an integral charge, but the
probability of nding that charge state depends on the velocity. However, an average
charge per heavy ion is often used in place of Z in stopping power formulas such as Eq. (1).
Experimental studies have led to several empirical equations for the average eective
charge, Z
e
, of heavy ions. The following formula is accurate over a wide energy range of
greater than 0.3 MeV/amu [48,49].
Z
e
=Z
o
1 exp 6:35 v=Z
4=3
o
1=2
3
Figure 2 Range, csda, of some heavy ions [42] and electrons [44] in water as a function of energy.
The dotted lines show the range for heavy ions of equal velocity.
Copyright 2004 by Taylor & Francis Group, LLC
The energy, v, is in units of MeV/amu and Z
o
is the heavy ion atomic number. This for-
mula assumes that a proton at the same energy is completely stripped and should not be
used at energy less than 0.3 MeV/amu. Corrections can be applied to extend Eq. (3) to
lower energies [48]. Fig. 3 shows the relative eective charge for several heavy ions as a
function of their energy. It can be seen that the lighter ions are generally stripped at ener-
gies above 1 MeV/amu. The core electrons of heavy nuclei are tightly bound and con-
siderably more energy is required to strip them in a collision process.
Energy loss due to charge exchange processes and the equilibrium charge are in-
cluded in the total stopping power found in compilations. This energy does not have to be
considered explicitly. There have been several references to track end eects that are
presumably due to charge exchange or other processes occurring at the end of the heavy
ion track. Schuler [54] once suggested that the excess production of molecular hydrogen
in benzene is due to charge exchange processes. However, later calculations using more
reliable cross sections indicate that charge exchange processes are insucient to explain all
of the hydrogen produced [55]. No other chemistry has been shown to be due exclusively
to charge exchange processes.
3. FUNDAMENTALS OF TRACK STRUCTURE
Many of the observed radiation chemical eects in water are due to the geometry of the
physical energy deposition events, commonly referred to as the track structure. The obser-
vation that the energy absorbed in the medium was more important for determining the
resultant chemistry than the energy loss by the incident particle was rst proposed by
Zirkle et al. who coined the term linear energy transfer, LET, to indicate this signicance
[56]. Of course, the track structure is largely determined by the energy loss processes.
Linear energy transfer is usually assumed to be equivalent to the stopping power of the
Figure 3 Relative eective charge of some heavy ions in water as a function of incident energy
(Eq. (3)). The dotted lines show the relative eective charge for heavy ions of equal velocity.
Copyright 2004 by Taylor & Francis Group, LLC
medium, which is almost always valid for bulk liquid water. However, one should use
caution in heterogeneous systems or in isolated nanometer sites. Track structure eects
are commonly called LET eects because most of the early studies used this parameter
to characterize the dierent yields for the various heavy ions. Linear energy transfer has
continued to be a dominant parameter in the radiation chemistry of heavy ions, but even
in the earliest studies it was apparent that LET was not the only factor that determined
product yields [57,58]. The variation in product yields for two ions at the same LET was
correctly ascribed by early investigators to subtle dierences in the track structures [16,
57,58].
The medium absorbing the energy loss by the ionizing radiation is almost always
structured. Water is made up of nite dimensioned molecules. Stopping power formulas
such as Eq. (1) give average rates of energy loss to a continuum of bound electrons with no
regard to the molecularity of the medium. In normal systems, energy is lost by the incident
radiation in stochastic processes leading to isolated regions or clusters of excited states and
ion pairs of medium molecules. The early cloud chamber pictures of Wilson in the 1920s
[59,60] show that h-particle tracks are made up of well-separated ionization events, except
near the end of the tracks. The cloud chamber tracks of a-particles appear as solid,
continuous strings of ionization events. Theories developed in the 1940s and 1950s used
these visualizations as aids to describe radiation chemical events due to the nonhomoge-
neous distributions of reactive species [12,61,62].
The track structure is determined to a large extent by the distribution of energy ab-
sorption by the medium, which is well characterized by the dipole oscillator strength
distribution, f(e). Classically, the dipole oscillator strength distribution gives the number
of electrons of the medium capable of absorbing a photon of energy e. The complete inte-
gral of the dipole oscillator strength distribution over all energies is equal to the total num-
ber of electrons in a molecule, ZV. Phase eects and other characteristics of the medium
are included in the dipole oscillator strength distribution. A large number of molecules,
even in the dierent phases, have distributions comparable to that for liquid water as
shown in Fig. 4 [63]. Therefore the overall track structures of ionizing radiation in most
liquid media are similar. Subtle dierences will be found because the oscillator strength
distributions are not identical for all media and other mechanisms, such as channeling in
crystals, may have a signicant contribution. The density of the medium has a direct eect
on energy loss processes, which is the reason why there is little meaning of track structure
in the gas phase.
The dipole oscillator strength distribution does not directly correlate to energy loss
by charged particles. However, Monte Carlo simulations of ionizing radiation using cross
sections incorporating the dipole oscillator strength distribution can give this information
[63,64]. Fig. 4 shows the frequency of a given energy loss for 1-MeV electrons in liquid
water. The similarity between the energy loss distribution and the dipole oscillator strength
distribution is obvious with both showing peaks at about 20 eV. There is a wide range of
possible collisions due to the passage of a fast electron in water, but it can be seen from
Fig. 4 that the most probable events involve energy losses of less than 100 eV.
The mean energy loss by a fast electron in liquid water is about 60 eV and somewhat
independent of the phase and the initial electron energy [63,64]. Collisions involving this
magnitude of energy loss will produce secondary electrons that further lead to an average
of one or two ionizations [65]. Low-energy electrons have relatively large total elastic cross
sections with a substantial backscattering component [6668]. In other words, electrons of
a few tens of eV do not go very far in liquid water and show a nearly isotropic angular
distribution as they thermalize [69]. The electrons will thermalize and eventually become
Copyright 2004 by Taylor & Francis Group, LLC
hydrated, adding another spatial delocalization from the initial energy loss event. The net
result is that the initial energy loss event leads to a cluster, or spur, of two or three ion-
izations spatially localized. The concept of a spur in the liquid phase of water is very old
[61,62]. The existence of spurs in water manifests itself in the characteristic time depen-
dence of the decay of reactive species produced by water decomposition [70]. The spur is
still the main entity used by radiation chemists to explain kinetic eects with low LET
radiation. Solvation of the electron does not occur in the gas phase and the delocalization
of the electron during this process makes up a substantial fraction of the size of the spur.
The spur has no counterpart in the gas phase, which causes much confusion when applying
gas phase theory and cross sections to the liquid phase.
Following the initial ionization events, water decomposes within a few picoseconds
to give a spur composed of hydrated electrons, OH radicals, H atoms, H
2
, and H
2
O
2
[16].
The nonhomogeneous distribution of water decomposition products making up the spur
relaxes by diusion and reaction of the sibling radicals and molecules. A combination of
direct observations of the hydrated electron decay with model calculations can be used to
estimate the average geometry of the spur [71]. Obviously, each spur is slightly dierent.
However, even a single 1-MeV electron produces greater than 10
4
spurs so that its chem-
istry appears as an average over all spurs. Time decay measurements using low-energy
multiphoton ionization just above the ionization potential of water have also been used to
estimate the distribution of low-energy electrons in water [72]. These latter experiments
agree well with the predicted results for the typical spur produced by fast electrons. The
observed time decay of the hydrated electron coupled with Monte Carlo calculations
suggests that the typical spur in electron radiolysis is well approximated by a Gaussian
distribution with a characteristic radius of 45 nm [73].
The average energy loss per unit track length of a 1-MeV electron is about 0.2 eV/nm
[64]. With an average energy loss per collision event of 60 eV the mean separation of spurs
is 300 nm, which is much too far apart for interspur reactions. (It is assumed throughout
Figure 4 Relative dipole oscillator strength distribution, f(e), for liquid water [63] and frequency of
a given energy loss by 1-MeV electrons in liquid water [64].
Copyright 2004 by Taylor & Francis Group, LLC
this article that incident particles are isolated from each other, an assumption not neces-
sarily correct at very high dose rates or with very short pulses of intense beams.) On the
other hand, a 5-keV electron has an average LET of 3 eV/nm, which corresponds to an
initial average separation of only a few spur diameters. The spurs produced by an electron
of this energy are not initially overlapping, but they will overlap soon thereafter as they
develop in time. The resulting geometry leads to a short track as originally dened by
Mozumder and Magee [20]. These authors suggested that even lower energy electrons
form blobs, which are elliptical or stretched spurs. All of these geometric shapes are meant
to be aids in predicting the resultant chemistry and should not be considered as xed
entities of nite dimensions.
The tracks of heavy ions can be explained in much the same fashion as that for fast
electrons, e.g., considering the primary energy loss followed by the delocalization due to
the multiple secondary electron thermalization and solvation. If the LET is greater than 3
eV/nm the spurs will be formed overlapping or will overlap shortly thereafter. At LET
much less than 3 eV/nm the spurs will exist independently of each other and the observed
chemistry of a heavy ion should be much like that of a fast electron. Monte Carlo simu-
lations suggest that the energy loss distributions for high velocity heavy ions are very
similar to that for fast electrons, i.e., the frequency of a given energy loss for a heavy ion is
nearly the same as shown in Fig. 4 for fast electrons (S.M. Pimblott, unpublished results).
Collisions by heavy ions have slightly dierent kinematic parameters than fast electrons.
For instance, an electron can lose up to half its energy in a collision, whereas the maximum
energy loss by a heavy ion is 4mE/M, where m is the electron mass, and E and M are the
heavy ion energy and mass, respectively. For both electrons and heavy ions, the energy
loss is largely determined by the dipole oscillator strength distribution, which is signi-
cantly peaked at low energy. The result is that almost all types of ionizing radiation pro-
duce large amounts of low-energy electrons with distributions in water similar to that
shown in Fig. 4. However, the initial collision process strongly determines the track struc-
ture and resultant chemistry because it determines the spacing of the clusters of low-energy
electrons.
Recent studies have applied Monte Carlo techniques well established for electrons to
heavy ions [74]. The predicted density normalized mean free path, K, of several heavy ions
is given in Fig. 5. It can be seen that the mean free path mirrors the stopping power by
decreasing with decreasing energy and passing through a minimum at low energies. Spur
sizes can vary widely, but from the discussion above a typical spur can be assumed to have
a diameter of about 10 nm. Fig. 5 shows the relationship between a spur diameter and the
mean free path of several ions. Low-energy heavy ions will produce primary ionization
events well within the diameter of a spur. Spatial delocalization due to thermalization and
solvation of the secondary electrons will result in one continuous track with a columnar-
like structure. Experimentally it has been observed that radical and molecular yields with
protons above about 20 MeV, 3 eV/nm, are virtually the same as with fast electrons
[75,76]. This observation agrees well with predictions based on the mean free path shown
in Fig. 5. Even higher energies are required for a helium or carbon ion track to resemble
that of a fast electron. These expected trends are observed experimentally, as will be shown
below. It has been speculated that the charge on an ion may slightly alter the distribution
of species within the spur [77], but no such process has been conrmed experimentally.
For a given medium such as water, its density and the velocity of the incident heavy
ion contribute to the formation of the columnar track structure. The spacing of the pri-
mary energy loss events must be within the delocalization of the secondary electrons for
eects attributed to high LET to occur. For this reason, typical LET eects will not be
Copyright 2004 by Taylor & Francis Group, LLC
observed in the gas phase. Simple density scaling of experimental or theoretical data
between the gas and condensed phases should be performed with caution.
Fig. 5 shows that the mean free path of low-energy carbon ions is less than that of a
water molecular diameter. Calculations based on such short mean free paths may predict
two energy loss events within the same molecule. However, as discussed elsewhere in this
book, the double ionization cross section is much lower than predicted by these results.
The problem is that cross sections are normally based on a single isolated collision at
gaseous density. Extrapolation to the condensed phase can lead to unrealistic predictions.
The column of species dened by the overlapping spurs along the path of a high
LET particle makes up what is commonly referred to as the track core [21]. Its physical
parameters are dicult to quantify and it has no corresponding entity in the gas phase.
Obviously, some energy loss collisions will be violent enough to form true y-rays, i.e., sec-
ondary electrons with enough energy to make tracks of their own. Various attempts have
been made to separate the track core from the region of radiation eects due to secondary
electrons (a region sometimes called the penumbra) [21,22,78,79]. There are at least two
problems with such an attempt. First, there is no clear demarcation between the two regions,
either by energy or chemically. The result has been confusion and dierent core denitions
for various chemical systems. A second diculty in dening a track core, or any track
parameter, is that the track is dynamic. The track is constantly expanding in time due to
diusion of the transient species and reactions initially associated with the track core may
envelop those due to the secondary electrons. Despite the inherent problems, some radiation
chemical problems have been successfully addressed by treating each part of the track sepa-
rately [7983]. This approach to heavy ion track modeling is advantageous because of
its relative simplicity. However, Monte Carlo calculations are required to completely eluci-
date the track structure and resulting chemistry. The few published Monte Carlo studies on
heavy ions are quite promising [74,84,85].
Figure 5 Energy dependence of the density normalized mean free path for e
,
1
H,
4
He, and
12
C
ions in liquid water. The dotted line shows the water diameter and the dashed line shows the typical
spur diameter. (From Ref. 74.)
Copyright 2004 by Taylor & Francis Group, LLC
4. CHEMICAL EFFECTS OF TRACK STRUCTURE
This section will discuss general eects due to track structure that apply to most condensed
phases. Water radiolysis will be used in specic examples because more data exist for this
chemical system than any other. As discussed above, the earliest experiments showed that
there were dierences in product yields for various ionizing radiations in liquid water.
With increasing particle LET there is an increase in the yields of molecular products such
as H
2
and H
2
O
2
and a corresponding decrease in the yields of radicals such as e
aq
, H, and
OH [16,86]. The nonhomogeneous distribution of reactive species initially produced by the
energy deposition relaxes in time toward a homogeneous distribution. In pure water, a
competition is quickly established between radicalradical combination reactions among
the sibling radicals and radical diusion into the bulk medium. Increasing the particle LET
increases the probability of radicalradical combination reactions with a corresponding
increase in molecular products.
It has always been assumed that the initial decomposition of water is independent
of particle type [8789]. There are experimental results that seem to support this con-
clusion. Pulsed experiments with protons nd that the short time yield of the hydrated
electron is similar to that found with fast electrons [90,91]. In addition, W-values, the aver-
age energy required per ion pair, for high-energy particles in a given medium are nearly
independent of the particle type [92]. Ion pairs are the main initial species formed in the
radiolytic decomposition of water so one would expect similar yields for the products due
to these species. In special circumstances the initial product yield may be dependent on a
particle type, e.g., low-energy heavy ions with large contributions from nuclear scattering
processes or radiation with a high probability of causing multiple ionization of the medium
molecules. For the most part, observed variations in product yields of a given medium with
radiation type or energy are due to eects of track structure on the radiation chemistry.
There are a few important details that must be addressed in discussing product yields
with heavy ions because of their extremely short range. For example, a 10-MeV proton
(LET = 4.7 eV/nm) has a range in water of 1.2 mm while a 10-MeV carbon ion (LET = 700
eV/nm) has a range of 13 Am. Both particles would be completely stopped in the solution
in most practical applications. The chemistry measured in such a conguration is an aver-
age over all particle energies from the incident particle energy, E
o
, to zero. Radiation
chemical yields, G-values, are given as the number of molecules formed or lost per 100 eV
deposited, and track segment yields, G
i
, are the radiation chemical yields in a discrete seg-
ment of path length. Within this small path segment the energy, LET, and other param-
eters of the heavy ion are well dened and constant. Model calculations normally predict
track segment yields while experiments usually, but not always, give track average yields,
G
o
. The dierence between track average and track segment yields can sometimes be sig-
nicant. The denition of the track segment yield is G
i
=d( G
o
E
o
)/dE
o
= G
o
+E
o
dG
o
/dE
o
.
It can be seen that track segment yields approach track average yields as the initial par-
ticle energy approaches zero or if the track average yield is independent of energy. Track
segment yields are often equivalent to track average yields, but such equality should not
be assumed.
Fig. 6 shows a plot of G
o
E
o
for HO
2
production as a function of the initial energy
of protons [75]. The proton was completely stopped in all the experiments and the track
average yield of HO
2
measured. The traditional integral or track averaged G-value at
10 MeV is given by the slope of the solid line to be 0.030 molecules/100 eV. The dier-
ential or track segment yield at 10 MeV is given by the slope of the dotted line to be
0.023 molecules/100 eV. This 30% dierence can be signicant in some applications. As
Copyright 2004 by Taylor & Francis Group, LLC
the proton energy approaches zero, the two types of yields approach each other. However,
one must remember that the Bragg peak occurs at the lower energies and yields are not
expected to be the same on both sides of it. There is as yet no satisfactory way to deal with
this problem. The best technique for obtaining dierential yields is to t the data as pre-
sented in Fig. 6 and determine the derivative [75]. However, one needs accurate yield
measurements at appropriate energy intervals, data which are seldom available.
There is a considerable amount of eort required to experimentally measure the
energy dependence of product yields. One solution is to use high energies where the long
ranges allow the direct measurement of track segment yields. However, nuclear frag-
mentation also occurs at high energies, which complicates interpretation of the data. The
track average yields are often the only measured values, especially in the early literature.
Track segment yields are often presented as a function of LET, especially when comparing
the results for dierent types of ions. The correctness of using LET will be discussed later,
but it is denitely incorrect to couple track average G-values with track segment LET. The
track average LET is dened as 1=E
o
m
E
o
0
dE=dx dE, which correctly weighs the LET
over the entire track. This method of representation of the data is by far the most common
encountered in the literature.
Product yields with heavy ions are generally presented as a function of the dier-
ential or integral LET of the irradiating particle depending on the type of G-value. How-
ever, LET is not the sole parameter that describes the track structure and thereby the
observed yields [16,57,58]. Many experimental observations have shown that for two inci-
dent particles at the same LET radical yields are greater and molecular yields are lower for
the ion with a higher charge [57,58,77,93]. Fig. 7 shows an example of the LET eect on the
production of HO
2
and O
2
in the radiolysis of water [75,9496]. The experiments actually
convert HO
2
to O
2
so both are measured, but HO
2
is the dominant species. Although HO
2
is a radical, it behaves like a molecular species as its yield increases with increasing LET.
Figure 6 The production of HO
2
in the proton radiolysis of water as a function of initial ion
energy [75]. The slope of the solid line (0.030) is the track average yield while the slope of the dotted
line (0.023) is the track segment yield for 10-MeV protons.
Copyright 2004 by Taylor & Francis Group, LLC
Many of the expected track eects discussed above are observable with this system. For
instance, high-energy protons give about the same HO
2
yield as fast electrons because they
both deposit energy in isolated spurs. One can readily observe that LET is not a unique
parameter for describing yields.
Equation (1) predicts to a rst approximation that for two dierent types of particles
to have the same LET the one with the higher charge will have a higher velocity. A higher
incident particle velocity produces a larger radial distribution of energy deposition in part
because the distribution of secondary electrons increases with an increase in the heavy
ion velocity. Several theoretical studies have examined the average radial distribution of
energy about a heavy ion track and the results agree with the observations [21,22,79,97,98].
Although two heavy ions may deposit the same amount of energy per unit path length the
ion with higher charge will lose that energy in a larger volume, which leads to a less dense
track of reactive species. Dierent media and even dierent products in a given media can
exhibit a variety of dependencies on particle Z at a given LET.
Early studies on biological media suggested that there must be a better parameter
than LET for characterizing the response to heavy ions [99]. These observations led to the
proposal that Z
2
/b
2
would be a better parameter, where b is the ratio of the ion velocity to
the speed of light [100]. The rationale for such a parameter is that the radial energy dis-
tribution should be related to the maximum energy of the secondary electron, which is
proportional to b
2
or E/M. Equation (1) suggests that the rate of energy loss along the
particle path is proportional to Z
2
times the corresponding value for a proton of the same
velocity. Combining the estimates for radial and axial energy loss gives a rst approx-
imation to the energy density for nonrelativistic ions as Z
2
/b
2
. At very low energies, the
parameter Z should be replaced by the eective charge of the heavy ion, Z
e
, as discussed
above and given by Eq. (3). Several sets of radiation chemical data have been examined
Figure 7 Track average yields of HO
2
as a function of heavy ion LET: (n,E,.)
1
H,
4
He, and
12
C,
Ref. 75; (o)
12
C, Ref. 94; (O)
12
C, Ref. 96; (+)
58
Ni, Ref. 95. The dotted line is the limiting fast
electron yield of 0.02 [123].
Copyright 2004 by Taylor & Francis Group, LLC
with some success using this parameter [101104]. However, it was also pointed out that
this parameter does not work in some instances [102]. Monte Carlo techniques have been
applied to heavy ion tracks in liquid water in order to examine energy loss in detail [74].
Fig. 8 shows the results from that work for the radial energy density or the fraction of
energy loss as a function of the radial distance. The radial distributions of energy about
protons and helium ions are found to be similar for ions of equal velocity, but not for
equal LET [74]. The results also show that the response of the Fricke dosimeter more
Figure 8 Radial energy deposition rate and fraction of total energy loss in a track segment as a
function of radial distance for (solid line) 1 MeV
1
H and for
4
He ions of (oooo) equal velocity and
() equal LET. (From Ref. 74.)
Copyright 2004 by Taylor & Francis Group, LLC
closely follows the parameter MZ
2
/E than LET. This parameter is equivalent to Z
2
/b
2
but
easier to visualize from common units used to describe heavy ions. The next section will
show that the MZ
2
/E parameter gives nearly universal agreement for all track average
product yields of water radiolysis at long times.
The eects of track structure on product yields are often discussed in terms of one
parameter or another that is related to the energy loss of the heavy ion. This approach
implies a static track, which is not true. There is a tremendous amount of diusion and
reaction of reactive species following the energy loss processes. One of the biggest chal-
lenges for experimentalists and theoreticians is determining the temporal history of prod-
ucts in water radiolysis with heavy ions. Not only do the time dependencies put severe
limitations on track models but also they can be vital for predicting radiation damage in
concentrated systems that cannot be probed directly. Several pulsed heavy ion experiments
have directly measured water product yields at specic times [90,91,105113]. Except for
the work of Burns and coworkers who used water jets, these experiments are limited to
determining the product yields on the microsecond time domain. The major diculty with
pulsed experiments is the need for relatively large doses in narrow pulses and the very
short heavy ion range. These limitations restrict the number of systems that can be exam-
ined and the time scales. Competition kinetics can be used to determine the time scale of
chemical reactions at high LET where direct observations are not possible. Solutes that are
radical scavengers will stop the competition between radical reaction and diusion at a
time characteristic of the particular scavenging reaction. By changing the solute concen-
tration one can eectively probe the chemistry in the particle track at dierent times.
Even most of the pulsed experiments rely on scavengers to stop the chemistry at short
times so product yields can be observed with pulses at longer times. Several experiments
using scavenger techniques have been able to determine the temporal variation of radicals
and molecular products from nanoseconds to microseconds using a variety of heavy ions
[114118]. Considerably more work is required to obtain a complete set of data for all
products of water radiolysis.
5. HEAVY ION RADIOLYSIS OF WATER
The only true method to characterize a product yield in the radiolysis of water is to have the
complete temporal dependence for a particular heavy ion with a given energy. The track is
constantly evolving in time due to diusion and reaction of the reactive species and in many
instances the yields are never really constant. This result is very dierent than fast electron
or g-radiolysis where the yields of transients can be considered to be invariant with time
from a few microseconds to milliseconds or longer. Track processes are occurring at times
less than a microsecond, while homogeneous reactions begin to dominate after a few min-
utes and products approach steady-state concentrations. The transition from track pro-
cesses to homogeneous chemistry occurs quickly with heavy ions. It is sometimes dicult
to consider the yield of a product as a single valued entity.
Product yields in the radiolysis of water are required for a number of practical and
fundamental reasons. Model calculations require consistent sets of data to use as bench-
marks in their accuracy. These models essentially trace the chemistry from the passage of
the incident heavy ion to a specied point in time. Engineering and other applications often
need product yields to predict radiation damage at long times. Consistent sets of both the
oxidizing and reducing species produced in water are especially important to have in order
to maintain material balance. Finally, it is impossible to measure the yields of all water
Copyright 2004 by Taylor & Francis Group, LLC
products for every heavy ion and energy. Therefore one must to be able to interpolate to
other ions or energy for which no data exist.
The next few subsections will present data for the products in the radiolysis of water
with a wide variety of heavy ions. Product yields with heavy ions are rarely constant in time,
but most of them vary slowly in the microsecond region. Furthermore, most chemical
systems used to probe yields with heavy ions were developed for examining fast electron or
g-radiolysis in the microsecond region. The following discussion on product yields can be
assumed to apply to the microsecond time regime. Because a wide range of systems are used
to determine product yields, consistency between the dierent experiments can be obtained
by examination of the material balance. The net decomposition of water can be obtained by
setting the number of H atoms in each product equal to the number of H atoms in the net
water decomposed to form that product and similarly for O atoms [16].
2GH
2
O 2GH
2
Ge
aq
GH GHO
2
2GH
2
O
2
GOH 4
GH
2
O 2GH
2
O
2
GOH 2GHO
2
2GO
2
5
Combining these two equations gives the material balance between the oxidizing and
reducing species formed in the decomposition of water.
2GH
2
Ge
aq
GH 2GH
2
O
2
GOH 3GHO
2
4GO
2
6
Molecular oxygen formation is probably negligible at all but the highest LET and HO
2
can
be normally ignored with fast electrons and g-rays.
The discussion in the previous section suggests that the track of a heavy ion becomes
more like that of a fast electron with increasing velocity. Therefore one expects that in the
high velocity limit the yields of water products with heavy ions are the same as with fast
electrons or g-rays. The yields for the major products of water radiolysis in fast electron or
g-radiolysis are given in Table 1. These values were taken froma number of dierent sources
in conjunction with the results predicted by model calculations [73,116,119123]. Material
balance shows that almost four molecules of water are decomposed for every 100 eV of
energy absorbed by fast electrons or g-rays. Because only about six water molecules are
initially decomposed, most of the water products escape intraspur reactions in fast electron
or g-radiolysis.
Linear energy transfer has often been shown to be a poor parameter for character-
izing product yields. Therefore in order to compare the results for the dierent ions the
MZ
2
/E parameter has been used. This parameter has been successfully used in the past and
seems to work well for the data presented here. Note that an increase in MZ
2
/Ecorresponds
to an increase in LET so product trends as a function of MZ
2
/E presented here are
comparable to similar discussions in the literature using LET. Unfortunately, there is no
good formalism to t product yields as a function of MZ
2
/E. The solid lines in Figs. 914
Table 1 Product Yields in the Radiolysis of Water at about 1 Asec
Particle e
aq
H
2
H OH H
2
O
2
HO
2
H
2
O
Fast electron/g
(reference)
2.5
(73, 119)
0.45
(120)
0.56
(119, 121)
2.5
(73, 122)
0.70
(116)
0.02
(123)
3.94
1 MeV
1
H 0.48 0.94 0.31 0.67 0.91 0.06 2.61
5 MeV
4
He 0.26 1.12 0.12 0.38 0.95 0.10 2.48
l 0.12 2.02 0.0 0.0 0.97 0.56 3.06
Copyright 2004 by Taylor & Francis Group, LLC
Figure 10 Track average yields of H
2
as a function of MZ
2
/E: (n)
1
H,
4
He, Ref. 121; (x)
4
He,
7
Li,
Ref. 26; (.)
1
H,
4
He, Ref. 131; (o)
1
H,
4
He,
14
N,
20
Ne, Ref. 126; (5)
1
H,
4
He, Ref. 45; (+)
1
H, Ref.
127;()
14
N, Ref. 128; (1) , Ref. 129. The dotted line is the limiting fast electron yield of 0.45 [120].
The dashed lines are for 1 MeV
1
H and 5 MeV
4
He ions.
Figure 9 Track average yields of e
aq
as a function of MZ
2
/E: (n)
1
H,
4
He, Ref. 121; (x)
4
He,
7
Li,
Ref. 26; (.)
4
He, Ref. 115; (D)
10
B(n,a)
7
Li, Ref. 124; (5)
10
B(n,a)
7
Li, Ref. 125; (+,)
58
Ni,
238
U, Ref.
93. The dotted line is the limiting fast electron yield of 2.5 [73,119]. The dashed lines are for 1 MeV
1
H and 5 MeV
4
He ions.
Copyright 2004 by Taylor & Francis Group, LLC
Figure 12 Track average yields of OH radical as a function of MZ
2
/E: (x)
4
He,
7
Li, Ref. 26; (.)
4
He, Ref. 114; (.)
1
H,
12
C, LaVerne, unpublished results; (o)
1
H,
4
He,
14
N,
20
Ne, Ref. 126; (5)
1
H,
4
He, Ref. 45. The dotted line is the limiting fast electron yield of 2.5 [73,122]. The dashed lines are for
1 MeV
1
H and 5 MeV
4
He ions.
Figure 11 Track average yields of H atom as a function of MZ
2
/E: (n)
1
H,
4
He, Ref. 121; (x)
4
He,
7
Li, Ref. 26. The dotted line is the limiting fast electron yield of 0.56 [119,121]. The dashed lines are
for 1 MeV
1
H and 5 MeV
4
He ions.
Copyright 2004 by Taylor & Francis Group, LLC
Figure 13 Track average yields of H
2
O
2
as a function of MZ
2
/E: (n)
1
H,
4
He, Ref. 132; (x)
4
He,
7
Li, Ref. 26; (.)
1
H,
4
He,
12
C, Ref. 116; (o)
1
H,
4
He,
14
N,
20
Ne, Ref. 126; (5)
1
H,
4
He, Ref. 45; (1)
, Ref. 133. The dotted line is the limiting fast electron yield of 0.7 [116]. The dashed lines are for 1
MeV
1
H and 5 MeV
4
He ions.
Figure 14 Track average yields of HO
2
and O
2
as a function of MZ
2
/E: (.)
1
H,
4
He,
12
C, Ref. 75;
(o)
12
C, Ref. 94; (O)
12
C, Ref. 96; (+)
58
Ni, Ref. 95; (j)
1
H,
4
He,
14
N,
20
Ne, Ref. 126; (D)
1
H,
4
He,
Ref. 121; (+) , Ref. 133. The dotted line is the limiting fast electron yield of 0.02 [123]. The dashed
lines are for 1 MeV
1
H and 5 MeV
4
He ions.
Copyright 2004 by Taylor & Francis Group, LLC
were obtained by tting the data for each product Y to the function G( Y) = G
e
( Y) +
( G
l
( Y)G
e
( Y))(ax)
n
/(1 + (ax)
n
), where x is MZ
2
/E, a and n are tting parameters, and
G
e
and G
l
are fast electron and innite MZ
2
/E (LET) yields, respectively. This function
works well in the inection region where most of the data are found, but its approach to the
limiting values is not good. Therefore the values of the parameters a and n will not be given,
but the value of n generally varied from 0.4 to 1.4.
5.1. Hydrated Electron
Hydrated electrons have been widely studied in fast electron radiolysis using pulsed
techniques. Only a few pulse radiolysis studies have examined hydrated electron formation
with heavy ions [90,81]. Burns and coworkers used 3-MeV protons and reported yields out
to 70 nsec, which is too short to compare to the results with solutes [90]. The work of Nale-
way et al. measured track segment yields in the microsecond region using time resolved
techniques [81]. Track segment yields were also measured by Appleby and coworkers [103].
All of the yields for the hydrated electron presented in Fig. 9 are track average yields
obtained in scavenging experiments [26,115,121,124,125]. Nitrous oxide and nitrate are
most often used as scavengers of the hydrated electron. Fig. 9 shows that there is good
agreement for hydrated electron yields over a wide range of MZ
2
/E values. The major
deviation occurs for the
10
B(n,a)
7
Li recoil experiments [124,125]. An average value of MZ
2
/
E for the two recoils is used, which is probably giving too high a value. Dosimetry is
sometimes a problem in the use of recoil ions.
Hydrated electron yields decrease with increasing MZ
2
/E, but they do not seem to
decrease to zero. Experiments have been performed on aerated and deaerated Fricke
dosimeter solutions using
58
Ni and
238
U ions [93]. One half of the dierence in the ferric
ion yields of these two systems is equal to the H atom yield. The Fricke dosimeter is highly
acidic so the electrons are converted to H atoms and to a rst approximation the initial
H atom yield can be assumed to be zero (see below). There is considerable scatter in the
data of the very heavy ions, but they seem to indicate that hydrated electron yields de-
crease to a lower limit of about 0.1 electron/100 eV. The hydrated electron distribution is
wider than that of the other water products because of the delocalization due to solvation.
This dispersion probably allows some hydrated electrons to escape the heavy ion track at
even the highest value of MZ
2
/E.
5.2. Molecular Hydrogen
The yields of H
2
are usually straightforward to measure using gas detection techniques. A
small amount of scavenger is added to stop the OH radical back reaction from occurring,
which would decrease the yields. Fig. 10 shows the results for H
2
as a function of MZ
2
/E for
the various ions [26,45,121,126130]. A track segment yield was determined at very high
energies, but not included in the gure [131]. There is some scatter in the data, but the overall
agreement is good. Molecular hydrogen yields are extremely sensitive to the scavenging
capacity of system because this species evolves over a long period of time [130]. The
scavenging capacity is equal to the product of the scavenger concentration and the
scavenging reaction rate coecient and reects the time dependence of H
2
formation. Fur-
ther detailed experiments on the H
2
yield at dierent scavenging capacity will probably help
clarify the results.
Copyright 2004 by Taylor & Francis Group, LLC
As expected, the H
2
yield increases with increasing MZ
2
/Evalues. Measurements with
ssion fragments found a limiting yield of 2.0 H
2
molecules/100 eV [129]. The maximum
value of MZ
2
/E with ssion fragments was determined from the average velocity and
eective charge of the two main fragments. Although the ssion fragment data point may
be in some error because of dosimetry, a t of data suggests that the maximum H
2
yield
is about 2 molecules/100 eV. These results show that H
2
is the major product of water
radiolysis at high MZ
2
/E values.
5.3. Hydrogen Atom
Only two studies have determined H atom yields in neutral water with heavy ions [26,121].
H atom yields are extremely dicult to determine directly because of competition with
hydrated electrons or OH radicals. Yields of H atoms are usually estimated from dier-
ences in molecular hydrogen yields using various scavengers for the H atom. Fig. 11 gives
the H atom yields as a function of MZ
2
/E. There are not many data points, but they seem
to agree with each other and show a decrease with increasing MZ
2
/E values.
The measured H atom G-value is about 0.25 at MZ
2
/E = 1, while the equivalent
yield of hydrated electrons is found at MZ
2
/E = 10. The persistence of the hydrated
electron to higher MZ
2
/E values suggests that it does not decrease to zero at an innite
value of MZ
2
/E. Most H atoms are produced in conjunction with OH radicals in the core
of the heavy ion track. The recombination rate constant is high so there is a small prob-
ability that H atoms will escape the track at high LET (MZ
2
/E). H atoms can be formed
by hydrated electron reactions and their yield cannot decrease to zero if hydrated electron
yields do not. However, hydrated electron yields are low at high MZ
2
/E values so the H
atom yield can be considered negligible in this region.
5.4. Hydroxyl Radical
OH radical yields have been measured over a wide range of energy and particle type and
the results are shown in Fig. 12 [26,45,102,114,126]. There is generally good agreement
with the data considering that the OH radical is dicult to measure. Most studies use
formate or formic acid to give CO
2
, while others rely on material balance. There are always
experimental problems with H atom reactions contributing to the observed product yields.
Intratrack combination reactions cause the OH radical yields to decrease with in-
creasing MZ
2
/E values from a G-value of 2.5 with fast electrons and g-rays. The lower
limit of OH radical yield appears to be zero for high MZ
2
/E values. OH radicals are highly
reactive and they are formed in the center of the heavy ion track so it is not surprising
that there is little probability for them to escape the track. The results also suggest that
OH radical production reactions, e.g., H atom reaction with H
2
O
2
, are negligible at high
MZ
2
/E values.
5.5. Hydrogen Peroxide
Fig. 13 gives the results of several studies on the measurement of H
2
O
2
with heavy ions
[26,45,116,126,132,133]. The yield of H
2
O
2
never varies signicantly from its value with
fast electrons and g-rays. There is a signicant amount of scatter in the data so predicting
trends is dicult. Large scatter in the data is expected for yields determined by material
Copyright 2004 by Taylor & Francis Group, LLC
balance or other manipulation of two or more sets of measurements. However, the pro-
duction of H
2
O
2
is usually determined directly using spectroscopic techniques. Hydrogen
peroxide is fairly reactive with all the radical species produced in water radiolysis. The
production of H
2
O
2
is due to OH radical combination reactions, but the yields are sen-
sitive to added scavengers for all radicals. Even variations in the concentration of hydrated
electron scavengers have been shown to aect H
2
O
2
yields [134]. This sensitivity to the
particular radiolysis conditions makes comparison of dierent sets of data dicult.
Some studies have suggested that H
2
O
2
yields reach a maximum corresponding to
the LET of 5-MeV helium ions and then decrease with increasing LET [116]. The decrease
was thought to be due to OH radical reactions with H
2
O
2
in the dense tracks at high LET.
Experiments with ssion fragments suggest that the H
2
O
2
yield does not decrease to zero
at very high MZ
2
/E values [133]. The latter result was obtained in acidic solutions, but the
outcome should be somewhat similar for neutral solutions. If this assumption is valid, then
the limiting value of H
2
O
2
at high MZ
2
/E values is about 1 molecule/100 eV, which is not
signicantly dierent from the value of 0.7 found with fast electrons and g-rays. It should
be stressed that the small variation in H
2
O
2
yields with the MZ
2
/E parameter is due to
multiple compensating reactions and not due to its inertness.
5.6. Perhydroxyl Radical
The mechanism for the production of HO
2
is probably the most uncertain in water
radiolysis. Its yield is extremely small with fast electrons and gamma rays, but HO
2
is the
most abundant radical produced at high LET. A wide variety of studies have determined
HO
2
yields over a huge range of MZ
2
/E values as shown in Fig. 14 [75,9496,121,126,133].
The data agree very well, which is expected because all the sources used ferrous sulfate
cuprous sulfate aqueous solutions. Cuprous sulfate is reduced in this system by the HO
2
to
give O
2
, which is then measured by a variety of gaseous techniques. However, O
2
produced
directly or by track reactions is also determined using this technique. Therefore the yields
generally quoted for HO
2
also include O
2
. Molecular oxygen yield is expected to be low at
all but the highest MZ
2
/E values.
The yield of HO
2
seems to increase to a value of over 0.5 molecule/100 eV at high
MZ
2
/E values. Extrapolation of the present data suggests a maximum value of 0.56, which
agrees well with the measured value of 0.5 with ssion fragments [133]. A previous work
estimated 0.57 molecules/100 eV for the track segment yield of HO
2
at innite LET. The
limiting values agree well with each other. If HO
2
has a simple precursor, such as the O
atom, then the limiting value of HO
2
at high MZ
2
/E values represents the yield of this
precursor in the initial decomposition of water. Further experimental and model studies
are necessary to separate the relative yields of HO
2
and O
2
and the mechanisms for their
formation.
5.7. Protons and Helium Ions
Many applications of heavy ion radiolysis involve a-particles or neutrons. The former are
well represented by the results for 5-MeV helium ions while the latter mainly give protons
with a distribution of energy. (Neutrons will also produce low-energy oxygen ion recoils in
water.) For the sake of comparison in this work, 1-MeV protons are assumed to represent
neutron yields. Individual G-values at other proton energies can be determined from
Figs. 914 and used in specic applications involving well described energy distributions.
Copyright 2004 by Taylor & Francis Group, LLC
Each of the gures (Figs. 914) has dashed lines showing the corresponding MZ
2
/E values
for 1-MeV protons and 5-MeV helium ions.
Table 1 contains the tabulated G-values for water products estimated for 1-MeV
protons and 5-MeV helium ions. The data are consistent with the curves in Figs. 914 and
give material balance between the oxidizing and reducing species. The major uncertainties
are in the yields for H
2
O
2
because of the large amount of scatter in the data for this
product. Net water decomposition yields determined from the data of Table 1 and Eq. (5)
show a net decrease from fast electrons to 1-MeV protons to 5-MeV helium ions. The
increasing density of reactive species in the tracks of these particles leads to a net recom-
bination of water. However, the net water decomposition seems to increase at the highest
MZ
2
/E values. The decreased yields of radicals at high MZ
2
/E values allow the molecular
products to survive the intratrack reactions and diuse into the bulk. About half of the
initial water decomposed by heavy ions is not recovered on the microsecond time scale.
6. FUTURE
Time resolved and track segment yields are important for fundamental studies on water
radiolysis with heavy ions. Advances in accelerators and laser probe systems will make
direct time resolved measurements easier. Scavenger concentration studies for obtaining
time dependencies will remain the most common experimental heavy ion technique.
Scavenger systems are amendable to track segment yield determination and the use of
older nonpulsed accelerators. Some products such as molecular hydrogen may never be
determined using time resolved techniques. Second-order reactions in the high-density
regions of heavy ion tracks may be a problem with some solutes. However, there is con-
siderable condence in the use of some of the systems that have been examined thoroughly.
New systems are constantly being evaluated and developed. All of the studies probing the
fundamental water products will be invaluable in predicting radiation damage in unex-
amined systems and for the development of track models. Detailed Monte Carlo track codes
for the radiation chemistry in heavy ion tracks must be advanced beyond the present level.
Much of the future work on heavy ion radiolysis will be largely driven by practical
applications. Corrosion processes associated with nuclear reactors and waste storage
containers will be a popular area of research because of the impact to society and the
availability of funds [11,132]. Scavenger and temperature eects on the main oxidizing
species involved in corrosion, e.g., H
2
O
2
, will be examined for application to the cooling
water in nuclear reactors. Cosmic studies, especially those related to planetary evolution,
will require extensive examination of the radical reactions in ices. Health eects ranging
from space travel, evaluation of the hazards of radon, and medical therapy will dominate
much of the work on water with biological endpoints. A large amount of work will be
performed on water in heterogeneous situations, such as absorbed water or bulk water
near solid surfaces. The interfacial eects are especially challenging because of the
dierences in initial energy distribution and in the type of radiolysis products formed.
ACKNOWLEDGMENTS
The research described herein was supported by the Oce of Basic Energy Sciences of the
U.S. Department of Energy. This document is NDRL-4424 from the Notre Dame Radia-
tion Laboratory.
Copyright 2004 by Taylor & Francis Group, LLC
REFERENCES
1. Curie, P.; Debierne, A. Comptes Rendus 1901, 132, 768.
2. Geisel, F. Ber. Deut. Chem. Ges. 1902, 35, 3608.
3. Geisel, F. Ber. Deut. Chem. Ges. 1903, 36, 342.
4. Ramsay, W.; Soddy, F. Proc. R. Soc. 1903, 72, 204.
5. Ramsay, W. J. Chem. Soc. (London) 1907, 91, 931.
6. Cameron, A.T.; Ramsay, W. J. Chem. Soc. (London) 1908, 93, 966.
7. Kernbaum, M. Radium 1910, 7, 242.
8. Duane, W.; Scheuer, O. Radium 1913, 10, 33.
9. Duane, W.; Scheuer, O. Comptes Rendus 1913, 156, 466.
10. Fricke, H.; Brownscombe, E.R. Phys. Rev. 1933, 44, 240.
11. Lind, S.C. The Chemical Eects of Alpha Particles and Electrons, Second edition; The
Chemical Catalog Company: New York, 1928.
12. Lea, D.E. Actions of Radiations on Living Cells; Cambridge University Press: New York,
1946.
13. Burton, M. Chem. Eng. News 1948, 26, 1764.
14. Bacq, Z.M.; Alexander, P. Fundamentals of Radiobiology; Butterworths Scientic
Publications: London, 1955.
15. Hart, E.J. J. Chem. Educ. 1959, 36, 266.
16. Allen, A.O. The Radiation Chemistry of Water and Aqueous Solutions; Van Nostrand: New
York, 1961.
17. LaVerne, J.A.; Schuler, R.H.; Ross, A.B.; Helman, W.P. Radiat. Phys. Chem. 1981, 17, 5.
18. Platzman, R.L. In Symposiumon Radiobiology, the Basic Aspects of Radiation Eects on Living
Systems; Nickson, J.J., Ed.; John Wiley & Sons: New York, 1952; 97 pp.
19. Platzman, R.L. In Radiation Biology and Medicine; Claus, W.D., Ed.; Addison-Wesley
Publishing Company: Reading, Mass, 1958; 15 pp.
20. Mozumder, A.; Magee, J.L. Radiat. Res. 1966, 28, 203.
21. Mozumder, A.; Chatterjee, A.; Magee, J.L. In Advances in Chemistry Series 81; Gould,
R.F., Ed.; American Chemical society: Washington D.C., 1968; 27 pp.
22. Mozumder, A. In Advances in Radiation Chemistry; Burton, M. Magee, J.L., Eds.; Wiley-
Interscience: New York, 1969; 1 pp. Mozumder, A. Fundamentals of Radiation Chemistry.
Academic Press: San Diego, 1999.
23. LaVerne, J.A. Radiat. Res. 2000, 153, 487.
24. LaVerne, J.A. Nucl. Instrum. Meth. Phys. Res. B 1996, 107, 302.
25. Kudoh, H.; Katsumura, Y. In Ion-Beam Radiation Chemistry in Radiation Chemistry: Present
Status and Future Trends; Jonah, C.D. Rao, B.S.M., Eds.; Elsevier: London, 2002; 37 pp.
26. Elliot, A.J.; Chenier, M.P.; Ouellette, D.C.; Koslowsky, V.T. J. Phys. Chem. 1996, 100, 9014.
27. McCracken, D.R.; Tsang, K.T.; Laughton, P.J. Aspects of the Physics and Chemistry of Water
Radiolysis by Fast Neutrons and Fast Electrons in Nuclear Reactors. AECL Publication 11895,
Atomic Energy of Canada Limited: Chalk River, Ontario, 1998.
28. Schimmerling, W., Budinger, T.F., Eds.; Manned Exploration of Deep Space, Cosmic Particle
Biology and Health; Lawrence Berkeley Laboratory: Berkeley, California, 1989.
29. Raju, M.R. Heavy Particle Radiotherapy; Academic Press: New York, 1980.
30. Choppin, G. In Research Needs and Opportunities in Radiation Chemistry Workshop; U.S.
Department of Energy: New York, 1998; 25 pp.
31. Rutherford, E. Philos. Mag. 1911, 21, 669.
32. Bohr, N. Philos. Mag. 1913, 25, 10.
33. Bethe, H. Ann. Phys. 1930, 5, 325.
34. Bohr, N. Det. klg. Danske Videnskabernes Selskab. 1948, 18, 1.
35. Bethe, H.A.; Ashkin, J. In Experimental Nuclear Physics; Segre` , E., Ed.; John Wiley &
Sons: New York, 1953; 166 pp.
36. Fano, U. Ann. Rev. Nucl. Sci. 1963, 13, 1.
Copyright 2004 by Taylor & Francis Group, LLC
37. Northclie, L.C. Ann. Rev. Nucl. Sci. 1963, 13, 67.
38. Inokuti, M. Rev. Mod. Phys. 1971, 43, 297.
39. Linear Energy Transport, ICRU Report 16; International Commission on Radiation Units
and Measurements: Washington D.C., 1970.
40. Northclie, L.C.; Schilling, R.F. Nucl. Data. Sect. A 1970, 7, 233.
41. Hubert, F.; Fleury, A.; Bimbot, R.; Gardes, D. Ann. Phys. Suppl. 1980, 5, 1.
42. Ziegler, J.F.; Biersack, J.P.; Littmark, U. The Stopping and Range of Ions in Solids;
Pergamon: New York, 1985.
43. Bragg, W.H. Philos. Mag. 1905, 10, 318.
44. Stopping Powers for Electrons and Positrons, ICRU Report 37; International Commission on
Radiation Units and Measurements: Bethesda, MD, 1984.
45. Anderson, A.R.; Hart, E.J. Radiat. Res. 1961, 14, 689.
46. Bichsel, H.; Inokuti, M. Nucl. Inst. Meth. Phys. Res. B 1998, 134, 161.
47. Northclie, L.C. Phys. Rev. 1960, 120, 1744.
48. Pierce, T.E.; Blann, M. Phys. Rev. 1968, 173, 390.
49. Bloom, S.D.; Sauter, G.D. Phys. Rev. Lett. 1971, 26, 607.
50. Toburen, L.H. In High-Energy Ion-Atom Collisions; Bere nyi, D. Hock, G., Eds.; Elsevier:
New York, 1982; 53 pp.
51. Lo, H.H.; Fite, W.L. Atomic Data 1970, 1, 305.
52. Wittkower, A.B.; Betz, H.D. Atomic Data 1973, 5, 113.
53. Allison, S.K.; Garcia-Munoz, M. In Atomic and Molecular Processes; Bates, D.R., Ed.;
Academic Press: New York, 1962; 721 pp.
54. Schuler, R.H. Trans. Faraday Soc. 1965, 61, 100.
55. LaVerne, J.A.; Schuler, R.H. J. Phys. Chem. 1982, 86, 2282.
56. Zirkle, R.E.; Marchbank, D.F.; Kuck, K.D. J. Cell. Comp. Phys. 1952, 39, 75.
57. Miller, N. Radiat. Res. 1958, 9, 633.
58. Schuler, R.H.; Allen, A.O. J. Am. Chem. Soc. 1957, 79, 1565.
59. Wilson, C.T.R. Proc. R. Soc. A 1923, 104, 192.
60. Wilson, C.T.R. Proc. Camb. Philos. Soc. 1923, 21, 405.
61. Magee, J.L. J. Am. Chem. Soc. 1951, 73, 3270.
62. Samuel, A.H.; Magee, J.L. J. Chem. Phys. 1953, 21, 1080.
63. LaVerne, J.A.; Pimblott, S.M. Radiat. Res. 1995, 141, 208.
64. Pimblott, S.M.; LaVerne, J.A.; Mozumder, A. J. Phys. Chem. 1996, 100, 8595.
65. Pimblott, S.M.; Mozumder, A. J. Phys. Chem. 1991, 95, 7291.
66. Hayashi, M. In Atomic and Molecular Data for Radiotherapy; International Atomic Energy
Agency: Vienna, 1989; 193 pp.
67. Michaud, M.; Sanche, L. Phys. Rev. A 1987, 36, 4672.
68. Michaud, M.; Sanche, L. Phys. Rev. A 1987, 36, 4684.
69. LaVerne, J.A.; Pimblott, S.M. J. Phys. Chem. A 1997, 101, 4504.
70. Schwarz, H.A. J. Phys. Chem. 1969, 73, 1928.
71. Pimblott, S.M.; LaVerne, J.A.; Bartels, D.M.; Jonah, C.D. J. Phys. Chem. 1996, 100, 9412.
72. Crowell, R.A.; Bartels, D.M. J. Phys. Chem. 1996, 100, 17940.
73. Pimblott, S.M.; LaVerne, J.A. J. Phys. Chem. A 1997, 101, 5828.
74. Pimblott, S.M.; LaVerne, J.A. J. Phys. Chem. A 2002, 106, 9420.
75. LaVerne, J.A.; Schuler, R.H.; Burns, W.G. J. Phys. Chem. 1986, 90, 3238.
76. LaVerne, J.A.; Schuler, R.H. J. Phys. Chem. 1987, 91, 5770.
77. LaVerne, J.A.; Schuler, R.H. J. Phys. Chem. 1986, 90, 5995.
78. Chatterjee, A.; Schaefer, H.J. Radiat. Environ. Biophys. 1976, 13, 215.
79. Magee, J.L.; Chatterjee, A. J. Phys. Chem. 1980, 84, 3529.
80. Chatterjee, A.; Magee, J.L. J. Phys. Chem. 1980, 84, 3537.
81. Naleway, C.A.; Sauer, M.C., Jr.; Jonah, C.D.; Schmidt, K.H. Radiat. Res. 1979, 77, 47.
82. LaVerne, J.A.; Schuler, R.H. J. Phys. Chem. 1984, 88, 1200.
83. Burns, W.G.; Marsh, W.R. Trans. Faraday Soc 1968, 64, 2375.
Copyright 2004 by Taylor & Francis Group, LLC
84. Cobut, V.; Frongillo, Y.; Patau, J.P.; Goulet, T.; Fraser, M.-J.; Jay-Gerin, J.-P. Radiat. Phys.
Chem. 1998, 51, 229.
85. Frongillo, Y.; Goulet, T.; Fraser, M.-J.; Cobut, V.; Patau, J.P.; Jay-Gerin, J.-P. Radiat. Phys.
Chem. 1998, 51, 245.
86. Draganic, I.G.; Draganic, Z.D. The Radiation Chemistry of Water; Academic Press: New
York, 1971.
87. Allen, A.O. J. Phys. Colloid Chem. 1948, 52, 479.
88. Hart, E.J. Radiat. Res. 1954, 1, 53.
89. Allen, A.O. Radiat. Res. 1954, 1, 85.
90. Burns, W.G.; May, R.; Buxton, G.V.; Tough, G.S. Faraday Disc. Chem. Soc 1977, 63, 47.
91. Burns, W.G.; May, R.; Buxton, G.V.; Wilkinson-Tough, G.S. J. Chem. Soc. Faraday Trans.
1981, 77, 1543.
92. Average Energy Required to Produce an Ion Pair, ICRU Report 31; International
Commission on Radiation Units and Measurements: Washington, D. C., 1979.
93. LaVerne, J.A.; Schuler, R.H. J. Phys. Chem. 1996, 100, 16034.
94. LaVerne, J.A.; Schuler, R.H. J. Phys. Chem. 1985, 89, 4171.
95. LaVerne, J.A.; Schuler, R.H. J. Phys. Chem. 1987, 91, 6560.
96. LaVerne, J.A.; Schuler, R.H. J. Phys. Chem. 1992, 96, 7376.
97. Chatterjee, A.; Maccabee, H.D.; Tobias, C.A. Radiat. Res. 1973, 54, 479.
98. Fain, J.; Monnin, M.; Montret, M. Radiat. Res. 1974, 57, 379.
99. Turner, J.E.; Hollister, H. Health Phys. 1969, 17, 356.
100. Katz, R. Health Phys. 1970, 18, 175.
101. Christman, E.A.; Appleby, A.; Jayko, M. Radiat. Res. 1981, 85, 443.
102. Appleby, A.; Christman, E.A.; Jayko, M. Radiat. Res. 1985, 104, 263.
103. Appleby, A.; Christman, E.A.; Jayko, M. Radiat. Res. 1986, 106, 300.
104. Sauer, M.C., Jr.; Schmidt, K.H.; Hart, E.J.; Naleway, C.A.; Jonah, C.D. Radiat. Res. 1977, 70,
91.
105. Sauer, M.C., Jr.; Schmidt, K.H.; Hart, E.J.; Naleway, C.A.; Jonah, C.D. Radiat. Res. 1977, 77,
47.
106. Sauer, M.C., Jr.; Schmidt, K.H.; Jonah, C.D.; Naleway, C.A.; Hart, E.J. Radiat. Res. 1978, 75,
519.
107. Sauer, M.C., Jr.; Jonah, C.D.; Schmidt, K.H.; Naleway, C.A. Radiat. Res. 1983, 93, 40.
108. Chitose, N.; Katsumura, Y.; Zuo, Z.; Domae, M.; Ishigure, K.; Murakami, T. J. Chem. Soc.
Faraday Trans. 1997, 93, 3939.
109. Chitose, N.; Katsumura, Y.; Domae, M.; Zuo, Z.; Murakami, T. Radiat. Phys. Chem. 1999, 54,
385.
110. Chitose, N.; Katsumura, Y.; Domae, M.; Zuo, Z.; Murakami, T.; LaVerne, J.A. J. Phys.
Chem. A 1999, 103, 4769.
111. Chitose, N.; Katsumura, Y.; Domae, M.; Cai, Z.; Muroya, Y.; Murakami, T.; LaVerne, J.A.
J. Phys. Chem. A 2001, 105, 4902.
112. Baldacchino, G.; Bouard, S.; Balanzat, E.; Garde` s-Albert, M.; Abedinzadeh, Z.; Jore, D.;
Deycard, S.; Hickel, B. Nucl. Inst. Meth. Phys. Res. B 1998, 146, 528.
113. Baldacchino, G.; LeParc, D.; Hickel, B.; Garde` s-Albert, M.; Abedinzadeh, Z.; Jore, D.;
Deycard, S.; Bouard, S.; Mouton, V.; Balanzat, E. Radiat. Res. 1998, 149, 128.
114. LaVerne, J.A. Radiat. Res. 1989, 118, 201.
115. LaVerne, J.A.; Yoshida, H. J. Phys. Chem. 1993, 97, 10720.
116. Pastina, B.; LaVerne, J.A. J. Phys. Chem. A 1999, 103, 1592.
117. Pimblott, S.M.; LaVerne, J.A. J. Phys. Chem. 1992, 96, 746.
118. Pimblott, S.M.; LaVerne, J.A. J. Phys. Chem. 1994, 98, 6136.
119. LaVerne, J.A.; Pimblott, S.M. J. Phys. Chem. 1991, 95, 3196.
120. Pastina, B.; LaVerne, J.A.; Pimblott, S.M. J. Phys. Chem. A 1999, 103, 5841.
121. Appleby, A.; Schwarz, H.A. J. Phys. Chem. 1969, 73, 1937.
122. LaVerne, J.A. Radiat. Res. 2000, 153, 196.
Copyright 2004 by Taylor & Francis Group, LLC
123. Bjergbakke, E.; Hart, E.J. Radiat. Res. 1971, 45, 261.
124. Henglein, A.; Asmus, K.-D.; Scholes, G.; Simic, M. Z. Physik. Chem. 1965, 45, 39.
125. Yokohata, A.; Tsuda, S. Bul. Chem. Soc. Jpn. 1974, 47, 2869.
126. Burns, W.G.; Sims, H.E. J. Chem. Soc. Faraday Trans. 1981, 1, 77, 2803.
127. Nichiporov, F.G. High Energy Chem. 1986, 20, 233.
128. Baverstock, K.F.; Cundall, R.B.; Burns, W.G. In Proc. 3rd. Tihany Symp. On Radiation
Chemistry; Dobo, J. Hedvig, P., Eds.; Akademiai Kiado: Budapest, 1972; 1133 pp.
129. Sowden, R.G. Trans. Faraday Soc. 1959, 55, 2084.
130. LaVerne, J.A.; Pimblott, S.M. J. Phys. Chem. A 2000, 104, 9820.
131. Appleby, A.; Christman, E.A.; Jayko, M. Radiat. Res. 1989, 118, 401.
132. Schwarz, H.A.; Carey, J.M., Jr.; Scholes, G. J. Am. Chem. Soc. 1959, 81, 1801.
133. Bibler, N. J. Phys. Chem. 1975, 79, 1991.
134. Hiroki, A.; Pimblott, S.M.; LaVerne, J.A. J. Phys. Chem. A 2002, 106, 9352.
Copyright 2004 by Taylor & Francis Group, LLC
15
DNA Damage Dictates the Biological
Consequences of Ionizing Irradiation:
The Chemical Pathways
William A. Bernhard
University of Rochester, Rochester, New York, U.S.A.
David M. Close
East Tennessee State University, Johnson City, Tennessee, U.S.A.
1. BIOLOGICAL CONSEQUENCES OF EXPOSURE TO IONIZING
RADIATION
1.1. Cellular Endpoint
The eects of ionizing radiation on DNA have been studied for many years because of the
central role the DNA plays as the major cellular target. Cell killing, mutagenesis, and
transformation are all caused by damage to DNA.
The ability of external penetrating ionizing radiation to cause biological damage was
recognized soon after the discovery of x-rays in 1896. Radiation protection standards for
workers were introduced in the 1920s. These rst standards were based on the ability of
ionizing radiation to produce acute eects such as erythema and could be observed im-
mediately after exposure. At the same time, the appearance of cancers among radiation
workers showed that radiation could produce neoplastic eects. This led to the recogni-
tion of long-term epidemiological studies for understanding radiation risks and setting
standards.
1.2. DNA is the Critical Target
When ionizing radiation aects just one nucleotide in a sequence, this may produce a point
mutation. Most point mutations are of little consequence because the same protein or a
functional variation is made anyway. However, some point mutations result in a nonsense
message from which a nonfunctional protein is constructed, while other point mutations
give a meaningful but changed message leading to a protein with altered properties.
Radiation can also cause larger-scale errors involving many nucleotides. In some
cases, the change in the chromosome is visible with a microscope (called chromosome
aberrations). Some important types of chromosome aberrations are deletions, amplica-
tion, and translocation. Deletions are simply loss of a segment of DNAwith the consequence
Copyright 2004 by Taylor & Francis Group, LLC
that a number of dierent proteins, coded by the deleted region, are not made at all. This
kind of error is thought to occur in the childhood cancer retinoblastoma. The amplication
type of error, with repetition of a whole sequence of DNA bases, also occurs in some child-
hood cancers. Translocation is what happens when one piece of DNA is physically moved
and joined onto some unrelated piece of DNA. Some kinds of human leukemia may result
from this type of error.
One eect of complex DNA damage is a cell unable to produce copies of itself. This is
called reproductive death; the cell can no longer divide. In tissue, a very large number of cells
will be present, so death of a few cells makes little dierence to the function of the tissue. If
the radiation dose is increased, some signicant eects may occur quickly. In humans, such
eects include radiation sickness, skin burns, or failure of blood cell production. For each of
these eects to be large enough to be noticeable, a minimum radiation dose, or threshold,
has to be exceeded. For all doses above this, the severity of the eect becomes greater with
increasing dose. These are nonstochastic or deterministic eects.
Some radiation eects result from nonlethal damage to a single cell. These eects are
called stochastic. They have the property that there is no threshold for these eects to
occur. It is the probability of occurrence rather than its severity which increases with dose.
The causation of some cancers may be rooted in a stochastic eect.
The dose from low-LET (linear energy transfer) ionizing radiation is delivered by
high-speed electrons (Compton and photoelectrons) traveling through the cell and creating
primary ionization tracks. One track of ionizations is the minimum disturbance at the
cellular level. Paretzke [1] has shown how one can convert the energy of an x-ray or a g-ray
into electrons and their distribution. One can then determine how many photons of a given
energy are required to deposit a given dose. Tables are available which present tissue dose in
centi-Gray (cGy) when the average track rate per cell nucleus is 1. For example, there would
be, on average, one track per cell nucleus for a dose of 1.0 Gy from a medical x-ray. The
same would be true for a dose of 0.33 Gy from
137
Cs.
Some radiobiologists believe that cancer can be initiated as a result of a single
radiation track passing through a single cell nucleus. That would mean cancer induction
can be a unicellular process following the rules of chance. This could be taken to imply that
there is no safe dose or no safe dose rate. Of course, most DNA damage is corrected by
repair enzymes. So a key question then becomes this: does repair of carcinogenic injuries
operate awlessly when dose is suciently low? If every carcinogenic lesion to DNA was
faithfully repaired, then the net eect of a particular dose to cancer induction would be
zero. It follows then that many small doses, with corresponding times in between, could be
absorbed without increasing the cancer rate. Human epidemiological studies show that
repair fails to prevent radiation-induced cancer, even at doses where the repair system has
ample time to deal with only a few tracks at a time. Therefore, a question of considerable
current interest is whether or not there is an inherent inability of the repair system to x
certain types of complex damage to DNA.
1.3. Unique Characteristic of Radiation Damage
DNA damage is caused routinely, and frequently, by normal endogenous biochemistry.
This type of damage is restricted to single sites and therefore to a single strand. Damage to a
single strand is relatively easy to faithfully repair since the complementary strand is avail-
able as a template. Radiation damage to DNA is very dierent since much of the damage is
produced in clusters; one example of which are the double-strand breaks (dsb). Double-
strand breaks pose a greater risk of misrepair. A dsb can lead to loss of base sequence
Copyright 2004 by Taylor & Francis Group, LLC
information or can permit the two ends of a break to separate and rejoin with the wrong
partner. Therefore the repair of dsb is essential in preventing chromosomal fragmentation,
translocations, and deletions.
1.4. Genomic Instability
The Low Dose/Low-Dose Rate Conundrum
The current health risks associated with exposure to low-dose radiation are extrapolated
fromhigh-dose data taken fromthe Life Span Study of the Japanese atomic bomb survivors.
Currently, a linear no threshold extrapolation is recommended. The numerous technical
reports and scientic papers about the Japanese A-bomb survivors were widely interpreted
as showing that the eects of occupational exposures to radiation would be too small to
detect in epidemiological studies. However, questions about the reliability of the A-bomb
results were presented by Stewart and Kneale [2]. Their Oxford Childhood Study observed
that children whose in utero exposures were as little as 10 to 20 mSv had 40% more
childhood leukemias than those who were not exposed. No similar eects are reported in the
A-bomb data. Of course, the nding of no eect is not a compelling argument for or
against a safe dose.
During World War II, a dierent exposure situation was created in the United States
by the Manhattan Project. Several hundreds of thousands of workers were recruited to a
new industry in remote and secret locations. Since some of the hazards of working with
radiation were recognized in advance, workers at many sites were monitored for radiation
exposure. A new discipline, Health Physics, was created to provide radiological protection.
The systematic collection of dosimetry records created an opportunity to investigate
relationships between repeated exposure to small doses of radiation and disease.
Studies of nuclear workers have suggested that radiation risk estimates based on A-
bomb survivors could be underestimating the cancer risks fromextended low-level exposure
to radiation [3]. Due to such studies, and advances in our understanding of basic mech-
anisms in radiation biology, the importance of determining the consequences of low dose/
low dose rate exposure has become more apparent. The BEIR (Biological Eects of
Ionizing Radiation) VII Committee is currently evaluating new DOE Low-Dose Radiation
studies (http://lowdose.tricity.wsu.edu/). The current protection standard used by the
International Commission of Radiation Protection is 20 mSv whole body annual dose
equivalent; in the United States, 50 mSv per year is used.
Bystander Eect
Acentral tenet in radiation studies has been that energy must be deposited in a cell nucleus to
elicit a biological eect. Now, a number of nontargeted eects of radiation exposure have
been described which suggest that one must reevaluate this central tenet. One of these
nontargeted eects is the bystander eect, which describes the ability of the cell aected by
radiation to induce instability in neighboring cells that have not absorbed any radiation
energy. Convincing demonstrations of the bystander eect have come from studies using
charged particle microbeams.
Using the Columbia microbeam, Zhou et al. [4] passed one alpha particle through the
nuclei of mammalian cells. When 10% of the cells were irradiated with a single alpha
particle, the mutation yield was similar to that observed when 100% of the cells were
irradiated. Clearly, a single alpha particle can induce genomic instability in cells that were
not irradiated.
Copyright 2004 by Taylor & Francis Group, LLC
Hormesis
Working in the opposite direction of the bystander eect is hormesis, which is the stim-
ulation of a protective eect by small doses of radiation. It has been proposed that a low
priming dose may increase the number or quality of enzymes available to repair
subsequent radiation damage. But to date the underlying biochemical basis of hormesis
is not known. In addition, it is not clear how to extrapolate the information gained from
studies of lower organisms to humans. UNSCEAR (United Nation Scientic Committee
on the Eects of Atomic Radiation) [5] examined the notion of adaptive response in
1993 and concluded that while it was an interesting phenomenon that occurs in some cell
systems at various stages of development, it has little relevance in radiation protection.
While it seems likely that the biochemical trigger for hormesis is damage to the DNA, this
is not yet proven.
2. ENERGY DEPOSITION
2.1. Ionization and Excitation
Most of the energy associated with an incident x-ray or g-ray is absorbed by ejected
electrons. These secondary electrons are ejected with sucient energy to cause further
ionizations or excitations. The consequences of excitations may not represent permanent
change, as the molecule may just return to the ground state by emission or may dissipate
the excess energy by radiationless decay. In the gas phase, excitations often lead to
molecular dissociations. In condensed matter, new relaxation pathways combined with the
cage eect greatly curtail permanent dissociation. Specically in DNA, it is known that the
quantum yields for uorescence are very small and relaxation is very fast [6]. For these
reasons, the present emphasis will be on the eects of ionizations.
The initial chemical events involving the deposition of energy in DNA are conven-
iently divided into two parts: (1) energy deposited in water and (2) energy deposited in the
DNA itself. These are often called indirect and direct eects. Since some of the water in a
cell is intimately associated with the DNA, these terms must be used with caution. The
presence of DNA close to an energy deposition event in the water will eect the fate of the
species produced, and, likewise, water molecules closely surrounding the energy deposition
event in the DNA will modify the subsequent fate of the initial species. So the presence of
each component modies the behavior of the other.
2.2. Indirect Effects
The initial ionization of a water molecule produces an electron and the water radical
cation. The water radical cation is a strong acid and rapidly loses a proton to the nearest
available water molecule to produce an HO
.
radical and H
3
O
+
. The electron will lose
energy by causing further ionizations and excitations until it solvates (to produce the
solvated electron e
aq
, a smaller
quantity of H-atoms, H
2
O
2
, and H
2
are also produced.
Of the two radical species HO
.
and e
aq
loss and e
loss in
guanine bases is characterized by a broad EPRsinglet. The same feature is also evident in the
EPR spectrum of DNA irradiated and observed at 77 K.
The EPR spectrum of irradiated DNA is not very well resolved. Attempts to improve
the spectral resolution have used deuterated DNA and samples of oriented DNA. Using
oriented DNA, Gra slund et al. [35] suggested the thymine anion Thy
for the
anions and >90% Gua
.
+
for the cations. The analysis produced a small imbalance in
the cations (44%) and anions (56%). It was suggested that some holes remain trapped in
the solvation shell. It is also possible that this reects small errors in the treatment of the
basis spectra, or that some DNA radicals are not accounted for because their EPR signal is
too broad and poorly resolved.
Another approach by Hu ttermann et al. [50] has been to use pulsed EPR techniques
to study the radicals present in DNA bers equilibrated in D
2
O and then irradiated and
observed at 77 K. This work supports the conclusions that the primary radiation-induced
defects are Cyt
.
and Gua
.
+
. Also reported are contributions from Thy
.
and an allyl
radical found on thymine [Thy(MeH)
.
]. Also discussed are three components tentatively
assigned as adenine and guanine anions and a species whose dominant hyperne
interaction involves the N1 of cytosine.
More recently, this same group has studied DNA with high-eld EPR (245 GHz)
[51]. This study shows nice spectra of the Thy(MeH)
.
radical. Also, the authors discuss
the eects of hydration levels on the production of the various base radicals. A more
recent paper by Weiland and Hu ttermann [52] considers the same base radicals in DNA at
77 K and then looks at the transformation of these radicals into the more stable room-
temperature products. Among these are C1V and C3V sugar radicals, dRib(C1VH)
.
, and
dRib(C3VH)
.
, respectively.
3.3. Predictions Based on Computational Chemistry
Several recent papers have reported Density Functional Theory (DFT) calculations on the
primary oxidation and reduction products observed in irradiated single crystals of the
common nucleobases: thymine [53], cytosine [54], guanine [55], and adenine [56]. The theo-
retical calculations include estimates of spin densities and isotropic and anisotropic
hyperne couplings which can be compared with experimental results (obtained from
detailed EPR/ENDOR experiments).
It should also be pointed out that the calculation of accurate hyperne coupling
constants is rather dicult. Two factors are involved: the isotropic component (A
iso
) and
the anisotropic component (T
xx
, T
yy
, T
zz
). One must have a good description of electron
correlation and a well-dened basis set in order to calculate accurate isotropic hyperne
couplings. This is not easy to do with molecules the size of the DNA bases. Even when the
computational demands are met, the theoretical calculations may deviate more than 20%
from the experimental results. Recently, it has been shown that the calculation of aniso-
tropic hyperne couplings for hydrogens is often within 510% of the experimental values
[54]. The goal is to make comparisons of calculated and experimental isotropic and
anisotropic hyperne couplings a useful guide in identifying radiation-induced free
radicals.
Copyright 2004 by Taylor & Francis Group, LLC
The present level of theoretical work can be seen in a recent paper by Sevillas group.
They have employed DFT to study the details of proton transfer reactions in Gua:Cyt
base pairs [57]. Using the DFT functional B3LYP with the 631+G(d) basis set on the
entire Gua:Cyt molecule (19 heavy atoms), the results are presented which show that it is
energetically favorable for the N1 proton of guanine to transfer to the N3 of cytosine after
reduction of cytosine or after oxidation of guanine.
Several additional problems must also be considered. The calculations discussed here
are computationally challenging. They involve a single-point calculation on the optimized
structure using triple-zeta plus polarization functions [B3LYP/6311G(2df,p)] in order to
compute spin densities. Also, the detailed DFT calculations were performed on isolated
molecules, whereas some of the experimental results reported involve free radical formation
in the solid state, mainly in single crystals. Therefore the theoretical calculations are ignoring
the electrostatic environment of the radicals discussed, in particular, the intricate hydrogen-
bonding structure that the free radicals are imbedded in. This often leads to nonplanar
radicals which may or may not represent what is believed to be observed experimentally in
the solid state.
In many cases, the theoretical and experimental results agree rather well. In other
cases, there are discrepancies between the theoretical and experimental results. A review of
the successes and failures of using DFT to calculate spin densities and hyperne couplings
of the primary radiation-induced free radicals observed in the nucleobases has recently
appeared [58].
3.4. DNA Stopping Power
Invaluable to understanding the spatial distribution of damage in DNAare calculations that
employ Monte-Carlo track-structure simulations [5962]. Since most of the energy loss is
through inelastic interactions between fast electrons and the DNA, the cross sections for
energy loss by electrons are of critical importance. While most theoretical calculations are
based on the stopping power of gaseous water, it has been pointed out by La Verne and
Pimblott [63,64] that the most probable energy loss and the mean energy loss for electrons in
DNAdo not compare favorably with gaseous water but are reasonably close to liquid water.
Their calculation was based on inelastic cross sections that are obtained from the dipole
oscillator strength distribution of DNA. They calculate a mean energy loss for a 1-MeV
electron in DNA of 57.9 eV, close to the value of 56.8 eV for liquid water but diering
substantially from 50.9 eV for gaseous water. In order to compare experimentally deter-
mined yields and degree of clustering with the calculated track structures, it would help
considerably to have more concrete information on the stopping power of DNA.
One important application of track-structure calculations is in understanding exper-
imental measurements of free radical yields in DNA, particularly crystalline DNA. Cur-
rently, the yield of initial ionizations in DNAis unknown. As a proxy, one can use the values
for water (low LET): G(initial ion)
f
G(OH)+G(eaq)+G(H)=0.59+0.50+0.09=1.18
Amol/J [65]. If the stopping power of DNA is the same as liquid water, then samples such
as crystalline DNA should yield 1.18 Amol/J initial ionizations at low LET. At 4 K, the ob-
served yields of 0.550.78 Amol/J [66,67] would require that 5070%of the initial ionizations
are stably trapped. To the degree that the DNAstopping power might be greater than that of
water, this percentage goes downandthe range of electronor hole transfer must be increased.
If the data on cross sections of low-energy electrons with DNA [68] could be integrated into
the Monte-Carlo track-structure calculations, it should help close this gap between experi-
ment and theory.
Copyright 2004 by Taylor & Francis Group, LLC
4. TRACK EXPANSION AND CHARGE MOBILITY IN DNA
4.1. Track Expansion Affects the Spatial Distribution of Products
The nonhomogeneous distribution of energy absorption is a dening characteristic of
ionizing radiation. If this characteristic had been absent, the unique impact of ionizing
radiation in biology and medicine would almost certainly not exist. But the fact that
energy deposition is spatially nonhomogeneous is not, in itself, sucient to explain the
high toxicity of ionizing radiation. The high toxicity (lethality, mutagenicity, and
carcinogenicity) is a consequence of the fact that the nal damage produced in DNA is
spatially nonhomogeneous. The severity of the biological consequences depends on the
degree to which stable products are formed in clusters [69] or, in dierent words, occur as
multiply damaged sites [70]. An inevitable conclusion is therefore that the rate of product
formation must be competitive with the rate of track expansion. Clustered products evolve
from clustered ionizations.
Fig. 2 depicts, in schematic, the two extremes plus an intermediate case of track
expansion. InFig. 2A, the initial distributiononions is shown at zerotime, where t=0 is after
thermalizationbut before charge transfer. The initial cluster density is a functionof LET, and
even at very lowLET (e.g.,
60
Co g-rays), a large percentage of energy is deposited in clusters
[71]. In Fig. 2B, ion mobility is high, which would be the case if DNAwere a good conductor.
After a very short time, combination reactions fully annihilate the charges in DNA. In Fig.
2C, the holes and excess electrons have low mobility, moving a short distance before
trapping. Closely spaced, opposite charges combine. In Fig. 2D, the mobility is zero; the
track is etched in place. We consider case B rst, the case where holes and electrons freely
conduct through the base stacks, consequently leaving no damage on the base stack.
Figure 2 A schematic representation of track expansion is shown assuming dierent mobilities. (A)
Spatial distribution of radical ions at the sites of initial thermalization. (B) When there is high
mobility, that of a conductor is assumed, no radical ions are trapped on DNA. (C) If there is low
mobility, some radical ions are trapped while others recombine. (D) At zero mobility, the spatial
distribution of trapped radicals is the same as in the initial track, i.e., the same as in (A).
Copyright 2004 by Taylor & Francis Group, LLC
If DNA has the properties of a molecular wire, as proposed in 1993 [72] and sub-
sequent work (e.g., Refs. 7376), then the spatial distribution in DNAwould look something
like that in Fig. 2B. If this was the case, it is very dicult to reconcile a large body of
knowledge in radiation biology because not only does it dramatically reduce the degree of
clustering, it would also make DNA remarkably radiation-resistant. Indeed, base damage
produced via the direct eect would not exist. Publications by radiation chemists and theo-
reticians, before and after the 1993 proposal, provide rather strong evidence that DNAis not
a molecular wire [7779]. Joining that body of work is a recent series of photochemistry
investigations that shows that charge mobility through DNAis that expected of an insulator
(a poor semiconductor) [8082]. Therefore the situation under Fig. 2B does not occur.
The other extreme shown in Fig. 2D also does not occur. Holes are mobile. We know
this because, as discussed above, the probability of nding a hole on a particular component
does not correlate with the probability of ionization. Holes are trapped preferentially at
guanine [83], to a lesser extent at the sugar [84], and rarely at the other bases. This
redistribution requires hole transfer. Trapping of the excess electron is also selective,
occurring primarily at cytosine [49,85,86]. This implies that after thermalization, the electron
transfers between bases until nding a stable trapping site. Therefore the situation is as
depicted in Fig. 2C. Akey question is What is the range of both the excess electron and the
hole within DNA and through its local environment?
4.2. Hole and Electron Transfer
Studies of the direct eect have been largely conned to DNA samples in the solid state.
This is done in order to maximize direct-type damage and minimize indirect-type damage.
In addition, low temperatures are often employed both as a means of sequestering the
DNA from the bulk water and as a means of stabilizing free radical intermediates. In
frozen DNA samples, the mobility of holes and excess electrons diers for the dierent
sample components: ice, solvation shell, DNA backbone, and base stacks. We start with
the ice phase.
Ice
In an aqueous solution of DNA, the water outside of the solvation shell is referred to as bulk
water. When DNAsolutions are frozen, the bulk water crystallizes as a separate phaseice.
Ice does not formif the concentration of DNAis brought to a level where only the solvation
shell remains, about 2022 waters/nucleotide. If brought to this concentration slowly, a lm
is formed. Freezing a lm does not create ice. Another type of sample is prepared by rst
lyophilizing DNAand then letting it sit at a preselected humidity that determines the level of
hydration, typically 2.5<G<22. Subsequent freezing of these cotton-like samples does not
yield ice.
During irradiation, the holes produced in the ice phase are sequestered by the ice. This
is not strictly true for the excess electrons; some escape the ice phase and are selectively
scavenged by DNA. The degree to which this increases the yield of electrons trapped by
DNA is considered to be relatively small and therefore when computing yields of direct
damage in DNA, the mass of the ice phase is usually excluded fromthe target mass. While ice
is a relatively passive component of the system, it does provide an important reference point
with respect to product yields. Electrons and holes are quite mobile in ice [87,88]; it is a
decent conductor. Consequently, even at low temperatures, the free radical yields are quite
low, e.g., the yield of HO
.
in ice at 77 K is 0.037 Amol/J [83]. This is just one example of a
Copyright 2004 by Taylor & Francis Group, LLC
general principle; high charge mobility promotes recombination and therefore results in low
yields of trapped free radicals.
DNA Solvation Shell
Ionization of DNAs solvation shell produces water radical cations (H
2
O
.
+
) and fast
electrons. The fate of the hole is dictated by two competing reactions: hole transfer to
DNA and formation of HO
.
via proton transfer. If the ionized water is in direct contact
with the DNA (G<10), hole transfer dominates. If the ionized water is in the next layer
out (9<G<22), HO
.
formation dominates [67,89,90]. The thermalized excess electrons
attach preferentially to bases, regardless of their origin. Thus the yield of one-electron
reduced bases per DNA mass increases in lockstep with increasing G, up to an G of 20
25. This means that when G exceeds 9, there will be an imbalance between holes and
electrons trapped on DNA, the balance of the holes being trapped as HO
.
. At G = 17, an
example where the water and DNA masses are about equal, the solvation shell doubles
the number of electron adducts, increasing the DNA-centered holes by a bit over 50%
[9193].
Samples that are prepared by freezing DNAsolutions result in a DNAaggregate phase
in which ice formation draws water from the solvation shell. The net result is that the DNA
aggregates have an average G of
f
15 [89,90]. For DNA prepared in this manner, the target
mass is DNA+15 waters [91]. Since the solvation shell is not pure water, containing about
one cation per nucleotide, it is more accurate if one replaces water mass by the equivalent
cation mass. Taking Na
+
as an example, 15 waters would be replaced by the sum of 1
Na+13.5 waters. A rule of thumb is that if one assumes that only the DNA is the target
mass, the solvation layer of DNA doubles the direct-type damage.
DNA Backbone
Forward reactions of the one-electron oxidized sugar-phosphate backbone are of two types:
hole transfer to the base stack or formation of a neutral sugar radical by irreversible
deprotonation. In much of the older literature, it was assumed that the former eclipsed the
latter. This was largely because free radicals centered on the sugar moiety were not detected
by EPR. As has been pointed out, this does not prove that sugar radicals are not formed
[94,95], and indeed, computed simulations have shown that such radicals would give EPR
spectra that would be dicult to detect in the presence of the intense signal from the base-
centered radicals [20,96]. Recent results leave little doubt that a substantial fraction (
f
25%)
of the holes are irreversibly trapped on the sugar [84,9799].
Evidence that direct-type damage occurs at the sugar consists of the following. Becker
et al. [100] observed sugar radicals in DNA exposed to high LET radiation. Swarts et al.
[101] observed free base release in DNA and Razskazovskiy et al. [98] observed free base
release in crystalline oligodeoxynucleotides. Free base release correlates with sugar damage.
The yields of strand-break products have been measured by Debije et al. [97] and
Razskazovskiy et al. [99] in crystalline oligomers x-irradiated at 4 K. Debije has identied
a 3V -centered sugar radical in crystals of d(CTCTCGAGAG)
2
, a B-form DNA [84]. Broadly
speaking, 1020% of the radicals initially stabilized on DNA at 4 K are sugar radicals. This
corresponds to about half of the backbone-centered holes (formed directly and indirectly by
transfer from water) leading to neutral sugar radicals via deprotonation of the one-electron
oxidized sugar-phosphate backbone. Also, it correlates well with the observed yields for
direct strand breaks, 0.050.15 Amol/J, in high molecular weight DNA (Sec. 5).
Copyright 2004 by Taylor & Francis Group, LLC
DNA Bases
Radiation Chemistry. It has been known for decades that charge migrates through
the stacked bases of DNA [102,103]. Compelling evidence comes from studies employing
EPR [30,77,104106], pulse radiolysis [107,108], and product analysis [101,109,110]. The
range of migration for electron and holes at 477 Kis at least 2 bp and the mean is between 3
and 11 bp [106,111]. Warming samples above 150 K extends thermally activated migration
to 30 bp or more [106]. Clearly, the DNA bases are very eective traps for both holes and
excess electrons. Because the activation energy needed to detrap these species is very disperse
[104], charge migration by hopping can be observed over a wide temperature rage,<20 to
>150 K. The dielectric relaxation that creates traps most assuredly occurs even at room
temperature, competing with other processes such as tunneling. While it may be dicult to
observe these trapping events at room temperature, one should anticipate that they
moderate the speed and range of charge transport, even at high temperatures and in
dilute aqueous solution.
Photochemistry. The proposal by Murphy et al. [72] that DNA is a molecular wire
stimulated newinterest in the possibility that DNAconducts charge through its base stacks,
particularly among photochemists. Based on an observed absence of uorescence, they
concluded that electron transfer occurs through a length of stacked bases (k-ways) >4.0
nm in<10
9
sec. They calculated h
e
f
0.2 A
1
for the equation k
ET
=k
0
exp{h
e
(RR
0
)},
where, for the case of tunneling, the electron transfer rate is expected to drop o
exponentially as a function of distance (RR
0
). Conductors do have very low values of
h
e
, such as Murphy et al.s group observed. The high conductivity of DNA was eventually
promoted as a paradigm shift [112,113]. Others, using the tools of photochemistry, have
measured the values of h, making sure of high quantum yields and conrming transfer with
corroborating evidence. For electron transfer and for hole transfer, h
e
f
1.0 A
1
and
h
h
f
0.6 A
1
are obtained, respectively, values typical of insulators [114116]. If
experiments are designed that are permissive for hopping, then low values of h
h
,<0.2
A
1
, can be observed, but under these circumstances, electron transfer is not fast [117]. In
our opinion, the evidence is conclusive; DNA is an insulator. The current questions of
interest focus on how excess electrons and holes move through DNA and what reactions
compete with charge transfer. Indeed, electron transfer in DNA proceeds independently of
the k-ways, the main function of which is to promote base stacking and thereby stabilize
the linearity of DNA.
Interstrand and Interduplex
DNADNA. While it is dicult to design an experiment that proves that electron/
hole transfer occurs between the strands of a single duplex (interstrand), it is rather simple to
prove that transfer occurs from one duplex to another (interduplex). This was done by
monitoring the concentration of holes and excess electrons trapped in crystalline DNA as a
function of temperature. Given that the annealing characteristics are relatively insensitive to
DNA conformation, sequence, or base stacking continuity, it is readily shown that electron
and/or hole transfer must be intermolecular [104]. In more elegant experiments (see below),
it is shown that interduplex tunneling occurs for both electrons and holes [118121]. These
newresults make it clear that, with regard to charge transfer via tunneling, the key variable is
distance, not the intervening covalent or k-bonds. Thus in a crystalline lattice, where the
tunneling distance between helices is comparable to that within a strand, the hole and
electron move in three dimensions.
Copyright 2004 by Taylor & Francis Group, LLC
If hole and electron transfer occurs interduplex, it follows that it will also occur
interstrand. The tunneling distances are comparable. This is elaborated in Sec. 4.3.
DNAProtein. A large number of proteins in nature perform their function by
expediting electron transfer, and there is an extensive literature on electron transfer through
proteins (see Refs.122124 for reviews). Relevant here are the observations that the excess
electron has a large range (not readily trapped) [125] while the hole is relatively immobile
(trapped by deprotonation at the peptide bond giving amido radicals) [126]. This raises the
expectation that electrons but not holes could be transferred fromprotein to DNA. This has
been observed by a number of groups [127131].
In one of the more detailed studies to date, Weiland and Hu ttermann [131] compare
the trapped radical distribution of DNA alone with that of dry chromatin. Using high-eld
EPR (285 GHz), these experiments signicantly reduce the spectral overlap between one-
electron oxidized guanine (Gua
ox
) and the other DNA radicals thereby improving the
precision over EPR measurements made at lower elds (9 GHz). Following x-irradiation at
77 K, the DNA samples trap out 55% of the radicals as one-electron reduced pyrimidines
(Pry
re
) and 22% as one-electron oxidized guanine. Under the same conditions, chromatin
gives 26% and 5%, respectively. The radical concentration at a specic dose increases by
1.4 in chromatin vs. DNA alone. Given that
f
47% of the radicals are preferentially on
DNA and that the chromatin samples consisted of
f
30% DNA and
f
70% histone
proteins, the authors nd that the concentration of radicals trapped by DNA in chromatin
is
f
2 that of DNA alone. The data can be used to estimate the inux of electrons (x) and
presumed eux of holes ( y). Solving the two equations, 100 +x y =200 and 26/5 =(55
+x)/(22 y), one obtains x =94 and y =6. Thus these results indicate that when DNAis
packaged in chromatin (dry), it traps twice as many electrons while hole trapping is relatively
unchanged. Warming the 77 K irradiated chromatin to 300 K reduces the total radical
concentration by at least 95% while not causing any marked changes in the DNAprotein
radical ratio. This means that electrons or holes are thermally mobilized in both compo-
nents, but there is no net exchange between the components.
The amount of information is still insucient to predict how packaging of DNA in
chromatin modies the nal product distribution in DNA. If the stable end products
reect a doubling in reductive damage while oxidative damage remains the same, then one
would expect an increase in frequency and complexity of clustered lesions. If that proves
true, the histone proteins and attending nucleosomes structure would act as radiation
sensitizers.
4.3. Tunneling and Hopping
Charge transfer in DNA occurs under two distinctly dierent conditions: one which is
temperature-independent and the other which is not. The mechanism underpinning the
former is tunneling, while the latter is controlled by an activation barrier that must be
hopped over. Of these two, tunneling is the one better characterized.
Tunneling
Tunneling as a mechanism of charge transfer in DNA has received widespread attention
(see, for example, Refs. 117 and 132). Recently, denitive proof that tunneling in DNA
occurs has been reported by Sevilla et al. [118121,133]. Key to this proof is the demon-
stration that the observed rates of hole and electron transfer are independent of temperature
from 4 to 130 K. We summarize their ndings here.
Copyright 2004 by Taylor & Francis Group, LLC
DNA, laced with an intercalator characterized by a high electron anity, is g-ir-
radiated and observed by EPR. The one-electron reduced intercalator presents an EPR
spectrum that is readily distinguishable from that of the DNA-trapped radicals. A key
example is mitoxantrone (MX), with an electron anity of 6.25 eV and a radical anion
spectrum that is a sharp singlet. Charges are injected into the DNA by g-irradiation at a
preselected temperature (4130 K). Holding the temperature constant, the EPR spectrum
changes as a function of time (0.530 h). Thereby, a direct measure of the rate of electron
transfer from one-electron reduced pyrimidines (Pry
re
) to the intercalator, e.g., MX,
is measured. The tunneling rate is observed to depend on the electron anity (EA) of the
acceptor (intercalator): for MX, h
e
= 0.9F0.1 A
1
and k
0
f
10
11
sec
1
, and for ethidium
bromide (EA=4.32 eV), h
e
=1.2 F0.1 A
1
and k
0
f
10
5
sec
1
. It also depends on the EAof
the donor (Pry
re
). The calculated EA of the Ade:Thy and Gua:Cyt base pairs are 0.30 and
0.49 eV, respectively [57,134]. The values of h
e
are 0.7F0.1 and 1.4F0.1 A
1
for Thy
re
and Cyt
re
, respectively. In addition, the rate depends on DNADNA distance, proving that
tunneling is intermolecular as well as intramolecular. In addition, the values of h are not
strongly dependent on the medium lying between the donor and acceptor. If intermolecular
transfer is not takenintoaccount, calculationof the transfer distance gives a deceptively large
number, e.g., 31 base pairs reported by Pezeshk et al. [135]. Taking intermolecular transfer
into account, the tunneling distance at long times (1 hr) is remarkably reproducible for MX-
DNA over a wide range of DNA packing congurations, 3538 A
1
assuming a value of k
0
of 110
11
sec
1
.
Two important insights into the spatial separation of end products grow out of the
recent ndings on tunneling in DNA. (1) In order for the distance between the stable
products derived from oxidized guanine and reduced pyrimidines to be less than 3040 A
(<10 bp), a competing irreversible reaction must trap the radical at a rate competitive with
that of tunneling. While this is highly improbable at low temperatures,<120 K, it is likely
to be signicant at RT. More needs to be learned about the rates of irreversible trapping.
(2) The rate of irreversible trapping of holes on the sugar must be fast relative to hole
tunneling to the bases. This means that the distance between a strand break (derived from
the sugar-centered hole) and base damage may be substantially less that 30 A
. The neutral
sugar radical will not be a strong getter for base radicals. One predicts therefore that a
clustering of strand breaks with base damages will be more prevalent than clustering of
base damage with base damage when DNA is irradiated at low temperatures. How this
shifts, when irradiation is at RT instead of<120 K, depends on the rates of base damage
xation. These rates are governed by an activation energy, which is our next topic.
Hopping
The trapping rate, for both holes and excess electrons, is competitive with the tunneling rate.
This must be the case because DNAis a very eective trap of both holes and electrons; there
are no other materials that are better traps that we are aware of. Thus the way to maximize
trapped radical density is to maximize DNA density, and this is why crystalline DNA tops
the scale in trapping capacity, with trapped radical yields of 0.50.7 Amol/J at 4 K [66,67].
Remarkably, only 3050% of all initial ionizations are lost to recombination. This is re-
markable because tunneling distances, either GuaGua or PyrPyr, are typically short,<
f
7
A
, in all directions [104]. This means that the trapping rate must be fast [k > 10
11
sec
1
exp(1.0 A
1
7 A
)
f
10
9
sec
1
], and the trap depth must be low enough to shut down
tunneling. In order to mobilize trapped charges, an activation energy is required. Charge
transfer governed by an activation energy is referred to as hopping.
Copyright 2004 by Taylor & Francis Group, LLC
Trapping entails dielectric relaxation about the charge altered site. This is analogous to
the Marcus reorganization energy in the liquid state [136]. But in polar solids at 4 K, large
displacements such as molecular translation and rotation cannot occur as a part of the
reorganization. It is constrained to relatively small displacements of the atoms surrounding
the newly charged site. The displacements optimize dipoledipole, dipolemonopole, and
monopolemonopole interactions. The most pliable particles in the DNA matrix (at <120
K) are the protons participating in highly polar bonds (NUHand OUH). Nearly every such
bond in a crystal lattice participates in a hydrogen bond, most notably the NUH
. . .
Oand the
NUH
. . .
N WatsonCrick bonds that zip together the two strands. When a charge attaches
to a particular site, the surrounding hydrogen-bonded protons are relatively free to shift
in response to the newcharge distribution. In the case of hole trapping, this proton shift is so
pervasive that we do not knowof any case where an organic radical cation has been trapped
within a hydrogen-bonded lattice; the radical cations are observed to consistently deprot-
onate. Furthermore, if lattice structure does not provide an opportunity for proton
displacements across hydrogen bonds, the probability of stabilizing the hole decreases
and, if stabilized, it is through carbon deprotonation [137139]. Insight into what makes
DNA such an ecient electron and hole trap has been obtained by studying the annealing
properties of traps.
Annealing DNA irradiated at 4 K leads to a steady decline in EPR signal intensity;
the radical concentration is approximately a monotonic function of temperature. If the
sample is held at any given annealing temperature, the concentration initially decreases and
then plateaus. Detrapping of the radicals is therefore characterized by a large dispersion in
activation energies. This is, in contrast to the annealing properties of a number of
crystalline DNA components, where discrete activation energies can be measured [140
142]. In our working model, the DNAs annealing behavior is explained by three classes of
traps: (a) very shallow, (b) shallow, and (c) deep. The rst two are reversible states and the
third is irreversible.
Very Shallow Traps. It has been proposed that the neutral Gua(N1H)
.
radical,
formed by proton transfer from the Gua
.
+
radical by proton transfer from N1 of Gua to
N3 of Cyt, is a shallow trap [143,144]. This proposal is based on projections from pK
a
made on monomers in dilute aqueous solution, which predict that proton transfer is
favored by 2.3 kJ/mol [22,145]. Ab initio calculations are in excellent agreement with this
value [146,147]. So one expects that an energy of at least 0.025 eV is needed to activate the
return of the proton to N1 Gua, reforming Gua
.
+
. Once Gua
.
+
is reformed, tunneling to
nearby guanines is reestablished as a competitive pathway. Proton transfer therefore is a
gate for hole transfer. Proton-coupled hole transfer describes the thermally driven transfer
of holes from one Gua:Cyt base pair to another.
Charge mobilization is initiated at very low temperatures,<20 K [104]. It has been
proposed that the mobile species at these low temperatures is the guanine-centered hole
[144]. But if activation of hole detrapping entailed only reverse proton transfer, one would
anticipate a relatively discrete activation energy (
f
0.025 eV) and instead a large dispersion
is observed. Detrapping, presumably of holes, occurs from
f
10 to
f
120 K. The energetics
of reorganization (dielectric relaxation) therefore must vary signicantly from site to site.
This observation means that the N1UH
. . .
N3 proton transfer is just one of a large number
of proton shifts. The hydration layer of DNA, even in crystals of DNA oligomers, is
disordered. It is therefore not dicult to envisage a wide range of trapping depths that
depend on not only the energy of proton transfer within the WatsonCrick hydrogen
bond, but also on the proton shifts within the immediate environment.
Copyright 2004 by Taylor & Francis Group, LLC
Shallow Traps. The depth of the traps for the excess electron is expected to be
deeper than the holes. Once more, using the ndings of Steenken [22,145], the change in
free energy for proton transfer can be estimated. In this case, one-electron reduced Cyt
(Cyt
.
and permits
tunneling to resume. Just like hole transfer, electron transfer is gated by proton transfer.
Proton-coupled electron transfer has been observed in aqueous DNA systems [148].
There is good evidence that the excess electron is mobilized at temperatures between
140 and 180 K. At these temperatures, the Cyt(N3+H)
.
concentration decreases, a de-
crease that correlates with the formation of a neutral radical, Thy(C6+H)
.
, produced by
protonation of one-electron reduced thymine (Thy
)
.
at C6. In order for this to occur, elec-
trons trapped at cytosine must be mobilized [30,49,106]. This increases the probability that
Thy
.
will be formed and, if the temperature is high enough to activate proton addition to
C6, formation of the Thy(C6+H)
.
radical becomes competitive with tunneling. While the
activation energy for electron detrapping shows dispersion, it is less so than for the hole.
This makes sense in that the energy of proton transfer is an order of magnitude larger than
that of the hole. This relatively discrete component would therefore be more dominant
relative to the more disperse contributions to reorganization by the local environment.
Deep Traps. A hole produced by ionization of the sugar-phosphate backbone is
permanently trapped by deprotonation from any of the ve ribose carbons, producing a
neutral sugar radical. The probability that any of the sugar radicals [dRib(C1V H)
.
, etc., in
Fig. 1a] will return to the native structure is very small. This is in part a consequence of
these radicals being relatively stable [142,149151] and in part due to the deprotonation
step being irreversible. The lost proton may attach to water in the hydration shell or to the
DNA bases, e.g., cytosine [30]. Either way, the increased distance between the unpaired
electron and site of altered charge helps protect the radical from recombination events
[137]. The sugar radicals are therefore relatively radiation-resistant. While annealing may
activate unimolecular reactions [150153], the reactions remain centered at the initial site
of hole formation. In this case, the damage is xed at the site of ionization.
The hole formed by ionizations of the backbone, if not trapped by deprotonation,
will transfer to the bases. If the temperature is low enough (<
f
120 K), the hole will be
trapped by guanine in a very shallow trap. But if the temperature is high enough, another
reaction pathway is activated, HO
addition, this
reaction is known to occur at room temperature [154].
Thus we nd that the hole is deeply trapped by two distinctly dierent mechanisms.
One that requires no activation energy, sugar radical formation occurs even at 4 K, and
the other that does, HO
+ H
+
!Cyt(C6 + H)
.
(see Fig. 1e for structures). But this reaction is not
observed when DNA, irradiated at<80 K, is warmed [159]. It is presumably quenched by
the faster reaction of Cyt
.
+ H
+
! Cyt(N3 + H)
.
. But a reaction exists by which the
excess electron is deeply trapped by cytosine, and it is not thermally activated. The Cyt(C6
+ H, N3 + H
+
)
.
radical is formed in DNA irradiated at 4 K [144,159]. In addition,
there is evidence of the closely related radical formed by the net gain of hydrogen at C5
of cytosine: Cyt(C5 + H)
.
or Cyt(C5 + H, N3 + H
+
)
.
(the protonation state at N3 is
undetermined). The proposed mechanism for formation of Cyt(C6 + H)
.
invokes spatially
correlated events: electron capture by cytosine accompanied by proton transfer to N3
correlated with the release of a proton by a nearby sugar-phosphate radical cation. Cy-
tosine, thereby, gains one electron and two protons. This radical would be, thereby, a
signature of a clustered lesion. If correct, one would expect to nd two end products, dihy-
drouracil and a single-strand break, in close proximity to one another.
General Scheme
Charge transfer in DNA is summarized by the scheme shown in Fig. 3. Holes on the sugar-
phosphate (SP) tunnel to the nearest base at a rate k
t
in competition with the deproto-
nation rate, k
irr
, forming sugar products (spr
n
). Holes on the bases (B
n
) tunnel between
Figure 3 A working model for charge transfer and trapping in DNA is shown for three deo-
xynucleotides (base B
n
attached to sugar phosphate SP) juxtaposed in space. These deoxynucleotides
may or may not be in the same strand or the same duplex. Hole tunneling occurs from SP to B
n
at a
rate k
t
in competition with the hole trapping rate k
irr
, the rate of deprotonation at one of the sugar
carbons resulting in stable sugar product radicals (spr
n
). Holes transferred to B
n
or holes (excess
electrons) initially formed on B
n
tunnel to other bases at a rate k(n F 1)
t
governed primarily the
through-space distance and the ionization energy (electron anity) of B(n F1). Competing with hole
(excess electron) transfer is reversible proton transfer, forming thermally labile trap sites, B(n FH
+
),
at a rate k
r
. Also competing with hole (excess electron) transfer is irreversible hydroxide (proton)
addition, forming stably trapped base product radicals, bpr
n
, at a rate k V
irr
. Not shown in this scheme
is the fact that the k values for holes are dierent than the k values for excess electrons.
Copyright 2004 by Taylor & Francis Group, LLC
guanines at rate k(n)
t
that depends on the distance to the nearest guanine; otherwise, the
rate is relatively independent of whether the transfer is intrastrand, interstrand, or
interduplex. Competing with hole tunneling between the bases is reversible proton transfer
at a rate k
r
. At low temperatures, the hole is locked in place by this reaction but, if the
temperature is suciently high, back transfer of the proton reenables tunneling. At even
higher temperatures, irreversible reactions (kV
irr
) lead to base products (bp
n
).
The excess electron follows the scheme in a similar way, except that there is no attach-
ment or reaction involving the sugar-phosphate backbone. It attaches to the pyrimidines
where its tunneling rate [k(n)
t
] is controlled by the distance to neighboring pyrimidines and
by the competing reversible proton transfer from guanine to the cytosine radical anion. At
low temperatures, proton transfer shuts down tunneling. At higher temperatures, back
transfer permits tunneling. If the temperature is suciently high, irreversible protonation
terminates transfer, forming base products (bp
n
).
Ultimately, if the end products are formed in clusters, the rates k
irr
and kV
irr
must be
fast relative to k(n)
t
. It will be important to determine these rates.
5. DISTRIBUTION OF STABLE END PRODUCTS IN MODEL SYSTEMS
The ultimate goal is to use our understanding of DNA radiation physics and chemistry to
predict what types of damage will occur in DNA in vivo, including yields and spatial
distribution. A step toward achieving this goal is measuring nonradical (diamagnetic)
damage produced by the direct eect. Until recently, there has been a paucity of information
on direct-type end products because of the diculties inherent to isolating and analyzing
products produced in very small amounts, on the order of 1 part in a thousand. In this
section, we review some of the relevant ndings on stable products produced in directly
ionized DNA, then outline a model by which end product distribution can be linked to our
current knowledge of free radical intermediates and conclude with a fewremarks on howthe
reaction pathways will be inuenced by a chromatin environment.
5.1. High Molecular Weight DNA
Much of the early work on measuring direct damage in DNA focused on cleavage of the
sugar-phosphate backbone, called strand breaks. While strand breaks represent an
important class of products, the specic chemical structure is generally left undetermined.
The importance of this type of lesion is primarily due to the fact that it can be detected
with very high sensitivity and secondarily due to observed correlations with biological
endpoints. High detection sensitivity is a consequence of the fact that strand breaks lead to
large changes in DNA conformation.
Among the earliest work, properties such as viscosity and molecular weight were used
to demonstrate that single- and double-strand breaks are produced by direct ionization and
that these products are inuenced by water content [160]. Progress on the measurement of
strand-break yields accelerated with the discovery by W. D. Taylor that closed circular
DNA (BX174 viral DNA) provided an exquisitely sensitive system for detection of single-
strand (ssb) and double-strand breaks (dsb) [161]. Closed circular DNA adopts a tight
supercoil conformation, which relaxes into an open circle when one strand is broken and
into a linear form when two strands are broken within close proximity (
f
10 base pairs) of
each other. Separation and quantication of these three forms were rst performed by
ultracentrifugation and subsequently by gel electrophoresis.
Copyright 2004 by Taylor & Francis Group, LLC
Products Characterized by a Single-Strand Break
The yieldof single-strandbreaks [G(ssb)] anddouble-strandbreaks [G(dsb)] measuredunder
nitrogen using low LET radiation are summarized in Table 1. The reported G(ssb) is
remarkably consistent considering that these results come from ve dierent laboratories.
The top nine entries report prompt strand breaks. The rst four are for relatively dry DNA,
irradiated at dierent temperatures: 77 K, 6jC, and 25jC. The values cluster tightly around
G(ssb)=0.05 Amol/J. The next ve values are for DNAthat is approximately fully hydrated;
in each case, there is a large uncertainty as to the actual hydration level. It is important
to note that these yields were calculated assuming that only DNA comprises the target
mass; for example, if the sample contains an average of 20 waters (MW=18)/nucleotide
(average nucleotide MW=308), then G(dry DNA)=G(hydrated DNA) (308 +20 18)/
308=2.2 G(hydrated DNA). Inclusion of the counter ion in the hydration layer increases
this factor. Thus if the energy absorbed by the DNA hydration layer is converted to strand
breaks with the same eciency as the energy absorbed by DNA itself, G(ssb) should more
than double in going fromdry to fully hydrated. For hydrated DNA, the observed values of
G(ssb) give an average of 0.07 F 0.02 Amol/J. Using this average, one obtains G(dry
DNA)=1.4G(hydrated DNA), which is consistent with transfer of holes from the
solvation shell to the DNA increasing strand breakage by 40%. If all the holes from the
tightly bound water (G<11) transfer to DNA, as is observed at low temperatures, and 50%
of those holes localize on the sugar so as to give strand breaks, then one would expect an
enhancement of roughly (10/20)0.5, i.e., about a 25%increase due to energy transfer from
the hydration layer. Within this framework of logic, a 40% increase can be explained by
either hole transfer occurring froman even larger sphere of water (G
f
16) or by 80%of holes
localizing on the backbone or a combination of these.
In addition to prompt strand breaks, there are other types of damage that, under
certain conditions, will develop into strand breaks. One such class is alkali labile sites; the
existence of this class of damage is revealed by the early work shown in lines 10 and 11 of
Table 1. Another class involves specic types of base damage that are substrates for
enzymatic repair and, consequently, treatment with the appropriate enzyme results in an
ssb. Entries 1215 compared to 14 showthat the yield of this class of damage is 1 to 4 that of
prompt strand breaks. Yokoya et al. [162] employed endonuclease III (Nth) and formami-
dopyrimidine-DNA glycosylase (Fpg); these enzymes act on products stemming from the
one-electron reduction of cytosine and thymine and one-electron oxidation of guanine,
respectively. These results are discussed further below in the section on base damage.
Products Characterized by Double-Strand Break
In Table 1, we see that the spread in values of G(dsb) is larger than in values of G(ssb).
This is not surprising given the inherent experimental diculties that revolve around
measuring the formation of relatively small amounts of linear DNA in the presence of
large amounts of open circular DNA. These diculties aside, the reported ratios for ssb/
dsb fall primarily in the 1020 range. It has been noted [163,164] that this ratio is much too
small to be accounted for by two ssb occurring independently on opposite strands within
close proximity (about 10 base pairs) of each other. While Boon et al. [163] suggest a
mechanism where base-centered radicals give rise to strand breaks, Ito et al. [164] suggest a
mechanism where localized multiple ionizations result in spatially correlated ssb, resulting
in a dsb. There is now considerable evidence that the base radicals, predominantly
oxidized guanine and reduced pyrimidines, are relatively ineective in causing strand
breaks [98,154], making the later explanation more plausible. Indeed, evidence continues
Copyright 2004 by Taylor & Francis Group, LLC
Table 1 Single- and Double-Strand Break Yields for Low LET Radiation in the Absence of O
2
DNA Conditions
a
G(ssb)
b
(Amol/J) G(dsb)
b
(Amol/J) ssb/dsb Reference
1 .pBR322 Dry/77 K 0.047 0.0014 35 [164]
2 .pBR322 Dry/25jC 0.056 0.0031 18 [164]
3 .pUC18 rh<0.1 (G
f
4)/6jC 0.054 0.0053 10 [162]
4 .pUC18 rh<0.1 (G
f
4)/6jC+37jC 0.047 0.0043 11 [162]
5 .qpBR322 Humid (G
f
20)/77 K 0.034 0.001 33 [164]
6 .pBR322 humid (G
f
20)/298 K 0.120 0.011 11 [164]
7 .pUC18 rh=0.97 (G
f
34)/6jC 0.067 0.0072 10 [162]
8 .pUC18 (G
f
34)/6jC+37jC 0.067 0.0072 9 [162]
9 .pBR322 Frozen aqueous/77 K 0.040.07 1020 [163]
10 .BX174 Alkaline sucrose assay 0.070 0.013 5 [188]
11 .SV40 Alkaline sucrose assay 0.110 0.011 10 [189]
12 .pUC18 G
f
4/6jC+37jC+Nth 0.10 0.0069 15 [162]
13 .pUC18 G
f
34/6jC+37jC+Nth 0.18 0.0171 11 [162]
14 .pUC18 G
f
4/6jC+37jC+Fpg 0.094 0.0103 9 [162]
15 .pUC18 G
f
34/6jC+37jC+Fpg 0.16 0.0193 8 [162]
16 .pUC18 G
f
34/6jC+37jC+Nth+Fpg 0.24 0.0237 10 [162]
a
For the rst two entries, alkaline sucrose gradients were used; all others employed gel electrophoresis. Degree of hydration, deduced from known relative humidity, is given
as G, the molar ratio of water to mononucleotide. Sample temperature during radiation.
b
Yields are based on dry mass of DNA.
Copyright 2004 by Taylor & Francis Group, LLC
to build in favor of dsb arising from two radicals, spatially correlated, on opposing strands
[165].
Complimenting the strand-break studies on DNA irradiated in the solid phase are
studies that use high concentrations of radical scavengers to divide out the indirect from
the direct eect in DNA irradiated in aqueous solution (see Refs. 15, 166, and 167). The
work of Krisch et al. was one of the early studies that provided persuasive evidence that
the direct eect is responsible for about half of the dsb produced under in vivo conditions.
The absolute yields obtained from these cell mimetic conditions, however, are dicult to
compare with the solid-state results.
Production of Strand Breaks by Monoenergetic Electrons and Photons
An incident beam, whether comprised of high-energy photons or fast charged particles,
deposits most of its energy via fast electrons. Because the energy of fast electrons is
dissipated by producing secondary electrons, a large fraction of the radiation chemistry is
due to electrons with very low energies,<20 eV [168]. It is of considerable value therefore
to investigate the eciency by which low-energy electrons create strand breaks.
Pioneering the investigation of strand breaks using monoenergetic electrons, Folkard
et al. [169] discovered that energies<25 eV are sucient to cause ssb, and between 25 and
50 eV, the onset of dsb occurs. They employed pBR322 DNA, under vacuum, spread
approximately as a monolayer. At 50 eV, the eciency for ssb/dsb is 2.0; it climbs to 5 for
100500 eV, and it plateaus around 10 for 20004000 eV. More recently, the energy of
electron beams has been extended down to 3 eV by the Sanche laboratory, measuring ssb
and dsb in plasmid DNA under ultrahigh vacuum [68,170,171]. Surprisingly, strand
breaks are observed in substantial yields even at energies below ionization thresholds,
1.510 eV. At energies below 20 eV, the authors provide convincing evidence that
dissociative electron attachment (DEA) is the major pathway for producing a wide range
of damage, including ssb and dsb. Just what fraction of strand breaks, under in vivo
condition, is due to DEAvs. one-electron oxidation is an important question that remains to
be answered.
Hieda and Ito [172] measured strand-break yields as a function of monochromatic x-
ray energy, using 2.146, 2.153, and 2.160 keV, corresponding to below, on, and above the K-
edge for phosphorus. While the eciency of single-strand breaks did not vary signicantly,
the eciency of double-strand breaks did increase by about 30% at the K resonance peak.
These authors raise the possibility that dsb eciency is enhanced by an Auger cascade in
phosphorus. The enhancement of dsb/ssb by resonant excitation of phosphorus provides a
method for assessing the relative contribution of the direct eect to biological endpoints such
as cell death and mutation [173].
Products Derived from the DNA Bases
The formation and identication of the free bases, and subsequently other types of base
damage, have been reported by Swarts et al. [101,174176]. These papers present the rst
measurements of specic products produced by the direct eect in DNA. Equally im-
portant, the sample type and preparation correspond closely to that employed in a wide
range of EPR studies that identify free radical intermediates trapped on DNA [14,30,83].
The free bases (thymine, cytosine, adenine, and guanine) are the simplest form of
product in this class. But the simplicity of these chemical products belies the complexity of
their origin. The yield of each free base was measured using HPLC on salmon sperm
Copyright 2004 by Taylor & Francis Group, LLC
DNA, variably hydrated (2<G<33), subjected to g-irradiation at RT under N
2
or O
2
, and
subsequently dissolved in water [101]. Focusing on the N
2
results, the yield of all four
bases combined holds relatively constant in the hydration range 2<G<33 (0.050.07
Amol/J,) where the G values are based on the weight of DNA plus its hydration shell. In
order to compare these values with those above (Table 1), one needs to recalculate these
based on the dry weight of DNA, which gives a plateau of 0.060.07 Amol/J for 2<G<13
and then climbs to 0.17 Amol/J at G=33. The yields for free base release therefore correlate
well with the ssb yields shown in Table 1, entries 19. Since it is well known that in dilute
aqueous solutions, free base is released as a consequence of radical formation on the
sugar-phosphate backbone [177], it is likely that some (or perhaps all) of the released bases
are derived from sugar-centered radicals. When Swarts et al. analyzed the yields of each of
the free bases separately, they found Gua to be about 60% that of Thy, Cyt, and Ade, with
the yields of the latter three being the same. It is important to reconcile this with the
observation that in crystalline deoxynucleotides [98], the yield of guanine is not reduced
relative to the other three bases, but when DNA is irradiated in dilute aqueous solution
[178], the yield of guanine is reduced. One possibility is that in high molecular weight
DNA, both solid and in dilute solution, the radical formed by hydrogen abstraction from
C4Vis the predominant sugar-centered radical. The C4Vradical undergoes fast h-elimination
of the 3V phosphate, forming an enolether radical cation [179] which is capable of
oxidizing an adjacent guanine via electron transfer [180]. Such an electron transfer would
prevent release of guanine as a free base. Additionally, one must propose that in crystalline
oligodeoxynucleotides, either hydrogen abstraction from C4V is not predominant or 3V
phosphate h-elimination is inhibited.
Swarts et al. [175] measured the yields of 14 base end products in addition to the free
bases. For 2<G<13, the yields (based on total weight) are relatively constant and are
largest for 8-oxoGua (0.0840.101 Amol/J), FapyGua (0.0120.021 Amol/J), and 5,6-
diHThy (0.0260.067 Amol/J). This result is consistent with holes and electrons produced
in the tightly bound water transferring to the DNA and creating the same spectrum of
products, at approximately the same yields, as for holes/electrons formed by direct
ionization of the DNA molecule. The ratio of yields for products stemming from one-
electron oxidation vs. one-electron loss is unexpectedly large: 0.201/0.028, 0.158/0.068,
and 0.150/0.052, for G of 2.5, 10.3, and 13.2, respectively. Since this ratio is expected to be
close to 1.0, the authors conclude that a large fraction of the reduction products were not
detected. The authors present a working model for oxidation product formation by direct
eects in DNA that combines their ndings with the preexisting knowledge on free radical
intermediates. This model, representing a major advance in the eld, proposes that (1) hole
transfer from the sugar-phosphate backbone to the bases competes with damage local-
ization on the sugar, (2) damage localized on the sugar will give rise to a free base, (3) holes
transferred to the base stack are trapped preferentially by guanine, and (4) irreversible hole
trapping occurs primarily by OH
; this
mobile state of the excess electron is formed by back transfer of the proton, from N3 of
cytosine to N1 of guanine. The electron walk is terminated by combination reactions with
base-trapped radicals or by irreversible protonation at the C5C6 bond of cytosine or
thymine. These later reactions are relatively minor because once permissive temperatures
(>140 K) are reached, the concentration of Cyt
.
aq
), which behaves as a free radical, diuses, and interacts with other
atoms until captured. The creation of hydrated electrons is regarded as the start of the
chemical stage at which time the systemis in thermal equilibrium(10
12
sec and <0.025 eV).
A simple consequence of free radical diusion and reactions with other molecules is the
modication and amplication of existing molecular damage from the physical stage as the
consequence of direct interactions in the target site.
Chemical alterations due to direct ionization of the molecules and free radical reac-
tions lead to degradation of the biomolecule and induction of cross-linking, and intercross-
linking of the DNA and proteins [25]. These and other changes to the conformation of the
biomolecules may change the enzymatic activity of the cell [26]. Table 1 provides a quan-
titative summary of the number of events produced by 1 Gy of radiation of dierent quality
in a mammalian cell. The dierences observed in biological eects of radiations of dierent
quality may arise not only fromdierences in their track structures but also from dierences
in physical and physiological conditions. It is noted that although energy deposited in the
cell by 1 Gy of radiation produces a large number of events, only a fraction of this leads to
induction of molecular damage. In mammalian cells, the majority of single-strand breaks
(SSB), in the form of single nucleotide gaps, and damage to DNA bases, in the form of base
products, are readily repaired [27]. Although the frequencies of observed changes in a single
cell induced by ionizing radiations are very low, it is widely believed that most cancers arise
Figure 1 Schematic sequence of events following DNA damage.
Copyright 2004 by Taylor & Francis Group, LLC
from multiple changes in a single cell, in which at least one of the initial events could have
been induced by radiation in the damaged cell or its progeny [28].
3. SIZE SCALE
3.1. Model of DNA
Dimensions of the DNA and its higher-order molecular structures, the nucleosome and the
basic ber, are given in Table 2. It is assumed that the genome of the mammalian cell con-
Table 1 Average Yield of Damage in a Single Mammalian Cell after 1 Gy of Radiation
Radiation Low-LET High-LET Comments
Tracks in nucleus 1,000 4 100 keV electrons and
3.2 MeV a
a
Ionizations in nucleus 100,000 810
5
100 keV electrons and
3.2 MeV a
b
Excitations in nucleus 100,000 810
5
100 keV electrons and
3.2 MeV a
b
Base damage (BD): SSB 3.3 3.4 100 keV electrons and
3.2 MeV a
c
8-Hydroxyadenine 700 [222]
Thymine damage 250 [222]
DSB: initial 38 46 3.4 MeV a [56]
DNA protein cross link 150 [222]
Chromosome aberration 1 3 [220]
Dicentric per cell 0.1 0.4 [220]
HPRT mutation 6.910
6
8.310
5
[219]
HPRT mutation (BG) 3.210
6
310
5
[219]
Lethal lesions 0.5 2.6 [221]
Lethal lesions per DSB 0.01 0.045 250 KV X, 100 keV/Am a [221]
DSB per lethal lesion 87 22 250 KV X, 100 keV/Am a [221]
Cell inactivation 30% 85% [223,224]
Complex SSB 4% 30% 100 keV electrons,
3.2 MeV a [7981]
Complex SSB
B
40% 70% 100 keV electrons,
3.2 MeV a [7981]
Complex DSB 20% 70% 100 keV electrons,
3.2 MeV a [7981]
Complex DAB
B
60% 90% 100 keV electrons,
3.2 MeV a [7981]
Fract hybrid DSB 0.38 0.25 100 keV electrons,
3.2 MeV a [7981]
d
a
Data from authors Monte Carlo track structure calculations.
b
For a nucleus 10-Am diameter.
c
See Table 5.
d
Hybrid DSB: when a SSB induced by direct energy deposition converted to a DSB by OH.
Copyright 2004 by Taylor & Francis Group, LLC
tains 6 pg of DNA in a nucleus, a physical size of 6349 Mbp [29], a molecular weight of an
average base pair of 660, and a length of a base pair of about 0.34 nm.
Models of DNAof various degrees of sophistication have been used in the calculations
of radiation-induced DNA damage. These can be classied into three groups. The simplest
of DNA models is a linear segment in the form of a cylinder [30]. These segments can be
generated along random chords cutting a convex body (spherical nucleus) using the method
of A-randomness [31]. This model has mainly been used to obtain frequencies of energy
depositions in macromolecular structures without a priori assumption of the role of the
atomic structure in determining the biological responses. A more realistic model used in our
early studies of DNA damage is the volume model of DNA [32]. In the volume model, DNA
is in its native B-formwith a diameter of 2.3 nmand divided into 0.34-nmslices. In turn, each
slice is divided into three volumes comprising the central core representing the volumes of
the complementary paired nucleobases and the two arches, each representing the volumes
occupied by the deoxyribosophosphate backbone of each strand. This model does not take
into account the detailed atomic structure of DNA. The third category is the sophisticated
atomic models of DNA that have been available for a number of years, and can now readily
be generated using commercially available programs such as Newhelix [33], Curves, and
MidasPlus [34,35]. Further renements of the latter models include the distribution of sol-
vent molecules around the sugarphosphate backbone and the nucleobases [36 37]. Mod-
eling of higher-order structures of DNA, nucleosome, chromatin, and other forms such
as triplexes, have been made by several authors [3841] and also available from the PDB
library.
3.2. A Model of the First Hydration Shell of DNA
The following describes the development of an atomistic model of DNA including the rst
hydration shell of a B-DNA [37]. Water plays an important role in the three-dimensional
structure of DNA, the conformation it adopts, and its response to radiation. Inuence on
the conformation of DNA by the solvent has been known since the earliest ber diraction
studies [42]. The native conformation of the macromolecule to a large degree is inuenced by
inter- and intramolecular forces, such as hydrogen bonding forces, repulsive and dispersive
forces, and interaction with the solvent. At high relative humidity or water activity, cor-
responding to low salt or low atomic organic solvent content, DNA adopts the B-form,
which is considered to be the most biologically relevant form of DNA. If the relative hu-
midity is lowered, DNA transforms from B-form to A-form.
Table 2
Structure
Diameter
(nm)
Segment
Length (nm)
Number per
Genome Comment
DNA segment 2.3 2.3 9.310
8
Assuming 6 pg of DNA in
nucleus, MW of 660, DNA
duplex 1.86-m length
Nucleosome 10 5 single
nucleosome
2.910
7
190 bp containing 50 nm of
DNA including the linker
Chromatin
segment
30 30 1.910
6
Solenoid of nucleosomes
Copyright 2004 by Taylor & Francis Group, LLC
The basic building block of DNA is the nucleotide. The backbones run in opposite
directions with each base pair having a purine base (A/G) hydrogen bonded to a pyrimidine
base (T/C). The purine and pyrimidine bases are on the inside of the helix while the phos-
phate and sugar are on the outside. The two chains are held together by hydrogen bonds
between the base pairs. Adenine is paired to thymine by means of two hydrogen bonds while
guanine is paired to cytosine by three hydrogen bonds. A single nucleotide consists of three
chemical parts; a simple sugar molecule, an inorganic phosphate and a nitrogen-containing
base. Successive nucleotides are linked together via a phosphodiester bond between the
sugar and phosphate of adjacent nucleotides. The nitrogen-containing bases are not in-
volved in any covalent linkages other than their attachment to the sugarphosphate back-
bone. It is the sequence of these nitrogen-containing bases along the sugarphosphate
backbone that constitutes the unique structural and functional individuality of the DNA
molecule. The nitrogen bases are linked at the C
1
position while the C
2
and C
4
hydroxyl
groups participate in phosphodiester bonds. The nitrogen-containing bases are either the
double-ringed purines or the single-ring pyrimidines. The phosphate group is present as part
of a diester linkage. The DNA bers can exist in three major forms, as B-DNA under
conditions of high humidity, as A-DNA, and a left-handed family of Z-DNA. The dier-
ences between the families occur in the way the sugarphosphate backbone is wrapped
around the helix axis, the way in which the base pairs are stacked and in the pucker of the
furanose ring [43]. The A-DNA form with 11 base pairs per pitch of 28 A
is much more
compressed than the B-DNA form with 10 base pairs per pitch of 34 A
. Another dierence
between the A and B forms is that the sugar in the B-DNA is puckered in the C(2V)-endo
family region, extending from O(4V)-endo to C(1V)-exo, resulting in a wide interphosphate
separation of 6.6 A
along the chain. For the A-DNA, the sugar puckering is conned to the
C(3V)-endo resulting in a shorter interphosphate separation of approximately 6 A
. Thus
phosphate groups in A-DNA can be bridged by water molecules as the interphosphate
distances is not too large while all phosphate groups are individually hydrated because the
interphosphate separation is too large [44]. These dierences results in dierent hydration
patterns around the phosphate groups, which may oer a scheme for the transition from A-
to B-DNA. At lower humidity, the hydration becomes more economical and the sugar
puckering changes, eectively reducing the interphosphate separation and so allowing water
molecules to bridge the free phosphate oxygen atom [45].
DNA hydration studies with various techniques have shown that there are approx-
imately 10 and 20 water molecules in the rst hydration of A- and B-DNA, respectively.
These water molecules bind in decreasing order of strength to the anionic oxygens of the
phosphate group, to the ester oxygens of the phosphodiester linkage, to the O(4V) oxygen
of the furanose ring, and to the electronegative atoms of the base pairs [46].
Distribution of water molecules around the polar atoms of a decamer canonical
B-DNA structure were obtained from quantitative analysis of the solvent interactions
within hydrogen bonding distances of polar atoms of oligonucleotides using 12 B-DNA
oligonucleotide crystal structures [37]. Table 3 shows the distribution of water molecules
around the bases, the phosphate group, and the sugar. The data are presented in terms of
polar coordinates around a polar atom in the plane of abc as indicated in the table. Data
shows that water interactions around the bases mainly occur with the polar atoms of the
major and minor grooves. In this manner, the data in Table 3 have been used to generate
variable DNA segments. The rst atom in the table given in bold type is the atom at the
origin of the r, #, U coordinate system.*
*Atomic coordinates of the hydrated B-DNA can be obtained from the author.
Copyright 2004 by Taylor & Francis Group, LLC
4. APPLICATION OF TRACK-STRUCTURE CALCULATIONS
IN UNDERSTANDING RADIOBIOLOGICAL MECHANISMS
Motivation for understanding mechanism of cellular damage by ionizing radiation tracks
dates back to the pioneering work of Lea [47]. Investigators since then have tried to relate
the initial events produced when a track of ionizing radiation traverses the cell nucleus to
the observed biological lesions. In most cases, such a description starts with knowledge of
spatial distribution of energy loss in subcellular structures such as DNA. For example, for
practical reasons, assessment of risk of radiation exposure requires allowances to be made
for the dierences in biological eectiveness of dierent types of radiation. Because of the
stochastic nature of radiation, such assessments depend on the absorbed dose in the vol-
ume of the target, the quality of radiation, and the size of the target. We also know that
relevant biological eects of radiation are mostly because of damage in individual cells. To
this end, it has also been established that the spatial and temporal distribution of radiation
interactions within the cell or its nucleus has an important inuence on its biological
eectiveness. Therefore these distributions must be considered if we are seeking either
practical quantities for comparison of dierent radiations in risk assessment and therapy,
or seeking a greater understanding of the fundamental problems in the mechanism of ra-
diation action. In pursuit of a suitable description of the mechanism(s) of radiation eects,
a number of biophysical models have been put forward including the exchange theory
of Revel, the breakage rst theory of Lea and the lesionnonlesion interaction by
Chadwick and Leenhouts [20,48,49,82].
4.1. Energy Depositions in Molecular Targets
Although local energy depositions can be experimentally measured over subcellular dis-
tances of nearly 1-Amdiameter, biophysical theories require knowledge of energy deposition
Table 3 Distribution of Water Sites for B-DNA
Atom
abc r(A) hj /j
Phosphate O1PPO2P 2.91 53 145
2.63 32 103
O2PPO1P 2.95 48 179
2.76 46 74
Sugar O4VC1VC2V 2.84 53 174
Cytosine O2C2N3 2.94 22 160
N4C4C5 3.07 63 9
Thymine O2C2N3 2.77 22 123
2.84 47 104
O4C4C5 2.68 45 15
Adenine N3C4C9 2.84 43 23
N6C6N1 3.18 41 176
N7C5C6 2.70 43 7
Guanine N2C2N3 3.36 24 35
O6C6N1 2.70 43 176
N3C4N9 2.91 45 7
N7C5C6 2.77 46 9
Copyright 2004 by Taylor & Francis Group, LLC
in small targets of the dimensions of DNA molecules. To this end, Monte Carlo track
structure methods allow calculations of energy depositions and DNA damage at resolu-
tions down to subnanometer dimensions (see, e.g., Chap. 4). To obtain the distribution
of energy deposition in molecular targets, the conceptual framework of the method is to
place the track in a virtual volume big enough to contain the entire track and then randomly
place the target structure in the volume. The method ensures the establishment of an elec-
tronic equilibrium and accurate calculation of absorbed dose in the target. In this way,
the energy deposited in the target can be grouped together in slices of 0.1-nm length to give
the total energy in the target. Fig. 2 shows examples of such distributions in three vol-
umes pertaining to a segment of DNA, a nucleosome, and a chromatin ber segment for
dierent radiation qualities. The left ordinate shows comparison of distributions of abso-
lute frequency of energy depositions in the target, randomly positioned and oriented in
water and irradiated with 1 Gy of the given radiation. The right ordinate gives the corre-
sponding average number of events in the target volume in one typical mammalian cell.
Subsequently, more complex forms of DNA structures could be used to obtain more de-
tailed physical and chemical information on the nature of damage from the distributions of
energy depositions.
4.2. Summary
Frequency of energy deposition in volumes of dimensions similar to DNA and other bio-
logical molecules provides a method of interpretation of mechanism of radiation eects. In
this method, the assumption is that radiation quality eects are mostly because of the spatial
properties of the radiation on a microscopic scale. The database of absolute frequencies of
energy depositions in cylindrical targets of dimensions 1100 nm include:
Modeling and calculations show that with increasing LET, there is a shift toward
a greater complexity of clustered damage in DNA [30,81,87,119].
Figure 6 Fraction of complex double strand breaks for protons (open squares), alpha particles
(solid squares) as a function of LET, and electrons (triangle). The contribution of base damage is not
included in these calculations.
Table 5 Frequency Distribution of a Number of Base Damage
3.2 MeV alpha particles
No. of Base BDV3 bp BD>3 bp
Damage 0 1 2 3 4 z5 adj opp adj opp Total Hits
Model 1 No break 31.4 12.5 6.9 3.7 3.5 1.7 1.9 4.3 4.6 58.0
Model 2 SSB 5.1 4.8 4.0 4.8 2.6 2.6 1.1 0.9 1.3 1.6 23.9
Model 3 SSB+ 0.1 0.5 1.3 1.0 1.0 2.4 0.1 0.1 0.0 0.3 6.3
Model 4 2SSB 0.2 0.1 0.1 0.2 0.3 0.8 0.0 0.1 0.0 0.0 1.7
Model 5 DSB 0.2 0.6 0.7 1.1 1.0 1.7 5.3
Model 6 DSB+ 0.1 0.1 0.2 0.6 1.0 1.7 3.6
Model 7 DSB++ 0.0 0.1 0.1 0.0 0.2 0.8 1.2
Ratio of base to sugar/phosphate damage = 3.4.
Copyright 2004 by Taylor & Francis Group, LLC
Experiments as well as calculations show the yield of strand breaks per unit dose
is nearly constant over a wide range of LETs [75,80,81].
The majority of strand breaks is of simple type containing a single strand break;
however, if base damage is taken into account, the majority of single strand breaks
appears to be complex [81].
These clusters occur with high frequencies per unit dose [68].
For high-LET radiations, per unit dose fewer regions of genomes are hit, i.e.,
smaller number of tracks per unit dose (Table 1).
1
t ln
T
15
in which the unit of r is 10
16
cm
2
and that of T, the particle kinetic energy, is eV. The
tting parameters are given by
a 2:07201 b
1
0:271302 b
2
0:119638 r 1:46521 c 0:074
At energies higher than 10 keV, ionization cross sections were calculated using Selt-
zers formula [183]. Seltzers formula gives the partial ionization cross sections for ve
molecular orbitals of water. The total ionization cross sections in the energy range of 10 keV
to 10 MeV were obtained by summing up all the partial cross sections (Fig. 8).
Excitation Cross Sections
There are various modes of excitation for T > 7.4 eV. As reported experimental excitation
cross sections are fragmentary, we used the compiled data of Paretzke [184] that includes
all the major excitation modes. Paretzke tted the experimental cross sections for the 10
major individual states using a model function:
r
exc
U
3
T
4pa
2
0
R
T
M
2
a
ln
4c
s
T
R
2
Copyright 2004 by Taylor & Francis Group, LLC
in which U
3
(T) is an empirical correction factor of the form
U
3
T 1 e
0:25T=E
a
1
3
where a
0
is the Bohr radius = 0.529310
8
cm, and R is the Rydberg constant = 13.6 eV.
Parameters M
a
2
, E
a
, and c
s
for each level are given by Paretzke. An additional three levels
were tted by the analytical model given by Gren and Stolarski [185]. Table 6 lists eight
major excitations and excitation energies considered in this work. The total excitation
cross section in the energy range of 10 eV to 10 keV was obtained by summing up all the
individual cross sections. The total excitation cross section for the high-energy region up
to 10 MeV is given by an empirical formula derived from the Fano plot, with tting pa-
rameters determined by Berger and Wang [186]:
r
exc
4p
a
0
137b
_ _
2
12:30 1:26 ln
b
2
1 b
2
b
2
_ _ _ _
4
in which b is the ratio of electron velocity to the velocity of light. Total excitation cross
section in the energy range of 10 eV to 10 MeV is shown in Fig. 8.
Figure 7 Experimental total ionization cross section as a function of electron energy. The solid line
is the least-squares t to all data [177181]. The calculated data of Dingfelder et al. [182] for liquid
water is plotted for comparison.
Copyright 2004 by Taylor & Francis Group, LLC
Elastic Scattering Cross Sections
Elastic scattering cross sections were calculated using the Rutherford formula, taking into
account the screening parameter given by Moliere [187]. The dierential cross section (dr/
dX)
el
and total cross sections r
el
for each molecule are represented by
dr
dX
_ _
el
ZZ 1r
2
e
1 b
2
b
4
1
1 cosh 2g
2
5
r
el
pZZ 1r
2
e
1 b
2
b
4
1
gg 1
6
Figure 8 Total cross sections for ionization, excitation, and elastic scattering of electrons in water
vapor in the energy range of 10 eV to 10 MeV.
Table 6 Excitations and Transition
Energy Considered in the Kurbuc Code
Excitation state Energy (eV)
A
1
B
1
7.4
B
1
A
1
9.7
Diuse band 13.3
Rydberg (A+B) 10.0
Rydberg (C+D) 11.0
H* Lyman a 21.0
H* Balmer a 21.0
OH* 9.0
Copyright 2004 by Taylor & Francis Group, LLC
The screening parameter g is given by
g g
c
1:7 10
5
Z
2=3
1
ss 2
7
where the eective atomic number of the water molecule, Z, was assumed to be 7.42, r
e
is the classical electron radius, s = T/m
0
c
2
is the kinetic energy in units of the electron
rest mass.
g
c
1:198 T < 50keV 8a
1:13 3:76
Z
137b
_ _
2
Tz50keV 8b
_
_
The parameter g
c
was determined by the least-square tting to the experimental data
[188191]. Total elastic scattering cross section in the energy range of 10 eV to 10 MeV is
shown in Fig. 8. Angular distributions for elastic scattering below 1 keV were obtained by
direct sampling of various experimental data [189,190,192].
Secondary Electrons
The calculation of the secondary electron spectrum was carried out using the method of
Seltzer [183]. For the jth orbital of a molecule, the cross-section dierential in kinetic
energy w of the ejected electron is written as the sum of close and distant collisions.
dr
j
dw
dr
j
c
dw
dr
j
d
dw
9
The rst term describes collision between two electrons.
dr
j
c
dw
2pr
2
e
m
0
c
2
n
j
b
2
T
T B
j
U
j
1
E
2
1
T w
2
1
T
2
s
s 1
_ _
2
1
ET w
2s 1
s 1
2
G
j
_ _
10
G
j
8U
j
3p
1
E
3
1
T w
3
_ _
tan
1
y
p
y
p
y 1
y 1
2
_ _
11
in which n
j
is the number of electrons in the orbital, B
j
is the orbital binding energy, U
j
is the
mean kinetic energy of the target electron in the orbital, E is the energy transfer (=w+B
j
)
and y =w/U
j
. Table 7 shows electron number, binding energy [193], and kinetic energy [194]
Table 7 Electron Number n
j
, Binding Energy B
j
, and Kinetic
Energy U
j
for the jth Orbital of a Water Molecule
j Orbital n
j
B
j
(eV) U
j
(eV)
1 1b
1
2 12.62 61.91
2 3a
1
2 14.75 59.52
3 1b
2
2 18.51 48.36
4 2a
1
2 32.4 70.71
5 1a
1
2 539.7 1589.5
Copyright 2004 by Taylor & Francis Group, LLC
for the jth orbital of a water molecule. The second term is described in terms of the
interaction of the equivalent radiation eld with the orbital electrons.
dr
j
d
dw
n
j
IEr
j
PE
E 12
where r
PE
( j)
is the photoelectric cross section for the jth orbital (per orbital electron), for an
incident photon of energy E, I(E) is the virtual-photon spectrum. Fig. 9 shows the dis-
tribution of secondary electrons for 1 and 10 keV primary electrons calculated by Eqs.
(10)(12) in comparison with the calculated data of Paretzke [184] and the experimental data
of Vroom and Palmer [195] and Bolorizadeh and Rudd [180].
Angular distribution of secondary electrons for the ejected energy w z 200 eV was
calculated using the kinematical relationships
cos h
wT 2m
0
c
2
Tw 2m
0
c
2
13
For w<200 eV, regardless of the primary electron energy, angular distributions were ran-
domly sampled using the experimental data of Opal et al. [196].
Monte Carlo Electron Code
The Monte Carlo track structure code kurbuc simulates electron tracks in water vapor for
initial electron energies 10 eV10 MeV [174]. The code kurbuc provides all coordinates of
Figure 9 Energy spectra of secondary electrons for 1 and 10 keV electrons produced by the Seltzers
theory in comparison with the calculations by Paretzke [184] and the experimental data [180,195].
Copyright 2004 by Taylor & Francis Group, LLC
molecular interaction in water vapor, amount of energy deposited at each event, and the type
of interaction at each event. Using tracks generated by kurbuc, various physical parameters
such as penetration distances, W values, radial distributions of interactions, point kernel,
and frequencies of energy deposition in small cylindrical targets can be obtained. As an
example of track structure analysis, Fig. 10 shows results of W values of low-energy elec-
trons in water vapor. Wvalues are derived from scoring the number of electronH
2
O
+
pairs
because of ionization process. Calculated W value at the high-energy limit agrees with the
experimentally determined asymptotic limit of 30.4 eV.
9.2. Ions
Ionization Cross Sections
Here we use the termions limiting to protons H
+
and alpha particles He
2+
. For fast ions,
the majority of energy is transferred in ionizing collisions, resulting in energetic free
electrons and the potential energy of residual ions. Ionizing collisions involving bare ion
projectiles, unless they are very slow, can be treated by using a classical description for the
trajectory of the projectile with a quantal description of the target. When fast ions slowdown
around the Bragg peak (0.3 MeV/u), interactions involving electron capture and loss by the
moving ions become an increasingly important component of the energy loss process.
Detailed descriptions for charge transfer processes will be given in Sec. 9.3.
Total ionization cross sections for bare ions (H
+
and He
2+
) were obtained by tting
polynomial functions to the experimental data given by Rudd et al. [198] for protons, and
Figure 10 Calculated W values for electrons in water vapor as a function of the initial energy of
electrons in comparison with the data of Combecher [197].
Copyright 2004 by Taylor & Francis Group, LLC
Rudd et al. [199] and Toburen et al. [200] for alpha particles.
r
ion
j
c
j
log
10
T
jl
14
in which T is in keV/u. Where data were lacking, extrapolation was made by taking into
account the reproducibility of stopping powers. Fig. 11 shows total ionization cross sections
of water vapor by H
+
(left) and He
2+
(right).
Excitation Cross Sections
There are no experimental data of excitation cross sections for proton and alpha particle
impact for water. The proton cross sections were obtained by scaling of the electron exci-
tation cross sections for high-energy protons >500 keV [201]. For the lower-energy regions,
the semiempirical model developed by Miller and Green [202] was adopted, which is based
on the electron impact excitation. They assumed an analytical function for each excited level
of the form
r
e
r
0
Za
X
T W
v
J
Xv
T
Xv
15
where r
0
=10
16
cm
2
, Z=10, and the parameters W, a, J, X, and v for 28 excitation states are
given in Table 3 of Ref. 202. We assumed cross sections for protons can be applied to the
impact excitation cross sections for alpha particles as shown in Fig. 11.
Figure 11 Total cross sections due to proton (left) and alpha particle (right) impact on water vapor.
Total ionization cross sections were obtained by tting polynomial functions to the experimental data
[198200]. The curve for excitation was assumed to be the same between protons and alpha particles.
Elastic scattering was evaluated by the classical mechanics trajectory calculations [Eqs. (16) and (17)].
Copyright 2004 by Taylor & Francis Group, LLC
Elastic Scattering Cross Sections
High-energy ions proceed along a straight-line trajectory. However, the eects of elastic
nuclear collisions will become more prominent in the low-energy region. Elastic collisions
transfer little or no energy but can have a signicant eect on the spatial character of the
track structure at very low energies. Also, the nuclear energy loss cannot be neglected <10
keV/u. Nuclear elastic scattering was evaluated using the classical mechanics trajectory
calculations taking into account the screening by atomic electrons [203]. For a particle
scattered in a central potential V(r), the deection angle h is obtained as a function of the
impact parameter p [204] such that
h p 2
_
l
r
min
p
r
2
1 Vr=T
cm
p
2
=r
2
_ dr 16
where r
min
is the distance of closest approach, T
cm
is the particle energy in the center-of-mass
system. Equation (16) was numerically solved using procedures given by Everhart et al.
[205]. The dierential elastic scattering cross section can be obtained by the classical rela-
tionship [206],
dr
el
dX
p
sin h
dp
dh
17
To reduce the divergence of the total cross section, which is obtained by integral of Eq. (17),
the cuto angle was set so as to limit the increase in the scattering probability at low scat-
tering angle. Fig. 11 shows calculated total cross section for H
+
(left) and He
2+
(right).
Secondary Electrons Distribution
Energy spectra of secondary electrons ejected by ion impact were calculated using the
empirical model given by Rudd [207]
dr
j
e
de
S
B
F
1
F
2
w
1 w
3
f1 expaw w
c
=vg
18
where w=e/B, v =(T/kB)
1/2
, w
c
=4v
2
2vR/4B, S =4pa
2
0
Nz
2
(R/B)
2
, e =electron energy,
B = binding energy for each orbital, a
0
= 0.52910
8
cm, N = electron number in orbital,
z = charge of the projectile nucleus, and R = 13.6 eV,
F
1
v
L
1
H
1
L
1
C
1
v
D
1
=1 E
1
v
D
1
4
H
1
A
1
ln1 v
2
=v
2
B
1
=v
2
_
19
F
2
v
L
2
H
2
=L
2
H
2
L
2
C
2
v
D
2
H
2
A
2
=v
2
B
2
=v
4
_
_
The tting parameters for water vapor are given by Rudd et al. [208]
A
1
= 0.97 B
1
= 82 C
1
= 0.40 D
1
= 0.30 E
1
= 0.38
A
2
= 1.04 B
2
= 17.3 C
2
= 0.76 D
2
= 0.04 a = 0.64
Copyright 2004 by Taylor & Francis Group, LLC
This model gives the single dierential cross sections (SDCS) for secondary electron
production by ion impact, and also the partial ionization cross sections for each orbital. The
boundary energy for energetic secondary electrons generated by ionization was determined
as 1 eV. Fig. 12 shows the randomly sampled energy spectra of ejected electrons for various
energies of H
+
(left) and He
2+
(right) in comparison with experimental data [200,209].
Experimental angular distributions of electrons ejected from water by H
+
impact are
given by Bolorizadeh and Rudd [209], and by Toburen and Wilson [210] in the energy range
between 15 keV and 1 MeV. Available experimental angular distributions for He
2+
are lim-
ited. To cover the broad ion energy range, available H
+
data were used assuming that dif-
ferences of angular distributions between dierent ions are not very large. Interpolation was
made for the intermediate energies of electrons and ions.
9.3. Low-Energy Ions
Charge Transfer Cross Sections
Charge transfer can produce residual ions without the release of free electrons, and free
electrons can be ejected fromthe moving ion (or neutral) with no residual ions being formed.
Charge transfer cross sections are generally designated as r
if
where i is the initial and f
is the nal charge state of the moving particle. Table 8 lists the interactions for low-energy
proton (H
+
) and neutral hydrogen (H
0
) with water. Table 9 lists the interactions for low-
energy He
2+
, He
+
, and He
0
. To develop track structure models, cross sections are needed
for both electron capture and electron loss. For incident ions with speeds comparable to or
Figure 12 Randomly sampled energy spectra for secondary electrons ejected by proton (left) and
alpha particle (right) impact in comparison with experimental data [200,217]. The lines are the results
obtained from sampling 5 10
5
times using calculated data from the Rudd model [Eqs. (18) and
(19)].
Copyright 2004 by Taylor & Francis Group, LLC
lower than the orbital speeds of bound electrons in the target, the capture of electrons from
the target is a signicant mechanism for the ionization of the target without production of
secondary electrons. If the projectile is a dressed ion (i.e., an ion bearing one or more
electrons), a collision can eject electrons from either the projectile or the target. For a fast
projectile, one can consider the projectile as a simple collection of a nucleus and electrons
traveling together at the same speed. After the charge transfer, the dressed ion or the neutral
atom becomes either stripped or ionizes a water molecule. In the former case, stripped ion,
an electron is ejected in a forward direction with nearly the same velocity as the projectile. In
the latter case, ionization of water molecule without changing the ionic charge state, an
electron is ejected with a dierent distribution of electron energy and with a similar ejection
angle as that of the bare ions.
Fig. 13 shows charge exchange cross sections based on available experimental data
for H
+
and H
0
(upper left) [211,212], He
2+
(lower left) [199], He
+
(lower right) [213,214],
and He
0
(upper right) [214,215]. There is a probability of two-electron transfer for helium
atom, such as r
20
and r
02
. Total cross sections for electron capture r
10
for H
+
and
electron loss r
01
for H
0
were evaluated using the analytical functions developed by Miller
and Green [202]. Cross sections for He ions were least-square tted by a simple polynomial
function similar to Eq. (14). Smooth extrapolation was carried out where the experimental
Table 8 Interactions for Low-Energy H
+
and H
0
with Water
Interactions H
+
+ H
2
O! H
0
+ H
2
O!
(i) Elastic scattering H
+
+ H
2
O H
0
+ H
2
O
(ii) Target ionization H
+
+ H
2
O
+
+e H
0
+ H
2
O
+
+ e
Target excitation H
+
+ H
2
O
*
H
0
+ H
2
O
*
(iii) Electron capture H
0
+ H
2
O
+
(r
10
)
Electron loss H
+
+ e + H
2
O (r
01
)
(iv) Electron capture and
target ionization
H
0
+ H
2
O
2+
+e
Electron loss and
target ionization
H
+
+ e + H
2
O
+
+ e
Table 9 Interactions of Low-Energy He
2+
, He
+
, and He
0
with Water
Interactions He
2+
+ H
2
O! He
+
+ H
2
O! He
0
+H
2
O!
(i) Elastic scattering He
2+
+H
2
O He
+
+H
2
O He
0
+H
2
O
(ii) Target ionization He
2+
+H
2
O
+
+e He
+
+H
2
O
+
+e He
0
+H
2
O
+
+e
Target excitation He
2+
+H
2
O* He
+
+H
2
O* He
0
+H
2
O*
(iii) Two-electron capture He
0
+H
2
O
2+
(r
20
)
One-electron capture He
+
+H
2
O
+
(r
21
) He
0
+H
2
O
+
(r
10
)
One-electron loss He
2+
+e +H
2
O (r
12
) He
+
+e +H
2
O (r
01
)
Two-electron loss He
2+
+e +e +H
2
O (r
02
)
(iv) One-electron capture
and target ionization
He
+
+H
2
O
2+
+e He
0
+H
2
O
2+
+e
One-electron loss and
target ionization
He
2+
+e +H
2
O
+
+e He
+
+e +H
2
O
+
+e
Copyright 2004 by Taylor & Francis Group, LLC
data were lacking. Fig. 14 shows calculated equilibrium charge fractions as a function of
ion energy [215]. These fractions are useful for the analytical calculations of electronic
stopping powers [216].
Cross Sections for Dressed Ions
The total ionization cross section for neutral hydrogen impact was obtained by a least-
square tting the experimental data of Bolorizadeh and Rudd [217]. Assuming similarity
with proton impact ionization, cross sections were approximated by scaling proton ion-
ization to give smooth transitions for the energy regions <20 or >150 keV/u, taking into
account the reproducibility of electronic stopping powers. Total ionization cross sections
for He
+
were tted to the experimental data of Rudd et al. [213] and Toburen et al. [200].
Total ionization cross sections for He
0
at energies lower than 100 keV/u were adjusted to
t the electronic stopping powers tabulated in ICRU Report 49 [203]. In the region below
30 keV/u, the stopping powers for He
0
is the main contributor. At energies above 100 keV/u,
the total cross sections were assumed to be the same as those of He
+
. Fig. 15 shows the total
ionization cross sections for H
0
, He
+
, and He
0
.
Figure 13 Charge exchange cross sections due to H
+
and H
0
(upper left panel), He
2+
(lower left
panel), He
+
(lower right panel), and He
0
(upper right panel) impact on water vapor. The curves are
tted to experimental data [199,211215] by various model functions and extrapolated where data
are lacking.
Copyright 2004 by Taylor & Francis Group, LLC
Figure 15 Total ionization cross sections due to dressed ion impact on water vapor. Cross sections
for H
0
and He
0
were adjusted to reproduce stopping powers for the lower energy protons and alpha
particles. (Experimental data from Refs. 200, 213, and 217.)
Figure 14 Equilibrium charge fractions of hydrogen beam (left panel) and helium beam (right
panel) in H
2
O calculated by the theory of Allison [215].
Copyright 2004 by Taylor & Francis Group, LLC
It was assumed that the spectra and angular distributions of electrons ejected from
the target molecule by neutral hydrogen impact were equal to that for protons. Toburen et
al. [200] have provided the ratios of SDCSs for ionization of water vapor by He
+
to those
for He
2+
as a function of the ejected electron energy. This ratio is distributed linearly from
0.4 to 1 for 0<e<90 eV for various projectile energies between 200 and 500 keV/u. The
energy spectra for secondary electrons because of He
+
impact were obtained by modifying
the Rudds model [Eqs. (18) and (19)] using a simple function g(e) such that:
dr
He
i
e
de
ge
dr
i
e
de
20
in which
ge 0:4 e=150 for e < 90 eV
1 for e z 90 eV
21
Fig. 12 (right) shows a good agreement between the calculated spectra and the experimental
data for He
+
energies of 200 and 500 keV/u. Available experimental angular distributions of
electrons ejected from water by He
+
are limited. To cover the broad ion energy range,
available proton data were used assuming that dierences of angular distributions between
dierent ions are not very large. Again, we assumed the spectra and angular distributions of
electrons ejected from the target molecule by He
0
impact to be equal to that for He
+
. We
assumed both excitation cross sections for protons [Eq. (15)] and elastic scattering cross
Figure 16 Mean lineal energy for 1 MeV proton tracks (left panel) and 5 MeV alpha particle tracks
(right panel) passing through and near a 1-Am diameter site. (), present calculations; () pits.
(From Ref. 74.)
Copyright 2004 by Taylor & Francis Group, LLC
sections for bare ions [Eqs. (16) and (17)] can be applied to respective cross sections for three
dressed ions (H
0
, He
+
, and He
0
).
Monte Carlo Codes for Low-Energy Ions
Based on the evaluated cross sections, a new generation of Monte Carlo track structure
codes were developed for simulation of tracks of low energy protons (the code lephist)
[175] and alpha-particles (the code leahist) [218] in water. The eective energy ranges are 1
keV1 MeV for protons and 1 keV8 MeV for alpha particles. The tracks of secondary
electrons were generated using the electron track code kurbuc. The codes can be operated
to generate tracks both in the track segment and the full-slowing-down modes. Tracks
were analyzed to provide conrmation on the reliability of the code and information on
physical quantities, such as range, W values, restricted stopping power, and radial dose
proles.
Model Calculation
Tissue equivalent proportional counter (TEPC) is the instrument for measuring the dis-
tributions of energy deposited in a specic volume. The instrument is commonly designed as
a spherical or a cylindrical volume with varying sensitive volume. Fig. 16 shows the dis-
tribution of mean lineal energy as a function of impact parameter b normalized by the radius
of the sensitive volume. The parameter b (the eccentricity) is the distance from the center of
the cavity to the chord of the ion cutting the sphere and R
sv
is the radius of the counter.
Distribution of the mean lineal energy were simulated for 1 MeV protons and 5 MeV alpha
particles through a wall-less counter of 1-Am diameter.
REFERENCES
1. UNSCEAR, Sources and eects of ionizing radiation. UNSCEAR 2000 Report, Supplement
N-46(A/55/46) volume 2. United Nations Publications, 2000.
2. UNSCEAR. Sources and eects of ionizing radiation. United Nations Scientic Committee
on the Eects of Atomic Radiation. 1993 Report to the General Assembly. Annex F: Inuence
of dose and dose rate on stochastic eects of radiation, 1993.
3. The Low Dose Radiation Risks is part of The Sixth Framework Program of the European
Union. http://www.cordis.lu/fp6
4. The Low Dose Radiation Research Program of the U.S. Department of Energy http://low-
dose.org.
5. National Council on Radiation Protection and Measurements, NCRP Report 132, Bethesda
MD, 2000. http://www.ncrp.com/
6. Strategic program plan for space radiation health research, Life Sciences Division, Oce of
Life and Microgravity Sciences and Application, NASA, Washington DC 20546.
7. National Academy of Sciences Space Science Board, Report of the Task Group on the Bio-
logical Eects of Space Radiation. Radiation Hazards to Crews on Interplanetary Mission
National Academy of Sciences, Washington, DC, 1997.
8. Modeling Human Risk: Cell and Molecular Biology in Context. Report No. LBNL-40278.
DOE, National Technical Information Service. US Department of Commerce, 5285 Port
Royal Rd, Springeld, VA 22161, 1997.
9. Mothersill, C.; Seymour, C. Radiat. Res. 2001, 155, 759.
10. Sasaki, M.S.; Ejima, Y.; Tachibana, A., et al. DNA damage response pathway in radioadaptive
Copyright 2004 by Taylor & Francis Group, LLC
response. Mutat. Res. Fundam. Mol. Mech. Mutagen. JUL25 2002, 504 (12), 101118. Special
Issue.
11. Lorimore, S.A.; Wright, E.G. Int. J. Radiat. Biol. 2002. in press.
12. Baverstock, K. Mutat. Res. Fundam. Mol. Mech. Mutagen. 2000, 454 (12), 89.
13. Health Eects of Exposure to Low Levels of Ionizing Radiation: BEIR V Committee on the
Biological Eects of Ionizing Radiation (BEIR V). National Research Council, 1990.
14. ICRP, 1990, Recommendations of the International Commission on Radiological Protection.
Publication 60, Annals of the ICRP, Pergamon Press: Oxford, 1990; Vol. 21, 13.
15. Goodhead, D.T. Int. J. Radiat. Biol. 1994, 65, 7.
16. Ward, J.F. Int. J. Radiat. Biol. 1994, 66, 427.
17. Prise, K.M.; Pinto, M.; Newman, H.C.; Michael, B.D. Radiat. Res. 2001, 156, 572.
18. Savage, J.R.K. BI Radiol. 62, 507.
19. Sachs, R.K.; Chen, A.M.; Brenner, D.J. Int. J. Radiat. Biol. 1989, 71, 1.
20. Ottolenghi, A.; Ballarini, F.; Merzagora, M. Radiat. Environ. Biophys. 1999, 38, 1.
21. Edwards, A.A. Int. J. Radiat. Biol. 2002, 78, 551.
22. Bridges, B.A. Radiat. Res. 2001, 156, 631.
23. Sankaranarayanan, K.; Chakraborty, R. Radiat. Res. 2001, 156, 648.
24. Cox, R. Int. J. Radiat. Biol. 1994, 65, 57.
25. Sharer, O.D.; Jiricny, J. BioEssays 2001, 23, 270.
26. Ward, J.F. In DNA Damage and Repair; Nickolo, J.A. Hoekstra, M.F., Eds.; Human Press,
1998; 65 pp.
27. Harrison, L.; Hatahet, Z.; Wallace, S.S. J. Mol. Biol. 1999, 290, 667.
28. Goodhead, D.T. Int. J. Radiat. Biol. 1989, 56, 623.
29. Newton, E.; Morton, E. Proc. Natl. Acad. Sci. U. S. A. 1991, 88, 7474.
30. Nikjoo, H.; Goodhead, D.T.; Charlton, D.E.; Paretzke, H.G. Phys. Med. Biol. 1989, 34,
691.
31. Fitzsimons, C.J.; Nikjoo, H.; Bolton, C.E.; Goodhead, D.T. Math Biosci. 1998, 154, 103.
32. Charlton, D.E.; Nikjoo, H.; Humm, J.L. Int. J. Radiat. Biol. 1989, 56, 1.
33. Dickerson, R.E.; Drew, H.E.; Conner, B.N.; Wing, R.M.; Fratini, A.V.; Kopla, M.L. Science
1982, 216, 475.
34. Lavery, R.; Sklenar, H. J. Biomol. Struct. Dyn. 1988, 6, 63.
35. Ferrin, T.E.; Huang, C.C.; Jarvis, L.E.; Langridge, R. J. Mol. Graph. Model, 6, 1327, 3637,
1988. http://www.cgl.ucsf.edu/Outreach/midasplus/
36. Berman, H.M.; Olson, W.K.; Beveridge, D.L.; Westbrook, J.; Gelbin, A.; Demeny, T.; Hsieh,
S.-H.; Srinivasan, A.R.; Schneider, B. Biophys. J. 1992, 63, 751.
37. Umrania, Y.; Nikjoo, H.; Goodfellow, J.M. Int. J. Radiat. Biol. 1995, 67, 145.
38. Joshua-Tor, L.; Sussman, J. Curr. Opin. Struct. Biol. 1993, 3, 323.
39. Friedland, W.; Jacob, P.; Paretzke, H.G.; Stork, T. Radiat. Res. 1998, 150, 170.
40. Rydberg, B.; Holley, W.R.; Mian, S.M.; Chatterjee, A. J. Mol. Biol. 1988, 284, 71.
41. Nikjoo, H.; Laughton, C.A.; Terrissol, M.; Panyutin, I.G.; Goodhead, D.T. Int. J. Radiat.
Biol. 2000, 76, 1607.
42. Franklin, R.E.; Gosling, R.G. Nature 1953, 171, 740.
43. Dickerson, R.E.; Drew, H.R.; Conner, B.N.; Wing, R.M.; Fratini, A.V.; Kopka, M.L. Science
1982, 216, 475.
44. Saenger, W. Annu. Rev. Biophys. Biophys. Chem. 1987, 16, 93.
45. Saenger, W.; Hunter, W.N.; Kennard, O. Nature 1986, 324, 385.
46. Saenger, W. Principles of Nucleic Acid Structure. Springer: New York, 1983.
47. Lea, D.E. Action of Radiation on Living Cells. Cambridge University Press: London, 1955.
48. Savage, J.R.K. Mutat. Res. 1998, 404, 139.
49. Hlatky, H.; Sachs, R.K.; Vazquez, M.; Cornforth, M.N. BioEssays 2002, 24, 714.
50. Nikjoo, H.; Goodhead, D.T.; Charlton, D.E.; Paretzke, H.G. Energy deposition by mono-
energetic electrons in cylindrical targets. MRC Monograph, 1994.
51. Nikjoo, H.; Goodhead, D.T.; Charlton, D.E.; Paretzke, H.G. Energy deposition by C, Al, Ti
Copyright 2004 by Taylor & Francis Group, LLC
and Cu x-rays in cylindrical volumes within mammalian cells. MRC Radiobiology Unit Mo-
nograph, 1988.
52. Charlton, D.E.; Goodhead, D.T.; Wilson, W.E.; Paretzke, H.G. Energy deposition in cy-
lindrical volumes: (a) Protons, energy 0.3 MeV to 4 MeV, (b) Alpha particles, energy 1.2 MeV
to 20 MeV. MRC Radiobiology Unit Monograph, 1985.
53. Nikjoo, H.; Boyden-Pratt, N.; Uehara, S.; Goodhead, D.T. Energy deposition by mono-
energetic protons in cylindrical targets. MRC Monograph, 2001.
54. Nikjoo, H.; Boyden-Pratt, N.; Girard, P.; Uehara, S.; Goodhead D.T. Energy deposition by
alpha-particles in cylindrical targets. MRC Monograph, 2002.
55. Cucinotta, F.A.; Nikjoo, H.; Goodhead, D.T. Radiat. Res. 2000, 153, 459.
56. Frankenberg, D.; Brede, H.J.; Schrewe, U.J.; Steinmetz, C.; Frankenberg-Schwager, M.;
Kasten, G.; Pralle, E. Radiat. Res. 1999, 151, 540.
57. Prise, K.M.; Ahnstrom, G.; Belli, M.; Carlsson, J.; Frankenberg, D.; Keifer, J.; Lobrich, M.;
Michael, B.D.; Nygren, J.; Simone, G.; Stenerlow, B. Int. J. Radiat. Biol. 1998, 74, 173.
58. Belli, M.; Cherubini, R.; Dalla Vecchia, M.; Dini, V.; Moschini, G.; Signoretti, C.; Simone,
G.; Tabocchini, M.A.; Tiveron, P. Int. J. Radiat. Biol. 2000, 76, 1095.
59. Prise, K.M.; Folkard, M.; Davies, S.; Michael, B.D. Int. J. Radiat. Biol. 1990, 58, 261.
60. Nikjoo, H.; Uehara, S.; Wilson, W.E.; Hoshi, M.; Goodhead, D.T. Int. J. Radiat. Biol. 1998,
73, 355.
61. Frankenberg-Schwager, M.; Frankenberg, D. Int. J. Radiat. Biol. 1990, 58, 569.
62. Radford, I.R. Int. J. Radiat. Biol. 1985, 48, 45.
63. Blocher, D. Int. J. Radiat. Biol. 1988, 54, 761.
64. Foray, N.; Arlett, C.F.; Malaise, E.P. Biochemistry 1997, 79, 567.
65. Olive, P.L. Radiat. Res. 1998, 150, S42.
66. Jenner, T.J.; DeLara, C.M.; ONeill, P. Int. J. Radiat. Biol. 1993, 64, 265.
67. Ward, J.F. Radiat. Res. 1995, 142, 362.
68. Goodhead, D.T.; Nikjoo, H. Int. J. Radiat. Biol. 1989, 55, 513.
69. Leenhouts, H.P.; Chadwick, K.H. Radiat. Prot. Dosim. 1985, 13, 267.
70. Booz, J.; Finnendegen, L.E. Int. J. Radiat. Biol. 1988, 53, 13.
71. Cucinotta, F.A.; Nikjoo, H.; Goodhead, D.T. Radiat. Environ. Biophys. 1999, 38, 81.
72. Ballarini, F.; Biaggi, M.; Merzagora, M.; Ottolenghi, A.; Dingfelder, M.; Friedland, W.;
Jacob, P.; Paretzke, H.G. Radiat. Environ. Biophys. 2000, 39, 179.
73. Cobut, V.; Frongillo, Y.; Patau, J.P.; Goulet, T.; Fraser, M.-J.; Jay-Gerin, J.-P. Radiat. Phys.
Chem. 1998, 51, 229.
74. Wilson, W.E.; Nikjoo, H. Radiat. Environ. Biophys. 1999, 38, 97.
75. Rydberg, B.; Heilbronn, L.; Holley, W.R.; Lobrich, M.; Zeitlin, C.; Chatterjee, A.; Cooper,
P.K. Radiat. Res. 2002, 158, 32.
76. Watanabe, R.; Nikjoo, H. Int. J. Radiat. Biol. 2002, 78, 953.
77. Moiseenko, V.V.; Hamm, R.N.; Waker, A.J.; Prestwich, W.V. Int. J. Radiat. Biol. 1998, 74,
533.
78. Begusova, M.; Sy, D.; Charlier, M.; Spotheim-Maurizot, M. Int. J. Radiat. Biol. 2000, 76,
1063.
79. Nikjoo, H.; ONeill, P.; Goodhead, D.T.; Terrissol, M. Int. J. Radiat. Biol. 1997, 71, 467.
80. Nikjoo, H.; Bolton, C.E.; Watanabe, R.; Terrissol, M.; ONeill, P.; Goodhead, D.T. Radiat.
Prot. Dosim. 2002, 99, 77.
81. Nikjoo, H.; ONeill, P.; Wilson, W.E.; Goodhead, D.T. Radiat. Res. 2001, 156, 577.
82. Cucinotta, F.A.; Nikjoo, H.; ONeill, P.; Goodhead, D.T. Int. J. Radiat. Biol. 2000, 76, 1463.
83. Ottolenghi, A.; Ballarini, F.; Merzagora, M. Radiat. Environ. Biophys. 1999, 38, 1.
84. Friedland, W.; Li, W.B.; Jacob, P.; Paretzke, H.G. Radiat. Res. 2001, 155, 703.
85. van Rijn, K.; Mayer, T.; Blok, J.; Bverberne, J.; Loman, H. Int. J. Radiat. Biol. 1985, 47, 309.
86. Ito, T.In Radiation Damage; Fuciarelli, A.F., Zimbrick, J.D., Eds.; Battelle Press: Columbus,
1995; 259 pp.
87. Nikjoo, H.; ONeill, P.; Terrissol, M.; Goodhead, D.T. Radiat. Environ. Biophys. 1999, 38, 31.
Copyright 2004 by Taylor & Francis Group, LLC
88. Kandaiya, S.; Lobachevsky, P.; Martin, F. Acta Oncol. 1996, 35, 803.
89. Martin, R.F.; Haseltine, W.A. Science 1981, 213, 896.
90. Ninbar, A.; Goodfellow, J.M. Radiat. Environ. Biophys. 1999, 38, 23.
91. Holley, W.R.; Chatterjee, A.; Magee, J.L. Radiat. Res. 1990, 121, 161.
92. Charlton, D.E.; Humm, J.L. Int. J. Radiat. Biol. 1988, 53, 353.
93. Nikjoo, H.; Martin, R.F.; Charlton, D.E.; Terrissol, M.; Kandaiya, S.; Lobachevsky, P. Acta
Oncol. 1996, 35, 849.
94. Terrissol, M.; Demonchy, M.; Pomplun, E. In MicrodosimetryAn Interdisciplinary
Approach; Goodhead, D.T., ONeill, P., Menzel, H.G., Eds.; Royal Society of Chemistry:
Cambridge, 1997; 15 pp.
95. Michalik, V. Int. J. Radiat. Biol. 1992, 62, 9.
96. Scholes, G.; Willson, R.L.; Ebert, M. Pulse radiolysis of aqueous solutions of deoxyribonu-
cleotides and of DNA: reaction with hydroxyl radicals. Chem. Commun. 1969, 17.
97. Buxton, G.V.; Greenstock, C.L.; Helman, W.P.; Ross, A.B. J. Phys. Chem. 1988, 17, 513.
98. Milligan, J.R.; Aguilera, J.A.; Ward, J.F. Radiat. Res. 1993, 133, 151.
99. Chatterjee, A.; Magee, J.L. Radiat. Prot. Dosim. 1985, 13, 137.
100. Terrissol, M. Int. J. Radiat. Biol. 1994, 66, 447.
101. Tomita, H.; Kai, M.; Kusama, T.; Aoki, Y.; Ito, A. Int. J. Radiat. Biol. 1994, 66, 669.
102. Alden, C.J.; Kim, S.H. J. Mol. Biol. 1979, 132, 411.
103. Makrigiorgos, G.M.; Bump, E.; Huang, C.; Baranowska-Kortylewicz, J.; Kassis, A.I. Int. J.
Radiat. Biol. 1994, 66, 247.
104. Michalik, V.; Begusova, M. Int. J. Radiat. Biol. 1994, 66, 267.
105. Balasubramanian, B.; Pogozelski, W.K.; Tullius, T.D. Proc. Nat. Acad. Sci. U. S. A. 1998, 95,
9738.
106. Aydogan, B.; Marshall, D.T.; Swarts, S.G.; Turner, J.E.; Boone, A.J.; Richards, N.G.; Bolch,
W.E. Radiat. Res. 2002, 157, 38.
107. Folkard, M.; Prise, K.M.; Vojnovic, B.; Davies, S.; Roper, M.J.; Michael, B.D. Int. J. Radiat.
Biol. 1993, 64, 651.
108. Boudaia, B.; Cloutier, P.; Hunting, D.; Huels, M.A.; Sanche, L. Science 2000, 287, 1658.
109. Sanche, L. Radiat. Prot. Dosim. 2002, 99, 57.
110. Hieda, K. Int. J. Radiat. Biol. 1994, 66, 561.
111. Takakura, K.; Ishikawa, M.; Ito, T. Int. J. Radiat. Biol. 1987, 52, 667.
112. Nikjoo, H.; Goorley, T.; Fulford, J.; Takakura, K.; Ito, T. Radiat. Prot. Dosim. 2002, 99,
91.
113. Abdoul-Carime, H.; Dugal, P.C.; Sanche, L. Radiat. Res. 2000, 153, 23.
114. Goodhead, D.T. Can. J. Phys. 1990, 68, 872.
115. Nikjoo, H.; Goodhead, D.T. Phys. Med. Biol. 1991, 36, 229.
116. Brenner, D.J.; Ward, J.F. Int. J. Radiat. Biol. 1992, 61, 737.
117. Michalik, V.; Frankenberg, D. Radiat. Environ. Biophys. 1996, 35, 163.
118. Holley, W.R.; Chatterjee, A. Radiat. Res. 1996, 145, 188.
119. Ottolenghi, A.; Merzagora, M.; Tallone, L.; Durante, M.; Paretzke, H.G.; Wilson, W.E.
Radiat. Environ. Biophys. 1995, 34, 239.
120. Ottolenghi, A.; Merzagora, M.; Paretzke, H.G. Radiat. Environ. Biophys 1997, 36, 97.
121. Moiseenko, V.V.; Hamm, R.N.; Waker, A.J.; Prestwich, W.V. Radiat. Environ. Biophys.
2001, 40, 23.
122. Goodhead, D.T.; Nikjoo, H. Radiat. Res. 1997, 148, 485.
123. Weinfeld, M.; Rasouli-Nia, A.; Chaudhry, M.A.; Britten, R.A. Radiat. Res. 2001, 156, 584.
124. Sutherland, B.M.; Bennett, P.V.; Sutherland, J.C.; Laval, J. Radiat. Res. 2002, 157, 611.
125. Sutherland, B.M.; Bennett, P.V.; Sidorkina, O.; Laval, J. Biochemistry 2000, 39, 8026.
126. David-Cordonnier, M.H.; Laval, J.; ONeill, P. J. Biol. Chem. 2000, 275, 11865.
127. David-Cordonnier, M.H.; Boiteux, S.; ONeill, P. Nucleic Acids Res. 2001, 29, 1107.
128. Jenner, T.J.; Fulford, J.; ONeill, P. Radiat. Res. 2001, 156, 590.
129. Cunnie, S.; ONeill, P. Radiat. Res. 1999, 152, 421.
Copyright 2004 by Taylor & Francis Group, LLC
130. Watson, J.D.; Crick, F.H.C. Nature 1953, 171, 737.
131. Franklin, R.E.; Gosling, R.G. Nature 1953, 171, 740.
132. Saenger, W. Principles of Nucleic Acid Structure. Springer Verlag, 1986.
133. Neidle, S. DNA Structure and Recognition. Focus Series. Oxford: IRL Press, 1994.
134. Lilley, D.M.J. Methods in Enymology; Lilley, D.M., Dahlberg, J.E., Eds.; 1992; Vol. 212B,
133.
135. Regaud, C.; Lacassagne, A. Radiophysiol. Radither. 1927, 1, 95.
136. Frank-Kamenetskii, M.D.; Mirkin, S.M. Annu. Rev. Biochem. 1995, 64, 65.
137. Praseuth, D.; Guieysse, A.L.; He le` ne, C. Biochim. Biophys. Acta 1999, 1489, 181.
138. Felsenfeld, G.; Davis, D.R.; Rich, A. J. Am. Chem. Soc. 1957, 79, 2023.
139. Arnott, S.; Bond, P.J.; Selsing, E.; Smith, P.J. Nucleic Acids Res. 1976, 11, 4141.
140. Raghunathan, G.; Miles, H.T.; Sasisekharan, V. Biochemistry 1993, 32, 455.
141. Panyutin, I.G.; Neumann, R.D. Nucleic Acids Res. 1994, 22, 4979.
142. Charlton, D.E. Radiat. Res. 1986, 107, 163.
143. Hofer, K.G. Acta Oncol. 1996, 35, 789.
144. Adelstein, S.J.; Kassis, A. Acta Oncol. 1996, 35, 797.
145. Walicka, M.A.; Ding, Y.; Roy, A.M.; Harpanhalli, R.S.; Adelstein, S.J.; Kassis, A.I. Int. J.
Radiat. Biol. 1999, 75, 1579.
146. Lobachevsky, P.N.; Martin, R.F. Radiat. Res. 2000, 153, 263.
147. Pomplun, E.; Terrissol, M. Radiat. Environ. Biophys. 1994, 33, 279.
148. Karamychev, V.N.; Zhurkin, V.B.; Garges, S.; Neumann, R.D.; Panyutin, I.G. Nat. Struct.
Biol. 1999, 6, 747.
149. Panyutin, I.G.; Neumann, R.D. Nucleic Acids Res. 1997, 25, 883.
150. Panyutin, I.G.; Neumann, R.D. Acta Oncol. 1996, 35, 817.
151. Panyutin, I.V.; Luu, A.N.; Panyutin, I.G.; Neumann, R.D. Radiat. Res. 2001, 156, 158.
152. de Lara, C.M.; Hill, M.A.; Jenner, T.J.; Papworth, D.; ONeill, P. Radiat. Res. 2001, 155, 440.
153. NCRP, Report No. 136. Evaluation of the linear-nonthreshold doseresponse model for
ionizing radiation. National Council on radiation Protection and Measurements. Bethesda,
Maryland, 308145916.
154. Joiner, M.C. Int. J. Radiat. Biol. 1994, 65, 79.
155. Azzam, E.I.; Toledo, S.M.; Gooding, T.; Little, J.B. Radiat. Res. 1998, 150, 497.
156. Hei, T.K.; Wu, L.J.; Liu, S.X.; Vannia, D.; Waldren, A.; Randers-Pheterson, G. Proc. Natl.
Acad. Sci. U. S. A. 1997, 94, 3765.
157. Kadhim, M.A.; Macdonald, D.A.; Goodhead, D.T.; Lorimore, S.A.; Marsden, S.J.; Wright,
E.G. Nature 1992, 355, 738.
158. Mothersill, C.; Seymour, C. Int. J. Radiat. Biol. 1997, 71, 421.
159. Mothersill, C.; Seymour, C. Radiat. Res. 1998, 149, 256.
160. Nagasawa, H.; Little, J. Radiat. Res. 1999, 152, 552.
161. Wu, L.-J.; Randers-Pherson, G.; Xu, A.; Waldern, CA.; Geard, C.R.; Yu, Z.-L.; Heui, T.K.
Proc. Natl. Acad. Sci. U. S. A. 1999, 96, 4959.
162. Zhou, H.; Randers-Pherson, G.; Waldern, C.A.; Vannias, D.; Hall, E.J.; Hei, T.K. 2000, 97,
2099.
163. Deshpande, A.; Goodwin, E.H.; Bailey, S.M.; Marrone, B.L.; Lehnert, B.E. Radiat. Res. 1996,
145, 260.
164. Sawant, S.G.; Randers-Pehrso, G.; Geard, C.R.; Brenner, D.J.; Hall, E.J. Radiat. Res. 2001,
155, 397.
165. Miller, R.C.; Randers-Pehrson, G.; Geard, C.R.; Hall, E.J.; Brenner, D.J. Proc. Natl. Acad.
Sci. U. S. A. 1999, 96, 19.
166. Prise, K.M.; Belyakov, O.V.; Folkard, M.; Michael, B.D. Int. J. Radiat. Biol. 1998, 74, 793.
167. Azzam, E.I.; Raaphorst, G.P.; Mitchel, R.E.J. Radiat. Res. 1994, 138, S28.
168. Lorimore, S.A.; Kadhim, M.A.; Pocock, D.A.; Papworth, D.; Stevens, D.L.; Goodhead,
D.T.; Wright, E.G. Proc. Natl. Acad. Sci. U. S. A. 1998, 95, 5730.
169. Watson, G.E.; Lorimore, S.A.; Macdonald, D.A.; Wright, E.G. Cancer Res. 2000, 60, 5608.
Copyright 2004 by Taylor & Francis Group, LLC
170. Seymour, C.B.; Mothersill, C. Radiat. Res. 2000, 153, 508.
171. Brenner, D.J.; Little, J.B.; Sachs, R.K. Radiat. Res. 2001, 155, 402.
172. Nikjoo, H.; Khvostunov, I.K. Int. J. Radiat. Biol. 2002. in press.
173. Khvostnov, I.K.; Nikjoo, H. J. Radiol. Prot. 2002, 22, A33.
174. Uehara, S.; Nikjoo, H.; Goodhead, D.T. Phys. Med. Biol. 1993, 38, 1841.
175. Uehara, S.; Toburen, L.H.; Nikjoo, H. Int. J. Radiat. Biol. 2001, 77, 139.
176. Uehara, S.; Nikjoo, H. J. Phys. Chem. 2002, 106, 11051.
177. Hayashi, M. TECDOC-506. Vienna: IAEA, 1989; 193 pp.
178. Djuric, N.Lj.; Cadez, I.M.; Kurepa, M.V. Int. J. Mass Spectrom. Ion Process. 1988, 83, R7.
179. Olivero, J.J.; Stagat, R.W.; Green, A.E.S. J. Geophys. Res. 1972, 77, 4797.
180. Bolorizadeh, M.A.; Rudd, M.E. Phys. Rev. A 1986, 33, 882.
181. Schutten, J.; de Heer, F.J.; Moustafa, H.R.; Boerboom, A.J.H.; Kistemaker, J. J. Chem. Phys.
1966, 44, 3924.
182. Dingfelder, M.; Hantke, D.; Inokuti, M.; Paretzke, H.G. Radiat. Phys. Chem. 1998, 53, 1.
183. Seltzer, S.M. In Monte Carlo Transport of Electrons and Photons; Jenkins, T.M., Nelson,
W.R., Rindi, A., Eds.; Plenum Press: New York, 1988; 81 pp.
184. Paretzke, H.G. GSF-Bericht 24. Institut fur Strahlenschutz der Gesellschaft fur Strahlenund
Umweltforschung: Munich, 1988; 1 pp.
185. Green, A.E.S.; Stolarski, R.S. J. Atmos. Terr. Phys. 1972, 34, 1703.
186. Berger, M.; Wang, R. In Monte Carlo Transport of Electrons and Photons; Jenkins, T.M.,
Nelson, W.R., Rindi, A., Eds.; New York: Plenum Press, 1988; 21 pp.
187. Moliere, G. Naturforsch Z. 1948, 3a, 78.
188. Shyn, T.W.; Cho, S.Y. Phys. Rev. A 1987, 36, 5138.
189. Katase, A.; Ishibashi, K.; Matsumoto, Y.; Sakae, T.; Maezono, S.; Murakami, E.; Watanabe,
K.; Maki, H. J. Phys., B 1986, 19, 2715.
190. Nishimura, H. In Electronic and Atomic Collisions; Oda, N., Takayanagi, K., Eds.; Amsterdam:
North-Holland, 1979; 314 pp.
191. Danjo, A.; Nishimura, H. J. Phys. Soc. Jpn. 1985, 54, 1224.
192. Trajmar, S.; Williams, W.; Kuppermann, A. J. Chem. Phys. 1973, 58, 2521.
193. Zaider, M.; Brenner, D.J.; Wilson, W.E. Radiat. Res. 1983, 95, 231.
194. ICRU, Secondary Electron Spectra from Charged Particle Interations, Report 55. ICRU,
Bethesda, 1996.
195. Vroom, D.A.; Palmer, R.L. J. Chem. Phys. 1977, 66, 3720.
196. Opal, C.B.; Beaty, E.C.; Peterson, W.K. At. Data 1972, 4, 209.
197. Combecher, D. Radiat. Res. 1980, 84, 189.
198. Rudd, M.E.; Goe, T.V.; DuBois, R.D.; Toburen, L.H. Phys. Rev. A 1985, 31, 492.
199. Rudd, M.E.; Goe, T.V.; Itoh, A. Phys. Rev. A 1985, 32, 2128.
200. Toburen, L.H.; Wilson, W.E.; Popowich, R.J. Radiat. Res. 1980, 82, 27.
201. Dingfelder, M.; Inokuti, M.; Paretzke, H.G. Radiat. Phys. Chem 2000, 59, 255.
202. Miller, J.H.; Green, A.E.S. Radiat. Res. 1973, 54, 343.
203. ICRU, Stopping Powers and Ranges for Protons and Alpha Particles, Report 49. ICRU,
Bethesda, 1993.
204. Mott, N.F.; Massey, H.S.W. The Theory of Atomic Collisions, 3rd ed; Oxford University
Press, London, 1965.
205. Everhart, E.; Stone, G.; Carbone, R.J. Phys. Rev 1955, 99, 1287.
206. Goldstein, H. Classical Mechanics. Addison-Wesley: Reading, MA, 1950.
207. Rudd, M.E. Phys. Rev. A 1988, 38, 6129.
208. Rudd, M.E.; Kim, Y.K.; Madison, D.H.; Gay, T.J. Rev. Mod. Phys. 1992, 64, 441.
209. Bolorizadeh, M.A.; Rudd, M.E. Phys. Rev. A 1986, 33, 888.
210. Toburen, L.H.; Wilson, W.E. J. Chem. Phys 1977, 66, 5202.
211. Toburen, L.H.; Nakai, M.Y.; Langley, R.A. Phys. Rev. 1968, 171, 114.
212. Dagnac, R.; Blanc, D.; Molina, D. J. Phys. B 1970, 3, 1239.
213. Rudd, M.E.; Itoh, A.; Goe, T.V. Phys. Rev. A 1985, 32, 2499.
Copyright 2004 by Taylor & Francis Group, LLC
214. Sataka, M.; Yagishita, A.; Nakai, Y. J. Phys. B 1990, 23, 1225.
215. Allison, S.K. Rev. Mod. Phys. 1958, 30, 1137.
216. Uehara, S.; Toburen, L.H.; Wilson, W.E.; Goodhead, D.T.; Nikjoo, H. Radiat. Phys. Chem
2000, 59, 1.
217. Blorizadeh, M.A.; Rudd, M.E. Phys. Rev. A 1986, 33, 893.
218. Uehara, S.; Nikjoo, H. Radiat. Prot. Dosim. 2002, 99, 53.
219. Ward, J.F. The radiation-induced lesions which trigger the bystander eect. Mutat. Res.
Fundam. Mol. Mech. Mutagen 2002, 499, 151.
220. Moiseenko, V.V.; Edwards, A.A.; Nikjoo, H.; Prestwich, W.V. Radiat. Res 1997, 147, 208.
221. Prise, K.M. Int. J. Radiat. Biol 1994, 65, 43.
222. Ward, J.F. Prog. Nucleic Acids Res. Mol. Biol. 1988, 35, 95.
223. Botchway, S.W.; Stevens, D.L.; Hill, M.A.; Jenner, T.J.; ONeill, P. Radiat. Res. 1997, 148,
317.
224. Cox, R.; Thacker, J.; Goodhead, D.T. Int. J. Radiat. Biol. 1977, 31, 561.
Copyright 2004 by Taylor & Francis Group, LLC
18
Microdosimetry and Its Medical Applications
Marco Zaider
Memorial Sloan-Kettering Cancer Center, New York, New York, U.S.A.
John F. Dicello
Johns Hopkins University School of Medicine, Baltimore, Maryland, U.S.A.
1. INTRODUCTION
The quantity absorbed dose is used in medical physics on the assumption that when its
value is the same, irradiation of equal objects results in equal biological eects. In reality,
most of the biological eects of radiation depend on the microscopical pattern of energy
deposition and the absorbed dose is only the expected value (average) of the energy
deposited per unit mass. That the same dose of radiation may result in dierent eects,
when delivered by dierent types of radiation, is well documented. For instance, at low
doses, the relative biological eectiveness (RBE) of neutrons relative to g-rays to induce a
cell to become malignant is on the order of 100, which means that the ratio of absorbed
doses of g-rays and neutrons causing equal eects is about 100 [113].
The invention of microdosimetry some 50 years ago is a direct result of the
empirical observation that radiations have RBE p 1. This was unexpected because, on rst
principles, one would anticipate that biological eects would be proportional to the
(average) number of biomolecular disruptions produced by radiation in the sensitive sites
of the cell. On the other hand, all radiations act on biomatter primarily through the agency
of ionizations and excitations (presumably, the events that initiate biological damage).
Over a fairly large range of energies, the average energy expended to produce an ion pair
(
f
34 eV) is the same for all radiations. Thus, one expects proportionality between average
eect (E) and the average number of, say, ionizations in the cell:
Effect / z 1
In this expression, z, termed specic energy, is the energy imparted per unit mass to a
volume V; the expected value of z (see below) is the absorbed dose D. It follows that, on
average:
Effect / z D 2
Copyright 2004 by Taylor & Francis Group, LLC
which means that RBE=1. Conversely, RBE p 1 entails a nonlinear relationship between
eect and specic energy. For instance, if:
Effect / z
2
3
then:
Effect / z
2
p z
2
D
2
4
This means that in order to obtain a physical predictor of biological eect, one needs infor-
mation on the distribution f(z) of specic energy z. It also meansand this is the essential
pointthat the spatial pattern of energy deposition, which will change from one radiation
to another and thus account for dierences in f(z) among radiations, is an important pre-
dictor of biological eect.
The application of microdosimetry to medical physics derives from biological models
of radiation action (primarily, the Theory of Dual Radiation Action [14]) that explicitly
utilize for their predictions a microdosimetric description of the radiation eld.
Specically, they concern the following two problems:
(a) The RBE of a radiation eld with respect to conventional treatment modalities
(photon or high-energy electrons) for which extensive empirical data already exist.
This includes, as particular cases, RBE variations within the treatment eld as well
as brachytherapy, where low-energy photons (with RBE p 1) are frequently used.
(b) Predicting eect modications due to the temporal pattern of dose delivery.
We shall start with denitions of the principal microdosimetric quantities.
2. MICRODOSIMETRIC QUANTITIES
The basic concept of microdosimetry is the microdosimetric event (or, simply, event), which
is energy deposition by a charged particle and its statistically correlated particles (e.g.,
secondary electrons from the same particle track) [1517]. Events are important because
they are statistically independent entities. The principal microdosimetric quantity, specic
energy z, is dened as follows:
z
e
m
5
where e is the energy imparted by ionizing radiation to matter of mass m. The absorbed
dose D is the average energy imparted, per unit mass, by ionizing radiation. Formally:
D~r lim
V!0; ~roV
z 6
where z is the average specic energy in a domain of volume V centered at ~r, and the limit is
taken at constant density q.
A second quantity (physically equivalent to specic energy) is the lineal energy,
which is the quotient of e by l , where l is the mean chord length in that volume:
y
e
l
7
In this denition, e refers to energy deposition imparted in a single event, a restriction that
does not apply to z.
In the matter of mass density q, the relationship between z (in a single event) and y is:
z
4y
qS
8
Copyright 2004 by Taylor & Francis Group, LLC
This follows from a theorem by Cauchy [18] according to which the mean chord length l in
a convex site is given by 4V/S, where V and S are, respectively, the volume and surface
area of the site. For a spherical site of diameter d:
l
2
3
d 9
and thus:
z
4y
p d
2
q
10
which, for q=1 g/cm
3
and with units as indicated, becomes:
z
Gy
0:204
d
Am
_ _
2
y
keV=Am
11
The probability that the lineal energy delivered in single events is in the interval [ y,y+dy]
is denoted as f( y)dy. The corresponding distribution in z is f
1
(z)dz, where the subscript
makes explicit the fact that this distribution refers to single events only. The distributions
f(z), f
1
(z), and f( y) depend on the geometry of the site where specic energy is determined.
Two moments of these distributions that are often invoked are the frequency average y
F
(or z
F
) and the dose average y
D
(or z
D
). They are dened as follows:
y
F
_
l
0
yfydy; z
F
_
l
0
zf
1
dz
z
D
1
y
F
_
l
0
y
2
fydy; z
D
1
z
F
_
l
0
z
2
f
1
zdz
12
The rst and second moments of the multievent distribution f(z;n) are denoted z and z
2
,
respectively. Because, by denition, D z , the average number of events n at dose D is:
n
D
z
F
13
At a given dose D, the distribution f(z;n) [or, equivalently, f(z;D)] is determined uniquely
by f
1
(z). This can be understood from the following argument: By denition, the
probability of exactly m events is given by the Poisson distribution:
pm; n e
n
n
m
m!
14
If we denote by f
m
(z) the distribution of z in exactly m events, then f
m
(z) can be evaluated by
successive convolutions because, by denition, events are statistically independent:
f
0
z dz
f
1
z u f
1
z
f
2
z
_
l
0
f
1
zVf
1
z zVdzV
.
.
.
f
m
z
_
l
0
f
1
zVf
m1
z zVdzV
15
Copyright 2004 by Taylor & Francis Group, LLC
The distribution f(z;n) becomes:
fz; n
l
m0
pm; n f
m
z 16
which proves that if f
1
(z) is known, one can calculate f(z;n) for any value of n (i.e., D). One
can also demonstrate a general relationship between the moments of f(z;n) and f
1
(z) [17],
namely:
m
k1
M
k1
n
k1
j0
C
j
k
M
kj
m
j1
; k 0; 1; 2; . . . 17
Here, m
k
is the kth moment of f
1
(z) and M
k
is the kth moment of f(z;n) and:
C
j
k
kk 1 . . . k j 1
j!
18
With the notations used above (Eq. (12)):
z
F
m
1
; z
D
m
2
=m
1
19
In particular:
M
1
z nz
F
20
M
2
zz z
D
21
or, equivalently:
M
2
z
D
DD
2
22
an expression that provides the mathematical justication for the much used (and abused)
linearquadratic expression (see below).
The dose-averaged specic energy z
D
plays a fundamental role in microdosimetric-
based models of radiation action. For instance, it will be shown below that its magnitude
relative to D determines the extent to which doserate eects are important. If the micro-
dosimetric spectrum f
1
(z) is known, z
D
can be obtained by straightforward integration.
There is, however, a second more powerful procedure for obtaining this quantity and this
will be now explained.
z
D
depends on two elements: the structure of the charged particles tracks in the
radiation eld, and the geometry of the site where z is determined. One may adopt the
point of view that f
1
(z) obtains from the random overlap of two objects: the particle
track (represented as a collection of local energy depositions, ionizations, and excitations)
and the site of interest. Consider then two objects (three-dimensional domains) of volumes
V
1
and V
2
and let v represent the volume of their overlap when they are placed randomly
in space. One can show [19] that the rst two moments of v, E(v) and E(v
2
) satisfy:
Ev
2
Ev
V
1
V
2
_
p
1
rp
2
r
4pr
2
dr 23
where p
1
is the point-pair distribution of distances (ppdd) between pairs of points randomly
chosen in V
1
(and similarly for p
2
). This result is remarkable because it shows that the ratio
of the second to the rst moments of v can be factorized in functions (ppdd) that characterize
each object individually. This result can be immediately translated to microdosimetry [20].
Assume that in a charged particle track, the volume occupied by each transfer point
(ionization or excitation) equals the energy locally deposited. The overlap volume becomes
Copyright 2004 by Taylor & Francis Group, LLC
the energy e deposited in the site (V
2
), V
1
becomes the total energy T in the track, and V
2
is
the site volume. With this:
Ee
2
Ee
TV
2
_
p
track
rp
site
r
4pr
2
dr 24
The product between the ppdd and the volume of the site is termed the proximity function of
that domain. Thus, we have a proximity function of energy deposition:
tr track energy p
track
r 25
and the proximity function of the site:
sr site volume p
site
r 26
It is immediately recognized that:
z
D
1
m
_
trsr
4pr
2
dr 27
where m is the mass contained in the site. A few examples will illustrate the concept of
proximity function. For an object that occupies the entire space:
tr 4pr
2
28
For a track represented by a straight line of innite length:
tr 2 29
For a track of nite length L:
tr 2 1
r
L
_ _
r 2 0; L 30
For a spherical site of diameter d:
tr 4pr
2
1
3r
2d
r
3
2d
3
_ _
r 2 0; d 31
For other regular volumes, see Ref. 21.
In summary, microdosimetry is the study and quantication of the spatial and
temporal distributions of absorbed energy in irradiated matter [15,17,22,23]. One makes a
distinction between regional microdosimetry [the object of which is the study of micro-
dosimetric distributions f(z)] and structural microdosimetry (a mathematically more
advanced approach, which is concerned with characterizing the spatial distribution of
individual energy deposition events, i.e., ionizations and/or excitations). Regional micro-
dosimetry asserts that the eect is entirely determined by the amount of specic energy
deposited in the relevant site (typically, a cell nucleus). The two kinds of microdosim-
etry, regional and structural, were shown to be in fact mathematically equivalentonce
the sensitive site is judiciously determined [16].
3. RADIOBIOLOGICAL CONSIDERATIONS
3.1. The Theory of Dual Radiation Action
In order to obtain a quantity that relates to biological eect, one must postulate a functional
dependence of eect probability on specic energy. One such approach, the Theory of Dual
Copyright 2004 by Taylor & Francis Group, LLC
Radiation Action, takes as starting point the empirical observation that for numerous
eects, the yield of lesions (alterations responsible for the endpoint observed) depends
quadratically on specic energy [14]:
Ez bz
2
32
where b is a positive constant found to be only weakly dependent on radiation quality. The
form of this equation obtains by postulating that a lesion is the result of two molecular
sublesions, produced each at a rate proportional to z. On average:
ED bz
2
bz
D
DD
2
33
The last step follows fromEq. (22). The expression, Eq. (33), is the familiar linearquadratic
equation. In a more familiar notation:
ED aDbD
2
34
Thus:
z
D
a
b
35
Although not strictly valid, it is possible to extend this formulation to the probability of
cellular survival [reproductive inactivation S(D)] by making the assumptions that (1) one
lesion inactivates the cell, and (2) the number of lesions produced is Poisson-distributed.
When valid (mostly for low-LET radiation), it leads to:
SD e
aDbD
2
36
Examples of sublesions are single chromosome breaks that combine to produce a dicentric
aberration or single-strand DNA breaks, which, when in close proximity, result in double-
stranded DNA breaks. The expression, Eq. (35), provides the link between the micro-
dosimetric quantity z
D
, and the doseeect coecients a and b.
3.2. RBE
Within the linearquadratic formulation, the RBE of radiation H (a
H,
b
H
) relative to
radiation L (a
L,
b
L
) is given by [9,24,25]:
RBED
H
a
L
2b
L
D
H
1
4b
L
a
H
D
H
b
H
D
2
H
a
2
L
1
_ _
37
Note that the RBE is a decreasing function of D
H
. If b
H
gb
L
, and taking Eq. (35) into
account, one obtains the simplied expression:
RBE
1
2D
H
z
2
D;L
4D
H
z
2
D;H
D
H
_
z
D;L
_ _
38
If, furthermore, b
H
=b
L
c0, the RBE is independent of dose:
RBE
a
H
a
L
c
z
D;H
z
D;L
y
D;H
y
D;L
39
The ecacy of radiotherapy is generally considered to depend on the probability of cell
killing, 1S(D). Because RBE depends nonlinearly on absorbed dose, both the absorbed
Copyright 2004 by Taylor & Francis Group, LLC
dose and (through z
D
) the microdosimetric properties of the radiation eld must be known
at locations in the tumor and in the surrounding healthy tissues.
3.3. Temporal Effects
The temporal pattern of dose delivery, an important parameter in radiotherapy, further
changes the expressions derived thus far. To reinforce these ideas, consider the archetypical
lesion, a chromosomal dicentric aberration that obtains when two single chromosome
breaks (sublesions) misjoin. A pair of sublesions can be produced by a single track (single
event) or by two independent tracks. The yield of lesions that result via the rst mechanism
(termed intratrack action) is proportional to dose because the number of events is propor-
tional to dose. In contrast, and for analogous reasons, the second mechanism (intertrack
action) generates lesions at a rate proportional to D
2
. The two mechanisms are represented
in Eq. (34) by the linear (aD) and quadratic (bD
2
) terms, respectively.
An important consequence of this interpretation is the fact that the magnitude of the
quadratic term (but not the linear one) depends on dose rate. Indeed, in most cases,
sublesions undergo repair and thus the lower the dose rate [i.e., the larger the (average)
time interval that separates the formation of the two sublesions, the lower the yield of
intertrack lesions]. With this, the linearquadratic equation becomes:
ED aDbqtD
2
40
where q(t) is a functiontermed doserate factorthat modies the quadratic term to
account for the temporal distribution of dose.
Mathematical expressions for the function q(t) can be obtained for practically any
temporal pattern of dose delivery. It is commonly assumed that during the time interval, t
repair eliminates sublesion damage according to exp(t/t
0
), where t
0
is a characteristic
time required for repair (typically on the order of 1 hr). For instance [17]:
(a) For irradiations at a constant dose rate:
qr
2
r
2
r
2
1e
r
41
where r=t/t
0
. In particular, when t>>t
0
:
qi
2t
0
t
42
(b) For f well-separated fractions (complete sublesion repair in between fractions):
q
1
f
43
In a typical low-dose rate (LDR) treatment in brachytherapy, the total irradiation time is
on the order of several days and therefore qc0. It follows that the probability of cell
survival S(D), is quasi-exponential and the RBE is determined by the linear (a) coecient,
or, in microdosimetric terms, by z
D
.
(c) If the dose rate is decreasing exponentially (e.g., a temporary or permanent implant
with radioactive seeds), then [26,27]:
qt
2 kt
2
lt
2
1 k
2
=l
2
_ _
1 e
kt
2
e
klt
lt
1 e
2kt
2kt
_ _
1 e
2kt
2
_ _
44
Copyright 2004 by Taylor & Francis Group, LLC
Here l = log(2)/t
0
. At any given time t after the implantation, the absorbed dose D(t) is:
Dt
R
0
k
1 e
kt
_
45
where R
0
is the initial dose rate and k is the radioactive decay constant of the radioisotope
used.
(d) In the most general case [28]:
q
n
i1
a
2
i
t
0
c
i
t
0
e
c
i
=t
0
1
_ _ _ _
n
i1
n
ji1
a
i
a
j
t
2
0
e
b
j
b
i
=t
0
e
c
j
=2t
0
e
c
j
=2t
0
_ _
e
c
i
=2t
0
e
c
i
=2t
0
_ _
1
2
n
i1
a
2
i
c
2
i
n
i1
n
ji1
a
i
a
j
c
i
c
j
46
This is the dose rate factor for any arbitrary doserate distribution that consists of n con-
tiguous segments, and where a
i
, b
i
, and c
i
are, respectively, the dose rate, the center, and the
time required to deliver the doserate segment i.
3.4. Microdosimetric Constraints on Models of Radiation Action
An Upper Boundary for Eect as a Function of Dose
If each cell responds to radiation autonomously (i.e., independently of any other cell), an
upper limit of eect probability E(D) is the probability that at least one energy deposition
event actually occurred in it. This latter is given [by Poisson statistics, see Eqs. (13) and
(14)] by 1exp(D/z
F
) and thus, for any dose, one must have:
ED V1 e
D=z
F
SD u 1 ED z e
D=z
F
47
where S(D) is the survival probabilitymeaning here absence of eect. It is obvious
that at suciently large doses the linearquadratic expression, S(D) = exp(aDbD
2
)
does not satisfy this condition and thusas a matter of principle and independent of any
biological assumptioncan not be valid.
Denition of Low Dose and Low-Dose Rate
Microdosimetry provides an unambiguous denition of physical low dose, namely a dose
such that n = D/z
F
1. Indeed, under these conditions, each cell is essentially traversed by at
most one particle. Furthermore:
fz; Dce
n
dz e
n
nf
1
z 48
For any arbitrary expression e(z) describing the probability of eect as a function of z, one
has:
ED
_
ezfz; DdzcD
_
ez
z
F
f
1
zdz 49
Copyright 2004 by Taylor & Francis Group, LLC
which shows that at physical lowdoses, the eect probability E(D) must be a linear function
of dose irrespective of any specic biological mechanism. The expression, Eq. (49), is often
used in the form [29]:
ED D
_
Qz f
1
zdz 50
where Q(z) is taken as an empirical response function, the shape of which is obtained by
solving the system of integral equations:
E
i
D D
_
Qz f
1;i
zdz i 1; 2; . . . 51
where E
i
(D) is empirical (measured) probability of eect produced by radiation i with single-
event microdosimetric spectrum f
1,i
(z) [30,31].
A further consequence is thatwith no more than one event involvedeects must
be independent of dose rate if the cells respond autonomously.
In most situations, only a small fraction of events traversing the cell is causative. For
instance, only one 1 of 6000 electrons traversing a cell nucleus will produce a lethal lesion
[32]. A possibly more meaningful denition of low dose (biological low dose) can be
made in terms of these causative events (hits) by asking that the contribution of the
quadratic term in dose (bD
2
) be less than a certain fraction, say 10%, of the total eect.
Let D be this limiting dose. By denition:
bD
2
0:1aD bqD
2
52
When cells are irradiated at a constant dose rate Dover a time t, q is given by Eq. (45) and
this expression becomes:
1
1
18Dt
0
a
b
_ _
Dt
0
D
1 e
D=Dt
0
_ _
53
Note that this is an equation for D as a function of a/b, both in units of Dt
0
. For tumor cells
(a/b = 10 Gy, t
0
= 0.5 hr, D = 2 Gy/min), q c1 and D = 1.1 Gy, which is signicantly
larger than the physical low-dose limit of 2 mGy. Note that D is an upper limit for the doses
where Eq. (53) is valid.
The KellererHug Theorem
This theorem, given here without demonstration, sets up a lower limit for the mean
number of inactivating events m. Given a probability of eect E(D) or its complement
S(D) = 1E(D), dene the mean inactivation dose [33]:
d
_
SDdD 54
and its variance:
r
2
2
_
DSDdDd
2
55
The mathematical statement of this theorem is:
m z
d
2
r
2
56
The mean number of inactivating events is d/z
F
.
Copyright 2004 by Taylor & Francis Group, LLC
4. MICRODOSIMETRIC DISTRIBUTIONS: EXPERIMENTAL TECHNIQUES
The experimental techniques used in microdosimetry are quite complex, and a complete
description is beyond the scope of this chapter (see Ref. 17). Conventional experimental
techniques do not provide methods for measuring frequency distributions of energy
depositions in sites with dimensions in the micrometer range or smaller, and a number
of alternate methods that are capable of providing equivalent information for many
biomedical applications have been developed. The underlying concept for most of these
methods is as follows: The mean energy imparted by a particle traversing a thickness x of
specied material of density q depends only on x and q. One can obtain equal energy
depositions by manipulating x and q such that their product xq remains constant. Thus, a
microscopical volume (x
f
1 Am, q =1 g/cm
3
) can be simulated with a larger cavity (x a few
centimeters) by correspondingly reducing the density. For instance, a 2.5-cm diameter
sphere lled with a propane-based tissue-equivalent (TE) gas at 17 Torr is equivalent to a
1-Am diameter of unit density tissue. In practice, it is usually not possible to exactly match
the atomic compositions of the gas and the solid. A typical counter consists of a spherical
or cylindrical cavity (called sensitive volume) lled with gas and delimited by a solid wall
material, and a central wire placed along the counter diameter and isolated electrically from
the counter wall. The traversal of the counter by charged particles results in the production
of ion pairs. If a voltage is applied between the (conductive) wall and the central wire, the
ions will drift along the electrical eld and induce an electrical signal at the collecting elec-
trode. The amount of charge collected is a measure of the energy deposited in the counter.
Figure 1 Schematic diagram of the Rossi proportional counter (see text for details).
Copyright 2004 by Taylor & Francis Group, LLC
For x-rays or g-rays, the average signal thus obtained is very weak relative to background,
and typically one runs the detector in the so-called proportional mode. This is achieved by
setting a suciently large voltage for the drifting ions to acquire enough energy to produce
second-generation and higher-generation ions through collisions with gas molecules. A
schematic diagram of a typical detector, usually called a Rossi proportional counter in ref-
erence to H. H. Rossi whose group pioneered this eld, is shown in Fig. 1.
5. MICRODOSIMETRIC DISTRIBUTIONS: THEORETICAL METHODS
For various technical reasons [17], the empirical determination of microdosimetric spectra
is not always possible, particularly for site sizes signicantly smaller than 1 Am or for
volumes of irregular shape. In these situations, calculations may be used instead. A key
element facilitating this approach is the possibility of regarding the spectrum of energy
deposition as resulting from the random overlapping of two objects: the charged-par-
ticle track (essentially, a collection of energy transfer points), and the sensitive site [14,15,
22,23,34]. Theoretical microdosimetric distributions are generated by randomly placing a
site on a Monte Carlo-generated track. In a sense, this is the opposite of what happens
in an experiment where the detector (a xed object) is traversed by random tracks. To the
extent that accurate cross sections for the interaction of the ionizing particles with the
medium are known and the code can adequately simulate the spatial and structural geo-
metries, theoretical calculations potentially represent the best procedure for obtaining mi-
crodosimetric quantities.
A full description of computational techniques in microdosimetry is beyond the
charter of this chapter and the interested reader may consult the textbook by Rossi and
Zaider [17]. Briey, the calculation proceeds along the following lines:
For a given source, one generates at each point in space the electron distribution in
energy (slowing down spectrum) n(E); commercially available codes are available for this
purpose.
For each electron energy E, the interactions with tissue (frequently represented by
liquid water) are simulated event by event, which means that one obtains a detailed
description of the geometrical coordinates, energy locally deposited, and type of energy
transfer (ionization or excitation) for each interaction point, primary or secondary.
Microdosimetric spectra f( y;E) for events transferring energy E are generated by ran-
domly placing the volume of interest (commonly, spheres or cylinders) on each track and
then evaluating, at each instance, the total energy deposited in it.
The lineal energy spectrum of the brachytherapy source is evaluated from:
fy
_
l
0
nEE
y
F
E
fy; EdE
_
l
0
nEE
y
F
E
dE
57
6. MICRODOSIMETRIC DISTRIBUTIONS FOR RADIATIONS USED
IN RADIOTHERAPY
6.1. Brachytherapy
Brachytherapy physics is concerned with electrons, whether primary or secondary, as in
the case of photon sources (Table 1).
Copyright 2004 by Taylor & Francis Group, LLC
At low doses or low-dose rates, the RBE of 100 keV photons relative to, for example,
1-MeV photons may exceed 2 [35,36]. For instance, the RBE for
125
I has been extensively
studied and results of 1.22 have been reported for the doserate range of 0.039 Gy/hr
[37]. Similarly, for
103
Pd, a study performed at 0.070.8 Gy/hr reported RBE values of
1.9F0.7 [3740]. While considerably lower RBE values apply to the higher doses or dose
rates usually employed in external radiotherapy, dierences in biological eectiveness on
the order of 1015% remain.
Microdosimetric distributions have been determined for a wide range of photon
energies, including those of the photon sources quoted in Table 1. Kliauga and Dvorak
[41] measured lineal energy spectra for photon energies in the range 121250 keV and for
site diameters of 0.247.7 Am. Fig. 2 shows these spectra as a function of photon energy.
Microdosimetric distributions for
103
Pd-,
125
I-,
241
Am-, and
192
Ir-encapsulated
sources were published by Wuu et al. [36] and are shown in Fig. 3. Table 2 gives y
D
and RBE values for some of the distributions shown in Fig. 3.
These spectra have notable common characteristics. All the spectra have a modal en-
ergy deposition that is related to the average energy lost per unit path length by the primary
particles. Because the lower energy electrons have a higher mean dE/dx, the spectra for the
lower energy photons and electrons tend to peak at the higher lineal energies. Conversely,
the higher energy particles have peaks at lower lineal energies. There is a decreasing
probability of events at lower lineal energies characteristic of the frequency distribution
of ionizations associated with the production of a single electron with energy close to zero
(i.e., an energy deposition just sucient to produce an ion pair). Irrespective of the mean
energy deposited, there is a kinematic upper limit, which represents secondary electrons with
precisely enough energy to make it across the diameter, thus depositing their total energies in
the sensitive volume. Electrons of lower kinetic energy must deposit less total energy, and
higher energy electrons will deposit only a fraction of their energy with a lower dE/dx.
6.2. High-Energy Electrons, X-Rays, and gg-Rays External Beam Therapy
Particles emitted from radioactive isotopes are generally too low in energy to provide the
penetration required for conventional treatments of tumors with external radiation beams.
Most external beam radiation therapies are performed with high-energy x-rays or
electrons produced with compact linear accelerators with accelerating potentials between
about 4 and 20 MeV. One notable exception is certain devices designed for stereotactical
radiosurgery or radiation therapy of supercial tumors that use cobalt-60 g-rays. The
Table 1 Physical Characteristics of Radioactive Sources Currently Used
in Brachytherapy
E
c
[MeV] E
b
(max) [MeV] T
1/2
137
Cs 0.662 1.17 30. years
198
Au 0.416 0.96 2.7 days
192
Ir 0.380 0.67 74.2 days
125
I 0.028 60.2 days
103
Pd 0.021 17. days
131
I 0.364 0.61 8.06 days
90
Sr/
90
Y 2.27 28.9 years
32
P 1.71 14.3 days
106
Ru/
106
Rh 3.55 367. days
Copyright 2004 by Taylor & Francis Group, LLC
Figure 2 Microdosimetric distributions for photons of dierent energies. (From Ref. 41.) The
spectra were measured in a 1-Am diameter spherical cavity.
Figure 3 Microdosimetric distributions for several photon radiations commonly used in
brachytherapy. (From Ref. 36.)
Copyright 2004 by Taylor & Francis Group, LLC
higher average energy of these therapy beams provides reduced attenuation and, therefore,
more dose at depth. At the same time, this results in events of lower average lineal energy,
as shown in Fig. 4 [42].
6.3. Heavy Particles in Medicine
It was quickly recognized after the development of high-energy particle accelerators that
accelerator-produced particles provided better penetration and more dose to the treatment
Table 2 Dose-Averaged Lineal Energy for the
Distributions Shown in Fig. 3
Radionuclide y
D
[keV/Am] RBE
103
Pd 3.8 2.3
125
I 3.5 2.1
241
Am 3.5 2.1
192
Ir 2.0 1.3
60
Co 1.6 1.0
Figure 4 Microdosimetric distributions as a function of lineal energy in a 2-Am equivalent
spherical detector. The solid line is a spectrum for cobalt-60 g-rays, and the dashed and dotted lines
are the spectra for 10- and 15-MeV bremsstrahlung x-rays.
Copyright 2004 by Taylor & Francis Group, LLC
volume while sparing the surrounding normal tissues. In addition, dierent particle types
produced dierent biological responses in both the malignant and normal tissues. This led
research groups to speculate that particles other than photons or electrons might have
physical and/or biological advantages in comparison. Clinical trials have taken place using
proton beams, neutrons, energetic heavy ions, and even negative pions. The protons,
pions, and heavy ions, in contrast to the exponential attenuation associated with electrons
and photons, have a relatively constant dE/dx with depth until near the end of their range,
where there is a rapid increase in energy deposited, reaching a maximum at what is called
the Bragg peak and followed by a rapid decrease. This type of depthdose curve is better
suited to deposit more dose in the treatment volume while sparing the surrounding normal
tissues. Negative pions have the added characteristic that, at the end of their range, they
disintegrate, transforming most of their rest mass (about 140 MeV equivalent) into sec-
ondary radiations. This is literally a localized submicroscopical nuclear explosion. Today,
protons and, to a lesser extent, energetic heavy ions are being used routinely to treat cancer
patients in a limited number of facilities. Although such particles showed that dose local-
ization can be a major advantage for specic diseases, their higher costs in relation to
electron accelerators [43] have restricted their use to a few centers, instead stimulating the
development of improved technologies with photons to achieve increasingly better dose
distributions at, unfortunately, increasing costs.
Therapeutic beams of energetic protons and heavy ions result not only in a dierent
depth dose distribution but also in dierent microdosimetric spectra as well. Energies for
protons used for therapy are nearly minimum ionizing and their microdosimetric spectra
are similar in the range of lineal energy to that of higher-energy electrons and photons, as
seen in Fig. 5 [44]. For the same energy per nucleon, the average dE/dx of the particles
increases with increasing atomic mass, with a corresponding increase in the linear energies,
as seen in Fig. 6 [45].
Figure 5 Microdosimetric spectra at the proximal, central, and distal positions of a therapeutic
proton beam with an initial energy of 155 MeV.
Copyright 2004 by Taylor & Francis Group, LLC
Moreover, for monoenergetic heavy ions of the same energy per nucleon, the average
number of ion pairs increases with a corresponding decrease in the deviation about the
mean, so we have a higher modal value with a more sharply dened Bragg peak.
The previous theoretical discussion tells us that increasing ionization density, such as
that observed with increasing ion mass, could result in a higher biological eectiveness, at
least until saturation sets in. Saturation in this case refers to large energy depositions that
have such a high probability of producing the biological endpoint that increased energy
deposited has little, if any, increased eect. We might expect, then, that changing the particle
type and energy might change the relative biological eectiveness of the particle compared
with photons and, therefore, the clinical outcome. This is indeed the case, rendering the RBE
a clinically useful parameter to vary, with the expectation that particles with higher RBEs
might be more eective in eradicating tumor cells. This is in part the motivation for the use of
beams of energetic heavy ions.
Unfortunately, energetic heavy ions require expensive, large accelerators to produce.
If increased biological eectiveness were the main therapeutic benet, with improved dose
localization not a major issue, then neutron beams might be more desirable. Fig. 7 shows
microdosimetric spectra for neutron beams as a function of the nominal neutron energy.
Experiments generally show that energetic neutrons are at least as eective as most
heavy ions for producing most biological eects. Although they are generally no better than
photons in terms of dose localization, they are much less expensive to produce. Therefore, if
biological eectiveness is the main clinical requirement and dose localization is not im-
portant, then neutron beams may be the modality of choice. Neutrons have been the focus of
clinical investigations for at least six decades, and today they are used for conventional
Figure 6 Microdosimetric spectra for various heavy ion beams as noted with nominal energies of
approximately 600 MeV per nucleon. Corresponding spectra for cobalt-60 g-rays, protons, and 14.5-
MeV neutrons are included for comparison.
Copyright 2004 by Taylor & Francis Group, LLC
external beam therapy in a few therapy centers. Clinical investigations of boron neutron
capture therapy are underway to take advantage of the increased RBEs with neutrons and,
and the same time, to conne the dose primarily to malignant cells. Lower-energy neutrons
have an exceptionally high interactioncross section with boron, so the boron is preferentially
loaded into cancerous cells via appropriately selected antibodies.
We had mentioned that negative pions have been used in clinical trials for cancer
therapy. They approach the dose localization of protons with an enhanced RBE in com-
parison with protons or photons. Because the pion mass is lighter than that of protons and
heavier ions, the scattering is greater and edges are less sharp. Moreover, although the
eectiveness of pions can be greater than that of protons or photons, it is generally lower
than that of neutrons or heavy ions. Finally, pions are produced as secondary particles at
higher energies and, therefore, are more expensive to produce at intensities suitable for
therapy. Although a few institutions have used pions in clinical trials with results com-
parable to those of neutrons and heavy ions, there are no such studies underway at this time.
In summary, the variety of radiation types available for both external beam radio-
therapy and brachytherapy provides multiple means of localizing the dose for maximum
eectiveness in specic applications. The variation in the microdosimetric spectra results in
corresponding variety in radiation quality and biological eectiveness, providing insight
into the mechanistic processes of energy transfer and dierent mechanisms for attacking
persistent diseases. Building upon this cadre of radiations, after a hundred years in the clinic,
radiation therapy remains one of the primary oncological methods for controlling cancer.
REFERENCES
1. Rossi, H.H.; Bateman, J.L.; Biavati, B.J. NYO-2740-4. NYO Reports 1967; 121127.
2. Bateman, J.L.; Rossi, H.H.; Kellerer, A.M.; Robinson, C.V.; Bond, V.P.Radiat. Res. 1972, 51
(2), 381390.
3. Bateman, J.L.; Johnson, H.A.; Bond, V.P.; Rossi, H.H. Radiat. Res. 1968, 35 (1), 86101.
Figure 7 Microdosimetric spectra in a 2-Am equivalent sphere as a function of neutron energy as
noted.
Copyright 2004 by Taylor & Francis Group, LLC
4. Bateman, J.; Rossi, H.H. NYO-2740-6. NYO Reports 1969; 235237.
5. Hall, E.J.; Kellerer, A.M.; Rossi, H.H.; Lam, Y.M. Int. J. Radiat. Oncol. Biol. Phys. 1978, 4
(1112), 10091013.
6. Hall, E.J.; Kellerer, A.M. Radiat. Res. 1973, 55 (3), 422430.
7. Wolf, C.; Lafuma, J.; Masse, R.; Morin, M.; Kellerer, A.M. Radiat. Res. 2000, 154 (4), 412420.
8. Kellerer, A.M.; Rossi, H.H. Br. J. Radiol. 1972, 45 (536), 626.
9. Kellerer, A.M.; Rossi, H.H. Radiat. Res. 1971, 47 (1), 1534.
10. Bond, V.P.; Meinhold, C.B.; Rossi, H.H. Health Phys. 1978, 34 (5), 433438.
11. Hall, E.J.; Novak, J.K.; Kellerer, A.M.; Rossi, H.H.; Marino, S.; Goodman, L.J. Radiat. Res.
1975, 64 (2), 245255.
12. Kellerer, A.M.; Hall, E.J.; Rossi, H.H.; Teedla, P. Radiat. Res. 1976, 65 (1), 172186.
13. Wolf, C.; Lafuma, J.; Masse, R.; Morin, M.; Kellerer, A.M. Radiat. Res. 2000, 154 (4), 412420.
14. Kellerer, A.M.; Rossi, H.H. Radiat. Res. 1978, 75 (3), 471488.
15. International Commission on Radiation Units and Measurements. Microdosimetry 32; ICRU:
Bethesda, MD, 1983.
16. Zaider, M.; Rossi, H.H. Radiat. Res. 1988, 113 (1), 1524.
17. Rossi, H.H.; Zaider, M. Microdosimetry and Its Applications; Springer-Verlag: Berlin, 1996.
18. Cauchy, A. Memoire sur la Rectication des Courbes et la Quadrature des Sourface Courbes.
Qevres Completes; Gauthier Villard: Paris, 1908.
19. Robbins, H.E. Ann. Math. Stat. 1944, 16, 342347.
20. Kellerer, A.M.; Rossi, H.H. Radiat. Res. 1978, 75 (3), 471488.
21. National Council on Radiation Protection and Measurements; Conceptual Basis for Cal-
culations of Absorbed-Dose Distributions Recommendations 168; National Council on Radiation
Protection and Measurements: Bethesda, MD, 1991.
22. Kellerer, A.M.; Chmelevsky, D. Radiat. Environ. Biophys. 1975, 12 (3), 205216.
23. Kellerer, A.M.; Chmelevsky, D. Radiat. Environ. Biophys. 1975, 12 (4), 321335.
24. Rossi, H.H. Radiat. Environ. Biophys. 1979, 17 (1), 2940.
25. Rossi, H.H. Ann. N.Y. Acad. Sci. 1969, 161 (1), 260271.
26. Dale, R.G. Br. J. Radiol. 1985, 58 (690), 515528.
27. Zaider, M.; Wuu, C.S. Br. J. Radiol. 1995, 68, 5863.
28. Zaider, M.; Dicello, J. RBEOER: a Fortran program for the computation of RBEs, OERs,
survival ratios and the eects of fractionation using the Theory of Dual Radiation Action. Los
Alamos Scientic Laboratory Report LA-7196-MS, 1978; 119.
29. Zaider, M.; Brenner, D.J. Radiat. Res. 1985, 103 (3), 302316.
30. Bond, V.P.; Varma, M.; Feinendegen, L.E.; Wuu, C.S.; Zaider, M. Health Phys. 1995, 68 (5),
627631.
31. Zaider, M. Health Phys. 1996, 70 (6), 845851.
32. Zaider, M.; Rossi, H.H. Int. J. Radiat. Biol. 1998, 74 , 633637.
33. Kellerer, A.M.; Hug, O. Random factors in the survival curve. Adv. Biol. Med. Phys. 1968, 12,
353366.
34. Kellerer, A.M.; Chmelevsky, D. Radiat. Environ. Biophys. 1975, 12 (1), 6169.
35. Wuu, C.S.; Zaider, M. Med. Phys. 1998, 25 (11), 21862189.
36. Wuu, C.S.; Kliauga, P.; Zaider, M.; Amols, H.I. Int. J. Radiat. Oncol. Biol. Phys. 1996, 36 (3),
689697.
37. Ling, C.C.; Li, W.X.; Anderson, L.L. Int. J. Radiat. Oncol. Biol. Phys. 1995, 32 (2), 373378.
38. Zellmer, D.L.; Shadley, J.D.; Gillin, M.T. Radiat. Prot. Dosim. 1994, 52 (14), 395403.
39. Nath, R.; Meigooni, A.S.; Melillo, A. Int. J. Radiat. Oncol. Biol. Phys. 1992, 22 (5), 11311138.
40. Zellmer, D.L.; Gillin, M.T.; Wilson, J.F. Int. J. Radiat. Oncol. Biol. Phys. 1992, 23 (3), 627632.
41. Kliauga, P.; Dvorak, R. 1199. Radiat. Res. 1978, 73 (1), 120.
42. Amols, H.A.; Zellmer, D.L. Med. Phys. 1984, 11, 247253.
43. Dicello, J.; Slater, J.M. Trans. Am. Nucl. Soc. 1993, 68B, 3740.
44. Dicello, J. Chapman and Hall: New York, 1995; 3544.
45. Dicello, J.; Wasiolek, M.; Zaider, M. IEEE Trans. Nucl. Sci. 1991, 38, 12031209.
Copyright 2004 by Taylor & Francis Group, LLC
19
Charged Particle and Photon-Induced
Reactions in Polymers
S. Tagawa, S. Seki, and T. Kozawa
Osaka University, Osaka, Japan
SUMMARY
Basic and applied researches on charged particle and photon-induced reactions of polymers
are surveyed. The basic parts are fundamentals of radiation eects on polymers and pulse
radiolysis studies on polymers. The intermediate parts are a great diversity of radiation
eects on polymers and reaction mechanisms of electron beam (EB) and x-ray resists. The
applied parts are economic scale of utilization of radiation and industrial application of
radiation to polymers.
1. INTRODUCTION
Since the rst studies of radiation eects on polymeric materials, fundamental researches on
radiation-induced reactions in polymers have been widespread, and the eld of their
applications has been growing until now. Seeds of typical application of dierent types of
radiation were introduced by Chapiro and by Charlesby during the 1950s, and great deal of
work has been published in this eld concerning various polymeric materials and radiation
sources. This is due to quick or sensitive responses of macromolecules to radiation, because
few reactions can dramatically change such macroscopic properties of polymeric materials
as solubility, mechanical properties, and thermal stability. Cross-linking and chain-scission
reactions caused by irradiation have been treated as the key reactions controlling the
properties of macromolecules. Decision of the predominant reaction in a polymer for a
radiation source has been very important to predict the possibility of modication of the
polymer for engineering applications. This approach to the combination of radiation and
polymers has been practically useful because polymeric materials have been the leading
structural materials, so that polymers are now used in every stage of our daily life.
However, recent progress in both radiation and polymer researches indicates a
dramatic change in the circumstance of this research eld, and the trend has been rapidly
expanding, especiallyinthe last decade. Polymers will be usednot onlyas structural materials
but also as electronic, optical, medical, and pharmaceutical materials, and radiation will be a
powerful candidate to manipulate the structure andproperty of polymers. In this section, the
Copyright 2004 by Taylor & Francis Group, LLC
potentials of radiation and polymers are reviewed for the forecast of technology in the
twenty-rst century.
2. ECONOMIC SCALE OF UTILIZATION OF RADIATION
The economic scale of utilization of radiation has been surveyed several times by many
organizations. Mainly from the viewpoint of public acceptance (PA), Management Infor-
mation Services, Inc., in the United States conducted a survey to clarify the size of those
economic scale [1] and the results were presented by Alan E. Walter, the former president of
the American Nuclear Society, at conferences [2]. Similar studies were done in Japan, but the
data had poor consistency for comparison. To overcome this shortcoming, in 1999, the
Special Committee for Evaluation of Economy on Utilization of Radiation examined
the Japanese economic scale of utilization of radiation, consistently and systematically, for
the rst time, based on reliable published data and reasonable assumption both under the
sponsorship of the Ministry of Education, Culture, Sports, Science and Technology
(MEXT) and with the support of the Japan Atomic Energy Research Institute (JAERI).
The evaluation was based on economic indices indicating the magnitude of the market
created by products manufactured through use of radiation.
We performed a detailed comparison of the economic scale of radiation use in the
elds of industry [4], agriculture [5], and medicine [6] in Japan and the United States. The
comparison between Japan and the United States was also done for an economic index
regarding market creation of products obtained from use of both radiation and nuclear
energy [7]. The results of our study are compared for the year 1997.
The economic scale is compared between the United States and Japan with selected
industrial parameters such as sterilization, semiconductors, radiographic testing, and radial
tire production, because the very large industrial markets make a whole comparison
dicult. The economic scale revealed in selected industrial elds was about $56 billion for
the United States and $39 billion for Japan. The former is larger by a factor of
f
1.4 [4].
The economic scale of the application of radiation in the eld of agriculture in Japan
was estimated from public documents to be about $964 million in 1997. The economic scale
survey in food irradiation and mutation breeding was extended to the United States for a
direct comparison to the situation in Japan. The maximum estimation amounted to $3.2
billion for food irradiation and $11.2 billion for mutation breeding. The economic scale for
products in selected agricultural elds was $14.5 billion for the United States and about
$0.8 billion for Japan, implying that the former is larger in magnitude by a factor of about
18 [5].
The economic scale of the use of radiological technology in medicine in Japan in 1997
was about $10 billion. The economic scale in 1997 was compared between Japan and the
United States [6]. Within the same limited parameters, the economic scale of the use of
radiological technology in medicine was approximately $49 billion for the United States and
$12 billion for Japan. The former is larger in magnitude by a factor of 4, implying that
practitioners in the United States do not hesitate to use radiological technologies for medical
applications. This tendency was clearly observed by the expenditure on prostate cancer,
FDG-PET, and so on.
In 1997, the economic scale of radiation usage was estimated to be $119 billion (1.4%
of GDP) in the United States and $52 billion ($1 =121 yen) (1.2%of GDP) in Japan. In the
United States, the use of radiation technology not only in industry but also in medicine is
signicantly advanced. The usage in agriculture is relatively small compared to those in
Copyright 2004 by Taylor & Francis Group, LLC
industry and medicine both in the United States and Japan. Usage in the agricultural sector
in Japan is much smaller than in the United States [7].
3. RADIATION-INDUCED REACTIONS IN POLYMERS
3.1. Fundamentals of Radiation Effects on Polymers
Early processes caused by high-energy charged particles and photons have already been
described in other chapters. Some characteristic processes occurring in the polymeric
systems are focused on in the present chapter. It is well known that in some radiation
chemical processes, the energy consumption in the process is far smaller than the energy
deposited by incident radiation, indicating that most of the energy is consumed by radiative
and/or nonradiative decay of excited states via energy transfer processes instead of pro-
ducing reactive intermediates. One very important example of this is the so-called protection
eects of polyolen by addition of aromatic molecules [8]. Energy and charge transfer pro-
cesses are in relation to each other, and these are systematized theoretically in the condensed
phases including polymeric systems. It is because of the high mobility, hence, of the large
diusion constant of charged species even in the polymeric system. It provides a validity of
the formulation developed in the liquid or solid phases for extension to polymeric systems.
In contrast, diusion processes of neutral reactive intermediates strongly depend on the
structure, molecular mobility, dielectric constants, etc. of solid materials. Although the neu-
tral species including (free) radicals bear strict limitation in their migration they are very
important promoters of radiation-induced reaction in the polymeric systems.
Energy Transfer Process
Several dierent processes have been accounted for transfer of electronic excitation energy
between molecules or functional groups in macromolecules: (1) electron exchange inter-
action, (2) radiative excitation transfer, (3) excimer migration, (4) exciplex excitation
transfer, and (5) long-range excitation transfer. Electron exchange interaction occurs over
a very short distance via some overlap of molecular orbitals of the excitation donor and
acceptor, called a Dexter-type energy transfer process [9]. Only the limitation of this transfer
process is the conservation of the total spin of the system. Radiative transfer is due to the
reabsorption of emission from a donor molecule by an acceptor via a photon. It needs the
overlap between an acceptors optical absorption and emission spectrum of a donor. There
is no total spin conservation in the system; however, the triplet state of the acceptor is not
formed because direct transition is usually spin forbidden. Excimers and exciplexes are the
pairs of an excited molecule and an unexcited molecule [10], and the excimer is the pair
formed between two consistent molecules [10]. Excimer formation has been reported in
many systems of solids, liquids, and gases. One example of the excimer formation in
radiation chemical processes is the excimer of phenyl groups in polystyrene of which the
yield, absorption, kinetics, and emission have been precisely reported to date [11,12]. Long-
range excitation transfer is mainly caused by a Coulomb interaction between an excited and
an unexcited molecule that are making simultaneous coupled electronic transitions of
almost equal energy. This is the so-called Fo rster-type energy transfer (inductive resonance
or vibrational relaxation transfer) (for a review, see Ref. 13). The transfer occurs with a
relatively slow rate constant because the motive force of the transfer, namely, the dipole
dipole interaction (coupling) of molecules, is very small compared to the width of the
vibrational band of the excited state of the donor molecule. This leads to the transfer from
Copyright 2004 by Taylor & Francis Group, LLC
the lowest vibrational excited state of donor molecules after vibrational relaxation, but
enables long-range excitation transfer over
f
10 nm in a single step.
In the real polymeric system, a large number of modeling eorts have been made for
energy transfer processes caused by radiation as well as low-energy photons, and direct or
indirect observation of excited-state dynamics has also been performed for a variety of
polymeric systems. The earliest modeling work of excitation dynamics in polymeric systems
was given by Simpson for polyenes by treating them as the chained analogs of simple
molecules having k conjugation [14]. The developed modeling and experimental evidence is
noted later. To date, the formation of the excited states, the energy transfer, and decom-
positionfromexcitedstates inthe radiation chemical processes have been studiedover a wide
range of polymeric systems: hydrocarbon polymers without double bonds (polyethylene,
polypropylene, etc. [1519]), saturated polymers with aromatics (e.g., polystyrene, poly-a-
methylstyrene, polyvinylcarbazole [11,12,2022]), k-conjugated polymers (e.g., polypheny-
lenevinylene, polythiophene [2225]), and j-conjugated polymers (e.g., polysilane, poly-
germane [26,27]). LET and temperature eects on excited states of solid-state polystyrene
[28a,b] and solid-state polysilanes [28c] were studied by ion beam pulse radiolysis.
Radical Processes
When a polymeric system is exposed to high-energy radiation, the system undergoes main-
chain scission and the creation of cross-links, end-links, double bonds, free radicals, etc. The
neutral radicals are the key promoters of the reactions above, and the structure, reactivity
(stability), migration, etc., have been extensively investigated by many techniques including
ESR. Details of the techniques and their results are described in other chapters or reviews
(for a book, see Ref. 29), and the eects of the above reactions on polymeric systems are
mainly discussed in the present section.
On the basis of no dierence in the probability of main-chain scission in any part of
the polymer backbone (no chain end eect), an approach to the statistical treatment of
main-chain scission was given by Charlesby in 1954 [30]. Although main-chain scission is
directly related to the degradation of a polymeric system, cross-linking and end-linking
reactions cause more dynamic changes in its macroscopic behavior such as gelation of the
polymers, which have produced many industrial applications: radiation curing, stiening,
grafting, etc. It is also a reasonable assumption that cross-linking and end-linking reactions
are produced randomly by low-LET radiation and the number of links is proportional to
the radiation dose, so long as the number of cross-links is suciently small in comparison to
the total number of structural units. Classical statistical theory of main-chain scission and
cross-linking were given by Kuhn in 1930 for chain scission [31], by Flory for cross-linking
in 1941 [32], and modied by Charlesby in 1953 [33]. The initial formulation of main-chain
scission and cross-linking probability is given as CharlesbyPinner relationship by
1=M
n
1= M
n
0
p
0
0:5q
0
D=m 1
1=M
w
1= M
w
0
0:5p
0
q
0
D=m 2
where p
0
is the probability of scission and q
0
is the probability of cross-linking. M
n
and M
w
denote the number and weight average molecular weight with their initial values of (M
n
)
0
and (M
w
)
0
, respectively, m is the molecular weight of a unit monomer, and D (MGy) is
absorbed dose. Gvalues of main-chain scission and cross-linking are related to the values of
p
0
and q
0
as follows,
Gs 9:6 10
3
p
0
3
Gx 4:8 10
3
q
0
4
Copyright 2004 by Taylor & Francis Group, LLC
where G(x) is the Gvalue of cross-linking and G(s) is the Gvalue of main-chain scission. The
above statistical equations were superior for the ease of calculation to other classical
theories; however, they were not sucient to give an entire solution of main-chain scission
and cross-linking of a polymeric systembecause of the assumption of the theory of an initial
arbitrary molecular size. This gave some diculties regarding average molecular weight,
solubility, viscosity, etc. Problems with the molecular weight distribution and its change
with the irradiation were solved by Saito [34] and by Inokuti [35]. In their theory, molecular
weight distribution was expanded by the Poisson and/or SchulzZimm distributions, and
changes in the distribution by the simultaneous reactions of main-chain scission and cross-
linking were analytically traced. All of these initial works were done in parallel, and the
following work developed on the starting points so provided.
ODonnell et al. used a binominal expansion of the initial molecular weight distribu-
tion and discarded higher terms than a cubic factor [36]. This leads to a computer simulation
to trace the changes in the molecular weight of polymers without using a mathematical table.
In the 1990s, modied formulations of chain scission and cross-linking were proposed by
Zhang et al. [37] and Olejniczak et al. [38]. Sun introduced a parameter h, which was related
to the glass transition temperature of the polymer, and took the eects of molecular stiness
of the polymers into account. Rosiak proposed a deductive distribution function of
molecular weight on the basis of an arbitrary distribution and extended the validity of the
CharlesbyPinner relationship. Major diculties have been steadily solved by many
attempts to improve the formulation; however, some new problems have also arisen in
recent progress in the investigation of radiation-induced reactions in some polymeric
systems. Nowadays, approach to main-chain scission and cross-linking reactions still
contains some assumptions: (1) G(s) and G(x) are constant with radiation dose, (2)
occurrence of scission and cross-linking may be treated separately and independently, (3)
random scission and cross-linking occur in the system, (4) not more than one cross-link
connects any two polymer molecules, etc. Some of these assumptions have been revealed to
be incorrect in some polymeric systems and for high-LET radiation. The problems will be
discussed in the following sections.
3.2. Pulse Radiolysis Studies on Polymers
Recent pulse radiolysis studies on polymers [39,40] and monomers [41] are reviewed from
1990. Very high quality of pulse radiolysis systems is required for pulse radiolysis studies on
polymers and polymeric systems, especially polymer and polymeric intermediates, because
of the complexity of reactions and small absorption coecients of reactive intermediates of
both polymers and polymeric systems compared with the reactive intermediates of aromatic
molecules such as pyrene and biphenyl. Therefore, most of pulse radiolysis studies on
polymers are studies on reactive intermediates of solute molecules, such as pyrene in polymer
and polymeric systems. Recent progress in pulse radiolysis studies on polymers and poly-
meric systems has been enhanced mainly by the development of both pulse radiolysis tech-
niques and advanced technology such as lithography. The development of pulse radiolysis
techniques has beendone by improvement of electronbeamintensityandstability, sensitivity
and dynamic range of photodetectors, time resolution of pulse radiolysis, wavelength region
of monitor light, and experimental temperature. Especially, mention may be made of the
development of laser synchronized picosecond [42] and subpicosecond [43] pulse radiolysis
systems and the application of low-temperature nanosecond pulse radiolysis technique over
a wide range of temperature [44a] with the monitoring wavelength region from 300 to 1600
nm [44b] to polymers and polymeric systems. Details of the progress in laser synchronized
picosecond and subpicosecond pulse radiolysis are described in another chapter.
Copyright 2004 by Taylor & Francis Group, LLC
Polystyrene and Related Compounds
The radiation chemistry of polystyrene (PS) in solution has been studied by many groups.
Absorption peaks of the intramolecular excimers of PS at 530 nmand poly-a-methylstyrene
at 520 nm in p-dioxane solutions [45a], and absorption peaks of a charge transfer (CT)
complex between a polystyrene phenyl ring and Cl atom at 320 and about 500 nm in
chloroformsolutions [45b], were observed for the rst time by using laser ash photolysis in
1980. Later, the CTcomplex between poly-a-methylstyrene and Cl atom[45c] was observed.
Intermediate species of PS in solutions, such as the CTcomplex [21a] and the intramolecular
excimer [46], were studied by using pulse radiolysis. The lifetime of the excimer absorption
and uorescence of polystyrene is 20 nsec [21b,45a,46]. The intramolecular excimer
formation and energy migration of oligostyrenes [21c,21d] were studied by using pulse
radiolysis. The CT complex between a PS phenyl ring and a Cl atom was formed through a
geminate ion recombination [21a] whose kinetics was dierent from that between the CT
complex and the excimer. Lifetimes of the CT complexes in CCl
4
and CHCl
3
were about 200
and 420 nsec, respectively, dominated by hydrogen abstraction fromPS by Cl atom[21a,46].
In the pulse radiolysis study on polystyrene solution, only the tail of absorption band due to
polystyrene intramolecular dimer cation radical was observed in less than 1000 nm by
previous laser ash photolysis and pulse radiolysis of PS in both solution and solid lm
[21b].
Recently, by using improved nanosecond pulse radiolysis with the monitoring wave-
length region from 300 to 1600 nm [44], absorption spectra due to main reactive inter-
mediates such as the intramolecular dimer cation radical in the near-IR wavelength region
were clearly observed in the pulse radiolysis of polystyrene in various solutions [47]. For
example, Fig. 1 shows the absorption spectra observed in the pulse radiolysis of polystyrene
solutions in CH
2
Cl
2
.
Saturated Hydrocarbon Polymers
Absorption due to main intermediates such as polymer cation radicals and excited states,
electrons, and alkyl radicals of saturated hydrocarbon polymers had not been observed for a
long time by pulse radiolysis [39]. In 1989, absorption due to the main intermediates was
observed clearly in pulse radiolysis of saturated hydrocarbon polymer model compounds
except for electrons [39,48]. In 1989, the broad absorption bands due to polymer excited
states in the visible region and the tail parts of radical cation and electrons were observed in
pulse radiolysis of ethylenepropylene copolymers and the decay of the polymer radical
cations were clearly observed [49]. Recently, absorption band due to electrons in saturated
hydrocarbon polymer model compounds was observed clearly by pulse radiolysis [49] as
shown in Fig. 2. In addition, very broad absorption bands in the infrared region were
observed clearly in the pulse radiolysis of ethylenepropylene copolymers [50] as shown in
Fig. 3. Radiation protection eects [51] and detailed geminate ion recombination processes
[52] of model compounds were studied by nano-, pico-, and subpicosecond pulse radiolyses.
Polysilanes
Very intense and sharp near-UV absorption bands due to radical ions of polysilanes [53a,b]
and polygermanes [53c] were observed by nanosecond pulse radiolysis. Broad visible and IR
absorption spectra due to the radical ions of polysilyne [54] and polygermyne [54] were also
observed. Very systematic pulse radiolysis studies on many dierent kinds of polysilanes [55]
have been made by our improved nanosecond pulse radiolysis system over a wide range of
Copyright 2004 by Taylor & Francis Group, LLC
Figure 1 Transient absorption spectra obtained in the pulse radiolysis in 200 mM (base mM unit)
polystyrene solutions in CH
2
Cl
2
at the pulse end (
x
) and 100 nsec after the pulse (D). Inset: time-
dependent behavior observed at 1200 nm.
Figure 2 Transient absorption spectra obtained in the pulse radiolysis of neat n-dodecane.
Copyright 2004 by Taylor & Francis Group, LLC
temperature and with the monitoring wavelength region from 300 to 1600 nm [44]. Both
radical cations and anions of polysilanes have very intense sharp near-UV and very intense
and broad absorption bands as shown in Fig. 4 [55a]. Transient absorption spectra due to
excited states [26,27a,b] and radical cations [27c] of polysilanes were also observed by laser
ash photolysis. Detailed studies on radical ions of oligogermanes were performed by
nanosecond pulse radiolysis [56].
Poly(Methyl Methacrylate)
A broad, structureless absorption in the 350- to 500-nm region was observed in pulse
radiolysis of additive-free poly(methyl methacrylate) (PMMA) in 1969, but the assignment
of the absorption was not possible at that time [57]. Then absorption peaks at 725 and 440
nmwere ascribed to the radical anion and the radical cation of PMMA, respectively, in 1983
[58]. However, these assignments were not compatible with the results obtained in pulse
radiolysis of solutions of PMMA and its substituted analogues [59]. No distinct absorption
peak at 440 nm was observed and a short-lived absorption component around 440 nm was
attributed to PMMA radical anions [59]. For pulse-irradiated solid PMMA, no distinct
absorption peak around 440 nm was observed by the pulse radiolysis of solid PMMA with
0.1-Asec time resolution, from 320 to 620 nm, and the short-lived absorption component
around 440 nm was attributed to PMMA radical anions [60]. Transient absorption spectra
of irradiated additive-free solid PMMAcontaining MMAmonomers were clearly observed
by our improved nanosecond pulse radiolysis system over a wide range of temperature and
with the monitoring wavelength region from 300 to 1600 nm [44]. Absorption bands due to
PMMA radical anions and MMA dimer anions [61] were observed in the UV and near-IR
regions, respectively, as shown in Fig. 5.
3.3. Diversity of Radiation Effects on Polymers
Classical investigations of radiation eects on polymers have been summarized in some
books or reviewarticles in the 1970s to the 1990s [29,6264]. Many of the works were focused
Figure 3 Transient absorption spectra obtained in the pulse radiolysis of neat n-polyolen.
Copyright 2004 by Taylor & Francis Group, LLC
Figure 4 Transient absorption spectra of radical anions (a) and radical cations (b) of polymethyl-
phenylsilane at 15, 100, and 250 nsec after a pulse. Superimposed gures indicate the kinetics traces of
transient absorption.
Copyright 2004 by Taylor & Francis Group, LLC
on the reaction mechanisms including main-chain scission and cross-linking induced by the
irradiation of low-LET radiation such as
60
Co g-rays, high-energy electron beams, etc.
Throughout these studies, the mainandgeneral objective was todevelopanunderstanding of
the dierent chemical reactions that occur when a polymer is subjected to radiation and to
relate these to observed changes in physical properties. Pioneering studies of the simplest
polymeric material, polyethylene, were started by Dole [65] and Charlesby [66] in 19481952,
demonstrating the drastic improvement in the thermal stability and insolubility for solvents.
Since this pioneering study, radiation-induced reaction mechanisms in polyethylene have
been vigorously and extensively investigated in detail, which brought up many questions in
the fundamental aspects of radiation chemical processes in polymeric materials.
Dose Rate. The dose rate factor is closely related to the atmosphere, temperature,
shape of irradiated specimen, etc.; thus the eect of dose rate is hardly discussed separately
from the others. An accumulation of reactive intermediates such as radicals basically plays
an important role in the dose rate eects in the ratio of main-chain scission and cross-linking
reactions. The accumulation, namely, the density of reactive intermediates, is a slowprocess
competing with the consumption by reaction with the molecules in the atmosphere, radical
recombination, etc., as well as the consumption by other reactions such as double bond
formation. In many cases of irradiation with relatively high dose rate by high-energy EB,
these eects may often be neglected because the consumption by atmospheric molecules
occurs via diusion in a polymeric material. However, the eects become crucial for cross-
linking in reactive atmosphere such as hydrogel formation or grafting [67], radiation-
induced conductivity measurement [68], or sterilization [69] by ionizing radiation. These
eects also play an important role in the ultra-high-density energy deposition by charged
particles, and will be discussed later.
Figure 5 Transient absorption spectra of irradiated additive-free solid PMMA containing MMA
monomers observed by the authors improved nanosecond pulse radiolysis system over a wide range
of temperatures and with the monitoring wavelength region from 300 to 1600 nm.
Copyright 2004 by Taylor & Francis Group, LLC
Temperature. Temperature determines the chain motion of irradiated polymer
molecules and is easily predicted to be a crucial factor controlling the eciency of the
reactions induced by radiation. Cross-linking reactions in polyethylene showed a large
dependence of the yield on its crystallinity, suggesting the motion of the chain (ends) directly
controls the eciency of cross-linking and main-chain scission. The yield of cross-linking
tends to be higher with an increase in the irradiation temperature. One of the recent typical
examples was reported by Oshima et al. for the radiation-induced reactions in poly(tetra-
uoroethylene) (PTFE) [70]. PTFE was so sensitive to ionizing radiation that its mechanical
properties degrade with either a very low dose in air or even under vacuum irradiation, and
this had been considered a typical main-chain scission type polymeric system. Oshima et al.
reported that the cross-linking reactions of PTFE became predominant by the irradiation at
its molten state just above the melting temperature (608 K) under oxygen-free atmosphere,
and antichemical solvents, high frictional properties, and cross-linked PTFE shows remark-
able improvement in radiation resistance, mechanical properties, frictional properties,
visible light transparency, and so on, compared with non-cross-linked PTFE. There is no
other way to attain the cross-linked PTFEthan by the above technique. Similar temperature
eects were also observed for polycaprolactone, which showed a remarkable increase in the
yield of cross-linking at the temperature above its melting point (333 K), keeping the high
yield even at RT in the quenched specimen [71]. The inverse eects of temperature on the
radiation-induced cross-linking were observed in poly(di-n-hexylsilane) of which liquid
crystalline transition temperature was 313 K. The polymer indicated ecient formation of
cross-linking for EB and ion beam irradiation at RT; however, the predominant reaction
abruptly changed into main-chain scission at above that temperature for the same radiation
sources. This was mainly attributed to the reduction of very ecient main-chain scission
reaction by radical recombination within the tightly packed crystalline phase of the
polymers, providing an exceptional example of radiation-induced cross-linking reactions
[72].
Structures. Stereotacticity is also an important factor that determines the eciency
of main-chain scission and cross-linking. Polypropylene is one of the sensitive polymers in
which radiation-induced reaction depends on the stereoregularity of the polymer backbone.
Wei et al. reported that G values of H
2
and CH
4
became smaller with increasing isotacticity
of polypropylene, suggesting the higher yield of cross-linking reaction in polypropylene with
higher isotacticity [73]. Reduced yield of main-chain scission reaction was also reported in
syndiotactic polystyrene in comparison with that in stactic polystyrene by Takashika et al.
[74]. Remarkable dierences in the G values of H
2
were also recorded in atactic and
syndiotactic polystyrene. The tacticity itself was also aected by the irradiation. High-
temperature irradiation with g-rays caused the reduction of isotacticity of isotactic PMMA
[75]. The changes were mainly caused by temporal main-chain scission followed by
recombination. The trace of the tacticity change enabled the evaluation of the eciency
of temporal main-chain scission, G(TCS), giving G(TCS) = 18.6, which was much higher
than the G value of permanent main-chain scission. The high value of G(TCS) is also very
important to reveal the amount of the potential reactive intermediates, hence for the fate of
energy deposited by radiation in the polymeric system.
3.4. Reaction Mechanisms of Electron Beam and X-Ray Resists
Electron beam resist has been a key material for mask fabrication in the semiconductor
industry. EB and x-ray lithography have recently attracted much attention as not only a
next-generation lithography in the semiconductor industry but also a nanofabrication tool
Copyright 2004 by Taylor & Francis Group, LLC
in the eld of nanotechnology and science. Under these circumstances, the research and
development of resist materials for EB and x-ray lithography have become more important
than ever. Many types of EB and x-ray resists have been developed. Details of these resists
will be discussed in Sec. 19.4.1. In this section, the radiation-induced reaction mechanisms of
resists are described.
Conventional resists such as PMMA and chlorinated poly(methylstyrene) (CMS) uti-
lize main-chain scission or cross-linking reactions induced by radiation. As the reaction
mechanisms of conventional resists have been reviewed in many books [76], this section will
be focused on the reaction mechanisms of chemically amplied EBand x-ray resists expected
as a next-generation resist for mass production. For chemically amplied resists, the latent
images are at rst formed with the acids generated by an exposure. The acids catalyze the
reactions such as a deblocking reaction to form real pattern in the following lithographic
process, typically during postexposure bake (PEB). In this process, the chain reactions by
acid catalysis enhance resist sensitivity. Since the role of radiation is to generate acids in
polymer matrix, acid-generation processes will be discussed.
Onium salts have been widely used as an acid generator for photo-, EB, and x-ray
resist. In addition, aromatic polymers such as novolak and polyhydroxystyrene have been
often used as a base polymer for EB and x-ray resist. The reaction mechanisms in a typical
resist system have been investigated by pulse radiolysis [43,52,7788], SR exposure
[79,80,8385], and product analysis [88]. Figure 6 shows the acid-generation mechanisms
induced by ionizing radiation in triphenylsulfonium triate solution in acetonitrile. The
yields of products from electron beam and KrF excimer laser irradiation of 10 mM
triphenylsulfonium triate solution in acetonitrile are shown in Fig. 7 to clarify the
Figure 6 Acid-generation reaction mechanisms induced by ionizing radiation triphenylsulfonium
triate solution. Percentages in the scheme are contributions of each reaction to total acid generation
for the 10 mM triphenylsulfonium triate in acetonitrile.
Copyright 2004 by Taylor & Francis Group, LLC
dierence between photoresist and radiation resist. In the case of KrF excimer resist, the
acid-generation processes mostly start from the direct excitation of onium salts. The main
products from the decomposition of excited sulfonium salts are in-cage recombination
products. This means the main proton source is hydrogen in the phenyl moiety in sulfonium
salts. Contrary to the photoinduced reaction, the reaction path from the excitation of
sulfonium salts accounts little for the yields of acids by EB and x-ray exposure [88].
In EB and x-ray lithography, the energy of radiation is deposited on resist materials
mostly via an ionization process. Cation radicals of base resin and electrons are generated by
the ionization. The electrons generated by ionization lose their energy through the
interaction with surrounding molecules and eventually thermalized. The initial separation
distance on average is thought to be approximately several nanometers, depending on the
type of materials. Cation radicals of base resin produce proton adducts via ion molecular
reaction with the base resin. In the case of novolak, the protons originate in the hydroxyl
groups of novolak. Therefore, the yield of acids is thought to be inuenced by the protection
of hydroxyl groups, which are often partially protected in the polarity change type resists.
For chemically amplied EB and x-ray resists, not only the dissolution characteristics of the
resists, but also the eciency of acid generation should be considered when hydroxyl groups
are protected in the design of the chemically amplied resists.
Onium salts scavenge electrons generated by ionization and release counteranions of
acids. As for various kinds of acid generators including nonionic acid generators, the rate
constant of their reactions with solvated electrons in methanol was evaluated [85]. All of
these are greater than 1.0 10
10
M
1
sec
1
and are considered to be diusion-controlled
reactions. The acids are thought to be formed through the recombination reaction of the
proton adducts with decomposed anions of onium salts. The electron- scavenging eect of
onium salts delays the recombination of cationic intermediates with electrons and prolongs
the lifetime of cationic intermediates. Yields of proton adducts increase with this eect.
Figure 7 Yields of products from electron beam and KrF excimer laser irradiation of 10 mM
triphenylsulfonium triate solution in acetonitrile.
Copyright 2004 by Taylor & Francis Group, LLC
In the chemically amplied resists, both the cation radicals of base resin and the
electrons generated by the ionization play an important role in the formation process of
acids [77]. Therefore, the geminate recombination of the cation radical of base resin with the
thermalized electrons is predicted to cause the decrease of the sensitivity. In addition, the
migration of thermalized electrons is predicted to cause the severe degradation of space
resolution of resists. The high eciency of reaction of oniumsalts with thermalized electrons
blocks both the geminate recombination and the migration of thermalized electrons in the
resist matrix. This prevents the degradation of sensitivity and spatial resolution of the resist
caused by the behavior of the thermalized electrons. This mechanism works well in the
current chemically amplied resists with spatial resolution around 100 nm. As the
dimension of resist pattern becomes closer to the initial separation distance of thermalized
electrons, the thermalization process and the subsequent reaction become more important.
The initial separation between cation radicals and electrons should be considered in the
resist design for nanolithography, as well as the scattering of secondary electrons and the
diusion of acids.
4. INDUSTRIAL APPLICATION OF RADIATION TO POLYMERS
4.1. Electron Beam and X-Ray Resists
With the advent of microelectronics, the human society has progressed to a highly
sophisticated information-oriented state. Microelectronic circuits have been fabricated
with a high resolution printing technique called lithography. The advance of micro-
electronics has highly depended on the technological innovation of lithography. Light
sources have changed toward shorter wavelength from the g-line (436 nm) of a Hg lamp to
the i-line (365 nm) of a Hg lamp to a KrF excimer laser (248 nm). An ArF excimer laser
(193 nm) is ready to be deployed as an exposure tool for mass production. An F
2
excimer
laser (157 nm) is being researched as the next-generation light source following the ArF
excimer laser. However, the resolving power of light sources will fall short of the market
demand in the near future. Radiation sources such as electron beam and x-ray are
expected to take the place of the light sources. Furthermore, nanotechnology has recently
attracted enthusiastic attention in many elds such as materials science, bioscience, and
information technology. EB and x-ray lithography are also promising tools for the mass
production of nanostructures. Research and development of each element such as a resist,
optics, and a mask for EB and x-ray lithography have been pursued. Especially, the resist
is a key technology for mass production and has been investigated energetically.
Mass production poses strict requirements on resist materials, most important of
which are sensitivity, spatial resolution, contrast, and etch resistance. The sensitivity of the
next-generation resists is required to be less than 10 AC/cm
2
(EB), 100 mJ/cm
2
(x-ray), and 25
mJ/cm
2
(EUV) [89]. It should be noted that the resist sensitivity is traditionally expressed not
by absorbed dose but exposure charge or energy per unit area. As for the spatial resolution,
45 nm is needed for the production of dynamic random access memory (DRAM) in 2010
[89]. Although resist patterns below 10 nmare presently fabricated by some kinds of resists,
they do not have enough sensitivity required for the mass production [90,91].
Electron beam lithography has been widely used in mask fabrication for photolitho-
graphy. X-raylithographyhas beenalsousedfor the fabricationof micromachines suchas the
LIGA(Lithographie, Galvanoformung, Abformung) process [92]. Manytypes of resists have
been developed for these purposes. Some positive and negative resists for EBlithography are
listed in Tables 1 and 2, respectively.
Copyright 2004 by Taylor & Francis Group, LLC
From the necessity of higher sensitivity resists than conventional ones, a chemically
amplied resist was developed originally for photolithography by Ito of IBM [93]. The
concept of chemically amplied made a great impact on lithography. Chemically
amplied resists have been used on a full scale in the mass-production line of semi-
conductors since the KrF excimer laser was deployed as an exposure source. The use of
chemically amplied resists is also one of the most promising technologies in EB and x-ray
lithography. Many types of chemically amplied resists have been developed and many
kinds of acid generators, cross-linkers, dissolution inhibitors, and resins have been
reported as shown in Table 3. Although chemically amplied resists based on acid
catalytic chain reaction mechanisms show high performance such as high sensitivity and
high contrast, they have very narrow process latitudes compared with conventional resists.
The sensitivities of chemically amplied resists tend to be degraded by impurities as these
depend on prebake temperature, prebake time, PEB temperature, PEB time, delay time
from exposure to development, and shelf life, among others. These problems are common
among photo, EB, and x-ray resists. Early researches focused on these problems to make
the chemically amplied resists applicable to the process lines, while the development of
new chemically amplied resists have been pursued [94100]. The feature required for the
next generation is so small that line-edge roughness and pattern collapse during develop-
ment are also problems and have been investigated [101,102]. Line-edge roughness should
be less than 5.5 nm (3 sigma) in the production of DRAM in 2010 [89].
Although single-layer resists are preferred from the viewpoint of simplicity of the
process, two- or three-layer resists are also promising [103]. These resists have an advantage
Table 1 Typical Conventional Positive EB Resists
Poly(methyl methacrylate)
Poly(butenesulfone)
Poly(hexauorobutylmethacrylate)
Poly(triuoroethyl-a-chloroacrylate)
Poly(methacrylic acid-co-phenylmethacrylate)
Poly(2-methylpentene-1-sulfone)/novolac
Poly(dimethyltetrauoropropylmethacrylate)
Poly(glycidylmethacrylate-co-hexauoropropylmethacrylate)
Poly(a-chloroacrylate)
Poly(isobutylene-co-methylmethacrylate)
Poly(methyl-isopropenylketone)
Poly(acrylonitrile-co-methylmetacrylate)
Table 2 Typical Conventional Negative
EB Resists
Poly(glycidylmethacrylate)
Epoxy poly(butadiene)
Poly(ethylacrylate-co-glycidylmethacrylate)
Chlorinated poly(methylstyrene)
Poly(iodostyrene)
Chlorinated poly(methylnaphthylmethacrylate)
Copyright 2004 by Taylor & Francis Group, LLC
over the single-layer resists in that multilayer resists can allot dierent roles to each layer.
Typically, a top layer bears the role of patterning and lower layers have the responsibilities
for etch resistance and planation of the substrate surface. Surface imaging technique is also
one of the candidates of the next-generation processes [104]. Imaging by silylation of
commercially available resists was reported. These techniques are expected to be useful
especially for EUV and a low-energy (a few kilovolts) EB lithography such as low energy e-
beam proximity projection lithography (LEEPLE).
In recent progress, calixarene resists have been prominent [105]. Hexaacetate p-
methylcalix[6]arene was demonstrated to work as a high-resolution negative resist. This
resist also shows high etch resistance. Calixarene resist has an advantage in its molecular
size (about 1 nm). Liquid crystal resists and inorganic resists show high resolution [106].
These resists are suitable for the fabrication of nanostructures. However, the resist
sensitivities are lower than those of chemically amplied resists, even PMMA.
It has been reported that the incorporation of nanoparticles such as fullerene and
oligosilsesquioxane (POSS) to resist matrix improves the resist performance such as etch
resistance and mechanical properties [107]. These resists are called nanocomposite resists.
A single-component chemically amplied resist that incorporates not only POSS but also
an acid generator into its main chain has been reported [108].
Although many types of excellent resists have been developed based on novel ideas,
the requirements for the next-generation resists have not yet been achieved. The quest for
the ultimate resist materials will continue.
4.2. Other Applications of Radiation to Polymers
Preparation, characterization, and fabrication of nanostructured materials have elicited
great interest in viewof their possibility of innovation in science, industry, environment, and
in our daily lives in the twenty-rst century. The research eld called nanoscience or
nanotechnology has been expanded and widespread since the end of the last century, and
the trend is expected to be wider and accelerated in this and in the next decade. Current
interest in the nanostructured materials predominantly comes from their unique physico-
chemical properties on account of their nite small sizes giving peculiar eects such as
Table 3 Typical Chemically Amplied EB Resists
Resist Component Type
No name Poly(4-tert-butoxycarbonyloxystyrene) Positive
[IBM] Acid generator
SAL series Novolak resin Negative
[Shipley] Cross-linker
Acid generator
RAY-PN (AZPN100) Novolak resin Negative
[Hoechst AG] Melamine cross-linker
EXP Novolak resin Positive
[NTT] 2,2-bis(t-Butoxycarbonyloxyphenyl)-propane
bis( p-t-Butylphenyl)iodonium triuoromethanesulfonate
PTBSS Poly(4-tert-butoxycarbonyloxystyrene-co-sulfur dioxide) Positive
[AT & T]
Copyright 2004 by Taylor & Francis Group, LLC
quantum size eect and quantum entrapment. Recent progress on the nature of nano-
structured materials has revealed unique electronic, magnetic, optical, and mechanical
properties, and the application of these properties has already started in industry. As
introduced in the previous section, nanoscale materials, to say nothing of their importance
in the semiconductor industry, are also expected to play a crucial role in the integrated
electronic circuit in the near future, which is obvious in the well-known Moores law and its
leading road map of semiconductor manufacturing.
In view of the preparation and fabrication of nanostructured materials, there are
apparently two kinds of trends. One is the assembling of molecules or atoms by using their
nature such as self-organization, self-reproduction, selective reaction, and condensation,
which is called bottom-up technology, to prepare the nanostructured materials. The other
approach, called the top-down technique, producing very small structures by using the
reaction caused by charged particles and photons, has been the major trend of micro- and
nanofabrication to date, and will be a key technology even in the future, because it is
necessary to prepare the interface of nanostructured materials for their real application even
if the materials themselves are produced by the bottom-up technique.
It has been considered that the most important advantage of radiation-induced re-
actions in the fundamental research is their nonselectivity in the media or spatial homoge-
neity of deposited energy, hence the distribution of reactive intermediates. It has been the
basic feature of radiation chemistry, not only in polymeric systems but also in organic and
inorganic systems, which has given much benet to us as an accurate quantitativeness
typically represented by G values. However, this nonselective and homogeneous concept is
no longer valid for the reaction induced by high-LET radiation including ion beams, which
has been suggested by the theoretical aspects on track structure in the condensed media. The
ultrahigh density of excitation and ionization within a charged particle track produce
extremely nonhomogeneous distribution of reactive intermediates along a particle trajec-
tory. The density of the excitation, ionization, and intermediates cannot be realized by any
other physicochemical techniques, and has potentials to cause brand-new chemical
reactions in the matter. If the yield of a chemical reaction can be controlled by the density
of the intermediates, we can freely control the size of the nonhomogeneous eld of chemical
reaction, which is called a chemical core, because the density of deposited energy is
basically presented as the function of radial distance from a particle trajectory.
Several experimental studies on the particle track suggested that the size of the
chemical core is controllable by selecting the target media and/or chemical treatment after
irradiation within the range froma fewto a fewtens of nanometers, which shows the best t
to produce the nanosized materials. An incident charged particle promotes a top-down-style
nonhomogeneous eld of chemical reaction, and its high controllability is provided by the
bottom-up nature of the target material. This is the concept of the complex of top-down and
bottom-up styles, which is one of the practical examples of radiation chemistry contributing
to nanoscience and nanotechnology.
Basics. It has been suggested that spatial distribution of deposited energy by charged
ions has played a signicant role in chemical reactions occurring in the target materials.
Models of the energy distribution were proposed experimentally and theoretically such as
Track Core and Penumbra models by Magee and Chattarjee (for a review, see Ref.
109), Katz et al. [110], Waligorski et al. [111], Varma et al. [112,113], Wingate and Baum
[114], Wilson [115], and other groups. In spite of the theoretical modeling eort, there are
still unknown factors in the relationship between the ion track structure and the values of
track radii that were experimentally obtained by the analysis of irradiation products
[72,116118].
Copyright 2004 by Taylor & Francis Group, LLC
As one of the basic formulations accounting for the radial dose distribution in an ion
track, the following formulas were given [92]:
q
c
r
LET
2
pr
2
c
1
LET
2
2pr
2
c
ln
e
1=2
r
p
r
c
h i
1
rVr
c
q
p
r
LET
2
2pr
2
ln
e
1=2
r
p
r
c
h i
1
r
c
<rVr
p
5
where q
c
and q
p
are the deposited energy density at the core and the penumbra region,
respectively, r
c
and r
p
are the radii of the core and the penumbra, respectively, and e is the
base of natural logarithm.
Experimental approaches to measure the radial dose distribution are also in progress
[119], and it was found that the distribution follows r
2
law in the inner region of a critical
distance and obeys r
3
law outside of that region. LaVerne and Schuler reported the
considerable decrease in the radiation chemical yield for ferric production in the Fricke
dosimeter, suggesting a model of a deposited energy density in an ion track, which depends
on the LET and the atomic number of an irradiation particle [120,121].
Puglisi and Licciardello [122], Licciardello and Puglisi [123], and Calcagno et al. [124]
also reported the eects of ion beambombardment to polystyrene leading to the aggregation
of molecules and cross-linking reactions. The abnormal change in molecular weight
distribution was ascribed to the intratrack reaction; however, the estimated size of ion
tracks was also larger than that of the track core. In many cases, the eects of the intratrack
nonhomogeneous reaction have been presented as the LET eects on the chemical yield of
radiation-induced reaction in the polymeric system. The LET eects on the yield of cross-
linking reaction were reported in polystyrene [125], polyethylene [126], polysulfone
[126,127], and polysilane [72,116,128], showing remarkable increase in the yield with an
increase in the value of LET of incident particles. Especially, polysulfone and polysilane
showed a drastic conversion from scission type (for low LET) to cross-linking (for high
LET) with the threshold value of LET [72,126]. Decrease in the yield of main-chain scission
was also observed in PMMAwith an increase in the LET. This experimental evidence can be
explained in part by the model of the nonhomogeneous reaction in an ion track in which
high-density reactive intermediates cause an increase in the yield of radical recombination,
cross-linking reaction, excited-state relaxation, cross-recombination of charges, etc. Ion-
beam-induced chemical reaction in other media was reviewed in other books [129] or
proceedings, and the present section introduces the present status of application of charged
particles in relation with nanoscience and nanotechnology.
Charged Particles for Nanofabrication. The basic approaches to nanofabrication by
using the nonhomogeneous reaction for nanofabrication can be divided into the next two
concepts: (1) making pores in polymer lm along a particle track and lling with metals,
semiconductors, ceramics, polymers, etc. and (2) directly making a wire along an ion track
based on the target polymer materials. The schematic view of these ideas is displayed in
Fig. 8. The rst approach surpasses the second in the choice of the materials to make, and the
second can prepare more precise structures of sizes smaller than 10 nm. The rst approach
began from the preparation of membranes having several microsized pores produced by
etching of charged particle tracks in nuclear materials (for reviews, see Refs. 130 and 131).
The concept of etched ion track and its lling was rst presented by Vetter and Spohr in 1993
[132] and at almost the same time by Ferain and Legras [133]. Their investigations were done
in parallel and overlapped with each other.
Copyright 2004 by Taylor & Francis Group, LLC
The attempt to use the membrane as the template of metallic replicas was successful for
preparation of copper whiskers on the substrate [132]. Vetter and Spohr used polycarbonate
membranes with a thickness of 10150 Am, and the lms were irradiated by heavy-ion beam
(Xe and heavier). The irradiated lms were etched by alkaline solution, and the etched pore
sizes were controlled by the time of the treatment. The copper metal was deposited by
electrolysis in a galvanic cell, and after the deposition, the membrane was removed
mechanically or by a suitable solvent. Figure 9 shows the SEM image of copper
microwhiskers on the substrate prepared by the above technique. The size of the whiskers
was a few micrometers as for the diameter of the cross section in this image. Fundamental
studies on the mechanism of etching processes have shown remarkable progress in the
controllability of the size of the etched pores [134], and are going on still now with the
collaboration of other groups to produce a drug-releasing system [135], thermoresponsive
membrane [136], etc.
Similar approach has also been taken by Ferain and Legras [133,137,138] and De Pra et
al. [139] to produce nanostructured materials based on the template of the membrane with
etched pores. Polycarbonate lm was also of use as the base membrane of the template, and
micro- and nanopores were formed by precise control of the etching procedure. Their most
resent report showed the successful formation of ultrasmall pores and electrodeposited
materials of which sizes were as much as 20 nm [139]. Another attractive point of these
studies is the deposited materials in the etched pores. Electrochemical polymerization of
conjugated polymer materials was demonstrated in these studies, and the nanowires based
on polypyrrole or polyaniline were formed with a fairly cylindrical shape reecting the side
wall structure of the etched pores. Figure 10 indicates the shape of the polypyrrole
microwires with their dimension changes by the limitation of the thickness of the template.
Figure 8 Schematic view of the two dierent concepts to prepare nanostructured materials using
chemical reactions in a particle track (chemical core). Bar length: 10 Am.
Copyright 2004 by Taylor & Francis Group, LLC
The direct formation of nanostructured materials was rst reported by Seki et al. as the
nonhomogeneous cross-linking reactions of polysilane giving nanogels or nanowires
along a particle trajectory in the polymer [72,116,140]. The schematic procedure to obtain
isolated nanowires is presented in Fig. 8. An incident ion deposits its energy densely within
a limited area, and a cross-linked, cylinder-like structure is formed along the projectile. The
uncross-linked part of a lm is washed by solvent, leading to the isolated nanowires of
cross-linked polysilanes. Figure 11 shows a series of atomic force microscope (AFM)
images for nanowires formed on a Si substrate, together with the enlarged view of a wire
and the schematic of a cross section. The section of a wire indicates a complete cylindrical
shape. The number of observed nanowires in a unit area shows good agreement with that of
incident particles. In spite of a development procedure by solvents, the wires remain
adhered to the Si substrate and the nanowires remain isolated from each other. This
indicates that one end of the wire is tightly connected to the substrate by chemical bonds.
Another interesting feature of the wires is their length. The lm with 170-nm initial
thickness is used to obtain the images in Fig. 11. The mean length of the nanowires in Fig.
11 is 180 F9 nm, suggesting that the length of a wire is fairly controlled by simply changing
the thickness of a target lm. Figure 12 shows the AFM images of the nanowires produced
in polysilanes with a variety of molecular weights and by the dierent ion beams. The
structure of the nanowire apparently changes from a thick rodlike to a ne wormlike wire
with a decrease in the molecular weight, and reects the lm thickness as the length of
nanowires. The thickness of the nanowire obviously depends also on the LET of ion beams.
Figure 9 Overview of copper whisker array prepared by electrodeposition in the membrane with
micropores, and dissolution of the polymer matrix by a solvent.
Copyright 2004 by Taylor & Francis Group, LLC
The formation of nanowire by the nonhomogeneous reaction brought up the need for a
new model dealing with the cross-linking reaction and gelation caused by high-LET
radiation on the basis of the nonrandom, nonhomogeneous spatial distribution of cross-
links. A preliminary expression of form for gel and sol fractions is given by [72]
s 1 r dr f g
f
g 1 1 r dr f g
f
6
where s denotes the fraction of a soluble part of a lm, g is the fraction of insoluble parts
reecting the total volume of the wires, r is the radius of a chemical track determined by
the energy deposition rate of an incident ion, dr is the dierential radius of the track, which
depends on the shape and size of an individual polymer molecule, and f is the uence of the
incident ion beams. The trace of the total volume of nanowires (gels) give the value of r in
the polymeric system for the incident charged particle based on the Eq. (6). The obtained
values of r show good agreement with those measured from AFM images of nanowires for
polysilane and heavy ions with high LET [140].
Figure 10 FE-SEM images of polypyrrole/C1O4 nanotubes obtained by electropolymerization in
the pores of supported nanoporous template with thickness of 350 nm (a, b, c) and 1.3 Am (d).
Copyright 2004 by Taylor & Francis Group, LLC
Figure 11 AFMimages of the nanowires. (a) Top viewobserved after development of nonirradiated
lms. Images (b), (c), and (d) indicate the surface morphology of the developed lms of poly(methyl-
phenylsilane) after ion irradiation by 450-MeV
129
Xe
23+
at 2.2 10
9
, 7.1 10
9
, and 1.1 10
10
ions/
cm
2
, respectively. The tone changing fromdark to bright in this gure implies the height as much as 24
nm. The enlarged view of a nanowire is observed for the same specimen as in (c).
Copyright 2004 by Taylor & Francis Group, LLC
The nonhomogeneous reaction caused by high-LET charged particles is a unique
technique to produce nanosized cylindrical structures. It not only produces isolated
nanowires on substrate but also controls the size and length of wires. It is needless to say that
homogeneous processes by radiation have played a very important role in the radiation
chemistry; however, nonhomogeneous processes might take part in future technologies as a
powerful tool for the preparation of nanostructured materials, although the process is a
rather extraordinary case in the eld of conventional radiation chemistry.
REFERENCES
1a. Management Information Services, In The Untold Story: Economic and Employment Benets
of the Use of Radioactive Materials; Organizations United for Responsible Low-Level Radio-
active Waste Solutions, 1994.
Figure 12 Changes in the sizes of the nanowires. Images (a), (b), and (c) are the top view of
nanowires formed by the irradiation of 450-MeV
129
Xe
23+
to a poly(methylphenylsilane) (PS) (Mn =
4.2 10
5
) lm at 700-nmthickness, a PS (Mn =1.5 10
5
) lm at 400-nmthickness, and a PS (Mn =
5.0 10
3
) lm at 180-nm thickness, respectively. (d) Observed for a PS (Mn = 4.2 10
5
) lm at 310-
nm thickness irradiated by 520-MeV
84
Kr
20+
.
Copyright 2004 by Taylor & Francis Group, LLC
1b. Economic and Employment Benets of the Use of Nuclear Energy to Produce Electricity; The
U.S. Council for Energy Awareness: Washington, DC, 1994.
1c. In The Untold Story: Economic and Employment Benets of Nuclear Technologies; Orga-
nizations United for Responsible Low-Level Radioactive Waste Solutions, 1997.
2. Walter, A.E. The visible but not imaginary numbers. In ANS International Conference on
Mathematics and Computations, Reactor Physics and Environmental Analyses: Portland,
Oregon, 1995.
3. Non-reactor nuclear technology applications in the 21st Century, In ANS Meeting, Albu-
querque: New Mexico, 1997.
4. Tagawa, S.; Kashiwagi, M.; Kamada, T.; Sekiguchi, M.; Hosobuchi, K.; Tominaga, H.;
Makuuchi, K. J. Nucl. Sci. Technol. 2002, 39, 1002.
5. Kume, T.; Amano, E.; Nakanishi, T.M.; Chino, M. J. Nucl. Sci. Technol. 2002, 39, 1106.
6. Inoue, T.; Hayakawa, K.; Shiotari, H.; Takada, E.; Torikoshi, M.; Nagasawa, K.; Hagiwara,
K.; Yanagisawa, K. J. Nucl. Sci. Technol. 2002, 39, 1114.
7. Yanagisawa, K.; Kume, T.; Makuuchi, K.; Tagawa, S.; Chino, M.; Inoue, T.; Takehisa, M.;
Hagiwara, M.; Shimizu, M. J. Nucl. Sci. Technol. 2002, 39, 1120.
8. Burton, M.; Lipsky, S. J. Phys. Chem. 1957, 61, 1461.
9a. Dextor, D.L. J. Chem. Phys. 1953, 21, 836.
9b. Inokuti, M.; Hirayama, F. J. Chem. Phys. 1965, 43, 1978.
10a. Birks, J.B.; Dyson, D.J.; Munro, I.H. Proc. Roy. Soc. A 1963, 275, 575.
10b. Birks, J.B. Nature 1967, 214, 1187.
11a. Basile, L.J. Trans. Faraday Soc. 1964, 52, 4987.
11b. Kistiakowsky, G.B.; Parmenter, C.S. J. Chem. Phys. 1965, 42, 2942.
12. Vala, M.T.; Haebig, J.; Rice, S.A. J. Chem. Phys. 1965, 43, 886.
13a. Fo rster, T. In Modern Quantum Chemistry; Sinanoglu, O. Ed.; Academic Press: New York,
1966; Vol. 3, 93 pp.
13b. Fo rster, T. In Energetics and Mechanisms in Radiation Biology; Phillips, G.O. Ed.; Academic
Press: New York, 1968; 183 pp.
14a. Simpson, W.T. J. Am. Chem. Soc. 1951, 73, 5363.
14b. Simpson, W.T. J. Am. Chem. Soc. 1955, 77, 6164.
15. Dole, M.; Fallgatter, M.B.; Katsuura, K. J. Phys. Chem. 1966, 70, 628.
16a. Holroyd, R.A.; Yang, J.Y.; Servedio, F.M. J. Chem. Phys. 1967, 46, 4540.
16b. Holroyd, R.A.; Yang, J.Y.; Servedio, F.M. J. Chem. Phys. 1968, 48, 1331.
17a. Partridge, R.F. J. Chem. Phys. 1970, 52, 1277.
17b. Partridge, R.F. J. Chem. Phys. 1970, 52, 2501.
18. Sauer, M.C.; Mani, I. J. Phys. Chem. 1968, 72, 3856.
19. Zeman, A.; Heusinger, H. Radiochim. Acta. 1967, 8, 149.
20a. Harrah, L.A. Mol. Cryst. Liq. Cryst. 1969, 9, 197.
20b. Wilske, J.; Heusinger, H. J. Polym. Sci. Part A 1969, 7, 995.
21a. Tagawa, S.; Schnabel, W.; Washio, M.; Tabata, Y. Radiat. Phys. Chem. 1981, 18, 1087.
21b. Tagawa, S. Radiat. Phys. Chem. 1986, 27, 455.
21c. Itagaki, H.; Horie, K.; Mita, I.; Washio, M.; Tagawa, S.; Tabata, Y. J. Chem. Phys. 1983, 79,
3996.
21d. Itagaki, H.; Horie, K.; Mita, I.; Washio, M.; Tagawa, S.; Tabata, Y.; Sato, H.; Tanaka, Y.
Macromolecules 1987, 20, 2774.
22. Tagawa, S.; Washio, M.; Tabata, Y. Chem. Phys. Lett. 1979, 68, 276.
23. Rauscher, H.; Ba ssler, H.; Bradley, D.D.C.; Hennecke, M. Phys. Rev. B 1990, 42, 9830.
24. Hennecke, M.; Damerau, T.; Mu llen, K. Macromolecules 1993, 26, 3411.
25. Heun, S.; Mahrt, R.F.; Greiner, A.; Lemmer, U.; Ba ssler, H.; Halliday, D.A.; Bradley,
D.D.C.; Burn, P.L.; Holmes, A.B. J. Phys. Condens. Matter 1993, 5, 247.
26a. Thorne, J.R.G.; Hochstrasser, R.M.; Zeigler, J.M. J. Phys. Chem. 1988, 92, 4275.
26b. Kim, Y.R.; Lee, M.; Thorne, J.R.G.; Hochstrasser, R.M. Chem. Phys. Lett. 1988, 145, 75.
27a. Matsui, Y.; Seki, S.; Iwamoto, T.; Tagawa, S. Chem. Lett. 1998, 861.
27b. Matsui, Y.; Seki, S.; Tagawa, S. Chem. Phys. Lett. 2002, 357, 346.
Copyright 2004 by Taylor & Francis Group, LLC
27c. Seki, S.; Tsuji, S.; Nishida, K.; Matsui, Y.; Yoshida, Y.; Tagawa, S. Chem. Lett. 2002,
1187.
28a. Kouchi, N.; Tagawa, S.; Kobayashi, H.; Tabata, Y. Radiat. Phys. Chem. 1989, 34, 453.
28b. Kouchi, N.; Aoki, Y.; Shibata, H.; Tagawa, S.; Kobayashi, H.; Tabata, Y. Radiat. Phys.
Chem. 1989, 34, 759.
28c. Shibata, H.; Seki, S.; Tagawa, S.; Yoshida, Y.; Ishigure, K. Nucl. Instrum. Methods Phys.
Res. B 1995, 105, 42.
29. Dole, M. The Radiation Chemistry of Macromolecules. Academic Press: New York, 1972.
30a. Charlesby, A. Proc. R. Soc. Lond. A 1954, 222, 60.
30b. Charlesby, A. Proc. R. Soc. Lond., A 1954, 224, 120.
31. Kuhn, W. Berichte der Deutschen Chemischen Gesellschaft 1930, 63, 1503.
32. Flory, P.J. J. Am. Chem. Soc. 1941, 63, 3083, 3091, 3096.
33. Charlesby, A. J. Polym. Sci. 1953, 11, 513.
34a. Saito, O. J. Phys. Soc. Jpn. 1958, 13, 198, 1451, 1465.
34b. Saito, O.; Kang, H.Y.; Dole, M. J. Chem. Phys. 1967, 46, 3607.
35a. Inokuti, M. J. Chem. Phys. 1960, 33, 1607.
35b. Inokuti, M. J. Chem. Phys. 1963, 38, 2999.
36. ODonnell, J.H.; Winzor, C.L.; Winzor, D.J. Macromolecules 1990, 23, 167.
37. Zhang, Y.F.; Ge, X.W.; Sun, J.Z. Radiat. Phys. Chem. 1990, 35, 163.
38. Olejniczak, J.; Rosiak, J.; Charlesby, A. Radiat. Phys. Chem. 1991, 37, 499.
39. Tagawa, S. In Radiation Eects on Polymers: ACS Symposium Series 475; Clough, R.L.,
Shalaby, S.W., Eds.; American Chemical Society: Washington, DC, 1991; 2 pp.
40. Tagawa, S. In CRC Handbook of Radiation Chemistry; Tabata, Y., Ito, Y., Tagawa, S., Eds.;
CRC Press, Inc.: Boston, 1991; Chap. XIV, 739 pp.
41. Tagawa, S. In CRC Handbook of Radiation Chemistry; Tabata, Y., Ito, Y., Tagawa, S., Eds.;
CRC Press, Inc.: Boston, 1991.
42. Yoshida, Y.; Mizutani, Y.; Kozawa, T.; Saeki, A.; Seki, S.; Tagawa, S.; Ushida, K. Radiat.
Phys. Chem. 2001, 60, 313.
43a. Kozawa, T.; Mizutani, Y.; Miki, M.; Yamamoto, T.; Suemine, S.; Yoshida, Y.; Tagawa, S.
Nucl. Instrum. Methods, A 2000, 440, 251.
43b. Kosawa, T.; Saeki, A.; Yoshida, Y.; Tagawa, S. Jpn. J. Appl. Phys. 2002, 41, 4208.
44a. Seki, S.; Kunimi, Y.; Nishida, K.; Yoshida, Y.; Tagawa, S. J. Phys. Chem. B 2001, 105, 900.
44b. Seki, S.; Yoshida, Y.; Tagawa, S.; Asai, K. Macromolecules 1999, 32, 1080.
45a. Tagawa, S.; Schnabel, W. Chem. Phys. Lett. 1980, 75, 120.
45b. Tagawa, S.; Schnabel, W. Makromol. Chem. Rapid Commun. 1980, 1, 345.
45c. Tagawa, S.; Schnabel, W. Polym. Photochem. 1983, 3, 203.
46. Washio, M.; Tagawa, S.; Tabata, Y. Radiat. Phys. Chem. 1983, 21, 239.
47. Okamoto, K.; Kozawa, T.; Yoshida, Y.; Tagawa, S. Radiat. Phys. Chem. 2001, 60, 417.
48. Tagawa, S.; Hayashi, N.; Yoshida, Y.; Washio, M.; Tabata, Y. Radiat. Phys. Chem. 1989, 34,
503.
49. Yoshida, Y.; Ueda, T.; Kobayashi, T.; Tagawa, S. J. Photopolym. Sci. Tech. 1991, 4, 171.
50. Tagawa, S.; Yoshida, Y. Unpublished data.
51. Tabuse, S.; Izumi, Y.; Kojima, T.; Yoshida, Y.; Kozawa, T.; Miki, M.; Tagawa, S. Radiat.
Phys. Chem. 2001, 62, 179.
52a. Saeki, A.; Kozawa, T.; Yoshida, Y.; Tagawa, S. Radiat. Phys. Chem. 2001, 60, 319.
52b. Saeki, A.; Kozawa, T.; Yoshida, Y.; Tagawa, S. Jpn. J. Appl. Phys. 2002, 41, 4213.
53a. Ban, H.; Sukegawa, K.; Tagawa, S. Macromolecules 1987, 20, 177.
53b. Ban, H.; Sukegawa, K.; Tagawa, S. Macromolecules 1988, 21, 45.
53c. Ban, H.; Tanaka, A.; Hayashi, N.; Tagawa, S.; Tabata, Y. Radiat. Phys. Chem. 1989, 34, 587.
54. Watanabe, A.; Komatsubara, T.; Matsuda, M.; Yoshida, Y.; Tagawa, S. Macromol. Chem.
Phys. 1995, 196, 1229.
55a. Seki, S.; Yoshida, Y.; Tagawa, S. Macromolecules 1999, 32, 1080.
55b. Seki, S.; Yoshida, Y.; Tagawa, S. Radiat. Phys. Chem. 2001, 60, 411.
55c. Seki, S.; Kunimi, Y.; Nishida, K.; Yoshida, Y.; Tagawa, S. J. Phys. Chem. B 2001, 105, 900.
Copyright 2004 by Taylor & Francis Group, LLC
55d. Seki, S.; Yoshida, Y.; Tagawa, S. J. Phys. Chem. B 2002, 106, 6849.
56a. Mochida, K.; Hata, R.; Chiba, H.; Seki, S.; Yoshida, Y.; Tagawa, S. Chem. Lett. 1998, 263.
56b. Mochida, K.; Kuwano, N.; Nagao, H.; Seki, S.; Yoshida, Y.; Tagawa, S. Chem. Lett. 1999, 3.
56c. Mochida, K.; Kuwano, N.; Nagao, H.; Seki, S.; Yoshida, Y.; Tagawa, S. Inorg. Chem.
Commun. 1999, 2, 238.
57. Ho, S.K.; Siegel, S. J. Chem. Phys. 1969, 50, 1142.
58. Tabata, M.; Nilsson, G.; Lund, A.; Shoma, J. J. Polym. Sci. Polym. Chem. Ed. 1983, 21, 3257.
59. Ogasawara, M.; Tanaka, M.; Kobayashi, H. J. Phys. Chem. 1987, 91, 937.
60. Szadkowska-Nicze, M.; Kiszka, M.; Mayer, J. J. Polym. Sci. Polym. Chem. Ed. 1983, 21, 3257.
61. Nakano, A.; Okamoto, K.; Kozawa, T.; Tagawa, S. Unpublished data.
62a. Clough, R.L.; Shalaby, S.W. In Radiation Eects on Polymer: ACS Symposium Series 475.
American Chemical Society: Washington, DC, 1991.
62b. Clough, R.L.; Shalaby, S.W. In Irradiation of Polymers: ACS Symposium Series 620.
American Chemical Society: Washington, DC, 1996.
63. Tabata, Y.; Ito, Y.; Tagawa, S. In CRC Handbook of Radiation Chemistry. CRC Press:
Boston.
64. Jonah, C.D.; Rao, B.S.M. In Radiation Chemistry Present Status and Future Trends. Elsevier:
Amsterdam, 2001.
65. Dole, M. In Early Developments in Radiation Chemistry; Kroh, J. Ed.; Royal Society of
Chemistry: London, 1989; Chap. 2. 81 pp.
66. Charlesby, A. In Early Developments in Radiation Chemistry; Kroh, J. Ed.; Royal Society of
Chemistry: London, 1989; Chap. 3. 29 pp.
67. Wach, R.A.; Mitomo, H.; Toshii, F.; Kume, T. J. Appl. Polym. Sci. 2001, 81, 3030.
68. Zhutayeva, Y.A.; Khatipov, S.A. Nucl. Instrum. Methods Phys. Res. B 1999, 151, 372.
69. Woo, L.; Ling, M.T.K.; Ding, S.Y.; Westphal, S.P. Thermochim. Acta 1998, 324, 179.
70a. Oshima, A.; Tabata, Y.; Kudoh, H.; Seguchi, T. Radiat. Phys. Chem. 1995, 45, 269.
70b. Oshima, A.; Ikeda, S.; Seguchi, T.; Tabata, Y. Radiat. Phys. Chem. 1997, 50, 519.
70c. Oshima, A.; Seguchi, T.; Kudoh, H.; Tabata, Y. Radiat. Phys. Chem. 1997, 50, 611.
71. Darwis, D.; Mitomo, H.; Enjoji, T.; Yoshii, F.; Makuuchi, K. J. Appl. Polym. Sci. 1998, 68,
581.
72a. Seki, S.; Kanzaki, K.; Yoshida, Y.; Tagawa, S.; Shibata, H.; Asai, K.; Ishigure, K. Jpn. J.
Appl. Phys. 1997, 36, 5361.
72b. Seki, S.; Maeda, K.; Kunimi, Y.; Yoshida, Y.; Tagawa, S.; Kudoh, H.; Sugimoto, M.; Sasuga,
T.; Seguchi, T.; Iwai, T.; Shibata, H.; Asai, K.; Ishigure, K. J. Phys. Chem. B 1999, 103, 3043.
73. Wei, G.; Qiao, J.; Xuan, H.; Zhang, F.; Wu, J. Radiat. Phys. Chem. 1998, 52, 237.
74. Takashika, K.; Oshima, A.; Kuramoto, M.; Seguchi, T.; Tabata, Y. Radiat. Phys. Chem. 1999,
55, 399.
75. Dong, L.; Hill, D.J.T.; ODonnell, J.H.; Carswell-Pomerantz, T.G.; Pomery, P.J.; Whittaker,
A.K.; Hatada, K. Macromolecules 1995, 28, 3681.
76. For example, Polymeric Materials for Microelectronic Applications, eds. Ito, H.; Tagawa, S.;
Horie, K. American Chemical Society: Washington, DC, 1994.
77. Kozawa, T.; Yoshida, Y.; Uesaka, M.; Tagawa, S. Jpn. J. Appl. Phys. 1992, 31, 4301.
78. Kozawa, T.; Uesaka, M.; Yoshida, Y.; Tagawa, S. Jpn. J. Appl. Phys. 1993, 32, 6049.
79. Watanabe, T.; Yamashita, Y.; Kozawa, T.; Yoshida, Y.; Tagawa, S. Polymeric Materials for
Microelectronic Applications; Ito, H., Tagawa, S., Horie, K., Eds.; American Chemical Society:
Washington, DC, 1994; Chap. 8 110120.
80. Watanabe, T.; Yamashita, Y.; Kozawa, T.; Yoshida, Y.; Tagawa, S. J. Vac. Sci. Technol. B
1994, 12, 3879.
81. Kozawa, T.; Uesaka, M.; Watanabe, T.; Yamashita, Y.; Shibata, H.; Yoshida, Y.; Tagawa, S.
Polymeric Materials for Microelectronic Applications; Ito, H., Tagawa, S., Horie, K., Eds.;
American Chemical Society: Washington, DC, 1994; Chap. 9 121129.
82. Kozawa, T.; Uesaka, M.; Watanabe, T.; Yamashita, Y.; Yoshida, Y.; Tagawa, S. J. Photo-
polym. Sci. Technol. 1995, 8, 37.
Copyright 2004 by Taylor & Francis Group, LLC
83. Nagahara, S.; Yamashita, Y.; Taguchi, T.; Kozawa, T.; Yoshida, Y.; Tagawa, S. Jpn. J. Appl.
Phys. 1996, 35, 6491.
84. Nagahara, S.; Yamashita, Y.; Taguchi, T.; Kozawa, T.; Tagawa, S. J. Photopolym. Sci.
Technol. 1996, 9, 619.
85. Kozawa, T.; Nagahara, S.; Yoshida, Y.; Tagawa, S.; Watanabe, T.; Yamashita, Y. J. Vac. Sci.
Technol. B 1997, 15, 2582.
86. Nagahara, S.; Kozawa, T.; Yamamoto, Y.; Tagawa, S. J. Photopolym. Sci. Technol. 1998, 11,
577.
87. Tsuji, S.; Kozawa, T.; Yamamoto, Y.; Tagawa, S. J. Photopolym. Sci. Technol. 2000, 13, 733.
88. Tagawa, S.; Nagahara, S.; Iwamoto, T.; Wakita, M.; Kozawa, T.; Yamamoto, Y.; Werst, D.;
Trifunac, A.D. Proc. SPIE 3999: California, 2000; 204 pp.
89. International Technology Roadmap for Semiconductors; 2001 Edition; Semiconductor Industry
Association (SIA) et al.
90. Emoto, F.; Gamo, K.; Namba, S.; Samoto, N.; Shimizu, R. Jpn. J. Appl. Phys. 1985, 24,
L809.
91. Fujita, J.; Watanabe, H.; Ochiai, Y.; Manako, S.; Tsai, J.S.; Matsui, S. Appl. Phys. Lett. 1995,
66, 3065.
92. Beckera, E.W.; Ehrfeldb, W.; Hagmannc, P.; Manerd, A.; Mu nchmeyer, D. Microelectron.
Eng. 1986, 4, 35.
93. Ito, H.; Willson, C.G. Polym. Eng. Sci. 1983, 23, 1012.
94. Ito, H. Jpn. J. Appl. Phys. 1992, 31, 4273.
95. Blum, L.; Perkins, M.E.; Liu, H.-Y. J. Vac. Sci. Technol. B 1988, 6, 2280.
96. Fedynyshyn, T.H.; Cronin, M.F.; Szmanda, C.R. J. Vac. Sci. Technol. B 1991, 9, 3380.
97. Fedynyshyn, T.H.; Cronin, M.F.; Poli, L.C.; Kondek, C. J. Vac. Sci. Technol. B 1990, 8, 1454.
98. Novembre, A.E.; Tai, W.W.; Kometani, J.M.; Hanson, J.E.; Nalamasu, O.; Taylor, G.N.;
Reichmanis, E.; Thompson, L.F.; Tomes, D.N. J. Vac. Sci. Technol. B 1991, 9, 3338.
99. Pan, S.-W.; Reilly, M.T.; Taylor, J.W.; Cerrina, F. J. Vac. Sci. Technol. B 1993, 11, 2845.
100. Ban, H.; Nakamura, J.; Deguchi, K.; Tanaka, A. J. Vac. Sci. Technol. B 1991, 9, 3387.
101. Ocola, L.E.; Orphanos, P.A.; Li, W.F.; Waskiewicz, W.; Novembre, A.E.; Sato, M. J. Vac.
Sci. Technol. B 2000, 18, 3435.
102. Nakasugi, T.; Ando, A.; Inanami, R.; Sasaki, N.; Sugihara, K.; Miyoshi, M.; Fujioka, H. Jpn.
J. Appl. Phys. 2002, 41, 4157.
103. Lee, K.Y.; Hsu, Y.; Le, P.; Tan, Z.C.H.; Chang, T.H.P.; Elian, K. J. Vac. Sci. Technol. B
2000, 18, 3408.
104. Rao, V.; Hutchinson, J.; Holl, S.; Langston, J.; Henderson, C.; Wheeler, D.R.; Cardinale, G.;
OConnell, D.; Bohland, J.; Taylor, G.; Sinta, R. J. Vac. Sci. Technol. B 1998, 16, 3722.
105. Fujita, J.; Onishi, Y.; Ochiai, Y.; Matsui, S. Appl. Phys. Lett. 1996, 68, 1297.
106. Robinson, A.P.G.; Palmer, R.E.; Tada, T.; Kanayama, T.; Allen, M.T.; Preece, J.A.; Harris,
K.D.M. J. Phys. D 1999, 32, L75.
107. Ishii, T.; Nozawa, H.; Tamamura, T.; Ozawa, A. J. Vac. Sci. Technol. B 1997, 15, 2570.
108. Wu, H.; Gonsalves, K.E. Adv. Mater 2001, 13, 670.
109. Magee, J.L.; Chattarjee, A. In Kinetics of Nonhomogeneous Processes; Freeman, G.R., Ed.;
John Wiley and Sons: New York, 1987; Chap. 4 171 pp.
110. Katz, R.; Sinclair, G.L.; Waligorski, M.P.R. Radiat. Meas. 1986, 11, 301.
111. Waligorski, M.P.R.; Hamm, R.N.; Katz, R. Radiat. Meas. 1986, 11, 309.
112. Varma, M.N.; Baum, J.W.; Kuehner, A.J. Radiat. Res. 1975, 62, 1.
113. Varma, M.N.; Baum, J.W.; Kuehner, A.J. Radiat. Res. 1977, 70, 511.
114. Wingate, C.L.; Baum, J.W. Radiat. Res. 1976, 65, 1.
115. Wilson, W.E. Radiat. Res. 1994, 140, 375.
116. Seki, S.; Kanzaki, K.; Kunimi, Y.; Tagawa, S.; Yoshida, Y.; Kudoh, H.; Sugimoto, M.;
Sasuga, T.; Seguchi, T.; Shibata, H. Radiat. Phys. Chem. 1997, 50, 423.
117. Koizumi, H.; Ichikawa, T.; Yoshida, H.; Shibata, H.; Tagawa, S.; Yoshida, Y. Nucl. Instrum.
Methods Phys. Res. B 1996, 117, 269.
Copyright 2004 by Taylor & Francis Group, LLC
118. Taguchi, M.; Matsumoto, Y.; Namba, H.; Aoki, Y.; Hiratsuka, H. Nucl. Instrum. Methods
Phys. Res. B 1998, 134, 427.
119. Furukawa, K.; Ohno, S.; Namba, H.; Taguchi, M.; Watanabe, R. Radiat. Phys. Chem. 1997,
49, 641.
120. LaVerne, J.A.; Schuler, R.H. J. Phys. Chem. 1987, 91, 5770.
121. LaVerne, J.A.; Schuler, R.H. J. Phys. Chem. 1996, 100, 16034.
122. Puglisi, O.; Licciardello, A. Nucl. Instrum. Methods Phys. Res. B 1994, 91, 431.
123. Licciardello, A.; Puglisi, O. Nucl. Instrum. Methods Phys. Res. B 1994, 91, 436.
124. Calcagno, L.; Percolla, R.; Foti, G. Nucl. Instrum. Methods Phys. Res. B 1994, 91, 426.
125. Aoki, Y.; Kouchi, N.; Shibata, H.; Tagawa, S.; Tabata, Y. Nucl. Instrum. Methods Phys. Res.
B 1988, B33, 799.
126. Sasuga, T.; Kudoh, H.; Seguchi, T. Polymer 1999, 40, 5095.
127. Sasuga, T.; Kawanishi, S.; Seguchi, T.; Kohno, I. Polymer 1989, 30, 2054.
128. Seki, S.; Shibata, H.; Yoshida, Y.; Ishigure, K.; Tagawa, S. Radiat. Phys. Chem. 1996, 48,
539.
129. Kudoh, H.; Katsumura, Y. Ion-beam radiation chemistry. In Radiation Chemistry Present
Status and Future Trends; Jonah, C.D., Rao, B.S.M., Eds.; Elsevier: Amsterdam, 2001, 3766.
130. Fleisher, R.L.; Price, P.B.; Walker, R.M. Nuclear Tracks in Solids: Principles and Applications;
Fleisher, R.L. Ed.; University of California Press: Berkeley, 1975; 1 pp.
131. Spohr, R. Ion Tracks and Microtechnology; Vieweg: Anfahrt, 1990.
132. Vetter, J.; Spohr, R. Nucl. Instrum. Methods Phys. Res. B 1993, 79, 691.
133. Ferain, F.; Legras, R. Nucl. Instrum. Methods Phys. Res. B 1993, 82, 539.
134. Molarres, M.E.T.; Buschmann, V.; Dobrev, D.; Neumann, R.; Scholz, R.; Schuchert, I.U.;
Vetter, J. Adv. Mater. 2001, 13, 62.
135. Spohr, R.; Reber, N.; Wolf, A.; Alder, G.M.; Ang, V.; Bashford, C.L.; Pasternak, C.A.;
Omichi, H.; Yoshida, M. J. Controlled Release 1998, 50, 1.
136. Reber, N.; Omichi, H.; Spohr, R.; Wolf, A.; Yoshida, M. Nucl. Instrum. Methods Phys. Res.
B 1999, 151, 146.
137. Ferain, F.; Legras, R. Nucl. Instrum. Methods Phys. Res. B 2001, 174, 116.
138. Ferain, F.; Legras, R. Radiat. Meas. 2001, 34, 585.
139. De Pra, L.D.; Ferain, F.; Legras, R.; Champagne, S.D. Nucl. Instrum. Methods Phys. Res. B
2002, 196, 81.
140. Seki, S.; Maeda, K.; Tagawa, S.; Kudoh, H.; Sugimoto, M.; Morita, Y.; Shibata, H. Adv.
Mater. 2001, 13, 1663.
Copyright 2004 by Taylor & Francis Group, LLC
20
Charged Particle and Photon Interactions
in Metal Clusters and Photographic
Systems Studies
Jacqueline Belloni and Mehran Mostafavi
UMR CNRSUPS, Universite Paris-Sud, Orsay, France
1. INTRODUCTION
A century ago, the reduction of silver ions in photographic plates helped W. Roentgen [1],
then H. Becquerel [2], to discover x-rays and radiation of radioactive elements, respec-
tively. Various metal ions were subsequently used widely in aqueous solutions as radical
scavengers and redox indicators of the short-lived primary radiolytic species, allowing
their identication and the calibration of their yield of formation [35]. Some underwent
reduction by g
[6] or pulse radiolysis [7] to the zero-valence metal, to form colloids and
then precipitates [7,8].
Due to the accurate knowledge of the dose used, a control of the progressive extent of
the radiolytic reduction was achieved. However, quite often, puzzling data were reported
when the zero-valent metal was formed, such as an induction time for precipitation, a
sensitivity of the radiolytic yields to the initial presence of added particles, and an unusually
weak reproducibility [9,10]. Moreover, oxidation of silver atoms by molecular oxygen was
observed [7], although the process was thermodynamically improbable for a noble metal
such as silver.
It was also observed, in 1973, that the fast reduction of Cu
+
ions by solvated electrons
in liquid ammonia did not yield the metal and that, instead, molecular hydrogen was
evolved [11]. These results were explained by assigning to the quasi-atomic state of the
nascent metal, specic thermodynamical properties distinct from those of the bulk metal,
which is stable under the same conditions. This concept implied that, as soon as formed,
atoms and small clusters of a metal, even a noble metal, may exhibit much stronger
reducing properties than the bulk metal, and may be spontaneously corroded by the solvent
with simultaneous hydrogen evolution. It also implied that for a given metal the
thermodynamics depended on the particle nuclearity (number of atoms reduced per
particle), and it therefore provided a rationalized interpretation of other previous data
[7,9,10]. Furthermore, experiments on the photoionization of silver atoms in solution
demonstrated that their ionization potential was much lower than that of the bulk metal
[12]. Moreover, it was shown that the redox potential of isolated silver atoms in water must
Copyright 2004 by Taylor & Francis Group, LLC
be lower than that of the silver electrode, Ej(Ag
+
/Ag
met
)=0.79 V
NHE
, by the sublimation
energy of the metal equal to 2.6 V and Ej(Ag
+
/Ag
0
)=1.8 V
NHE
[13]. In the early 1980s,
an increasing number of experimental work emphasized, for metal or semiconductor
particles prepared by various ways in the gaseous and condensed phases, the nuclearity-
dependent properties of clusters of atoms or molecules [1417], theoretically predicted
earlier by Kubo [18]. For two decades, the importance of these fundamental aspects and the
consequences for various applications, from electronics to material science, catalysis,
photography, nonlinear optics, etc. have given rise to a new research domain focused on
the synthesis and properties of these mesoscopic phases.
The radiation-induced method, in the g- or pulse regime [19,20] and, to a certain
extent, the photo-induced method [21], provide a particularly powerful means to produce in
condensed media, metal, and semiconductor clusters frommonomers as precursors, to study
their properties and to understand the exotic phenomena which occur whenever a newphase
of oligomeric particles is formed in the bulk of a homogeneous mother phase [2226],
phenomena which are therefore rather frequent in physics and chemistry. Unlike a recent
reviewconcerning semiconductor clusters [19], the present chapter is specically focused on
metal clusters, induced by ionizing radiation or ultraviolet (UV)visible photons.
2. METAL CLUSTER NUCLEATION AND GROWTH
2.1. Principles of Metal Atom Formation
The specicity of the radiation-induced reduction of metal ion precursors into metal atoms,
which then coalesce into clusters, is attributable to strong reducing agents such as the
Figure 1 Arrhenius plot of the rate constant of the hydrated electron with silver cation Ag
+
.
(From Ref. 50.)
Copyright 2004 by Taylor & Francis Group, LLC
species solvated electrons e
solv
, and H
)=2.87 V
NHE
and Ej(H
+
/H
)=2.3 V
NHE
in water) [27], they easily reduce
metal ions up to the zero-valent state (Fig. 1). The reduction is achieved at room
temperature, so that the inclusion of clusters into thermo-sensitive systems is possible.
Generally, the reduction is achieved under deaerated conditions to avoid a com-
petitive scavenging of e
solv
and H
OH, COO
)
are strong reducing free radicals (Ej((CH
3
)
2
COH/(CH
3
)
2
C
OH)=1.8 V
NHE
[29] at pH 7
and Ej(CO
2
/COO
)=1.9 V
NHE
) [30]. H
radicals (Ej(H
/H
2
)=+2.3 V
NHE
) oxidize
these molecules as well by H abstraction. Alcohol and formyl radicals can reduce, directly,
or via complexation with the ion, the metal ions into lower valencies or into atoms for
monovalent cations.
Some low-valency cations produced in early steps and at partial reduction are
immediately neutralized by the counter anion if the solubility product is quite low, and
they behave indeed as transient monomers able to coalesce into semiconductor clusters [19].
2.2. Cluster Nucleation
As the early species produced after reduction such as metal atoms, dimers and oligomers are
short-lived; time-resolved observations of the reactions of these transients are carried out by
the pulse radiolysis method, coupled with optical absorption or conductivity. Generally,
kinetics are studied in the absence of oxygen or a stabilizer, unless their specic interaction
has to be known. The earliest [7,8,20] and most complete data [3139] on the nucleation
mechanism were obtained on silver clusters. Indeed, silver may be considered as a model
system owing to the one-step reduction of the monovalent silver ions, hydrated or
complexed with various ligands, and to the intense absorption bands of the transient
oligomers (Fig. 2) and nal clusters (Fig. 3). The wavelength of the absorption band maxima
of the atom Ag
0
and of the charged dimer Ag
2
+
in aqueous solutions and the rate constants
of their formation are given in Table 1. Atoms and charged dimers of other metals, such as
Tl [40,41], In [42], Au [43,44], and Cu [42], are formed by homolog reactions (Table 1).
Figure 3 Left: Optical absorption spectrum and STM image of radiation-induced oligomers Ag
7
3+
at partial reduction and stabilized by PA [85,86]. Right: Absorption spectrum of radiation-induced
silver clusters Ag
n
in the presence of EDTA. (a) After partial reduction (10 krad), (b) after 4 days, (c)
after 8 days. Inset: TEM micrographs of (a), (b), (c), and electron diraction pattern of sample c.
(From Ref. 145.)
Copyright 2004 by Taylor & Francis Group, LLC
The band maxima of metal species are dierent in the gaseous and condensed phases.
The optical absorption bands of Ag
0
and Ag
2
+
are highly dependent on the environment
[45]. As shown in Table 1 and Fig. 2, the inuence of the interaction of ligands CN
, NH
3
,
or EDTA with the atom or the dimer is important [46]. The transient product of the
reduction in water of complexed silver ions is not the isolated atom but a complexed silver
atom, Ag
0
(CN)
2
2
[37], Ag
0
(NH
3
)
2
[47], or Ag
0
(EDTA) [37,48], respectively (Fig. 2), as well
as for Au
0
(CN)
2
2
[44] or CuCl
3
3
[42].
The optical absorption spectra depend on the solvent. They are red-shifted with the
decreasing polarity of the solvents as in EDA, liquid NH
3
, where they appear at a longer
wavelength than in water [49], and blue-shifted in methanol [38]. Moreover, the maximum
in NH
3
is red-shifted with the increase of temperature. In water, the Ag
0
band is almost
unchanged in the range 20200 jC, while that of Ag
2
+
is markedly shifted to the red [50].
Electron spin echo modulation analysis of Ag
0
in ice or methanol glasses has concluded to
a charge transfer character to solvent (CTTS) of the absorption band [51].
The multistep reduction mechanism of multivalent cations is known partially from
pulse radiolysis studies [19]. For example, the reduction of Au
III
Cl
4
into Au
II
Cl
3
and the
disproportionation of Au
II
into Au
I
and Au
III
have been directly observed and the rate
constants determined [52]. However, the last step of reduction of Au
I
complexed by Cl
into Au
0
is not observed by pulse radiolysis because the e
aq
k (nm) of M
0
k
M
+
+M
0
k (nm) of M
2
+
Ag
+
3.6 10
10
360 [13,35] 8 10
9
290, 315 [13,35]
Tl
+
3 10
10
450, 260 [40,41] 1.4 10
9
700, 420, 245 [40,41]
In
+
500 [42] 1.5 10
9
310, 460 [42]
Ag(CN)
2
5 10
9
450, 500 [37] 2 10
10
350, 410, 490 [37]
Ag(EDTA)
3
1.7 10
9
400, 450 [37] 1.6 10
9
310, 340, 400, 475 [37]
Ag(NH
3
)
2
+
3.2 10
10
350 [8], 385 [47] 315, 340 [48]
Au(CN)
2
1.1 10
10
420 [43,44]
Ag
I
,PA 3.6 10
9
360 [39] 8.9 10
9
310, 450 [39]
Ag
I
, gelatin 1.1 10
10
360 [34] 1.1 10
10
290, 315, 308, 325 [34]
Cu(Cl)
3
2
2.7 10
10
380 [42] 4.9 10
7
360 [42]
Copyright 2004 by Taylor & Francis Group, LLC
The radius of M
n
increases as n
1/3
. The fast reactions of ion association with atoms or
clusters play an important role in the cluster growth mechanism. First, the homolog charge
of clusters slow down their coalescence. Second, the subsequent reduction in situ of the ions
xed on the clusters favors their growth rather than the generation of new isolated atoms.
The competition between reduction of free ions and of adsorbed ions is controlled by the
formation rate of the reducing radicals, that is, by the dose rate. Coalescence reactions obey
second-order kinetics. Therefore the cluster formation by reduction into isolated atoms
followed by coalescence is predominant at high irradiation dose rate, particularly in the
pulsed regime compared to continuous irradiation at much lower dose rate.
Moreover, almost in all the early steps, the redox potential of the clusters, which
decreases with the nuclearity, is quite negative. Therefore the growth process undergoes
another competition with a spontaneous corrosion by the solvent and the radiolytic protons,
corrosion which may even prevent the formation of clusters, as mostly in the case of
nonnoble metals. Monomeric atoms and oligomers of these elements are so fragile to reverse
oxidation by the mediumthat H
2
is evolved and the zerovalent metal is not formed [11]. For
that reason, it is preferable in these systems to scavenge the protons by adding a base to the
solution and to favor the coalescence by a reduction faster than the oxidation [53].
Figure 4 Growth kinetics of silver clusters observed through their absorbance at 400 nm in the
presence of (a) cyanide or (b) sulfate. (From Ref. 54.)
Copyright 2004 by Taylor & Francis Group, LLC
After the formation of Ag
0
and Ag
2
+
, followed by the reaction of Ag
2
+
with Ag
+
yielding Ag
3
2+
(k
max
=315 and 260 nm) [35], or its dimerization into Ag
4
2+
(k
max
=265 nm),
the multistep coalescence of oligomers result in clusters of increasing nuclearity Ag
n
. The
absorption spectrum is shifted to the surface plasmon band of Ag
n
at 380400 nm, the
coalescence rate depending on the ligand [54]. At the same initial concentration of atoms,
the plateau is reached at an almost 10
3
longer time for Ag
n
(CN
) [54] than Ag
n
(SO
4
2
) [31]
(Fig. 4). In methanol, the coalescence rate constants are also much lower [38]. Indeed, it is
known, according to the Mie theory [55] and its extension [56], that the interaction of light
with the electrons of small metal particles results in an absorption band whose shape and
intensity depend on the cluster size, the complex dielectric constants of the metal and the
environment (solvent, ligand, molecules adsorbed). During the coalescence, the total
amount of silver atoms formed by the pulse is constant, but they are aggregated into
clusters of increasing nuclearity n with a decreasing concentration. Thus the absorbance
increase observed at 400 nm (Fig. 4) is assigned to the increase with the nuclearity of the
extinction coecient per silver atom. It was shown from the kinetics analysis that the
plasmon band is totally developed with the constant value per atom e = 1.510
4
l mol
1
cm
1
beyond n=13. [57] Then, the coalescence into larger clusters is still continuing,
although the spectrum is unchanged.
3. REACTIVITY OF TRANSIENT METAL CLUSTERS
A number of rate constants for reactions of transients derived from the reduction of metal
ions and metal complexes were determined by pulse radiolysis [58]. Because of the short-
lived character of atoms and oligomers, the determination of their redox potential is
possible only by kinetic methods using pulse radiolysis. In the couple M
n
+
/M
n
, the reducing
properties of M
n
as electron donor as well as oxidizing properties of M
n
+
as electron
acceptor are deduced from the occurrence of an electron transfer reaction with a reference
reactant of known potential. These reactions obviously occur in competition with the
cascade of coalescence processes. The unknown potential Ej(M
n
+
/M
n
) is derived by
comparing the action of several reference systems of dierent potentials.
3.1. Redox Potentials of Metal Atoms and Dimers
The oxidation reactions of Ag
0
and Ag
2
+
[7,20], of Au
0
(CN)
2
2
[44], or of Tl
0
and Tl
2
+
[41,59], by even mild oxidizing molecules (O
2
, CCl
4
, CH
3
NO
2
, N
2
O, Fe
III
, for example) were
observed by pulse radiolysis. The high rate constants found indicate the strong electron
donating character of atomic silver, gold, or thallium. This conrms the evaluation of
Ej(Ag
+
/Ag
0
)=1.8 V
NHE
[13] and Ej(Tl
+
/Tl
0
)=1.9 V
NHE
[41], which were derived
from the dierence between the metal electrode potential, Ej(M
+
/M
met
), and the metal
sublimation energy (Table 2). More generally, because of the high value of the sublimation
energy, the potential of any metal Ej(M
+
/M
0
) is expected to be quite negative.
Table 2 Redox Potentials (V
NHF
) of the Hydrated and Complexed M
I
/M
0
Couple in Water
Couple
M
I
/M
o
Au
+
/
Au
0
Cu
+
/
Cu
0
Tl
+
/
Tl
0
Ag
+
/
Ag
0
Ag(CN)
2
/
Ag
0
(CN)
2
2
Ag(NH
3
)
2
+
/
Ag
0
(NH
3
)
2
Ag(EDTA)
3
/
Ag
0
(EDTA)
4
Ej 1.4 [44] 2.7 [13] 1.9 [41] 1.8 [13] 2.6 [36,61] 2.4 [47] 2.2 [37]
Copyright 2004 by Taylor & Francis Group, LLC
The redox potential values of all metal atoms, except alkaline and earth-alkaline
metals [60], are higher than that of Ej(H
2
O/e
aq
)=2.87 V
NHE
. However, some complexed
ions are not reducible by alcohol radicals under basic conditions and thus Ej(M
+
L/
M
0
L)<2.1 V
NHE
(Table 2). The results were conrmed by SCF calculations of Ag
+
L
and Ag
0
L structures associated with the solvation eect given by the cavity model for
L=CN
[61] or NH
3
[47], respectively.
3.2. Redox Potentials of Oligomeric Metal Clusters
For clusters of higher nuclearity too, the kinetic method for determining the redox potential
Ej(M
n
+
/M
n
) is based on electron transfer, for example, from mild reductants of known
potential which are used as reference systems, towards charged clusters M
n
+
. [31] Note that
the redox potential diers from the microelectrode potential Ej(M
+
, M
n
/M
n
) by the
adsorption energy of M
+
on M
n
(except for n=1). The principle [31] is to observe at which
step n of the cascade of coalescence reactions, a reaction of electron transfer occurring
between a donor S
is
produced by the same pulse as the atoms M
0
, the radiolytic radicals being shared between
M
+
and S. One form at least in the couple S/S
to M
n
+
, which is thus indirectly reduced, and the S
behaves as an electron relay. After the adsorption of another cation, the potential of the
new reduced cluster Ej(M
n+1
+
/M
n+1
) is more positive and another electron is transferred
from S
), and it should be
taken into account in the simulation. This corrosion process is generally negligible because
Copyright 2004 by Taylor & Francis Group, LLC
the step of M
+
release is not fast enough. However, it is observed when the coalescence is
also slow, for example in the case of ions included in cavities of a Naon membrane, where
the coalescence of Ag
n
is controlled by the very slow diusion between the cavities (10
4
l mol
1
sec
1
) [64,65]. It has been found that the smallest clusters could be oxidized by the
protons H
3
O
+
, which are highly concentrated at the cavity surface. In contrast, when
clusters reach by coalescence the critical nuclearity for which their potential is higher than
Ej(H
3
O
+
/H
2
)=0 V
NHE
, they escape corrosion and are observed by optical absorption. The
Figure 5 Comparison between simulated and experimental decay of MV
+.
in the presence of
growing clusters Ag
n
,
CN
or Au
I
(CN)
2
in the
presence of the methylviologen redox couple MV
2+
/MV
+.
, oxidation of the smallest
clusters is also observed, because coalescence in cyanide solutions is slow (Fig. 4) [54,66].
While supercritical silver clusters (n > 6 F 1) (Table 3) accept electrons from MV
+.
with a
progressive increase of their nuclearity, the subcritical clusters undergo a progressive
oxidation by MV
2+
(Fig. 5). Actually, the reduced ions MV
+.
so produced act as an
electron relay favoring the growth of large clusters at the expense of the small ones. If
cyano-silver clusters require a higher nuclearity than aquo clusters to react with MV
+.
(Table 3), it means that the ligand CN
n
clusters Ej(Ag
n
+
/Ag
n
)
aq
together with the eect on ionization potentials IP
g
(Ag
n
) of
bare silver clusters in the gas phase [67,68]. The asymptotic value of the redox potential is
reached at the nuclearity around n = 500 (diameter c 2 nm), which thus represents, for
the system, the transition between the mesoscopic and the macroscopic phase of the bulk
metal. The density of values available so far is not sucient to prove the existence of odd
even oscillations as for IP
g
. However, it is obvious from this gure that the variation of Ej
and IP
g
do exhibit opposite trends vs. n, for the solution (Table 5) and the gas phase,
respectively. The dierence between ionization potentials of bare and solvated clusters
decreases with increasing n as n
1/3
which corresponds fairly well to the solvation free
energy of the cation M
n
+
deduced from the Born solvation model [45] (for the single atom,
the dierence of 5 eV represents the solvation energy of the silver cation) [31].
A variation of Ej(Cu
n
+
/Cu
n
), similar to that of silver, is expected for copper because
the atomic potential is Ej(Cu
+
/Cu
0
)=2.7 V
NHE
[13] (Table 2), that of Cu, is Ej(Cu
7
+
/
Cu
7
) = 0.4 V
NHE
[69] (Table 5), and the bulk electrode potential is at 0.52 V
NHE
.
Table 3 Nuclearity-Dependence of Ej(M
n
+
/M
n
) (in V
NHE
) [63]
Reference System
(Electron Donor)
Ej(S/S
)
(Ref. Couple)
Metal Cluster
(Electron Acceptor) n
c
(Reduced Atoms/Cluster)
Ni
+
1.9 Ag
n
1
(CH
3
)
2
C
.
OH (pH 5) 1.8 Ag
n
1
SPV
.
0.41 Ag
n
4
Cu
+
0.16 Ag
n
11
Q
.
(pH=4.8)
a
0.22 Ag
n
85 F5
Q
.
(pH=3.9) 0.33 Ag
n
500 F30
MV
+
.
0.41 Ag
n.CN
56
MV
+.
0.41 Ag
n.PA
4
MV
+
.
0.41 Cu
n.Cl
6 F1
a
Q
.
is the semiquinone of naphtazarin.
Copyright 2004 by Taylor & Francis Group, LLC
Some calculations [70] were made to derive the microelectrode potential Ej(M
+
,
M
n1
/M
n
) for silver and copper from the data in the gas phase [nuclearity-dependent M
M
n
bond energy and IP
g
(M
n
)]. The potential Ej(M
+
,M
n1
/M
n
) presents oddeven
oscillations with n (more stable for n even) as for IP
g
, but again, the general trends are
opposite, and an increase is found in solution because to the solvation energy.
3.3. Nucleation, Growth, and Reactivity of Transient Bimetallic Clusters
The spectra of silver and gold nanoclusters are intense and distinct (Table 4). They are
thus particularly suitable to detect the evolution of a cluster composition during the
construction of a bimetallic cluster in mixed solution. The system studied by pulse
radiolysis was the radiolytic reduction of a mixed solution of two monovalent ions, the
cyano-silver and the cyano-gold ions Ag(CN)
2
and Au(CN)
2
and Au(CN)
2
at
an alloyed nucleus surface as for monometallic clusters (Section 20.4.4). The reaction
competing with the intermetal exchange is now the chemical reduction, which can be
very fast provided nuclei are present. In a pulse radiolysis study, the transient reduced
viologen formed by an electron pulse has been observed to react with the small alloyed
(AuAg)
n,CN
clusters and the developed alloyed clusters are stable toward segregation [66].
Their nal size is quite large because the ions reduced by the donor are concentrated on a
small number of nuclei. Moreover, it was found [66] that the redox potential of such
critically alloyed AgAu clusters is almost the same (0.4 V
NHE
at n
c
= 6) as that of the
more noble metal (silver in cyanide environment) [54].
4. SYNTHESIS OF STABLE MONOMETALLIC NANOCLUSTERS
Stability means that clusters do not undergo coalescence nor corrosion by the medium, at
least in the absence of oxygen. The quite negative value of Ej(M
I
/M
0
) and the dependence
of the cluster redox potential on the nuclearity have crucial consequences in the formation
of early nuclei, their possible corrosion or their growth. As an example, the faster the
coalescence, the lower is the probability of corrosion of the small clusters by the medium.
The property of stability oers the means to apply to these clusters a larger amount of
suitable characterization techniques than to transient oligomers.
Various applications (Sections 20.5.3 and 20.7.3) require that the synthesis of stable
clusters be as small as possible, and the coalescence should be prevented somewhat by a
stabilizer or a support. Moreover, stable metal clusters can be used as precursors to
synthesize, by functionalizing their surface, highly elaborated materials, acquiring specic
properties [21], For example, fullerene- [71,72] or pyrene-functionalized gold clusters [73]
are used as self-assembled photoactive antenna units which could serve as building blocks
in the design of light-harvesting systems.
4.1. Cluster Size Stabilization
Metal atoms formed by irradiation or any other method tend to coalesce into oligomers,
which themselves progressively grow into larger clusters and eventually into precipitates,
as found in early radiolytic experiments. However, for studying stable clusters or for
applications, the coalescence must be limited as in other chemical methods of cluster
synthesis [74], either by adding a polymeric molecule acting as a cluster stabilizer or by
generating the atoms in a conned structure where diusion is prevented. Some ligands
such as CN
become particularly
sensitive to oxidation in air, because the Fermi level is shifted to a more negative potential
by the electron donation [24]. Likewise, binding of electronegative groups to the gold
surface leads to damping of the surface plasmon band of gold nanoparticles [127]. It was
shown that (SCN)
2
j
[128].
Although the mechanism of the photo-induced generation of mono- and bimetallic
metal clusters, except for the photographic application (Section 20.6), has been studied
with considerably less detail than for the radiolytic route, some stable clusters, mostly of
noble metals (Ag, Au, Pt, Pd, Rh), have also been prepared by UV excitation of metal ion
solutions [129141]. Generally, halides and pseudo-halides counter anions are known to
release, when excited, solvated electrons, which reduce the metal ions up to the zerovalent
state. Oxalate excitation yields the strong reducing carbonyl radical COO
.
[30]. Photo-
sensitizers are likewise often added [142]. Metal clusters are photo-induced as well at the
surface of photo-excited semiconductors in contact with metal ions [143,144].
4.4. Developed Clusters
Actually, the kinetics study of the redox potential of transient clusters (Section 20.3.2) has
shown that beyond the critical nuclearity, they receive electrons without delay from an
electron donor already present. The critical nuclearity depends on the donor potential and
then the autocatalytic growth does not stop until the metal ions or the electron donor are not
exhausted (Fig. 8c). An extreme case of the size development occurs, despite the presence of
the polymer, when the nucleation induced by radiolytic reduction is followed by a chemical
reduction. The donor D does not create new nuclei but allows the supercritical clusters to
develop. This process may be used to select the cluster nal size by the choice of the
radiolytic/chemical reduction ratio. But it also occurs spontaneously any time when even a
mild reducing agent is present during the radiolytic synthesis. The specicity of this method
is to combine the ion reduction successively:
(i) By radiolytic reduction. The atoms independently formed in the bulk coalesce
into oligomers.
(ii) By chemical reduction beyond a certain oligomer nuclearity. This chemical
agent is unable thermodynamically to reduce the ions directly into atoms
(Section 20.3.1), but it achieves the reduction of the rest of the ions after they
have been adsorbed on the radiation-induced clusters acting as nuclei of
catalytic reduction and growth. Note that the polymer restricts the coalescence
of the nuclei or of the nal particles but does not prevent at all the development
by diusion and adsorption of excess ions and electron transfer from the donor
as shown for some of the following systems.
The nal silver cluster diameter increases, at given initial Ag
+
and PVA concen-
trations, for example from 15 to 50 nm (n 50 times larger), when the part of reduction is
increasingly achieved by the donor SPV
ions accumulate and the cluster appearance is delayed by an induction time [108]. With
higher doses, Au
I
ions become more abundant than Au
III
and are also reduced into atoms
and clusters. In the presence of alcohols or of preformed clusters, Au
I
ions do not
accumulate because the potential order of (Au
I
/Au
0
) and (Au
II
/Au
I
) couples is now
inverted as a result of complexation or adsorption, respectively. Au
I
ions are thus allowed
to disproportionate, giving rise to atoms and growing clusters. In addition, a very slow
chemical reduction by the alcohol is found, which occurs exclusively at the surface of the
clusters formed by irradiation so that all gold ions may be reduced, whatever the initial
dose [108]. As seen by AFM observations, the cluster size remains constant and their
concentration increases with dose during the g-irradiation, while during the postirradia-
tion reduction by alcohol the size increases at constant density of particles. The lower the
dose, the larger the nal size of the homodisperse cubic crystallites, which may range from
10 to 500 nm, depending on the respective parts of radiolytic and chemical types of
reduction.
Gold ions Au
I
(CN)
2
in the
presence of alcohol have shown that, rst, the semiconductor spectrum at 360 nm is
bleached, then silver atoms and clusters at 450600 nm are formed [164,165]. The electron
Copyright 2004 by Taylor & Francis Group, LLC
Figure 9 Left: Schematic representation of chemical or UV-induced synthesis of Pt nanowires and
nanoparticles in zeolites FSM-16. Right: Comparison of temperature dependence of magnetization
for Pt nanowires and nanoparticles in FSM-16. (From Ref. 137.)
Figure 10 Size-dependence of the melting point and diusion coecient of silica-encapsulated
gold particles. The dotted curve is calculated by the equation of Buat and Borel. The bulk melting
temperature of Au is indicated by the double arrow as T
m
(l). The solid curve (right-hand side axis)
is a calculated Au self-diusion coecient. (From Ref. 146.)
Copyright 2004 by Taylor & Francis Group, LLC
transfer between the couple MV
2+
/MV
+.
and Ag
n
clusters formed on an AgCl crystallite
was studied by pulse radiolysis [166]. The coalescence of Ag atoms at the AgCl surface is
slow so that, as in the presence of CN
, Au(CN)
2
, Au(CN)
2
EB, 7.9 10
3
AgAu, Alloy [102]
Ag
+
, AuCl
4
, PtCl
6
2
EB, 7.9 10
3
; g, 0.540 Au/Pt, Bilayer [184]
Au(CN)
2
, PtCl
6
2
EB, 7.9 10
3
; g, 0.540 Pt/Au, Bilayer [184]
Cu
2+
, Cd
2+
g, 0.42, 0.48 CuCd, Alloy [83,182]
Cu
2+
, Pd
2+
g, 30 CuPd, Alloy [53]
AuCl
4
, Pd
2+
g 35 Au/Pd, AuPd, Bilayer [181]
AuCl
4
, Pd
2+
EB, 7.9 10
3
Au/Pd, AuPd, Alloy [181]
Ag
+
,Rh
2+
g, 35 Rh, Ag, Segregation [185]
Ag
+
, Rh
2+
EB, 7.9 10
3
Rh, Ag, Segregation [185]
Copyright 2004 by Taylor & Francis Group, LLC
structure of the composite cluster should be derived indeed from the evolution observation
of the absorption spectrum at increasing dose during the cluster building [102,180,181].
Because the spectrumis attributable to surface phenomena, it changes with dose if one metal
is preferentially reduced rst. The surface composition of a bilayered cluster is shifted at
increasing dose from the spectrum of the more noble metal to that of a cluster coated by the
second. In contrast, the spectrum shape of alloyed clusters is unchanged at increasing dose
while the intensity increases. X-ray analysis of clusters at partial reduction (increasing dose)
clearly indicates at each step whether atoms of only one or of both types of metals are present
in the particles [102]. In the case of an alloyed cluster, electron and X-diraction methods
may reveal superlattices that correspond to a perfectly ordered atom arrangement
[4,99,140]. Provided that the lattice constants of the metals are somewhat dierent, the
distances between atomic plans in HRTEM images may indicate the possible alloyed
character of clusters.
For example, when the mixed solution of Ag(CN)
2
and Au(CN)
2
is irradiated by
g-radiolysis at increasing dose, the spectrum of pure silver clusters is observed rst at
400 nm, because Ag is more noble than Au due to the CN
, Ag
+
, gold initially appears at
520 nm. Therefore, Ag
+
ions essentially act as an electron scavenger, and as an electron
relay toward more noble gold ions as far as gold ions are not totally reduced. Then silver-
coated gold clusters are formed and the maximumis shifted to 400 nm, which is that of silver
(Fig. 11) [102]. But at higher g- or EB dose rate (irradiation time of a few seconds), the
electron transfer is too slowto compete with coalescence and the spectrumof alloyed clusters
Figure 11 Evolution with increasing irradiation dose of the absorption spectra of a mixed solution
AuCl
4
/Ag
+
(50/50). Left: with PVA at a dose rate of 3.8 kGy hr
1
. Right: 35 kGy hr
1
. Dotted
line: spectrum with PA. Inset: Evolution with increasing dose of the absorbance at k=480 nm.
(From Ref. 102.)
Copyright 2004 by Taylor & Francis Group, LLC
surrounded by Cl
develops at 480 nm without any shift from the lowest doses (Fig. 11).
Once this quenching radiolytic reduction has consumed all the ions, the intermetal displace-
ment becomes in fact excluded at roomtemperature. In both cases, low- and high-dose rates,
a transverse analysis of a single cluster in a scanning transmission electron microscope
equipped with a eld emission gun and x-ray detector conrmed the cluster composition in
depth, bilayered and alloyed, respectively [179].
The same evolution of the absorption spectrum with the dose has been found in a
high dose rate for various values of the Ag and Au ion fraction in the initial solution.
Clusters Ag
1x
Au
x
are alloyed with the same composition. The maximum wavelength k
max
and the extinction coecient e
max
of the alloy depend on x. The experimental spectra are
in good agreement with the surface plasmon spectra calculated from the Mie model at x
values for which optical data are available (Fig. 12) [102]. Similar calculations were carried
out for the alloy Ag
x
Pd
1x
obtained at a moderate dose rate [180].
Figure 12 Top: Maximum wavelength of the plasmon band of alloyed goldsilver clusters as a
function of the mole fraction x of gold in alloyed goldsilver clusters, produced at the dose rate 7.9
MGy hr
1
and the dose 20 kGy. . Experiments, 5 calculated values by Mie model. Bottom:
Extinction coecient at the maximum of the plasmon band as a function of the mole fraction x of
gold in alloyed gold-silver clusters. . Experiments, 5 calculated values from Kreibig equation [74]
with r = 5 nm; D with r = 3 nm. (From Ref. 102.)
Copyright 2004 by Taylor & Francis Group, LLC
The same contrast between bilayered and alloyed clusters of AuPd has been observed
at low- and high-dose rate, respectively [180], by using spectrophotometry, EDAX, and XPS
analysis at partial and complete reduction.
Other examples of bimetallic clusters are presented in Table 5. Depending on the rate
of the intermetal electron transfer competing with the dose rate-dependent coalescence
(Fig. 13 and Table 5):
1. Alloying may occur spontaneously, even at low reduction rate (g-irradiation
with low dose rate) [53], as for alloyed Cu3Pd [53], CuPd [53], NiPt [53], CuCd
[84,182], AgCd [177], Ag
x
Pd
1x
[180], and AgPt clusters [183].
2. Alloying may require higher g-dose rates as for Ag
x
Au
1x
[102].
3. Alloying is only obtained through a short and intense irradiation provided by an
electron accelerator as for Ag
x
Au
1x
[102], and AuPd [180].
4. Alloying is even not obtained. In extreme cases, the electron transfer is achieved
within times shorter than the irradiation time required for complete reduction
Figure 13 Scheme of the inuence of the dose rate on the competition between the inter-metal
electron transfer and the coalescence processes during the radiolytic reduction of mixed metal ion
solutions. Sudden irradiation at high dose rates favor alloying, whereas low dose rates favor core
shell segregation of the metals because of metal displacement in the clusters.
Copyright 2004 by Taylor & Francis Group, LLC
(a few seconds with EB) and, in spite of the very fast reduction, metals are
segregated into coreshell structures (Au/Pt, [184]) or into distinct clusters (Ag
n
and Rh
n
) [185]. For multivalent ion precursors, intermetal electron transfer
reactions may also occur between the transient lower valency ions.
5.3. Catalytic Properties of Metal Clusters
One of the important applications of mono- and multimetallic clusters is to be used as
catalysts [186]. Their catalytic properties depend on the nature of metal atoms accessible to
the reactants at the surface. The possible control through the radiolytic synthesis of the
alloying of various metals, all present at the surface, is therefore particularly important for
the catalysis of multistep reactions. The role of the size is twofold. It governs the kinetics
by the number of active sites, which increase with the specic area. However, the most
crucial role is played by the cluster potential, which depends on the nuclearity and controls
the thermodynamics, possibly with a threshold. For example, in the catalysis of electron
transfer (Fig. 14), the cluster is able to eciently relay electrons from a donor to an
acceptor, provided the potential value is intermediate between those of the reactants [49].
Below or above these two thresholds, the transfer to or from the cluster, respectively, is
thermodynamically inhibited and the cluster is unable to act as a relay. The optimum
range is adjustable by the size [63].
Figure 14 Mechanism of catalytic electron transfer involving metal clusters as relay. The thermo-
dynamic conditions to be fullled are that the cluster redox potential be higher than the donor D and
lower than the acceptor A potential, which implies that the cluster itself is a size range that oers the
ecient redox potential. (From Ref. 63.)
Copyright 2004 by Taylor & Francis Group, LLC
Clusters, as possible catalytic reactors, are perfectly dispersed in solutions. They are
thus suitable systems for observing, under quasi-homogeneous conditions by time-
resolved techniques, the kinetics of catalyzed electron transfer, which would be inacces-
sible on a solid catalyst. It was demonstrated that the reaction of radiation-induced free
radicals CO
2
.
and (CH
3
)
2
C
by Pt clusters in solution or
supported on colloidal TiO
2
particles [189], have been studied by pulse radiolysis.
In some examples, the reaction catalyzed by radiation-induced clusters was studied
during long periods under the conditions of a pilot system to compare their properties with
catalysts prepared by chemical reduction. They easily fulll the conditions of controlled,
nanometric and homodisperse size, of alloying for multimetallic clusters and of strong
grafting on various supports. They were found ecient catalysts, for example, in the case of
alumina-supported iridium clusters for hydrazine decomposition in thrusters developed for
spacecraft orbit and attitude control systems [88]. Radiation-induced mono- and multi-
metallic clusters imbedded in NaonR [64] or grafted on metal [190], on SnO
2
transparent
counterelectrodes [152,152], or on carbon electrodes [25,100] were found to be quite active
in electrochemical, photovoltaic, and fuel cells (methanol or hydrogen oxidation), respec-
tively. Multimetallic clusters deposited on gelatine silver bromide emulsions are excellent
catalysts for silver reduction in the oset process [191] The catalytic performances of photo-
induced Pt nanowires in zeolites were tested by using the watergas shift reaction [137].
6. PHOTOGRAPHY
Silver photography [192195], and later, radiography [1,2], are the oldest applications of
the unique properties of oligomers, long before the mechanisms were understood. The time-
resolved techniques were particularly helpful to improve our understanding of the intimate
reaction process. Since its invention, innumerable improvements to photography have been
introduced to capture an image as close as possible to our vision of the light intensities and
the colors of a scene, and to enhance the photosensitivity of the photographic layer.
However, the basic principle of silver photography lies on the photosensitivity to visible
light of minute silver halide crystals, which are xed in a gelatine layer as a mosaic. Each
crystal will behave as a picture element (pixel). During exposure, the light generates clusters
of a few silver atoms on the crystals. Their ensemble constitutes the latent image which is of
an extremely weak intensity and therefore invisible. Then the development consists of
converting chemically into metal particles the crystals containing a cluster with a super-
critical number of photo-induced silver atoms and of catalytically transforming the latent
image into a visible picture. In the last step of xing, undeveloped crystals which are still
photosensitive should be eliminated and are dissolved by complexation. The result is the
negative image, which is then used as a mask to produce the positive image.
6.1. Photographic Development
It is well known from photographic experience that critical nuclearity, which is required
for the development, depends on the redox potential of the developer (it is an electron
Copyright 2004 by Taylor & Francis Group, LLC
donor in excess in the development bath). Subsequently, the autocatalytic growth does not
stop until all the metal ions of the exposed crystals containing supercritical clusters are not
exhausted and the crystals transformed into silver metal grains. For over a century,
numerous theoretical models have been proposed, without success [196,195], to explain
how the supercritical clusters created by the light act as nuclei to catalyze development.
Trautweiler [197] had the intuition in his speculative model that the ionization potential of
supercritical sizes should lie above that of the developer. However, the only data then
available did not conrm this view.
Actually, the kinetic study of the cluster redox potential by pulse radiolysis [31]
(Section 20.3.2) somewhat mimics the process of the black-and-white photographic
development, except that clusters are free in the solution (not xed on AgBr crystals),
and that they are produced by ionizing radiation (as in radiography and not by visible
photons; but the last choice had been incompatible with the time-resolved optical detection
in the visible. Beyond the critical nuclearity, they receive electrons without delay from the
developer already present (actually, the photographic development is achieved in a delayed
step).
However, the similarity of the features presented by the processes in both cases,
reduction catalyzed by supercritical clusters in solution and photographic development on
AgBr crystals, allows us to extend to photographic development the same mechanism
demonstrated by pulse radiolysis [31,198]. Consequently, the mechanism results from the
increase of the cluster redox potential with the nuclearity when the developer solution is in
contact with silver bromide crystals. It explains the existence of a critical size in the
photographic development, and the threshold of the discrimination between underexposed
and exposed crystals for complete reduction is imposed by the rst monoelectronic redox
potential of the developer. It also explains the sensitizing eect [66] of gold ions more
noble than Ag
+
, and the fragility to oxidation of the smallest clusters of the latent image
[5]. With standard developers and sulde-gold sensitization, the amplication of the
cluster size by the development is from the critical number [194] n
c
=3 to n=10
89
, that
is the total number of silver ions in the AgBr crystal (about 1 Am).
Apart from the development in photography, most of nucleation and growth
mechanisms based on a chemical reduction (Section 20.4.4) behave as development
processes, and are likewise controlled by the nuclearity dependence of the cluster redox
potential and by the potential of the electron donor.
6.2. Enhancement of Photographic Sensitivity
According to the GurneyMott model [199], the primary light eect on a AgBr crystal is to
produce as many electronhole pairs (e
h
+
) as photons absorbed. The hole corresponds
to the electron vacancy created by the electron photoejection. The electron reduces a silver
cation into an atom, generally close to a sensitizer adsorbed at specic surface sites of
AgBr (sulde or/and gold centers). Then a cation adjacent to the atom and constituting a
charged dimer is reduced by another electron, and so on, the result being a cluster of a few
atoms. But an important part of electrons are lost by direct recombination with the parent
holes, before these diuse to the surface where they are irreversibly scavenged by gelatine
and additives (Fig. 15, top, left). Holes are also able to oxidize the newly formed silver
atoms, so counterbalancing somewhat the reduction by electrons (indirect recombination).
If R is the fraction of initial pairs lost by both types of recombination, and U
theor
=
1 pair/photon the theoretical quantum yield, the eective quantum yield U
e
is U
e
=
U
theor
(1R). In sulde/gold sensitized crystals, the yield is about U
e
= 0.100.30
Copyright 2004 by Taylor & Francis Group, LLC
(the fraction R is thus 0.900.70). Thus, apart from the absorption properties and the area
of the crystal, the photosensitivity of a lm depends directly on the quantum yield of the
cluster formation.
A recent approach to enhance the sensitivity was to specically scavenge the holes,
faster than their possible recombination with the electrons or the atoms, with the help of a
dopant [200]. In addition to a small size and an ionic character to allow its inclusion in the
AgBr crystal, the dopant should obey other strict criteria concerning the redox potential.
The dopant should be a bielectronic donor with a very weak rst redox potential, in order to
let the dopant scavenge the holes, without spontaneously reducing free Ag
+
and producing
fogging in the dark, and with a very negative second redox potential, to avoid a possible
hole-like behavior of the dopant oxidized form, so blocking any reversible oxidation
(Fig. 15, top, center). In addition, a second silver ion can be reduced (Fig. 15, top, right).
From radiation chemical knowledge, the anion formate HCO
2
)=1.07 V
NHE
and Ej(CO
2
/CO
2
.
)=1.9 V
NHE
in water)
[30].
Sensitometric tests at variable light intensity I under conditions of photography,
achieved on emulsions doped at the relative concentration of 10
6
mol HCO
2
per mol
Ag
+
, conrmed the photo-induced bielectronic transfer (Fig. 15, bottom) [200]. The
emulsion is completely stable in the dark. The number of photons required to induce
development of the same grain population fraction is 5 times less (after immediate
development) or 10 times less (development delayed by 20 min after exposure) in doped
than in undoped emulsions where U
e
=0.20. The quantum yield is thus close to the
theoretical limit U
e
=1 atom/photon in immediately developed doped emulsion (R=0),
and U
e
=2 atom/photon if the development is delayed for 20 min, and the slow additional
reduction of a silver ion by CO
2
.
can occur, thus doubling the gain. An important sen-
sitivity enhancement is similarly observed for dye-sensitized crystals [200]. These doping
studies [200,201] not only provide a better understanding of the mechanism of the latent
image formation, but also oer a promising route for improving the performance of all
kinds of silver emulsions, for black-and-white and color photography, radiography, holog-
raphy, etc.
7. CLUSTER FORMATION AND CLUSTER EXCITATION BY POWERFUL
LASERS
Metal clusters exhibit intense optical absorption due to the surface plasmon band, some
in the visible spectrum (Table 4). The response to light excitation is linear as far as the
source is of weak intensity. However, other phenomena appear when pulses of extremely
high intensity are used. A burst activity has been seen in recent years in the area of inter-
action of light from powerful laser pulses with metal in order to produce nanoparticles by
laser ablation, or with nanoparticles in order to let them absorb suddenly a high amount of
energy and to observe the transient and the irreversible eects on the clusters. Eventually, a
change in the morphological shape induced by the laser or a fragmentation of nanoclusters
may occur (Fig. 16).
7.1. Nanoparticles Formation by Laser Ablation
The method of laser ablation from a metal has been used to prepare nanoparticles dis-
persed in a solution [202207]. Recently, colloidal gold nanoparticles in water having an
average diameter of 5.5 nm were prepared by laser ablation at 1064 nm (800 mJ/cm
2
) from
a gold metal plate [208]. The nal nanoparticle size depends on the laser uence and the
stabilizer concentration (Fig. 17).
7.2. Nanoparticles Melting and Fragmentation
Gold clusters can be tailored by subjecting them to melting or to fragmentation and
bimetallic nanoparticles can be alloyed after melting under irradiation by a pulsed laser at
532 nm in the vicinity of the wavelength of the surface plasmon band [209212]. In fact, the
electrons of the plasmon band are highly excited. The hot electrons transfer the energy to
the lattice in less than a fewpicoseconds. The transfer can occur before or after the electrons
Copyright 2004 by Taylor & Francis Group, LLC
have reached thermal equilibrium (Fig. 16). Eventually, the energy of the phonons is trans-
ferred to the surrounding solvent.
If the laser pulse is very short and its energy high enough, the temperature of the
lattice can reach several thousands degrees. Depending on the energy of the pulse, melting,
fragmentation, and even vaporization of the particles is thus possible. It was shown that
silver and gold clusters undergo signicant morphological change under laser irradiation.
Gold nanorods can be transformed into spherical particles by exciting the metal clusters
with femtosecond laser pulses (Fig. 18) [211]. The amount of energy needed to transform a
gold nanorod into a sphere is more than 60 fJ. The time corresponding to the melting is
around 35 psec, which is shorter than the period required for the energy transfer from the
lattice to the solvent (100 psec). Therefore, sucient energy is accumulated by the lattice
arising from photon absorption and the temperature increases enough to melt and change
the morphology of the gold nanorods. Therefore, if the laser energy increases above the
threshold of melting, the fragmentation channel becomes open. Transmission electron
microscopy (TEM) measurements of nanorods irradiated by a high-energy nanosecond
laser pulse clearly showed that the nanorods are fragmented into small spheres.
Figure 16 Scheme of the successive events following the interaction of a high uence laser pulse
with metal clusters. First, the energy is conned in a hot cluster, then the eects depend on the
uence and on the cluster surface to volume ratio. Low uence: the energy is transferred to the
solvent, which expands into bubbles scattering the light; meantime, the cluster is reversibly cooled
without change of the shape. Medium uence: the cluster temperature is high enough to obtain
cluster melting and changing to spherical shape. High uence: the energy conned within the cluster
creates a fast step of expansionvaporization of the metal particle which acts as a light scattering
center. After the energy has been transferred to the solvent, the solvent bubbles become secondary
light scattering centers and the metal vapor condenses into fragmented clusters. Eventually, the
solvent bubbles cool down by energy dissipation to the bulk.
Copyright 2004 by Taylor & Francis Group, LLC
7.3. Optical Limitation Induced by Nanoparticles
The nonlinear optical properties of nanoclusters were also studied as potential light
limitation systems [213]. The nonlinear optical response of gold particles (of 5-, 10-, or 30-
nm diameter, the latter being enlarged by development) prepared by g-radiolysis in water
solution and stabilized by PVA is size-dependent. The threshold and the amplitude of
limitation depend on the particle size (Fig. 19). At a given total atom concentration, the
light transmission at 532 nm of nanosecond laser pulses at very high uence is weakly
limited by 5-nm clusters, but strongly limited by the 30-nm clusters. The optical limiting
eect induced by gold clusters, measured as a function of the excitation wavelength,
reaches the highest eciency below 530 nm, and the wavelength dependence of the limiting
eect amplitude induced at a xed excitation wavelength increases monotonically to the
UV as expected from photo-induced scattering centers.
The time delay of the amplitude maximum and the uence threshold, where non-
linear eects are observed, dier for the time-resolved signals after nanosecond or
picosecond pulses [213]. Therefore, two types of scattering centers are successively
responsible for the optical limitation. The fast mechanism which reaches an amplitude
maximum in less than 1 nsec appears at relatively high uence (>0.94 J cm
2
, Fig. 20) and
with picosecond pulses, that is for high-energy connement. It is assigned to light
scattering by gold clusters that have expanded and vaporized after having absorbed an
important amount of light energy (Fig. 16). The slow limitation mechanism occurs in a
second step when the energy is transferred subsequently from the cluster to the solvent,
and is a result of light scattering by solvent bubbles. The second mechanism is observed
alone for 5-nm clusters even at high uences, or for 10- and 30-nm clusters at moderate
uences because the energy dissipation from small clusters into the solvent producing
bubbles is then more ecient and precludes metal particle vaporization.
Figure 17 Average diameter of gold nanoparticles after laser ablation at 532 nm and a laser
uence of (n) 320, (.) 480, and (x) 1200 mJ pulse
1
cm
2
as a function of the SDS concentration in
the solution. (From Ref. 208.)
Copyright 2004 by Taylor & Francis Group, LLC
The tests for limitation eciency as a function of the number of pulses (8 nsec, 5 J
cm
2
) have shown that the 10-nm clusters are irreversibly modied and lose their activity
after 20 pulses [213]. It was concluded that the limitation decrease is due to the decrease
after each pulse of the size of part of the clusters below the size threshold for optical
limitation. For 30-nmclusters, the condensation following the vaporization leads to smaller
clusters, but the size after relaxation is still large enough to cause optical limitation, and the
cluster is able to absorb again enough energy to vaporize. After 20 pulses, an equilibrium
size, with a two times lower eciency, seems to be reached.
8. CONCLUSION
Radiation chemical methods have been proven to have a potential to induce and to study
the dynamics of nucleation and growth and of the reactivity of metal clusters from the
monomer to the stable nanoparticle.
Pulse radiolysis permits the study, via the time-resolved technique, of the detailed
mechanisms of metal ion reduction in dierent environments (ligand, solvent, support, . . .),
Figure 18 Comparison of optical absorption data and TEM images (at the same magnication)
for two gold samples irradiated by laser pulses having the same uence (0.25 J cm
2
) and dierent
pulse width. Top: with a 7-nsec laser pulse at 800 nm. (a) Partial melting of the gold nanorods and
(b) selective optical hole burning in the near-IR band corresponding to the nanorods. Bottom: with
laser pulses of 100 fsec. (c) Complete melting of the gold nanorods into nanodots and (d) complete
depletion of the nanorod band. (From Ref. 211.)
Copyright 2004 by Taylor & Francis Group, LLC
of nucleation of atoms into oligomers, of their growth into stable metal clusters, and of the
reactivity of the transient species at each step, also in mixed solutions of two dierent metal
ions. Thus during this growth process, the nuclearity-dependent properties of oligomers
and clusters were determined, such as the redox potential, which is of crucial importance in
certain catalytic reactions and for the growth itself which can be autocatalyzed, and for the
stability of the clusters at the nal size.
Pulse radiolysis and pulsed laser methods are powerful tools to study the reactions at
the interface between clusters and solution, such as fast heterogeneous catalytic processes,
or transfer of the charges generated inside the particle.
The generation of clusters through irradiation of a homogeneous ionic solution has
numerous advantages: (1) an ideal instantaneous mixing of the radiolytic species, acting as
the reductant, with the ions; (2) the strong reducing properties of these species even at room
temperature and a formation of atoms without energetic barrier; (3) a strict control of the
reduction equivalents by the choice of the dose; (4) a control of the competition between the
creation of isolated atoms and the growth of nuclei by the choice of the dose rate from slow
to quasi-sudden regimes. The chances to create isolated atoms in the rst step are thus
Figure 19 Size dependence of the optical transmission obtained at 532 nm vs. the laser uence for
g-induced gold clusters. (.) 5-nm diameter (PVA 0.1 mol L
1
). (o): 10 nm diameter (PVA 10
3
mol
L
1
). (D): 30 nm diameter (PVA 0.1 mol L
1
, MV
2+
10
3
mol L
1
). (5): Reference solution
containing PVA and MV
2+
without clusters. TEM micrographs of the samples I, II, and III are on
the right. (From Ref. 213.)
Copyright 2004 by Taylor & Francis Group, LLC
higher than in any other method. (5) The possible synthesis in situ of clusters inside cavities
of opaque porous material. These conditions explain why the general concept of size eects
on cluster properties, which are the most drastic at low nuclearity, was indeed pointed out
rst in radiolytic experiments. They also explain that the growth can be restricted to very
small nal sizes and that their distribution is exceptionally narrow.
The knowledge of the kinetics mechanism, explaining the properties of the nal
particles from the conditions of the radiolytic synthesis, has been exploited for guiding the
procedure of obtaining clusters with denite size and composition according to the needs
of various studies and applications. Due to this mastering, an increasing number of data
have been obtained on the cluster size- and shape-dependent properties, such as optical
and magnetic properties, fusion temperature, cluster-to-solvent thermal energy transfer,
optical limitation, etc. The understanding of the detailed mechanism of bimetallic cluster
synthesis has also permitted the control of conditions favoring alloying, possibly with
ordered atom arrangement, or segregation in a coreshell structure. Moreover, except for
the reduction step, the mechanism is not specic to the radiation-induced method and will
be helpfully extended to other ways of chemical or photochemical synthesis. In photog-
raphy, this understanding resulted in a new interpretation of the critical nuclearity and
of the selective amplication occurring during the development. Besides, conditions were
Figure 20 Time proles of induced absorbance at 580 nm after focusing a 30-psec, 532-nm laser
pulse at various uences F onto gold clusters, 30 nm diameter. Dotted line: time prole of the pulse.
The fast amplitude increase, up to a maximum in less than 1 nsec for uences z0.9 J cm
2
, is
assigned to light scattering by expanded metal clusters. The delayed increase with a maximum at 3
nsec occurs at uences z0.18 J cm
2
already, and is assigned to light scattering by solvent bubbles.
(From Ref. 213.)
Copyright 2004 by Taylor & Francis Group, LLC
found of favoring the latent image formation and markedly enhancing the photographic
sensitivity by suppressing the electronhole recombination.
However, the variety of composite materials to be elaborated by the method is still
barely explored. For example, bi- and multimetallic nanoparticles, included in dierent
matrices (polymeric membranes, porous supports, . . .) or functionalized, have promising
applications. New methods of cluster characterization at this extremely low size scale are
developed and will improve their study.
Moreover, the interpretation of experimental data on clusters in solution requires
more elaborate theoretical models to include the solvation eects around the structure of a
small metal cluster. New kinetic models must be developed to describe nucleation, which
governs the phase transition from a solute to a small solid phase.
REFERENCES
1. Roentgen, W.C. Sitz.-Ber. Phys.-Med. Ges. Wuerzburg 1896, 11 pp.
2. Becquerel, H. C. R. Acad. Sci. 1896, 122, 420.
3. Fricke, H. J. Chem. Phys. 1934, 2, 556.
4. Hassinsky, M.; Pujo, A.M. C. R. Acad. Sci. 1955, 240, 2530.
5. Hassinsky, M. La Chimie Nucle aire et ses Applications. Masson et Cie: Paris, 1957.
6. Koulkes-Pujo, A.M.; Rashkov, S. J. Chim. Phys. 1967, 64, 534.
7. Baxendale, J.H.; Fielden, E.M.; Keene, J.P.; Ebert, M. In Pulse Radiolysis; Keene, J.P.,
Swallow, A., Baxendale, J.H., Eds.; Acad. Press: London, 1965; 207 pp.
8. Von Pukies, J.; Roebke, W.; Henglein, A. Ber. Bunsenges. Phys. Chem. 1968, 72, 842.
9. Hassinsky, M. In Radiation Chemistry; Dobo, J., Hedwig, P., Eds.; Akad. Kiado: Budapest,
1972; 2, 1353. And following discussion.
10. Belloni, J. In Actions Chimiques et Biologiques des Radiations; Hassinsky, M., Ed.; Masson et
Cie: Paris, 1971; 47 pp.
11. Delcourt, M.O.; Belloni, J. Radiochem. Radioanal. Lett. 1973, 13, 329.
12. Basco, N.; Vidyarthi, S.K.; Walker, D.C. Can. J. Chem. 1973, 51, 2497.
13. Henglein, A. Ber. Bunsenges. Phys. Chem. 1977, 81, 556.
14. Halperin, W.H. Rev. Mod. Phys. 1986, 58, 533.
15. Morse, M.D. Chem. Rev. 1986, 86, 1046.
16. Schumacher, E. Chimia 1988, 42, 357.
17. Haberland, H., Ed.; In Clusters of atoms and Molecules; Springer, 1994.
18. Kubo, R. J. Phys. Soc. Jpn. 1962, 17, 975.
19. Belloni, J.; Mostafavi, M. In Studies in physical and theoretical Chemistry 87. Radiation Che-
mistry: Present Status and Future Trends; Jonah, C.D., Rao, B.M., Eds.; Elsevier, 2001;
411 pp.
20. Henglein, A. Ber. Bunsenges. Phys. Chem. 1995, 99, 903.
21. Kamat, P.V. J. Phys. Chem. B 2002, 106, 7729.
22. Henglein, A. Chem. Rev. 1989, 89, 1861.
23. Belloni, J.; Amblard, J.; Marignier, J.L.; Mostafavi, M. In Clusters of Atoms and Molecules;
Haberland, H. Ed.; Springer-Verlag, 1994; 2, 290.
24. Tausch-Treml, R.; Henglein, A.; Lilie, J. Ber. Bunsenges. Phys. Chem. 1978, 82, 1335.
25. Belloni, J.; Mostafavi, M.; Remita, H.; Marignier, J.L.; Delcourt, M.O. New J. Chem.
1998, 22, 1239.
26. Belloni, J.; Mostafavi, M. In Metal Clusters in Chemistry; Braunstein, P., Oro, R., Raithby, J.,
Eds.; Wiley, 1999; 1213 pp.
27. Baxendale, J.H.; Busi, F. The Study of Fast Processes and Transient Species by Electron Pulse
Radiolysis. NATO ASI Ser. 86. D. Reidel, 1982.
Copyright 2004 by Taylor & Francis Group, LLC
28. Buxton, G.V.; Greenstock, C.L.; Helman, W.P.; Ross, A.B. J. Phys. Chem. Ref. Data 1988,
17, j2n.
29. Elliott, J.; Simon, A.S. Radiat. Phys. Chem. 1984, 24, 229.
30. Schwarz, H.A.; Dodson, R.W. J. Phys. Chem. 1989, 93, 409.
31. Mostafavi, M.; Marignier, J.L.; Amblard, J.; Belloni, J. Radiat. Phys. Chem. 1989, 34, 605.
32. Janata, E.; Lilie, J.; Martin, M. Radiat. Phys. Chem. 1994, 43, 353.
33. Janata, E. Radiat. Phys. Chem. 1994, 44, 449.
34. Kapoor, S.; Lawless, D.; Kennepohl, P.; Meisel, D.; Serpone, N. Langmuir 1994, 10, 3018.
35. Janata, E.; Henglein, A.; Ershov, B.G. J. Phys. Chem. 1994, 98, 10888.
36. Texier, I.; Mostafavi, M. Radiat. Phys. Chem. 1997, 49, 459.
37. Remita, S.; Mostafavi, M.; Delcourt, M.O. J. Phys. Chem. 1997, 100, 10187.
38. Mostafavi, M.; Dey, G.R.; Franc ois, L.; Belloni, J. J. Phys. Chem. 2002, 106, 10184.
39. Mostafavi, M.; Delcourt, M.O.; Keghouche, N.; Picq, G. Radiat. Phys. Chem. 1992, 40, 445.
40. Cercek, B.; Ebert, M.; Swallow, A.J. J. Chem. Soc. A, 1966, 612.
41. Butler, J.; Henglein, A. Radiat. Phys. Chem. 1980, 15, 603.
42. Ershov, B.G.; Sukhov, N.L. Radiat. Phys. Chem. 1990, 36, 93.
43. Ghosh-Mazumdar, A.S.; Hart, E.J. Adv. Chem. Ser. 1968, 81, 193.
44. Mosseri, S.; Henglein, A.; Janata, E. J. Phys. Chem. 1989, 93, 6791.
45. Belloni, J.; Khatouri, J.; Mostafavi, M.; Amblard, J. In Ultrafast Reaction Dynamics and
Solvent Eects; Rossky, P.J., Gauduel, Y., Eds. Am. Inst. Phys. 1993; 541 pp.
46. Mostafavi, M.; Belloni, J. Recent Res. Dev. Phys Chem. 1997, 1, 459.
47. Texier, I.; Remita, S.; Archirel, P.; Mostafavi, M. J. Phys. Chem. 1996, 100, 12472.
48. Mostafavi, M.; Remita, S.; Delcourt, M.O.; Belloni, J. J. Chem. Phys. 1996, 93, 1828.
49. Belloni, J.; Delcourt, M.O.; Marignier, J.L.; Amblard, J. In Radiation Chemistry; Hedwig, P.,
Schiller, L.R., Eds.; Akad. Kiado: Budapest, 1987; 89 pp.
50. Mostafavi, M.; Lin, M.; Wu, G.; Katsumura, Y.; Muroya, Y. J. Phys. Chem. A 2002, 106,
3123.
51. Kevan, L. J. Phys. Chem. 1981, 85, 1828.
52. Ghosh-Mazumdar, A.S.; Hart, E.J. Int. J. Radiat. Phys. Chem. 1969, 1, 165.
53. Marignier, J.L.; Belloni, J.; Delcourt, M.O.; Chevalier, J.P. Nature 1985, 317, 344.
54. de Cointet, C.; Mostafavi, M.; Khatouri, J.; Belloni, J. J. Phys. Chem. 1997, 101, 3512.
55. Mie, G. Ann. Phys. 1908, 25, 377.
56. Creighton, J.A.; Eadon, D.J. J. Chem. Soc. Faraday Trans. 1991, 87, 3881.
57. Henglein, A.; Tausch-Treml, R. J. Colloid Interface Sci. 1981, 80, 84.
58. Buxton, G.V.; Mulazzani, Q.G.; Ross, A.B. J. Phys. Chem. Ref. Data 1995, 24, 3.
59. Rao, P.S.; Hayon, E. J. Phys. Chem 1975, 79, 865.
60. Renou, F.; Mostafavi, M. Chem. Phys. Lett. 2001, 335, 363.
61. Remita, S.; Archirel, P.; Mostafavi, M. J. Phys. Chem. 1995, 99, 13198.
62. Khatouri, J.; Ridard, J.; Mostafavi, M.; Amblard, J.; Belloni, J. Z. Phys. D 1995, 34, 57.
63. Khatouri, J.; Mostafavi, M.; Belloni, J. In Photochemistry and Radiation Chemistry: Com-
plementary Method for the Study of Electron Transfer; Wishart, J., Nocera, D., Eds.; ACS, 1998;
293 pp.
64. Platzer, O.; Amblard, J.; Marignier, J.L.; Belloni, J. J. Phys. Chem. 1992, 96, 2334.
65. Amblard, J.; Platzer, O.; Ridard, J.; Belloni, J. J. Phys. Chem. 1992, 96, 2340.
66. de Cointet, C.; Khatouri, J.; Mostafavi, M.; Belloni, J. J. Phys. Chem. 1997, 101, 3517.
67. Jackschath, C.; Rabin, I.; Schulze, W. Z. Phys. D 1992, 22, 517.
68. Alameddin, G.; Hunter, J.; Cameron, D.; Kappes, M.M. Chem. Phys. Lett. 1992, 192, 122.
69. Khatouri, J.; Mostafavi, M.; Amblard, J.; Belloni, J. Z. Phys. D 1993, 26, 82.
70. Henglein, A. J. Phys. Chem. 1993, 97, 5457.
71. George Thomas, K.; Biju, V.; George, M.V.; Guldi, D.M.; Kamat, P.V. J. Phys. Chem. B
1999, 103, 8864.
72. Sudeep, P.K.; Ipe, B.I.; George Thomas, K.; George, M.V.; Barazzouk, S.; Hotchandani, S.;
Kamat, P.V. Nano Lett. 2002, 2, 29.
Copyright 2004 by Taylor & Francis Group, LLC
73. Ipe, B.I.; George Thomas, K.; Barazzouk, S.; Hotchandani, S.; Kamat, P.V. J. Phys. Chem. B
2002, 106, 18.
74. Kreibig, U.; Vollmer, M. Optical properties of metal clusters, Springer, Berlin, 1995. Sinzig, J.,
Radtke, U., Quinten, M., Kreibig, U., Eds. Z. Phys. D 1993, 26, 242.
75. Henglein, A. J. Phys. Chem. 1979, 83, 2209.
76. Zhu, Y.; Qian, Y.; Zhang, M.; Chen, Z.; Lu, B.; Wang, C. Mater. Lett. 1993, 17, 314.
77. Jonah, C.D.; Matheson, M.S.; Meisel, D. J. Phys. Chem. 1977, 81, 1805.
78. Rafaelo, R.; Haruvy, Y.; Binenboym, J.; Baruch, G.; Rajbenbach, L.A. J. Mol. Catal. 1983,
22, 219.
79. Zhu, Y.; Qian, Y.; Li, X.J.; Zhang, M. Chem. Commun. 1997, 12, 1081.
80. Graetzel, M. Acc. Chem. Res. 1981, 14, 376.
81. Ershov, B.G. Russ. Chem. Bull. 1994, 43, 16.
82. Sosebee, T.; Giersig, M.; Holzwarth, A.; Mulvaney, P. Ber. Bunsenges. Phys. Chem. 1995, 99,
40.
83. Henglein, A. Chem. Phys. Lett. 1989, 154, 473.
84. Kumar, M.; Kapoor, S.; Gopinathan, C. Radiat. Phys. Chem. 1997, 50, 465.
85. Remita, S.; Orts, J.M.; Feliu, J.M.; Mostafavi, M.; Delcourt, M.O. Chem. Phys. Lett. 1994,
218, 115.
86. Mostafavi, M.; Keghouche, N.; Delcourt, M.O.; Belloni, J. Chem. Phys. Lett. 1990, 167, 193.
87. Kurihara, K.; Kizling, J.; Stenius, P.; Fendler, J.H. J. Am. Chem. Soc. 1983, 105, 2574.
88. Belloni, J.; Delcourt, M.O.; Leclere, C. Nouv. J. Chim. 1982, 6, 507.
89. Mostafavi, M.; Delcourt, M.O.; Picq, G. Radiat. Phys. Chem. 1992, 41, 453.
90. Mostafavi, M.; Keghouche, N.; Delcourt, M.O. Chem. Phys. Lett. 1990, 169, 81.
91. Ershov, B.G.; Henglein, A. J. Phys. Chem. 1998, 102, 667 and 10663.
92. Keita, B.; Nadjo, L.; de Cointet, C.; Amblard, J.; Belloni, J. Chem. Phys. Lett. 1996, 249, 297.
93. Henglein, A.; Ershov, B.G.; Malow, M. J. Phys. Chem. 1995, 99, 14129.
94. Khatouri, J.; Mostafavi, M.; Amblard, J.; Belloni, J. Chem. Phys. Lett. 1992, 191, 351.
95. Sukhov, N.L.; Akinshin, M.A.; Ershov, B.G. Khim. Vys. Energ. 1986, 20, 292.
96. Lin, M.Z. Thesis, University of Paris-Sud (1996).
97. Remita, H.; Derai, R.; Delcourt, M.O. Radiat. Phys. Chem. 1991, 37, 221.
98. Belloni, J.; Mostafavi, M.; Remita, H.; Marignier, J.L.; Delcourt, M.O. In Synthesis Chem-
istry and Some Properties of Metal Nanoparticles; Bradley, J., Chaudret, B. Eds; New J. Chem.
1998, 22, 1239.
99. Treguer, M.; Remita, H.; Pernot, P.; Khatouri, J.; Belloni, J. J. Phys. Chem. A 2001, 105,
6102.
100. Le Gratiet, B.; Remita, H.; Picq, G.; Delcourt, M.O. J. Catal. 1996, 164, 36.
101. Lawless, D.; Kapoor, S.; Kennepohl, P.; Meisel, D.; Serpone, N. J. Phys. Chem. 1994, 98,
9616.
102. Treguer, M.; De Cointet, Ch.; Remita, H.; Khatouri, J.; Mostafavi, M.; Amblard, J.; Belloni,
J.; De Keyzer, R. J. Phys. Chem. 1998, 102, 4310.
103. Keita, B.; Nadjo, L.; Gachard, E.; Remita, H.; Khatouri, J.; Belloni, J. New J. Chem. 1997,
21, 851.
104. Belloni, J.; Marignier, J.L.; Delcourt, M.O.; Minana, M. US Patent No 4 629 709 (1986). CIP
No 4 745 094 (1987).
105. Mills, G.; Henglein, A. Radiat. Phys. Chem. 1985, 26, 391.
106. Delcourt, M.O.; Belloni, J.; Marignier, J.L.; Mory, C.; Colliex, C. Radiat. Phys. Chem. 1984,
23, 485.
107. Westerhausen, J.; Henglein, A.; Lilie, J. Ber. Bunsenges. Phys. Chem. 1981, 85, 182.
108. Gachard, E.; Remita, H.; Khatouri, J.; Belloni, J.; Keita, B.; Nadjo, L. New J. Chem. 1998,
22, 1257.
109. Henglein, A.; Meisel, D. Langmuir 1998, 14, 7392.
110. Ershov, B.G.; Sukhov, N.L.; Troistskii, D.I. Russ. J. Phys. Chem. 1994, 68, 734.
111. Michaelis, M.; Henglein, A. J. Phys. Chem. 1992, 96, 4719.
Copyright 2004 by Taylor & Francis Group, LLC
112. Ershov, B.G.; Sukhov, N.L.; Troistskii, D.I. Russ. J. Phys. Chem. 1994, 68, 734.
113. Ershov, B.G.; Janata, E.; Michaelis, M.; Henglein, A. J. Phys. Chem. 1991, 95, 8996.
114. Liu, Y.; Zhu, Y.; Zhang, Y.H.; Qian, Y.; Zhang, M.; Yang, L.; Wang, C. J. Mater. Chem.
1997, 7, 787.
115. Marignier, J.L.; Belloni, J. J. Chem. Phys. 1988, 85, 21.
116. Fujita, N.; Matsuura, C.; Hiroishi, D.; Saigo, K. Radiat. Phys. Chem. 1998, 53, 603.
117. Henglein, A.; Lilie, J. J. Phys. Chem. 1981, 85, 1246.
118. Henglein, A.; Gutierrez, M.; Janata, E.; Ershov, B.G. J. Phys. Chem. 1992, 96, 4598.
119. Henglein, A.; Giersig, M. J. Phys. Chem. 1994, 98, 6931.
120. Buxton, G.V.; Rhodes, T.; Seller, R. J. Chem. Soc. Faraday Trans. 1 1982, 78, 3341.
121. Breitenkamp, M.; Henglein, A.; Lilie, J. Ber. Bunsenges. Phys. Chem. 1976, 80, 973.
122. Henglein, A.; Janata, E.; Fojtik, A. J. Phys. Chem. 1992, 96, 4734.
123. Gutierrez, M.; Henglein, A. J. Phys. Chem. 1996, 100, 7656.
124. Lian R. Thesis, USTC, Hefe (2001).
125. Henglein, A.; Mulvaney, P.; Linnert, T. Faraday Discuss. 1991, 92, 31.
126. Strelow, F.; Henglein, A. J. Phys. Chem. 1995, 99, 11834.
127. Linnert, T.; Mulvaney, P.; Henglein, A. J. Phys. Chem. 1993, 97, 679.
128. Dawson, A.; Kamat, P.V. J. Phys. Chem. 2000, 104, 1184.
129. Teo, B.K.; Keating, K.; Kao, Y.-H. J. Am. Chem. Soc. 1987, 109, 3494.
130. Noshima, N. J. Macromol. Sci. Chem. A 1990, 27, 1225.
131. Sato, T.; Kuroda, S.; Takami, A.; Yonezawa, Y.; Hada, H. Appl. Organomet. Chem. 1991, 5,
261.
132. Kobayashi, M.; Sato, H. Chem. Lett. 1993, 1659.
133. Torigoe, K.; Esumi, K. Langmuir 1992, 8, 59 and 1993, 9, 1964.
134. Esumi, K.; Suzuki, A.; Aihara, N.; Usui, K.; Torigoe, K. Langmuir 1998, 14, 3157.
135. Sasaki, M.; Osada, M.; Higashimoto, N.; Inagaki, S.; Fukushima, Y.; Fukuoka, A.; Ishikawa,
M. Microporous Mesoporous Mater. 1998, 21, 597; J. Mol. Catal. A 1999, 141, 223.
136. Pal, T. Talanta 1998, 46, 583.
137. Fukuoka, A.; Higashimoto, N.; Sakamoto, Y.; Sasaki, M.; Sugimoto, N.; Inagaki, S.; Fuku-
shima, Y.; Ishikawa, M. Catal. Today 2001, 66, 23.
138. sau, T.K.; Pal, A.; Jana, N.R.; Wang, Z.L.; Pal, T. J. Nanopart. Res. 2001, 3, 257.
139. Kim, F.; Song, J.H.; Yang, P. J. Am. Chem. Soc. 2002, 124, 14316.
140. Mallik, K.; Mandal, M.; Pradhan, N.; Pal, T. Nano Lett. 2001, 1, 319.
141. Abid, J.P.; Wark, A.W.; Brevet, P.F.; Girault, H.H. Chem. Commun. 2002; 792.
142. Itakura, T.; Torigoe, K.; Esumi, K. Langmuir 1995, 11, 4129.
143. Fernandez, A.; Gonzalez-Elipe, A.R.; Real, C.; Caballero, A.; Munuera, G. Langmuir 1993,
9, 121.
144. Stathatos, E.; Lianos, P.; Falaras, P.; Siokou, A. Langmuir 2000, 16, 2398.
145. Remita, S.; Mostafavi, M.; Delcourt, M.O. New J. Chem. 1994, 18, 581.
146. Dick, K.; Dhanasekaran, T.; Zhang, Z.Y.; Meisel, D. J. Am. Chem. Soc. 2002, 124, 2312.
147. Michaelis, M.; Henglein, A. J. Phys. Chem. 1992, 96, 4719.
148. Marignier, J.L.; Dokuchaev, A.; Hautecloque, S.; Grand, D. Proc. 7th Int. Symp. Small Part.
Inorg. Clusters, 1994; 189 pp.
149. Belloni, J.; Delcourt, M.O.; Leclere, C. Nouv. J. Chim. 1982, 6, 507.
150. Belloni, J.; Lecheheb, M. Radiat. Phys. Chem. 1987, 29, 89.
151. Bruneaux, J.; Cachet, H.; Froment, M.; Amblard, J.; Belloni, J.; Mostafavi, M. Electrochim.
Acta 1987, 32, 1533.
152. Bruneaux, J.; Cachet, H.; Froment, M.; Amblard, J.; Mostafavi, M. J. Electroanal. Chem.
1989, 269, 375.
153. Gacoin, T.; Chaput, F.; Boilot, J.P.; Mostafavi, M.; Delcourt, M.O. In Eurogel 91; Vilminot,
S., Nass, R., Schmidt, H., Eds.; North Holland: E-MRS, 1991; 159 pp.
154. Zhu, X.; Qian, Y.; Zhang, M.; Chen, Z.; Zhou, G. J. Mater. Chem. 1994, 4, 1619.
155. Valentin, E.; Bernas, H.; Ricolleau, C.; Creuzet, F. Phys. Rev. Lett. 2001, 86, 99.
Copyright 2004 by Taylor & Francis Group, LLC
156. Keghouche N. Thesis, University Constantine, 1993.
157. Le Gratiet B. Thesis, University Paris-Sud, Orsay, 1996.
158. Torigoe, K.; Remita, H.; Picq, G.; Belloni, J.; Bazin, D. J. Phys. Chem. B 2000, 104, 7050.
159. Zhang, G.; Liu, X.; Thomas, J.K. Radiat. Phys. Chem. 1998, 51, 135.
160. Michalik, J.; Azuma, N.; Sadlo, J.; Kevan, L. J. Phys. Chem. 1995, 99, 4679.
161. Michalik, J.; Wasowicz, T.; Sadlo, J.; Reijerse, E.J.; Kevan, L. Radiat. Phys. Chem. 1996, 47,
75.
162. Gachard, E.; Belloni, J.; Subramanian, M.A. J. Mater. Chem. 1996, 6, 867.
163. Sasaki, M.; Osada, M.; Higashimoto, N.; Yamamoto, T.; Fukuoka, A.; Ishikawa, M. J. Mol.
Catal. A 1999, 141, 223.
164. Vucemilovic, M.I.; Micic, O.I. Radiat. Phys. Chem. 1988, 32, 79.
165. Micic, O.I.; Meglic, M.; Lawless, D.; Sharma, D.K.; Serpone, N. Langmuir 1990, 6, 487.
166. Marignier, J.L.; Ashokkumar; Mostafavi, M. Proc. 50th Annual IS&T Conf. on Silver
Halides, 1997; 67 pp.
167. Michaelis, M.; Henglein, A.; Mulvaney, P. J. Phys. Chem. 1994, 98, 6212.
168. Henglein, A.; Mulvaney, P.; Holzwarth, A.; Sosebee, T.E.; Fojtik, A. Ber. Bunsenges. Phys.
Chem. 1992, 96, 754.
169. Mulvaney, P.; Giersig, M.; Henglein, A. J. Phys. Chem. 1992, 96, 10419.
170. Henglein, F.; Henglein, A.; Mulvaney, P. Ber. Bunsenges. Phys. Chem. 1994, 98, 180.
171. Henglein, A.; Giersig, M. J. Phys. Chem. 1994, 98, 6931.
172. Shibata, T.; Bunker, B.A.; Zhang, Z.; Meisel, D.; Vardeman, C.F. Jr.; Gezelter, D. Am.
Chem. Soc. 2002, 124, 11989.
173. Belloni, J. Curr. Opin. Colloid Interface Sci. 1996, 1, 184.
174. Malkov, A.; Belloni, J. J. Chem. Phys. 1992, 89, 885.
175. Ershov, B.G.; Janata, E.; Henglein, A. J. Phys. Chem. 1994, 98, 7619 and 10891.
176. Ershov, B.G.; Janata, E.; Henglein, A. Radiat. Phys. Chem. 1996, 47, 59.
177. Henglein, A. J. Phys. Chem. 1992, 96, 2411.
178. Khatouri, J.; Mostafavi, M.; Amblard, J.; Belloni, J. Z. Phys. D 1993, 26, 82.
179. De Vyt, A.; Gijbels, R.; Davock, H.; Van Roost, C.; Geuens, I. J. Anal. At. Spectrom. 1999,
14, 499.
180. Remita, H.; Khatouri, J.; Treguer, M.; Amblard, J.; Belloni, J. Z. Phys. D 1997, 40, 127.
181. Remita, H.; Etcheberry, A.; Belloni, J. J. Phys. Chem. 2002, 106. in press.
182. Kumar, M.; Kapoor, S.; Gopinathan, C. Radiat. Phys. Chem. 1999, 54, 39.
183. Remita, S.; Mostafavi, M.; Delcourt, M.O. Radiat. Phys. Chem. 1996, 47, 275.
184. Remit, S.; Picq, G.; Khatouri, J.; Mostafavi, M. Radiat. Phys. Chem. 1999, 54, 463.
185. Torigoe, K.; Remita, H.; Beaunier, P.; Belloni, J. Radiat. Phys. Chem. 2002, 64, 215.
186. Braunstein, P.; Oro, R.; Raithby, J. Metal Clusters in Chemistry, Eds; Wiley, NY, 1999; 1213
pp.
187. Meisel, D.; Mulac, W.A.; Matheson, M.S. J. Phys. Chem. 1981, 85, 179.
188. Delcourt, M.O.; Keghouche, N.; Belloni, J. Nouv. J. Chim. 1983, 7, 131.
189. Belloni, J.; Lecheheb, A. Radiat. Phys. Chem. 1987, 29, 89.
190. Amblard, J.; Belloni, J.; Platzer, O. J. Chim. Phys. 1991, 88, 835.
191. Agfa GV. Eur. Patent 95203 1706 (1995).
192. Hamilton, F. In Theory of the Photographic Process, 4th Ed.: James, T.H., Ed.; MacMillan:
NY, 1977.
193. Glafkides, P. Chimie et Physique Photographiques, 5th Edition, Editions de lUsine Nouvelle,
1987.
194. Tani, T. Photographic Sensivity. Theory and Mechanisms. NY: Oxford University Press,
1995.
195. Belloni, J. In C. R. Acad. Sci. Phys.; Bre chignac, C., Cahuzac, P., Eds.; 2002, 3, 381.
196. Hamilton, J.F. In Growth and Properties of Metal Clusters; Bourdon, J., Ed., Elsevier Publ.,
1980; 289 pp.
197. Trautweiler, F. Photogr. Sci. Eng. 1968, 12, 138.
Copyright 2004 by Taylor & Francis Group, LLC
198. Mostafavi, M.; Marignier, J.L.; Amblard, J.; Belloni, J. J. Imaging Sci. 1991, 35, 68.
199. Gurney, R.W.; Mott, N.F. Proc. R. Soc. Lond. A 1938, 164, 485.
200. Belloni, J.; Treguer, M.; Remita, H.; de Keyzer, R. Nature 1999, 402, 865.
201. Treguer, M.; Remita, H.; Belloni, J.; De Keyzer, R. J. Imaging Sci. 2002, 46, 193.
202. Fojtik, A.; Henglein, A. Ber. Bunsenges. Phys. Chem. 1993, 97, 252.
203. Neddersen, J.; Chumanov, G.; Cotton, T.M. Appl. Spectrosc. 1993, 47, 1959.
204. Sibbald, M.S.; Chumanov, G.; Cotton, T.M. J. Phys. Chem. 1996, 100, 4672.
205. Yeh, M.S.; Yang, Y.S.; Lee, Y.P.; Lee, H.F.; Yeh, Y.H.; Yeh, C.S. J. Phys. Chem. 1999, 103,
6851.
206. Chen, Y.H.; Yeh, C.S. Colloids Surf. 2002, 197, 133.
207. Mafune , F.; Kohno, J.-Y.; Takeda, Y.; Kondow, T.; Sawabe, H. J. Phys. Chem. 2001, 105,
5114.
208. Mafune , F.; Kohno, J.-Y.; Takeda, Y.; Kondow, T. J. Phys. Chem. 2002, 106, 85555.
209. Kamat, P.V.; Flumiani, M.; Hartland, G.V. J. Phys. Chem. B 1998, 102, 3123.
210. Hodak, J.H.; Henglein, A.; Giersig, M.; Hartland, G.V. J. Phys. Chem. B 2000, 104, 11708.
211. El-Sayed, M.A. Acc. Chem. Res. 2001, 34, 257.
212. Kamat, P.V. J. Phys. Chem. B 2002, 106, 7729.
213. Mostafavi, M.; Franc ois, L.; Belloni, J.; Delaire, J. Phys. Chem. Chem. Phys. 2001, 3, 4965.
Copyright 2004 by Taylor & Francis Group, LLC
21
Application of Radiation Chemical Reactions
to the Molecular Design of Functional
Organic Materials
Tsuneki Ichikawa
Hokkaido University, Sapporo, Japan
Although radiation chemical reactions have been applied to many industrial elds, espe-
cially to polymer industries, they are mainly aimed at improving the mechanical, thermal,
and surface properties of materials [1]. Applications of radiation chemical reactions to the
construction of modern functional organic materials such as electronic and optical materials
are relatively few. However, radiation chemical reactions have the following characteristics
that are advantageous for studying and constructing functional organic materials:
1. The reactions take place even in a solid without structural deformation, which is
advantageous for modifying the chemical and physical properties of precisely
shaped solid substances.
2. The reactions take place within the region where the radiation energy is absorbed,
so that only a part of a solid substance can be modied with precise spatial
resolution. Moreover, the region is controllable three dimensionally by changing
the eld of exposure and the depth of penetration that depends on the energy of
ionizing radiation.
3. The reactions can be induced without additives, which is advantageous for main-
taining the purity of materials.
In the following section, we will show how radiation chemical reactions are useful for the
molecular design of functional organic materials by taking our recent studies on next-
generation resist polymers and organic semiconductors as examples.
1. APPLICATION OF DISSOCIATIVE ELECTRON ATTACHMENT TO THE
CONSTRUCTION OF RADIATION RESIST
Because of their short wavelength, ionizing radiations such as ultra-short-wave UV light,
x-rays, and electron and ion beams have been considered to be the light sources of next-
generation lithography for mass production of very large scale integrated circuits [25].
The lithography process is based on radiation-induced change in the dissolution rate of
Copyright 2004 by Taylor & Francis Group, LLC
resist polymers. Ionizing radiation induces either degradation or cross-linking of resist
polymers, which increases or decreases the dissolution rate, respectively.
Poly(methyl methacrylate) (PMMA) has been widely used as a radiation-degradable
resist polymer [6]. Although PMMA has high spatial resolution of less than 10 nm, the
sensitivity to ionizing radiation is poor. Several attempts to increase the sensitivity have been
made by modifying the molecular structure of polymers [710]. However, the sensitivity is
still not enough for mass production. Another shortcoming is the limited spatial resolution.
The main chains of PMMA are randomly destroyed by homolytic scission [1113], so that
the molecular weight of the decomposed polymer has a wide distribution, which limits the
resolution if a spatial resolution of less than the original molecular size is expected.
A novel approach to increase the sensitivity and the spatial resolution is to use a
polymer that can be cut just in half by ionizing radiation. Because the selective scission of
the polymer skeleton does not leave a longer fragment that is dicult to be dissolved, the
sensitivity and the resolution are expected to be improved [14,15] by using such a polymer
as a radiation resist (see Fig. 1).
Dissociative electron attachment is a radiation chemical reaction suitable for cutting
a polymer just in half. If two equivalent polymer skeletons R are connected with a
functional group XY that has a large cross section of dissociative electron attachment, the
polymer captures an ejected electron at the center of the polymer skeleton and is broken
into two fragments with similar molecular weight, as RXYR + e
!RX
+ YR. The
key to construct such a polymer is to nd a functional group that is possible to connect
two polymer chains, to capture an electron eciently, and to dissociate into two fragments
after the capture.
1.1. Molecular Design for Position-Selective Scission of Polymer Skeleton
The dissociation of a functional group connecting two polymer skeletons must be fast
enough for competing the charge recombination due to the migration of positive charges.
Assuming the linear free energy relationship, a functional group with larger heat of dis-
sociative electron attachment is expected to have lower activation energy of dissociation.
Esters of carboxylic acids are strong candidates for the dissociative electron attachment,
Figure 1 Schematic representation of the formation of polymer fragments by random scission
(left) and selective scission by dissociative electron attachment (right).
Copyright 2004 by Taylor & Francis Group, LLC
since the dissociation of ester anions is expected to be easy due to high stability of resultant
carboxylate anions [16].
Table 1 shows molecular products generated by electron attachment to several esters
at 77 K in 2-methyltetrahydrofuran (MTHF) matrices. The concentration of the esters is
ca. 0.05 mol/dm
3
. Table 1 also shows the changes in enthalpies for the dissociation of anion
radicals into carboxylate anions and neutral free radicals that are calculated from the
experimental values of electron anities of carboxylates and the dissociation energies of
ester bonds [17]. Within the solutes examined, only benzyl esters dissociate into carboxylate
anions and benzyl radicals that show intense absorption and uorescence bands with
maxima at 320 and 475 nm [18], respectively (see Fig. 2). Dissociative electron attachment
to alkyl esters is endothermic, so that the electron attachment does not result in the dis-
sociation but simply gives their molecular radical anions with broad absorption bands at
<600 nm, which is attributable to ketyl-type anions with the unpaired electron on the
carbonyl carbon and the excess charge on the carbonyl oxygen [19]. Although the heat of
dissociative electron attachment is higher for C
2
H
5
COOCH
2
C
16
H
9
and CF
3
COOC
2
H
5
than
for C
2
H
5
COOCH
2
C
6
H
5
, as shown in Fig. 3, electron attachment to C
2
H
5
COOCH
2
C
16
H
9
and CF
3
COOC
2
H
5
give the radical anions showing the absorption spectrum very similar to
the pyrenyl radical anion [20] and the ketyl-type anion, respectively.
These results indicate that the dissociation of an ester radical anion into a carboxylate
anion and a neutral free radical is controlled by two factors. One is the exothermicity of the
reaction and the other is the distribution of the excess electron on the radical anion.
Dissociation of an ester radical anion takes place if the change in the free energy is negative.
Because the change in entropy is positive for the dissociation, the change in the free energy,
DG = DHTDS, is always negative if the change in the enthalpy is negative. However,
the negative change in the free energy does not certify the dissociation of the radical anion.
The radical anion does not dissociate if the activation energy for the dissociation is too high.
The activation energy is lowered if the radical anion of the ester R
1
COOR
2
shares the
excess electron with R
1
COO and R
2
groups, because the sharing causes the formation of
Table 1 Reaction of Free Electron with Esters in MTHF
Solute
Expected
reaction DH/kJmol
1
Observed
radical
a
Yield
of acid
b
C
6
H
5
COOCH
2
C
6
H
5
C
6
H
5
COO
+
CH
2
C
6
H
5
48 CH
2
C
6
H
5
f2.5
C
2
H
5
COOCH
2
C
6
H
5
C
2
H
5
COO
+
CH
2
C
6
H
5
17 CH
2
C
6
H
5
f2.5
C
11
H
23
COO
CH(C
6
H
5
)C
8
H
17
C
11
H
23
COO
+
CH(C
6
H
5
)C
8
H
17
17 CH (C
6
H
5
)C
8
H
17
f2.3
C
2
H
5
COOCH
2
C
16
H
9
C
2
H
5
COO
+
CH
2
C
16
H
9
52 [C
2
H
5
COOCH
2
C
16
H
9
]
<0.15
CF
3
COOC
2
H
5
CF
3
COO
+
C
2
H
5
50 [CF
3
COOC
2
H
5
]
<0.15
C
2
H
5
COOC
2
H
5
C
2
H
5
COO
+
C
2
H
5
57 [C
2
H
5
COOC
2
H
5
]
<0.15
C
2
H
5
COOC(CH
3
)
3
C
2
H
5
COO
+
C(CH
3
)
3
40 [C
2
H
5
COOC(CH
3
)
3
]
<0.15
a
In solid matrices at 77 K.
b
Number of molecules per 100 eV radiation energy absorbed to MTHF solution at 293 K.
Copyright 2004 by Taylor & Francis Group, LLC
an antibonding orbital on the ester bond and therefore the reduction of the activation energy
for the dissociation. However, because the electron anity of C
16
H
9
or CF
3
CO is much
higher than those of their counterparts, the radical anion of C
2
H
5
COOCH
2
C
16
H
9
or
CF
3
COOC
2
H
5
does not share the excess electron but is localized on C
16
H
9
or CF
3
CO,
respectively. The radical anion of C
2
H
5
COOCH
2
C
16
H
9
or CF
3
COOC
2
H
5
is therefore
stable. The excess electron on benzyl esters may occupy the antibonding orbital of the ester
Figure 2 Excitation (broken line, observed wavelength = 475 nm) and uorescence (solid line,
excitation wavelength = 320 nm) spectra of the benzyl radical generated in g-irradiated MTHF at 77
K by dissociative electron attachment to C
2
H
5
COOCH
2
C
6
H
5
.
Figure 3 Absorption spectra for g-irradiated MTHF at 77 K containing C
2
H
5
COOCH
2
C
16
H
9
(solid line) and CF
3
COOC
2
H
5
(broken line).
Copyright 2004 by Taylor & Francis Group, LLC
bond, so that the activation energy is low enough to dissociate into the carboxylate anion
and the benzyl radical even at 77 K (see Fig. 4).
Figure 5 shows the relationship between the concentration of benzyl esters and the
yield of benzyl radicals that is determined from the absorption band at 320 nm. The yield
is saturated at 0.03 mol/dm
3
, which indicates that 0.03 mol/dm
3
of benzyl esters capture
all electrons generated in MTHF and toluene matrices at 77 K. It turns out that one benzyl
ester molecule can capture an electron that is generated within the radius of 2.4 nm.
Assuming the molecular weight of one monomer unit to be 100, one benzyl ester in 333
monomer units is enough for collecting all electrons generated by ionizing radiation from
a polymer with the specic density of unity. The yield of benzyl radicals in toluene is
about one fth of that in MTHF, which is due to the protection eect of the aromatic
solvent; the yield of ejected electrons available for the dissociative electron is much lower
in toluene. Assuming that the G value (the number of events per 100 eV radiation energy
absorbed) of free-electron formation is 2.5, the radiation dose for reducing the initial
molecular weight of 33,300 to 16,650 is 120 kGy.
1.2. Synthesis of Polymer by Living Radical Polymerization
Since the benzyl esters are found to be a suitable functional group for connecting two poly-
mer skeletons, the next problem is to connect two equivalent polymer skeletons with benzyl
esters. This can be achieved by applying recently developed living radical polymerization
[2123].
Synthesis of polymer molecules with the same molecular weight necessitates living
polymerization. Although living polymerization in homogeneous systems had been limited
to ionic polymerization, recent development of polymer science makes it possible to
synthesize polymers with narrow polydispersity by radical polymerization. Living polymer-
ization necessitates propagating polymer ends that do not deactivate during the polymer-
ization. Regular radical polymerization is not living because the propagating radicals
quickly combine with each other. Living radical polymerization protects terminal radicals
from the combination by using a radical capping agent such as stable nitroxyl radicals.
Heating a mixture of monomer Mand a initiator RONXwith a capping agent ONXstarts
the dissociation of the initiator into reactive and stable free radicals, and therefore starts the
polymerization at about the same time, as
RONX
!
RONX; RM!RM:
The capping agent protects the deactivation of propagating radicals during the polymer-
ization, because the steady-state concentration of reactive propagating radicals is lowered
by the reaction with the capping agent.
Figure 4 Schematic representation of the molecular orbitals of an excess electron on dissociative
(left) and nondissociative (right) functional groups.
Copyright 2004 by Taylor & Francis Group, LLC
RM
n
ONX !
p
RM
n
ONX
There are two strategies for constructing a polymer with benzyl ester in the middle of
the skeleton. One is to make polymer skeletons with the same molecular weight, and then
combine two skeletons with benzyl ester. The other is to synthesize a chemical with
propagation sites of polymerization at both sides of benzyl ester, and use the chemical as
an initiator of living polymerization. Since the former strategy did not work well, probably
due to low reactivity of polymer molecules with benzyl esters, the latter approach will be
mentioned.
We synthesized polystyrene with benzyl ester in the middle of the skeleton by using
the following initiator that dissociates into stable tetramethylpyperidinoxyl radical and
reactive benzyl-type radical.
Figure 6 shows the results of the living radical polymerization of styrene at 400 K.
The linear dependence of the molecular weight on the conversion ratio and the narrow
distribution of the molecular weight are the evidence of living radical polymerization.
Figure 7 shows the proles of the size exclusion chromatography (SEC) of coupled
polystyrene that was dissolved in MTHF and irradiated at 295 K with a Co 60 g-ray
source. The irradiation cuts the main chains just in half, which indicates that the scission
of the skeleton exclusively takes place in the middle of the skeleton by dissociative electron
attachment to the benzyl ester. Figure 8 shows the distribution of the molecular weight of
polystyrene after g-irradiation at 295 K of MTHF solution containing 10 wt.% of coupled
polystyrene. The G value of the dissociation into two polymer fragments is about 5, which
is twice as much higher as that of corresponding C
6
H
5
COOCH
2
C
6
H
5
. The eciency of the
scission in toluene solution is about one fth of that in MTHF, which is due to the
protection eect of the aromatic solvent.
Figure 5 Relation between the radiation chemical yield of the benzyl radical and the concentration
of benzyl esters in 2-methyltetrahydrofuran and toluene at 77 K.
Copyright 2004 by Taylor & Francis Group, LLC
Figure 6 Relation between the conversion ratio and the molecular weight of coupled polystyrene
generated by living radical polymerization at 300 K.
Figure 7 Size exclusion chromatography proles of coupled polystyrene before and after g-irradia-
tion in MTHF at 295 K.
Copyright 2004 by Taylor & Francis Group, LLC
As shown in Fig. 9, irradiation of neat polystyrene does not result in the ecient scis-
sion but increases the molecular weight due mainly to cross-linking, though a small bump
at the low molecular weight side of the SEC pattern indicates the dissociative electron
attachment. The cross-linking is suppressed by addition of hydroquinone as a radical scav-
enger, which suggests that combination reaction of polymer radicals takes place in the neat
polymer, as
CH
2
CHC
6
H
5
!CH
2
CHC
6
H
5
; CH
2
CC
6
H
5
H
CH
2
CHC
6
H
5
!CH
2
CC
6
H
5
H
CH
2
CHC
6
H
5
H !CH
2
CC
6
H
5
H
2
e
COOCHCH
3
C
6
H
5
!COO
CHCH
3
C
6
H
5
CH
2
CC
6
H
5
CH
2
CC
6
H
5
!CH
2
CC
6
H
5
2
cross link
CH
2
CC
6
H
5
CHCH
3
C
6
H
5
!
CH
2
CC
6
H
5
CHCH
3
C
6
H
5
graft
The cross-linking and the grafting of the polymer radicals are suppressed by the addition of
hydroquinone, as
CH
2
C C
6
H
5
HOC
6
H
4
OH !CH
2
C CH
6
H
5
OC
6
H
4
OH
CHCH
3
C
6
H
5
HOC
6
H
4
OH !CH
2
CH
3
C
6
H
5
OC
6
H
4
OH
Although the irradiation of 200 kGy decomposes about 80% of polystyrene in toluene by
the dissociative electron attachment, the yield of the decomposition is only 20% for solid
toluene. Because of its low eciency of scission, the coupled polystyrene may not be a
polymer suitable as a radiation resist. However, the present study has shown that a polymer
that can be decomposed into two equivalent skeletons by ionizing radiation is possible to be
Figure 8 Distribution of the molecular weight of polystyrene after g-irradiation at 295 K of
MTHF solution containing 10 wt.% of coupled polystyrene.
Copyright 2004 by Taylor & Francis Group, LLC
constructed by living radical polymerization. Our preliminary study on polymethylacrylate
containing benzyl ester in a part of the main chain showed that the eciency of the scission
for the neat polymer was comparable to that of the coupled polystyrene in MTHF, though
the scission does not take place without benzyl ester.
2. MOLECULAR DESIGN OF POLYSILANE AS A CONDUCTIVE POLYMER
Polysilanes are j-conjugated polymers composed of SiSi skeletons and organic pendant
groups. They are insulators with lled intramolecular valence bands and empty intra-
molecular conduction bands. However, because of strong j conjugation, they have rather
narrow band gaps of less than 4 eV [24,25] and are converted to conductors by photo-
excitation or by doping electron donors or acceptors. Recently they have attracted much
attention because of their potential utility as one-dimensional conductors, nonlinear
optical materials, and electroluminescent materials [2628].
The optical and electronic functions of polysilanes owe to their delocalized highest
occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO)
that are occupied by holes and conduction electrons, respectively. The polymer does not
show high conductivity or optical nonlinearity if the electrons or holes are localized on a
small part of the polymer chain. To elucidate the structure of HOMO and LUMO is
therefore important for the molecular design of polysilanes as functional materials.
Analyses of the electronic and electron spin resonance (ESR) spectra of the radical
cation and anion of polysilanes make it possible to elucidate the structure of HOMO and
LUMO, because an unpaired electron in the radical anion or cation occupies HOMO or
LUMO, respectively. As schematically depicted in Fig. 10, the radical ions of polysilanes
show absorption bands in UV and near-IR regions [2931]. The former band corresponds
to intraband transitions between valence and conduction bands. The latter band corre-
sponds to transitions within the valence or the conduction band [32,33]. Because the near-
Figure 9 Comparison of the SEC proles of coupled polystyrene g-irradiated at 295 K in vacuum
without additive (left) and with 0.07 mol/kg of hydroquinone (right).
Copyright 2004 by Taylor & Francis Group, LLC
IR band is characteristic to the transition of a hole in the radical cation or a conduction
electron in the radical anion, analysis of the near-IR band gives useful information on the
structure of these charge carriers. Analysis of the ESR spectra of radical ions is also useful
for elucidating the electronic structure of the carriers. The ESR spectra give direct infor-
mation on the structure of charge carriers in HOMO and LUMO, because ESR transitions
take place between the spin sublevels of ground-state unpaired electrons.
The radiation chemical matrix isolation is a useful method for the spectroscopic study
of radical ions. In this method, a solution containing a small amount of solute molecules is
rapidly frozen to make a rigid solid matrix and is then irradiated at cryogenic temperature.
The radical ions of solute molecules generated by ionizing radiation are immobilized in the
matrix, so that the lifetime of otherwise unstable radical ions is long enough for precise
spectroscopic measurements. The radical anion of solute molecules is selectively obtained by
using ether or alcohol, RH, as a solvent
RHionizing radiation !RH
RH !RH
2
R formation of solvent radical by proton transfer
S solute e
!S
RX !RX
RX !RXRX
hole transfer
RX
S !RXS
2
p r
SiSi
sp
3
r
SiSi
sp
3
2
p
1
2
p 3p
x
1
2
3p
x
1
2
3s
2
p 3p
x
1
2
3p
y
1
2
3s
3p
x
1
where j
SiSi
(sp
3
) and j
SiSi
V
(sp
3
) denote hybrid orbitals composed of two adjacent SiSi
j bonds. The HOMO is composed of only Si 3p
x
atomic orbitals, so that this is a pseudo-k
orbital. The LUMO locating at the bottom of the conduction band is given by a linear
combination of the antibonding orbitals with the same phase, so that the atomic orbitals
for the LUMO are expressed by
LUMO
1
2
p r*
SiSi
sp
3
r*
SiSi
sp
3
2
p
1
2
p 3p
x
1
2
3p
y
1
2
3s
1
2
p 3p
x
1
2
3p
y
1
2
3s
2
p 3p
y
3s 2
Figure 15 Solvatochromic shift for poly(cyclohexylmethylsilane) radical anion in the mixed sol-
vent of MTHF and methylcyclohexane at 77 K.
Copyright 2004 by Taylor & Francis Group, LLC
where j*
SiSi
(sp
3
) and j*
SiSiV
(sp
3
) denote hybrid orbitals composing two adjacent anti-
bonding SiSi j* bonds. The LUMO is further coupled with SiC antibonding orbitals to
eliminate Si 3s orbital, since the ESR spectra of permethylorigosilane and polysilane
radical anions do not show
29
Si hyperne splitting arising from Fermi contact interaction
by an unpaired electron in the Si 3s orbital [32]. The actual atomic orbital for the LUMO
is therefore 3p
y
. The reason for mixing with SiC antibonding orbitals will be given in the
next section.
The delocalization of the HOMO and LUMO arises from pseudo-k orbital inter-
actions between adjacent 3p
x
and 3p
y
atomic orbitals of Si atoms, respectively. Since the 3p
orbital is on the SiSiSi plane, the degree of the interaction depends on the dihedral angle
between adjacent SiSiSi planes. Fluctuation of the dihedral angles along the polymer
skeleton then causes the so-called Anderson localization of the HOMO and LUMO [41].
Anderson localization is the localization of electrons on low-dimensional materials,
which is induced by the irregularity of the periodic potential eld [43]. Figure 17 gives a
schematic representation of Anderson localization of a particle in one-dimensional box. The
same is true for an electron on a polymer skeleton. Alocalized state in a completely periodic
Figure 16 Schematic representation of the HOMO and LUMO of polysilanes.
Figure 17 Schematic representation of Anderson localization of a particle in one-dimensional box.
Copyright 2004 by Taylor & Francis Group, LLC
eld is expressed by the superposition of steady-state delocalized wave functions of ground
and excited states, so that the energy of a localized state is always higher than that of the
delocalized ground state. Increase of the irregularity of the periodic eld causes the increase
of the ambiguity of the energies of delocalized states, which reduces the energy dierences
between the delocalized ground and excited states. The energy for generating a localized
state therefore decreases with increasing irregularity, and a localized state becomes a ground
state if the irregularity exceeds the energy gaps between adjacent delocalized states. In other
words, the irregularity of the potential energy eld allows the mixing of delocalized ground
and excited states to generate a localized ground state. The Anderson localization is easier
for a longer polymer skeleton, because the energy gaps between adjacent energy levels de-
crease with increasing length of the skeleton.
Anderson localization is dicult to take place for conformationally constrained
polymers with narrow distributions of dihedral angles. Introduction of bulky side chains is
therefore expected to enhance the delocalization of the HOMO and LUMO of polysilanes
[44]. Figure 18 shows the absorption spectra of the radical ions of poly(n-decyl-(S)-2-
methylbutylsilane) that is a rigid rodlike polysilane with narrow distribution of the
torsional angles [45,46]. The energy of the near-IR bands is much lower than that of
poly(cyclohexylmethylsilane), which indicates that the Anderson localization is suppressed
in poly(n-decyl-(S)-2-methylbutylsilane). The absorption spectrum of the radical anion of
poly(n-decyl-(S)-2-methylbutylsilane) in nonpolar methylcyclohexane matrix is the same as
that in polar MTHF. Observation of no solvatochromic shift also indicates the suppression
of the Anderson localization, because the solvatochromic shift is not expected for a
polymer ion with extended charge distribution.
The sharpness of the UV absorption and emission bands corresponding to HOMO X
LUMO transition is also a measure of delocalization. The transition energy depends on the
location of a polymer skeleton where a photon is absorbed or emitted, if electron orbitals are
localized on parts of the skeleton. A sharp UV band of poly(n-decyl-(S)-2-methylbutylsi-
lane) indicates the delocalization of HOMO and LUMO. Table 2 summarizes the location
Figure 18 Absorption spectra of poly(n-decyl-(S)-2-methylbutylsilane) at 295 K in MTHF and its
radical cation and anion at 77 K in g-irradiated freon and MTHF matrices, respectively.
Copyright 2004 by Taylor & Francis Group, LLC
and the width of the UV absorption and emission bands for several polysilanes. Polysilanes
with pendant groups composed of zC
6
linear alkanes and branched alkanes with the branch
at the h position show sharp UV bands and low Stokes shifts, which indicates that these
pendant groups are eective for stretching the polymer skeleton [47] and therefore for
preventing the Anderson localization.
The structures of polysilanes with exible and rigid skeletons are schematically shown
in Fig. 19 together with the distribution of charge carriers on the skeletons. In conclusion, to
replace the pendant groups of polysilanes with bulky ones is an eective way of suppressing
the undesired Anderson localization of charge carriers on a part of the polymer skeleton. An
ideal polysilane quantum wire with high intramolecular charge mobility can be synthesized
by using bulky pendant groups.
Table 2 Absorption and Emission Bands of Polysilanes in MTHF at 296 K
Absorption band
a
Emission band
a
Pendant groups M
n
/10
4
M
w
/10
4
k
max
Width Stokes shift Width
1-Methylpropyl, decyl 0.39 0.52 no peak
(S)-2-methylbutyl, decyl 67.2 94.1 3.84 0.10 0.05 0.12
2-Methylpropyl, decyl 1.61 3.15 3.88 0.10 0.05 0.15
2-Ethylbutyl, decyl 1.25 2.66 3.83 0.1 0.04 0.10
(2-Methylpropyl)
2
0.22 0.24 3.97 0.22 0.25 0.21
3-Methylpentyl, decyl 29.5 41.3 3.87 0.22 0.13 0.28
n-Butyl, decyl 1.74 7.53 3.91 0.45 0.29 0.24
2-Methylpropyl, methyl 8.93 18.8 4.21 0.62 0.5 0.28
2-Methylpropyl, ethyl 3.55 5.23 3.97 0.28 0.27 0.32
2-Methylpropyl, propyl 2.38 2.93 4.05 0.28 0.20 0.27
2-Methylpropyl, butyl 3.60 5.30 3.98 0.18 0.12 0.19
2-Methylpropyl, pentyl 0.76 4.72 3.92 0.14 0.07 0.17
2-Methylpropyl, hexyl 1.94 3.55 3.88 0.10 0.04 0.15
a
In electron volts.
Figure 19 Schematic representation of the molecular structures of exible polysilane with Ander-
son-localized charge carrier (upper) and rigid polysilane with delocalized charge carrier.
Copyright 2004 by Taylor & Francis Group, LLC
2.2. Intermolecular Migration of Charge Carriers
It has been established from conductivity measurements that thermally activated and eld-
assisted hole hopping is responsible for the charge transport in solid polysilanes [48,49].
The mobility of the hole is as high as 10
8
m
2
/V sec, while the mobility of the electron is a
few orders of magnitude lower. In this section, we will show the reason why only the hole
is mobile in polysilanes and how we can construct electron-conductive polysilanes.
It has been shown in the previous section that the structure of conduction electrons on
polysilane skeletons is essentially the same as that of holes. The suppression of electron
conduction is therefore attributable to low electron conductivity of side chains, because
interchain charge hoppings through the side chains are necessary for the migration of con-
duction electrons between polymer molecules [50]. Figure 20 compares the ESR spectra of
the radical anion and cation of poly(cyclohexyl methylsilane) and poly(methylphenylsilane)
in MTHF and freon mixture, respectively. The line width for the radical anions primarily
arises from the anisotropy of the ESR g factors. The two radical anions show the same g
factors, which indicates that the helical structures of the polymer skeletons are the same [51].
The spectrum of the radical cation is more than ve times broader than that of the radical
anion. This is caused by stronger hyperne interactions between the unpaired electron and
the protons of the side-chain alkyl groups. The stronger hyperne interaction arises fromthe
delocalization of the hole onto the alkyl substituents. The lack of a detectable hyperne
interaction in the polysilane radical anion suggests that the excess electron in the LUMO or
the intramolecular conduction band is not delocalized onto the side chains so that the side
chains act as good intermolecular insulators for the conduction electron. On the other hand,
since the hole is delocalized onto the side chains, it can migrate to an adjacent main chain via
the side chains.
The optical absorption spectra of polysilane radical ions also support the validity of
the above-mentioned mechanism of charge migration. Figure 21 compares the absorption
Figure 20 Comparison of the ESR spectra of radical cations and anions of poly(cyclohexylme-
thylsilane) and poly(methylphenylsilane) at 77 K in g-irradiated freon and MTHF matrices,
respectively.
Copyright 2004 by Taylor & Francis Group, LLC
spectra of the radical anion and cation of both poly(cyclohexylmethylsilane) and poly-
(methylphenylsilane). The near-IR band of the radical anion is scarcely inuenced by the
substitution of the side chains from the cyclohexyl group to the phenyl group. On the
other hand, the near-IR band of the radical cation is blue-shifted by the substitution. This
indicates that the unpaired electron of the radical anion is conned within the main chain,
whereas that of the radical cation is extended onto the side chains.
The delocalization of the conduction electron onto the side chains would be expected
if the pendant groups were replaced with more electrophilic substituents than the phenyl
group. However, this is not the case. Figure 22 shows the absorption spectrum of poly-
(methylnaphthylsilane) radical anion. The absorption spectrum is very similar to that of
the naphthalene radical anion, which implies that the unpaired electron is localized on the
pendant group. Increase of the electron anity of pendant groups does not necessarily
cause the delocalization.
Replacing all the pendant groups with aryl groups attains the delocalization of the
unpaired electron. Figure 23 compares the absorption spectra of the radical anions of poly-
(4-ethylphenyl-phenylsilane) and poly(dicyclohexylsilane) in MTHF at 77 K. Although the
molecular structures of these polysilanes are similar, the absorption spectra are dierent.
The absorption spectrum of the poly(dicyclohexylsilane) radical anions is similar to that of
Figure 21 Comparison of the electronic absorption spectra of poly(cyclohexylmethylsilane) and
poly(methylphenylsilane) radical anions and cations at 77 K in g-irradiated MTHF and freon
matrices, respectively.
Copyright 2004 by Taylor & Francis Group, LLC
the poly(cyclohexylmethylsilane) radical anion; the unpaired electron is conned within the
main chain. The poly(4-ethylphenylphenylsilane) radical anion shows a blue-shifted near-
IR band and several peaks in visible and near-IR regions. The blue shift and the peaks
strongly suggest that the unpaired electron is extended onto the pendant groups. Recent
studies on the electroluminescence of polysilanes revealed that poly(di-4-t-butylphenylsi-
lane) is electron conductive [52], which also supports the delocalization of conduction
electron onto the pendant groups. Table 3 summarizes the experimental results on the near-
Figure 22 Comparison of the electronic absorption spectra of poly(methylnaphthylsilane) and
naphthalene radical anions at 77 K in g-irradiated MTHF matrices.
Figure 23 Comparison of the electronic absorption spectra of poly(4-ethylphenyl-phenylsilane)
and poly(dicyclohexylsilane) radical anions at 77 K in g-irradiated MTHF matrices.
Copyright 2004 by Taylor & Francis Group, LLC
IR bands of polysilane radical anions and the location of conduction electrons. Except for
poly(diarylsilanes), the conduction electron is conned either in the skeleton or in one of the
pendant groups.
The dierence of the spatial distribution of the conduction electron and hole on poly-
silanes can be explained by taking the shapes of the molecular orbitals on the main and
side chains into account. Figure 24 shows HOMO and LUMO of the model compounds of
polysilanes obtained by semiempirical PM3 calculations. The HOMO of all the model
compounds is extended to both the SiSiSi skeletons and pendant groups. The LUMO of
SiHC
2
H
5
(SiH
3
)
2
, the model compound of poly(dialkylsilane), and SiHC
6
H
5
(SiH
3
)
2
, the
model compound of poly(methylphenylsilane), is not extended onto the pendant group.
The LUMOof SiHC
10
F
7
(SiH
3
)
2
, the model compound of poly(methylnaphthylsilane), is on
the pendant group but not on the skeleton. Although the LUMO+1 of SiHC
10
F
7
(SiH
3
)
2
has approximately the same energy as that of LUMO, it does not mix with LUMO but is
localized on the pendant group. These results indicate that the mixing of the LUMO of the
skeleton and the pendant group is very dicult.
The diculty of the mixing arises fromthe symmetry of the atomic orbitals of LUMO
on the skeleton with respect to molecular orbitals on the pendant group. Let us assume
orbital interactions between polymer skeletons and pendant groups are much weaker than
those within the skeletons and within the pendant groups. The molecular orbitals of poly-
silanes are then divided into three parts, a delocalized orbital on a polymer skeleton, a
delocalized orbital on a pendant group, and a localized orbital on an SiC bond. The entire
orbitals are reconstructed from these partial orbitals by treating orbital interactions be-
tween the skeletons and the pendant groups as perturbations. Figure 25 illustrates the
shape of the partial orbitals before and after the reconstruction. As shown in Fig. 15, the
atomic orbitals of the delocalized HOMO and LUMO on the Si skeleton are given by 3p
x
and 3p
y
3s=
2
p
, respectively. The atomic orbital on the SiC carbon of the delocalized
HOMOof alkyl or aryl groups is 2p
x
, so that the delocalized HOMO on the Si skeleton and
pendant groups are mixed with each other through k-type interactions with 3p
x
and 2p
x
Table 3 Near-IR Spectra and the Location of Conduction Electron on Polysilanes
Pendant 1 Pendant 2 E
a
(pendant 2)/eV
Near-IR
band/eV
Electron
location
Methyl Phenyl 1.15 0.8 Main chain
Methyl p-Fluorophenyl 0.89 0.8 Main chain
Methyl Triuorophenyl Triuorophenyl
anion
Pendant 2
Methyl 1-Naphthyl 0.19 Naphthalene
anion
Pendant 2
Phenyl 4-Ethylphenyl 1.17 1.1 Main and
side chains
Phenyl 4-Propylphenyl Approx. 1.1 1.1 Main and
side chains
Phenyl 4-tert-Butylphenyl 1.06 1.1 Main and
side chains
Cyclohexyl Cyclohexyl 0.8 Main chain
Cyclohexyl Methyl 0.8 Main chain
Copyright 2004 by Taylor & Francis Group, LLC
atomic orbitals. The delocalized HOMO on the Si skeleton does not mix with the SiC j
orbitals, because 3p
x
is orthogonal to the j orbitals.
The LUMO on the Si skeleton interacts not with the LUMO on pendant groups but
with SiC antibonding j orbitals. The atomic orbital of the LUMO, 3p
y
3s=
2
p
, is
divided into two sp
3
j orbitals with the symmetry axes pointed toward two SiC bonds, as
1
2
p 3p
y
3s
2
p
1
2
3p
y
2
p 3p
z
1
2
3s
1
2
3p
y
2
p 3p
z
1
2
3s
3
These orbitals completely overlap with two SiC antibonding orbitals, since the antibond-
ing orbitals are expressed by
1
2
p
1
2
3p
y
F
1
2
p 3p
z
1
2
3s
2
p carbon sp
3
4
The LUMO on the Si skeleton therefore mixes up with the antibonding orbitals to generate
3p
y
atomic orbitals, as
Figure 24 Isosurfaces at F0.02 of molecular orbitals of the model compounds of polysilanes ob-
tained by semiempirical PM3 calculation.
Copyright 2004 by Taylor & Francis Group, LLC
1
2
p
1
2
p 3p
y
3s
2
p
1
2
3p
y
2
p 3p
z
1
2
3s
2
p carbon sp
3
1
2
p
1
2
3p
y
2
p 3p
z
1
2
3s
2
p carbon sp
3
2
5
3p
y
1
2
carbon sp
3
1
carbon sp
3
2
The LUMOon the Si skeleton is impossible to be extended onto the pendant group because
the SiCantibonding orbital is orthogonal to the delocalized LUMOon the pendant group.
Observation of no Si 3s component on the ESRspectra of oligosilane and polysilane radical
anions [32] supports the conclusion that the LUMO on the Si skeleton is mixed with the
antibonding SiC j orbitals. The energy diagram for the orbital interactions is also shown
in Fig. 25.
Figure 25 Energy diagrams for orbital interactions between polysilane skeleton and pendant group.
HOMO
Si
, HOMO
p
and LUMO
Si
, LUMO
p
denote the HOMO and LUMO delocalized on the skele-
ton and a pendant group, respectively. j
Si-C
and j*
Si-C
are bonding and antibonding SiC j orbitals,
respectively.
Copyright 2004 by Taylor & Francis Group, LLC
The mechanism of orbital interactions mentioned above is not applicable if each
monomer unit of polysilanes has sucient electron anity and two pendant groups of the
monomer unit have approximately the same electron anity. The LUMO of polysilanes is
then regarded as a series of the LUMOs of monomer units that is perturbed by SiSi bonds.
The LUMO of the monomer unit has some electron density on the Si atom, since the
unpaired electron in the LUMO is delocalized over the two pendant groups with approxi-
mately the same electron density. Connection of the monomer units with SiSi bonds
therefore causes the delocalization of the LUMO on adjacent monomer units and therefore
on both the Si skeleton and pendant groups.
In conclusion, radiation chemical matrix isolation technique gives precise informa-
tion on the electronic structure of charge carriers, which is useful for the molecular design
of organic semiconductors as electronic and optical materials.
REFERENCES
1. Clough, R.L. Nucl. Instrum. Methods B 2001, 185, 8.
2. Tolfree, D.W.L. Rep. Prog. Phys. 1998, 61, 313.
3. Melngaillis, J.; Mondelli, A.A.; Berry, I.L. III; Mohondro, R. J. Vac. Sci. Technol. B 1998, 16,
927.
4. Wallra, G.M.; Hinsberg, W.D. Chem. Rev. 1999, 99, 1801.
5. Cerrina, F. J. Phys. D: Appl. Phys. 2000, 33, R103.
6. Hatzakis, M. J. Electrochem. Soc. 1969, 116, 1033.
7. Bowden, M.J. Materials for microlithography. ACS Symp. Ser. 1984, 266, 39.
8. Reichmanis, E.; Thompson, L.F. Chem. Rev. 1989, 89, 1273.
9. Iwayanagi, T.; Ueno, T.; Nonogaki, S.; Itoh, H.; Willson, C.G. Electronic and photonic
applications of polymers. Adv. Chem. Ser. 1988, 218 (Chapter 3).
10. Bouden, M.J.; Thompson, L.F. J. Appl. Polym. Sci. 1973, 17, 3211.
11. Ichikawa, T.; Yoshida, H. J. Polym. Sci. Part A: Polym. Chem. 1990, 28, 1185.
12. Yates, B.M.; Shiozaki, D.M. J. Polym. Sci. Part B: Polym. Phys. 1991, 31, 1779.
13. Ichikawa, T.; Oyama, K.; Kondoh, T.; Yoshida, H. J. Polym. Sci. Part A: Polym. Chem. 1994,
32, 2487.
14. Bignozzi, M.C.; Ober, C.K.; Nobvembre, A.J.; Knurek, C. Polym. Bull. 1999, 43, 93.
15. Barclay, G.G.; King, M.; Orellana, A.; Malenfant, P.R.L.; Sinta, R.; Malmstrom, E.; Ito, H.;
Hawker, C.J. Organic thin lms. ACS Symp. Ser. 1998, 695, 360.
16. Ichikawa, T.; Ueda, H.; Koizumi, H. Chem. Phys. Lett. 2002, 363, 13.
17. Lias, S.G.; Bartmess, J.E.; Liebman, J.F.; Holmes, J.L.; Levin, R.D.; Mallard, W.G. J. Phys.
Chem. Ref. Data 1988, 17(Suppl. 1).
18. Izumida, T.; Ichikawa, T.; Yoshida, H. J. Phys. Chem. 1980, 84, 60.
19. Ogasawara, M.; Tanaka, M.; Yoshida, H. J. Phys. Chem. 1987, 91, 937.
20. Shida, T. Electron Absorption Spectra of Radical Ions. Elsevier: Amsterdam, 1998; 85 pp.
21. George, M.K.; Veregin, R.P.N.; Kazmainer, P.M.; Hamer, G.K. Trends Polym. Sci. 1994,
2, 66.
22. Hawker, C.J. Acc. Chem. Res. 1997, 30, 373.
23. Sawamoto, M.; Kamigaito, M. Trends Polym. Sci. 1996, 4, 371.
24. Takeda, K.; Shiraishi, K. Phys. Rev. B 1989, 39, 11028.
25. Yokoyama, K.; Yokoyama, M. Chem. Lett. 1989, 1989, 1005.
26. Miller, R.D.; Michl, J. Chem. Rev. 1989, 89, 1359.
27. Michl, J. Synth. Met. 1992, 4950, 367.
28. Suzuki, H.; Hoshino, S.; Furukawa, K.; Ebata, K.; Yuan, C.H.; Bleyl, I. Polym. Adv. Technol.
2000, 11, 460.
Copyright 2004 by Taylor & Francis Group, LLC
29. Irie, S.; Oka, K.; Nakao, R.; Irie, M. J. Organomet. Chem. 1990, 388, 253.
30. Irie, S.; Irie, S.M. Macromolecules 1992, 25, 1766.
31. Ushida, K.; Kira, A.; Tagawa, S.; Yoshida, Y.; Shibata, H. Proc. Am. Chem. Soc. Div. Polym.
Mater. 1992, 66, 299.
32. Kumagai, J.; Yoshida, H.; Koizumi, H.; Ichikawa, T. J. Phys. Chem. 1994, 98, 13117.
33. Kumagai, J.; Yoshida, H.; Ichikawa, T. J. Phys. Chem. 1995, 99, 7965.
34. Ichikawa, T. J. Phys. Chem. B 2002, 106, 10684.
35. Ban, H.; Sukegawa, K.; Tagawa, S. Macromolecules 1987, 20, 1775; 1988, 21, 45.
36. Seki, S.; Matsui, Y.; Yoshida, Y.; Tagawa, S.; Koe, JR.; Fujiki, M. J. Phys. Chem. B 2002, 106,
6849.
37. Seki, S.; Yoshida, Y.; Tagawa, S. Radiat. Phys. Chem. 2001, 60, 411.
38. Seki, S.; Kunimi, Y.; Nishida, K.; Yoshida, Y.; Tagawa, S. J. Phys. Chem. B 2001, 105, 900.
39. Seki, S.; Kunimi, Y.; Nishida, K.; Aramaki, K.; Tagawa, S. J. Organomet. Chem. 2000, 611,
64.
40. Seki, S.; Yoshida, Y.; Tagawa, S.; Asai, K. Macromolecules 1999, 32, 1080.
41. Ichikawa, T.; Kumagai, J.; Koizumi, H. J. Phys. Chem. B 1999, 103, 3812.
42. Ichikawa, T.; Koizumi, H.; Kumagai, J. J. Phys. Chem. B 1997, 101, 10698.
43. Skinner, J.L. J. Phys. Chem. 1994, 98, 2503.
44. Ichikawa, T.; Yamada, Y.; Kumagai, J.; Fujiki, M. Chem. Phys. Lett. 1999, 306, 275.
45. Fujiki, M. J. Am. Chem. Soc. 1994, 116, 6017.
46. Fujiki, M. Appl. Phys. Lett. 1994, 65, 3251.
47. Terao, K.; Terao, Y.; Teramoto, A.; Nakamura, N.; Terakawa, I.; Sato, T. Macromolecules
2001, 34, 2682.
48. Samuel, L.M.; Sanda, P.N.; Miller, R.D. Chem. Phys. Lett. 1989, 159, 227.
49. Vanderlaan, G.P.; Dehaas, M.P.; Hummel, A.; Frey, H.; Moller, M. J. Phys. Chem. 1996,
100, 5470.
50. Kumagai, J.; Tachikawa, H.; Yoshida, H.; Ichikawa, T. J. Phys. Chem. 1996, 100, 16777.
51. Ichikawa, T.; Sumita, M.; Kumagai, J. Chem. Phys. Lett. 1999, 307, 81.
Copyright 2004 by Taylor & Francis Group, LLC
22
Applications to Reaction Mechanism Studies
of Organic Systems
Tetsuro Majima
Osaka University, Osaka, Japan
1. INTRODUCTION
1.1. Radical Ions and the Generation Methods
Radical cation and anion (M
+/
+/
+/
+/
can generate M
+/
+
*
/
+
*
/
+/
+/
+/
+/
and M
+
, respectively. Laser ash photolysis (LFP) is a useful
technique for the kinetic study involving M
+/
. However, generation of M
+/
in
solution by PET must be accompanied by ion pair formation, e.g., M
+/
, M
+
/anion,
etc. Because solvated electron (e
s
+
in RTPI,
recombination of e
s
and M
+
decreases the yield of M
+
. Such accompanied species
inuence the transient behavior of M
+/
+
or M
+/
+
, M
, or M in the triplet
Copyright 2004 by Taylor & Francis Group, LLC
excited state (
3
M*) at nsec100 Asec after an electron pulse (e
+
is generated during PR of M in alkylhalides such as 1,2-dichloreethane (DCE) or
butyl chloride (BC), M
+/
are shown in Eqs. (16). Because electron and hole are trapped by DCE
and DMF, respectively, as stable species, the reactivities of M
+
and M
can be studied in
the time scale of nsec100 Asec.
DCE QDCE
s
1
DCE
M ! M
DCE 2
e
s
DCE ! CICH
2
CH
2
CI
3
DMF QDMF
s
4
e
s
M ! M
5
DMF
DMF ! DMF H
DMF H
6
Either M
+
or M
at 10
5
M is generated during PR as the only species absorbing
light in the visible and near-infrared (NIR) regions [4254]. Because free M
+
or M
can
be generated during PR, eects of counter ionic species and charge recombination are not
involved in the time scale of nsec100 Asec. Thus PR has been extensively used for the
kinetic study involving various M
+/
.
The g-R of glassy rigid matrix of degassed BC or MTHF solutions containing M at
77 K can generate M
+
or M
+
or M
+
or M
+
or M
with other
molecules, cannot be carried out because of the rigid matrix.
Because M
+/
+/
+
*
/
*. The photochemical
transformation of M
+/
+
*
/
. The single hv
532
has a
0.5-cm diameter, 5-nsec duration, 100 mJ, and an incident photon number of 1.4 10
18
pulse
1
cm
2
. The laser beam was directed in the opposite direction of the electron beam
through the sample solution. The laser beam intersected the electron beam at an angle of
20j. An almost collinear arrangement was used, such that the narrow laser beam passed
through the central part of the irradiated volume. Transient absorption was measured by
using the probe beam mentioned above. The PRLFP combined method at low temper-
atures over the range of 140300 Kwas performed in the cell that was mounted in a variable-
temperature liquid-nitrogen cryostat.
Transient uorescence was measured without the probe beam. Fluorescence from the
sample solution was collected with a 50-mmlens with focal length of 70 mmand was focused
into a quartz optical ber with diameter 250 mm and length of 10 m, which transported the
uorescence to a gated-multichannel/monochromatic spectrometer. The gate time of the
image intensier in the uorescence was set to be longer than 20 nsec, typically 50 nsec, to
eliminate the eect of the mode dispersion of the optical ber. The emission from the
sample was measured for e/hv
532
, e
, and hv
532
irradiations. Fluorescence spectra were
obtained by subtraction of the spectra for e
and hv
532
irradiations from that for e
/hv
532
irradiation to eliminate the Rayleigh and Raman scatterings and other emissions induced
by hv
532
irradiation. Because e
+
) have been
reported from the synthetic and mechanistic points of view [1,2,4,8,42,6470]. It is well
known that cis(c)trans(t) isomerization, dimerization, and addition of a nucleophile occur
in ArO
+
as typical reactions [1,2,4,8,42,6470]. The reactivities depend on the structure or
substituents of ArO
+
. No unimolecular ct isomerization occurs in stilbene
+
(St
+
) [64
66], although it does so in c-4,4V-dibromostilbene
+
and c-4,4V-dimethylstilbene
+
[66].
Dimerization of St
+
with St occurs at k
d
= (3.53.9) 10
8
M
1
sec
1
[42,67] to yield k-
dimer
+
(k-St
2
+
) with overlapping k-electrons between the two benzene rings, which
converts to j-dimer
+
(j-St
2
+
) with an acyclic linear structure having both a radical and a
cation on the 1- and 4-positions of the C
4
linkage, decomposing to t-St
+
and St as nal
products [42]. Although t-St
+
has little reactivity toward O
2
, a high reactivity of k
O
2
=2.6
10
8
M
1
sec
1
has been reported for (E)-2,3-diphenyl-2-butene
+
[70].
Three reactions of eight stilbene derivatives
+
(S
+
= RCH j CRVRU = 1
+
8
+
)
(Scheme 1), unimolecular isomerization from c-S
+
to t-S
+
[Eq. (7)], and bimolecular
reactions with O
2
[oxidation, Eq. (8)] and a neutral S [dimerization, Eq. (9)] during PR and
g-R in DCE or BC will be described here [44,45,71].
c S
! t S
; k
i
for c t unimolecular isomerization 7
S
O
2
! RC
HCOO
CRVRVV; k
O
2
for oxidation 8
S
S ! S
2
; k
d
for dimerization 9
Scheme 1 Eight stilbene derivatives (S).
Copyright 2004 by Taylor & Francis Group, LLC
Formation and Reactions of c-S
+
and t-S
+
It is well established that 1
+
is formed during PR and g-R in alkyl halide such as DCE and
BC[40,7274]. The absorption spectra of c-S
+
and t-S
+
(S
+
=1
+
8
+
) generated by g-
Rof c-S and t-S in BCat 77 Kshowbands similar to those of c-1
+
and t-1
+
. The transient
absorption spectra immediately after e
+
7
+
and t-1
+
8
+
[40,72,73], formed from capture of the hole generated in the initiation step during PR of S
in DCE (Figs. 1 and 2, Table 1). It is well known that t-1
+
shows two bands at 480 and
750 nm assigned to D
2
pD
0
and D
1
pD
0
transitions, respectively, while c-1
+
also shows
two bands at 515 and 780 nm, respectively [40,72,73]. The absorption bands of c-1
+
are
broader and weaker than those of t-1
+
. Delocalization of k-electrons is less in c-1
+
than
in t-1
+
.
The transient absorptions of t-S
+
decayed monotonously with rst-order kinetics
and did not show any shifts or spectral changes. The transient absorptions of c-1
+
, c-2
+
,
and c-6
+
without p-methoxyl group ( p-MeO) decayed with k
obs
, in a similar way to those
of t-1
+
8
+
. On the other hand, formation of the band of t-3
+
5
+
and t-8
+
after e
+
5
+
and c-8
+
with p-MeO, respectively. k
i
= 4.5 10
6
to 1.4 10
7
sec
1
for the unimolecular was isomerization of c-S
+
to t-S
+
was calculated from the rise of the peaks of t
+
, which did not depend on [c-S].
Dimerization was observed in 1
+
3
+
and t-6
+
, because the decay of the absorp-
tion of 1
+
3
+
and t-6
+
increased with increasing [S], respectively. The transient
absorption of c-1
+
shifted to that of t-1
+
during the decay because of the bimolecular
isomerization of c-1
+
to t-1
+
via j-1
2
+
. Similarly, the transient absorption of c-2
+
shifted to that of t-2
+
through bimolecular isomerization. k
d
= (2.04.3) 10
8
M
1
sec
1
were calculated from the dependence of k
obs
on [S] (Table 2). No dimerization of c-6
+
was
observed, because k
obs
did not depend on [c-6]. The transient absorption of c-7
+
with p-
MeO and methyl group (Me) on the olenic carbon decayed with k
obs
according to rst-
order kinetics without any shifts or dependence of [c-7]. In other words, neither unim-
olecular isomerization nor dimerization was observed in c-7
+
.
The k
obs
of t-S
+
(S =3
+
5
+
and 7
+
) with p-MeOwas accelerated in the presence
of O
2
, and k
O
2
= (1.24.5) 10
7
M
1
sec
1
for the oxidation was calculated (Table 2). On
the other hand, no oxidation was observed on a similar time scale in t-1
+
, t-2
+
, and t-6
+
without p-MeO nor in t-8
+
with two p-MeOs. Oxidation of c-7
+
was also observed with
k
O
2
similar to that for t-7
+
, while no oxidation was observed in c-1
+
, c-2
+
, and c-6
+
without p-MeO and in c-8
+
with two p-MeOs. In the case of c-3
+
5
+
with p-MeO,
k
O
2
could not be estimated because of the fast unimolecular isomerization of c-3
+
5
+
to
t-3
+
5
+
at k
i
= 4.5 10
6
1.3 10
7
M
1
sec
1
.
Dimerization Mechanism of c-S
+
and t-S
+
Decay of c-1
+
has been analyzed by neutralization with Cl
+
with c-1 occurs at k
d
c
=3.5
10
8
M
1
sec
1
at r.t. to formthe k-type dimer
+
of c-1 [k-(c-1)
2
+
, Eq. (11)], which changes
to j-type dimer
+
of 1 [j-1
2
+
, Eq. (12)] [42]. Such fast formation of j-1
2
+
was conrmed by
absorption measurement during g-R of c-1 in BC rigid matrix at 77 K(see Section 2.2). The
unimolecular isomerization of c-1
+
to t-1
+
does not occur, although the bimolecular
isomerization of c-1
+
to t-1
+
is proposed to occur through j-1
2
+
on the time scale of
100300 nsec at (510) 10
3
M at r.t. [42,67]. Thermally unstable j-1
2
+
decomposes
Copyright 2004 by Taylor & Francis Group, LLC
Figure 1 Transient absorption spectra of c-S
+
and t-S
+
(S
+
=1
+
4
+
) recorded at time t after
e
+
and t-S
+
(S
+
= 5
+
8
+
) recorded at time t after
e
+
and t-S
+
(S
+
=1
+
8
+
) Obtained by Pulse Radiolysis (PR) of c-S
and t-S (S=18) (5.010
3
M) in 1,2-Dichloroethane (DCE) at r.t.
a
k
max
[nm] (e
k
[M
1
cm
1
])
S
+
c-S
+
(D
2
) t-S
+
(D
2
) c-S
+
(D
1
) t-S
+
(D
1
)
1
+
515 480 780 750
(180,000) (65,000) (6,180) (11,400)
2
+
520 490 780 790
(30,900) (80,900) (11,500) (19,600)
3
+
530 470,500 nd 720,800
(13,200) (39,200; 74,800) (14,700; 41,700)
4
+
530 480 nd 680
(nd) (105,000) (14,700)
5
+
540 480 nd 700
(nd) (84,600) (14,700)
6
+
520 460,510 700 >900
(3,270) (24,500; 208,000) (1,240) (>11,000)
7
+
450,500 440,490 760,780 740,800
(8,820; 9,710) (18,400; 29,400) (7,940; 7,940) (9,810; 184,000)
8
+
nd 480,540 nd 780,880
(33,100; 76,000) (24,500; 61,300)
a
e
k
was estimated from DO.D.
E
and the optical length of 1 cm with the assumption of [S
+
] = 8 10
6
M which
was calculated from DO.D.
480
, the optical length of 1 cm, and e
480
= 6.5 10
4
M
1
cm
1
for 1
+
in DCE.
Table 2 Rate Constants for the Reactions of S
+
(S
+
=1
+
8
+
) During PR in DCE at r.t.
a
S
+
c-S
+
t-S
+
k
i
[sec
1
] k
d
[M
1
sec
1
] k
o
2
[M
1
sec
1
] k
d
[M
1
sec
1
] k
o
2
[M
1
sec
1
]
1
+
<10
6
3.9 10
8
<10
6
3.9 10
8
<10
6
2
+
<10
8
3.4 10
8
<10
6
4.3 10
8
<10
6
3
+
4.5 10
6
2.0 10
8
nd 3.0 10
8
4.5 10
7
4
+
1.3 10
7
nd nd <10
6
4.0 10
7
5
+
1.4 10
7
nd nd <10
6
1.2 10
7
6
+
<10
6
<10
6
<10
6
3.410
8
<10
6
7
+
<10
6
<10
6
2.5 10
7
<10
6
2.8 10
7
8
+
nd nd <10
6
<10
6
<10
6
a
k
i
, calculated from the formation of t-S
+
at 5.010
3
M of c-S; k
d
, measured from the dependence of the decay
of S
+
on [S]; k
O
2
, obtained from the dependence of the decay of c-S
+
or t-S
+
on [O
2
]. k
i
<10
6
sec
1
denotes
that the formation of t-S
+
was not observed. k
d
<10
6
M
1
sec
1
and k
O
2
<10
6
M
1
sec
1
denote that there was
no dependence of the decay of c-S
+
or t-S
+
on [c-S], [t-S], or [O
2
].
Copyright 2004 by Taylor & Francis Group, LLC
rapidly into the thermodynamically stable t-1
+
and t-1 [Eq. (13)], as proposed [42,67].
Therefore the bimolecular isomerization is enhanced with increasing [c-1].
S
Cl
! neutral products; k
n
for neutralization 10
c S
c S ! p c S
2
S 1 3; k
c
d
11
p c S
2
! r S
2
12
r S
2
! t S
t S 13
Concentration eects suggest that the bimolecular isomerization of c-2
+
and c-3
+
to t-2
+
and t-3
+
involve the dimerization of c-2
+
and c-3
+
to form k-(c-2)
2
+
and k-
(c-3)
2
+
and then j-2
2
+
and j-3
2
+
[Eqs. (11) and (12), respectively]. Similar k
d
c
for c-1
+
c-
3
+
indicates that p-Me and p-MeO substitutions have little eect on the dimerization of
c-S
+
. Decay of the absorption of other c-S
+
at higher [c-S] could not be detected, because
the decay was too fast to be detected within the experimental limits, or [c-48] could not be
increased because of low solubilities in DCE. k
d
c
for c-8
+
with p-MeO is expected to be 10
8
M
1
sec
1
from analogy with c-3
+
, while the steric hindrance resulting from the twisted
structure with the 3,5-dimethoxyphenyl ring rotating relative to the CjC double bond
and to Me on the olenic carbon probably inhibits the dimerization of c-6
+
and 7
+
,
respectively.
Similarly to c-1
+
, t-1
+
decays via neutralization with Cl
+
of t-1 [k(t1)
2
+
, Eq. (14)]
with overlapping of k-electrons between the two benzene rings of t-1
+
and t-1.
3d,11
Neither
the unimolecular nor the bimolecular isomerization of t-1
+
to c-1
+
occurs. It is found that
k-(t-1)
2
+
generated at 77 K converts to j-1
2
+
upon warming [Eq. (15)], and nally de-
composes into the thermodynamically stable t-1
+
and t-1 [Eq. (13)] as discussed above for
c-1
+
[42].
t S
c S ! p t S
2
S 1 3; k
d
V 14
p t S
2
! r S
2
15
t-2
+
, t-3
+
, and t-6
+
show behavior similar to that of t-1
+
(Table 2). Similar k
d
1
for t-1
+
t-3
+
and t-6
+
suggests that p-Me and m-MeO substitutions have little eect on
dimerization of t-S
+
and the similar dimerization mechanism [Eqs. (1113)]. On the other
hand, k
obs
of t-4
+
, t-5
+
, t-7
+
, and t-8
+
were much slower than those of t-1
+
3
+
and
t-6
+
because k
obs
did not change with varying concentration (5 10
3
10
1
M). This
indicates that no dimerization of t-4
+
, t-5
+
, t-7
+
, and t-8
+
occurs. It is suggested that
the dimerization involves the initial formation of k-complex with overlapping of the two
benzene rings and that the k-complex formation is inhibited by steric hindrance of the
substituents on the benzene rings and olenic carbons.
Enhancements of Unimolecular Isomerization and Oxidation by ChargeSpin
Separation
In contrast to c-1
+
and 2
+
, the unimolecular isomerization of c-3
+
5
+
and c-8
+
having
p-MeO occurs to yield t-3
+
5
+
and t-8
+
, respectively, k
i
for c-4
+
and c-5
+
are larger
than those for c-3
+
and c-8
+
. Neither the unimolecular nor bimolecular isomerization
occurs in c-6
+
and c-7
+
. The oxidation of t-3
+
5
+
and 7
+
having p-MeO with O
2
occurs at k
O
2
= (1.24.5) 10
7
M
1
sec
1
, while no oxidation occurs in c-1
+
, c-2
+
, or
c-6
+
without p-MeO and in t-8
+
with two p-MeOs. The unimolecular isomerization and
Copyright 2004 by Taylor & Francis Group, LLC
oxidation with O
2
are remarkably enhanced in 3
+
5
+
and 7
+
with p-MeOas an electron-
donating substituent on the benzene ring.
It is reported that t-1
+
has a planar structure with two benzene rings and a CjC
double bond as the most stable conformation, while c-1
+
has a twist angle of 26j between
the two benzene rings on the basis of MOcalculations [75]. Although the positive charge and
unpaired electron are delocalized on the SOMO involving k-orbital of sp
2
C atoms and
n-orbital of O atom of p-MeO in 1
+
8
+
, as shown in structure (A), a quinoid-type
structure (B) is considered to contribute in 3
+
5
+
, 7
+
, and 8
+
with p-MeO but not in
1
+
, 2
+
, and 6
+
without p-MeO (Scheme 2). Alternatively, the n-orbital of O atom of
the p-MeO and the k-orbital of the olenic h-carbon participate considerably in the SOMO
of 3
+
5
+
, 7
+
, and 8
+
. Therefore, it is suggested that p-MeO in 3
+
5
+
, 7
+
, and 8
+
causes separation and localization of the positive charge on O atom of the p-MeO and an
unpaired electron on the olenic h-carbon (chargespin separation), which induces a twisted
conformation with a large twist angle and CUC single-bond character for the central CjC
double bond in (B). Such chargespin separation is less important in 1
+
, 2
+
, and 6
+
than
in 3
+
5
+
, 7
+
, and 8
+
. It is suggested that distonic radical cations known in the gas phase
[76] can be used to explain the reactivities of radical cations even in solution [77].
Neither the unimolecular isomerization of c-7
+
nor the oxidation of t-8
+
occurred,
although 7
+
and 8
+
have p-MeO. These are possibly explained in terms of a barrier to the
twisting of the CjC double bond and the spin density on the olenic carbon, respectively.
The contribution of (B) is decreased by the electron-donating Me on the olenic carbon;
therefore, it is suggested that the single-bond character of the CjC double bond is lower
in c-7
+
than in c-3
+
and that the barrier to the twisting of the CjCdouble bond is higher
in c-7
+
than in c-3
+
. An unpaired electron appears on the olenic carbon on the side of
the p-MeOin 3
+
because of contribution of (B), while it appears on both olenic carbons in
Scheme 2 Chargespin delocalization (A) and separation (B) in S
+
.
Copyright 2004 by Taylor & Francis Group, LLC
8
+
with two symmetrical p-MeOs. Therefore, the density of an unpaired electron on the
olenic carbon in t-8
+
is lower than that in t-3
+
. Neither the unimolecular ct isomer-
ization nor the oxidation of c-6
+
with an extremely small molar absorption coecient (e) at
k
max
, but similar to that of 1,3-dimethoxybenzene
+
[78] may be interpreted by a twisted
structure with the 3,5-dimethoxyphenyl ring rotating relative to the CjC double bond,
because (B), having m-MeO, is not possible in c-6
+
.
The k
i
for c-8
+
was measured at various temperature. Fromthe Arrhenius plot of ln k
i
and the reciprocal of the temperature (T
1
), the activation energy (E
a
) and the pre-
exponential factor (A) were determined to be 1.3 kcal mol
1
and 10
7.5
sec
1
, respectively,
which are smaller than E
a
= 7.7 kcal mol
1
and A = 10
11F2
sec
1
for c-4,4V-
dibromostilbene
+
and E
a
= 3.3 kcal mol
1
and A = 10
9F2
sec
1
for c-4,4V-
dimethylstilbene
+
[66]. This result is consistent with the larger k
i
for c-8
+
than for c-
4,4V-dibromo- and c-4,4V-dimethylstilbene
+
, and suggests that E
a
predominantly controls
the isomerization. Based on the dierences in E
ox
between c-S and t-S (Table 3), E
a
is too
high to be overcome in the isomerization of t-S
+
to c-S
+
(1114 kcal mol
1
). This renders
the isomerization one-way from c-S
+
to t-S
+
for S=16.
The k
O
2
for free radical-O
2
reactions is found to be nearly equal to the diusion rate
constant (k
di
). For example, k
O
2
=2.4 10
9
M
1
sec
1
for the benzyl radical-O
2
reaction is
similar to k
di
= 6.7 10
9
M
1
sec
1
in cyclohexane. On the other hand, k
O
2
= (1.24.5)
10
7
M
1
sec
1
for the oxidation of t-3
+
5
+
and t-7
+
are found to be two orders of
magnitude smaller than k
O
2
for free radicals and k
di
= 7.8 10
9
M
1
sec
1
in DCE.
This suggests that an unpaired electron is not completely localized on the olenic carbon in
t-3
+
5
+
and t-7
+
, and that the positive charge interferes with the reactivity of t-3
+
5
+
and t-7
+
as a radical toward O
2
because of the electrophilic character of O
2
.
Chain Isomerization and Oxidation of S
+
In order to conrm the enhancement of ct isomerization and oxidation by chargespin
separation in S
+
with p-MeO, product analyses were carried out. Isomerization from c-S
to t-S was found to occur stoichiometrically with large G values (5121) and G
lim
(2010
4
)
in g-R of c-S in DCE at r.t. (Table 4). The larger G values for the yield of t-35 and t-8
during g-R of c-35 and c-8, respectively, with p-MeO than for c-1, 2, and 6 without p-
MeO suggest a smaller barrier to ct isomerization for c-3
+
5
+
and c-8
+
than for c-1
+
, 2
+
, and 6
+
because of the greater single bond character of the former central CjC
double bond for c-3
+
5
+
and 8
+
relative to c-1
+
, 2
+
, and 6
+
because of the con-
Table 3 Half-Peak Oxidation Potentials (E
1/2
ox
) of c-S and t-S (S=16) and Dierence
Between E
1/2
ox
(c-S) and E
1/2
ox
(t-S)
a
S R of RCHjCHPh E
1/2
ox
(c-S)[V] E
1/2
ox
(t-S)[V] DE
1/2
ox
(c-t)[V]
1 Ph 1.25 1.14 0.11
2 4-CH
3
C
6
H
4
1.16 1.05 0.11
3 4-CH
3
OC
6
H
4
0.94 0.82 0.12
4 2,4-(CH
3
O)
2
C
6
H
3
0.80 0.64 0.16
5 3,4-(CH
3
O)
2
C
6
H
3
0.87 0.72 0.15
6 3,5-(CH
3
O)
2
C
6
H
3
1.23 1.03 0.20
a
E
1/2
ox
(c-S) and E
1/2
ox
(t-S) of c-S and t-S at 1.0 10
2
M, reference electrode of Ag/AgNO
3
, supporting electro-
lyte of 10
1
M tetraethylammonium tetrauoroborate in acetonitrile; DE
1/2
ox
(c-t) = E
1/2
ox
(c-S)-E
1/2
ox
(t-S).
Copyright 2004 by Taylor & Francis Group, LLC
tribution of (B) with separation and localization of the positive charge on O atom of p-
MeO and an unpaired electron on the h-olenic carbon (Scheme 2). The isomerization of
c-35 and c-8 proceeds via a chain reaction mechanism involving ct unimolecular
isomerization and endergonic hole transfer (HT) from t-S
+
to c-S. On the other hand,
the isomerization of c-1 and -2 proceeds via a chain reaction mechanism involving
dimerization and decomposition into t-S
+
and t-S.
The formation of oxidation products ac in a range of Gvalues (0.73.8) during the g-
R of S in O
2
-saturated DCE suggests that ac would be produced from complicated
reactions of peroxy radicals with S(Table 5). On the other hand, the regioselective formation
of 3d with large G values (2.63.0) in oxidation of 3
+
with O
2
is explained by spin
localization on the h-olenic carbon because of the contribution of (B) in 3
+
. The results of
products analyses are essentially identical with prediction based on k
i
and k
O
2
for S
+
measured with PR. It should be emphasized that the reactivities of ct unimolecular
isomerization and reaction of S
+
with O
2
can be understood in terms of chargespin
separation induced by p-MeO.
In conclusion, c-3
+
5
+
and c-8
+
with p-MeO generated during PR or g-R
isomerize unimolecularly to the corresponding t-S
+
at k
i
= 4.5 10
6
1.4 10
7
sec
1
,
while t-3
+
5
+
and 7
+
with p-MeO are oxidized with O
2
at k
O
2
= (1.24.5) 10
7
M
1
sec
1
. On the other hand, neither the unimolecular isomerization from t-S
+
to c-S
+
nor
the oxidation with O
2
occurs in t-S
+
without p-MeO. It should be noted that an unpaired
electron is not completely localized on the olenic carbon in t-3
+
5
+
and 7
+
and that a
positive charge interferes the reactivity of t-3
+
5
+
and 7
+
as a radical toward O
2
because of the electrophilic character of O
2
.
2.2. Cycloreversion of Tetraphenylcyclobutane Radical Cations (TPCB
+
)
Cleavage of a CUC bond gives a distonic radical cation as an intermediate, while concerted
cleavage of two CUC bonds yields the corresponding ArO
+
and ArO in cycloreversion of
aryl-substituted cyclobutane
+
. Therefore, the cycloreversion mechanism is related to
dimerization of ArO
+
, where k- and j-dimers
+
are detected during PR of ArO such as
Table 4 Products From g-Radiolysis (g-R) of c-S (S=18)
a
and k
i
of c-S
+b
Product from g-R
S Conv. of c-S [%] Yield of t-S [%] G-value of t-S G
lim
k
i
[sec
1
]
1 6.0 94 18.9 30 <10
6
2 5.4 100 18.2 200 <10
6
3 32.5 76 81.5 >10
3
4.510
6
4 26.0 100 105.6 >10
4
1.310
7
5 22.7 97 94.7 >410
3
1.410
7
6 2.9 100 10.2 140 <10
6
7 2.0 85 5.4 20 <10
6
8 39.1 100 120.6 >10
4
5.510
6
a
Dose, 2.610
2
Gy; [S] = 1.0 10
2
M in DCE; r.t.; the yield of t-S based on conversion of c-S; G value of t-S,
the yield of t-S; G
lim
, the G value of t-S at [c-S]
l
in the plots of G
1
of t-S vs. [c-S]
-1
.
b
k
i
, calculated from the formation of t-S
+
during PR of c-S at 5.0 10
3
M in DCE.
Copyright 2004 by Taylor & Francis Group, LLC
Table 5 Products From g-R of S (S=18) in the Presence of O
2
a
and k
O
2
of S
+b
RCHjCRVPh G value or yield
S R RV PhCHO a RCHO b RCH[O]CHPh
c
c RCH
2
C(jO)Ph
d
d k
O
2
[M
1
sec
1
]
c-1 Ph H 1.7 0.9 0.1 <10
6
t-1 Ph H 1.6 1.1 0.3 <10
6
t-2 4-CH
3
C
6
H
4
H 1.1 1.3 0.7 0.05 <10
6
c-3 4-CH
3
OC
6
H
4
H 1.0 1.6 1.2 3.0 nd
t-3 4-CH
3
OC
6
H
4
H 0.9 1.4 1.4 2.6 4.5 10
7
t-4 2,4-(CH
3
O)
2
C
6
H
3
H 1.3 1.5 nd nd 4.0 10
7
t-5 3,4-(CH
3
O)
2
C
6
H
3
H 1.3 2.3 nd nd 1.2 10
7
t-6 3,5-(CH
3
O)
2
C
6
H
3
H 1.0 0.9 nd nd <10
6
t-7 4-CH
3
OC
6
H
4
CH
3
3.8 1.7 nd 0 2.8 10
7
c-7 4-CH
3
OC
6
H
4
CH
3
3.5 1.7 nd 0 2.5 10
7
t-8 (4-CH
3
OC
6
H
4
)
2
1.0 0.8 0.2 <10
6
a
Dose, 4.910
4
Gy; [S]=1.0 10
2
M in O
2
-saturated DCE; G value, the yield of oxidation products. Isomerization products were also observed with small G values at
higher conversions in prolonged g-R. nd denotes that the values were note measured.
b
k
O
2
, obtained from the dependence of the decay of S
+
on the concentration of O
2
during pulse radiolysis of S (5.0 10
3
M) in the presence of O
2
in DCE.
c
1-Aryl-2-phenyl-1,2-epoxyethane (aryl=R).
d
Arylmethyl phenyl ketone (aryl=R).
Copyright 2004 by Taylor & Francis Group, LLC
styrene derivatives, 1,1-diphenylethylene, and St at higher concentrations [32,67,7983].
Cycloreversion of r-1,c-2,t-3,t-4- and r-1,t-2,c-3,t-4-tetraphenylcyclobutanes
+
(t,c,t-
TPCB
+
and t,t,t-TPCB
+
, respectively) during g-R in the rigid matrix of glassy BC of
TPCB (5 10
3
M) at 77 K will be described here [42].
Cycloreversion of t,c,t-TPCB
+
Tetraphenylcyclobutane radical cation is expected to decompose into k-type t-St
2
+
, j-type
St
2
+
, or t-St
+
plus t-St. Abroad absorption spectrumof t,c,t-TPCB
+
was detected at 77 K
with weak peaks at 370, 390, 485, 550, and 780 nm, which changed to have clear peaks at
370, 485, 550, and 780 nmupon warming up to 90 K(Fig. 3a). The 48-nmpeak is assigned to
t-St
+
. Photoirradiation of t,c,t-TPCB
+
at 77 K at k over 390 nm gave bands at 400470,
470500, and 680800 nm(Fig. 3b). Warming this sample caused a large increase of the peak
at 350, 550, and 780 nm. The dierence spectrum was obtained by subtracting the spectra
after photoirradiation and warming and indicated the collapse of bands at 400470 and 470
500 nm and the formation of peaks at 350, 550, and 780 nm (Fig. 3c).
Observed spectral changes can be explained by four species: t,c,t-TPCB
+
, t-St
+
, k-
(t-St)
2
+
with peaks at 400470 and 680800 nm [84], and j-St
2
+
with peaks at 350, 550, and
780 nm. Thus t,c,t-TPCB
+
converts to j-St
2
+
upon warming [Eq. (16)], while photo-
irradiation of t,c,t-TPCB
+
at 77 K causes cycloreversion into k-(t-St)
2
+
[Eq. (17)], which
converts to j-St
2
+
upon warming [Eq. (18)]. Finally, j-St
2
+
decomposes into the thermo-
dynamically stable t-St
+
and t-St [Eq. (19)].
t; c; t TPCB
heat ! r St
2
16
t; c; t TPCB
hv ! p t St
2
17
p t St
2
heat ! r St
2
18
r St
2
heat ! t St
t St 19
Cycloreversion of t,t,t-TPCB
+
A broad absorption spectrum of t,t,t-TPCB
+
was observed at 77 K, at 400670 nm with a
peak at 480 nm, and collapsed without any particular spectral change upon warming up to
90 K (Fig. 4). However, the spectrum changed to have clear peaks at 485 and 780 nm upon
photoirradiation at 77 K (Fig. 4a). The absorption spectrum then changed to have a peak
at 550 nm upon warming. The dierence spectrum, obtained by subtracting the spectra
both before and after warming, clearly indicates the collapse of the peak at 485 nm and the
formation of bands at 370, 550, and 770 nm (Fig. 4b).
These results can be interpreted by four species: t,t,t-TPCB
+
, t-St
+
, t-St
+
/t-St
pair with peaks at 485 and 780 nm, and j-St
2
+
. As there was no overlapping of k-electrons
between two benzene rings in the t-St
+
/t-St pair, it is anticipated that the pair has the
same absorption as t-St
+
. Thus t,t,t-TPCB
+
is rather stable up to 90 K and does not
convert to k-(t-St)
2
+
nor j-St
2
+
. Photoirradiation of t,t,t-TPCB
+
at 77 K causes
cycloreversion into t-St
+
/t-St pair [Eq. (20)], which converts to j-St
2
+
upon warming [Eq.
(21)].
t; t; t TPCB
hv ! t St
=t St 20
t St
=t St heat ! j St
2
21
Copyright 2004 by Taylor & Francis Group, LLC
Intermediacy of Stilbene Dimer Radical Cations (St
2
+
)
Cycloreversion of TPCB
+
and dimerization of St
+
are summarized in Scheme 3, in which
k-(t-St)
2
+
, j-St
2
+
, and the t-St
+
/t-St pair are involved as the key intermediates. k-(t-St)
2
+
takes various overlapping arrangements of the k-electrons between two benzene rings, and
exhibits a large charge resonance (CR) band at shorter k of 680800 nm than those of
other dimer
+
reported [84]. On the other hand, j-St
2
+
has an acyclic linear structure,
Figure 3 Absorption spectral changes of t,c,t-TPCB
+
recorded immediately after g-R of t,c,t-
TPCB (5 10
3
M) in glassy BC matrices at 77 K and after warming (a), or after g-R and subsequent
photoirradiation at k longer than 390 nm (1) and after warning (in the order of 2 and 3) (b), and
dierence spectra between spectra after photoirradiation and after warming (DA=A
No. 1
A
No. 2
) (c).
Copyright 2004 by Taylor & Francis Group, LLC
having both a radical and a cation on the 1- and 4-positions of the C
4
linkage. Although
the c-St
2
+
was not detected, the bimolecular reaction of c-St
+
and c-St yields j-St
2
+
or
c-St
+
/c-St pair, which decomposes into thermodynamically stable t-St
+
and t-St.
It should be noted that t,c,t-TPCB
+
decomposes into j-St
2
+
, but t,t,t-TPCB
+
does
not upon warming up to 90 K. On the other hand, photochemical cycloreversion of both
TPCB
+
occurs through cleavage of two CUC bonds to give k-(t-St)
2
+
or t-St
+
/t-St pair.
This may account for the dierent interaction between two phenyl (Ph) groups through the
Figure 4 Absorption spectral changes of t,t,t-TPCB
+
recorded immediately after g-Rof t,t,t-TPCB
(510
3
M) in glassy BC matrices at 77 K and after warming (a), or after g-R and subsequent
photoirradiation at k longer than 390 nm (1) and after warming (in the order of 2 and 3) (b), and
dierence spectra between spectra after photoirradiation and after warming (DA=A
No. 1
A
No. 2
) (c).
Copyright 2004 by Taylor & Francis Group, LLC
CUC bond of cyclobutane ring of t,c,t-TPCB
+
and t,t,t-TPCB
+
in the ground and
excited states.
2.3. Reactions of Radical Cations of Organophosphorus Compounds
Little is known about the reactions of radical cations of organophosphorus compounds
(P
+
) such as aryl phosphates and phosphonates as caged compounds and reagents in
organic syntheses [8591]. To elucidate PET reactions including P
+
as the key inter-
mediate, PR and g-R are used here.
Intramolecular Reaction of Radical Cations of Trinaphthyl Phosphates and Dinaphthyl
Methylphosphonates
From product analysis, uorescence quenching, and LFP, initially formed trinaphthyl
phosphates (9
+
) and dinaphthyl methylphosphonates (10
+
) decompose into 1,1V-
binaphthyl (Np
2
) and 1-naphthyl phosphate or methyl phosphate, respectively. Although
it is suggested that 9
+
and 10
+
change to the intramolecular k-dimer
+
between two Np
Scheme 3 Cycloreversion of TPCB
+
and dimerization of St
+
.
Copyright 2004 by Taylor & Francis Group, LLC
groups and Np
2
+
is subsequently eliminated (Scheme 4), no direct evidence of the
transformation [90,91].
The assignments of 9
+
, 10
+
, and Np
2
+
can be conrmed by their absorption
spectra observed during g-R in BC rigid matrices at 77 K. Three peaks of 9
+
at 400, 680,
and 720 nm and two broad bands in the region of 500600 and 800850 nm were observed
after the g-R of 9 in BC rigid matrix at 77 K (Fig. 5). The absorption spectrum
Scheme 4 Elimination of Np
2
+
from 9
+
through the intramolecular k-dimer
+
.
Figure 5 Absorption spectral changes of 9
+
and 10
+
recorded immediately after g-R of 9 at 3
10
2
M (a) and 10 at 1 10
2
M (b) in BC rigid matrices at 77 K and after warming. Arrows show
decrease in the peak upon warming.
Copyright 2004 by Taylor & Francis Group, LLC
disappeared with no formation of a peak upon warming with independence on [9]. The
broad band in the region of 500600 nm, assigned to an intramolecular k-dimer
+
of 9
+
with face-to-face interaction between the two Np groups, disappeared faster than the
peaks of 9
+
at 400, 680, and 720 nm (Fig. 5a). Similar spectral changes of 10
+
were
observed during g-R of 10 in BC rigid matrix at 77 K (Fig. 5b).
The transient absorption spectrum assigned to 9
+
and intramolecular k-dimer
+
of
9
+
was observed immediately after 8-nsec e
+
(Fig. 6). This indicates that intramolecular dimerization
in 9
+
occurs to reversibly form k-dimer
+
at k
d
intra
= 1.0 10
8
1.0 10
9
sec
1
. In DCE,
9
+
and k-dimer
+
disappeared by neutralization with Cl
+
and decay of k-dimer
+
of 9
+
, k
r
= 5.3
10
5
sec
1
is estimated for the reaction.
Warming up to 90 K is not enough for the activation barrier for the formation of
Np
2
+
fromk-dimer
+
of 9
+
, while the reaction is completely attained even for e
duration
of 8 nsec at r.t. The peaks of k-dimer
+
of 9
+
were much broader than that of k-dimer
+
of
1,3-di-1-naphthylpropane
+
(720 nm) because of the dierent spacers, OUP(O)UO and
(CH
2
)
3
, between the two Np groups in 9
+
and 1,3-di-1-naphthylpropane
+
, respectively.
No elimination of Np
2
+
occurs in 1,3-di-1-naphthylpropane
+
and bis(1-naphthyloxy)
methane
+
. The electron withdrawing character of the P(O) group in the OUP(O)UO
spacer may be responsible for the elimination of Np
2
+
.
Figure 6 Transient absorption spectra of 9
+
observed immediately (open circle), 5 Asec (solid
triangle), and 20 Asec after e
+
reacts with nucleophiles such as water and
alcohol to yield phosphine oxides and phosphates [9295]. During a reaction of triphenyl-
phosphines (P(C
6
H
4
-X-p)
3
+
) with water to give phosphine oxides OjP(C
6
H
4
-X-p)
3
(Scheme 5), electron-donating character of the p-substituent shifted the bands of P(C
6
H
4
-
X-p)
3
+
to longer k (520610 nm) and decreased the yield of OjP(C
6
H
4
-X-p)
3
(0.260.02)
on the basis of LFP, PR, g-R, and product analysis [96].
The positive charge on the P atom of P(C
6
H
4
OMe-p)
3
+
is less than P(C
6
H
5
)
3
+
be-
cause of conjugation with the k-electrons of the benzene ring and the n-electrons of O atom
of p-MeO. Thus, the k
r
of P(C
6
H
4
OMe-p)
3
+
with H
2
O is smaller than that of P(C
6
H
5
)
3
+
.
The delocalization of the positive charge of 3
+
may be related to the longer k shift of the
bands with an increase in the electron-donating character of p-substituent. The dierence of
reactivities of P(C
6
H
4
-X-p)
3
+
fromthose of 9
+
and 10
+
is a consequence of the respective
HOMO, which are the phosphorus nonbonding orbital and Np k-orbital, respectively.
3. DIMERIZATION OF RADICAL ANIONS
Whether or not the radical anion of aromatic hydrocarbons (ArH
) is stabilized by CR k-
interaction with ArH, giving the dimer radical anion, (ArH)
2
, is an interesting subject.
Although much eort has been concentrated on detecting (ArH)
2
with a sandwich
structure between two aromatic groups of ArH
(has
not been observed [46,47,73,97111]. As an exceptional case, (anthracene)
2
has been
formed from the cleavage of a 4k+4k anthracene photodimer
in a rigid matrix at 77 K
[102]. Radical anions of several olens with electron-withdrawing substituents such as
fumaronitrile, maleic anhydride, and acrylonitrile have been observed to dimerize to give
the dimer
which are assumed to have a broad CR band in the 8002000 nm range [101].
In contrast to St
+
forming k- and/or j-(St)
2
+
, no dimerization of St
reacts with St
.
3.1. Dimer Radical Anions of Aromatic Acetylenes [48]
Diphenylacetylene
(11a
+
with water to give OjP(C
6
H
4
-X-p)
3
.
Copyright 2004 by Taylor & Francis Group, LLC
respectively) at higher [11a
]=(2.67.3) 10
2
and 3.5 10
3
M in THF[107] and
HMPA [108], respectively. The structure of (11a)
2
2
is assumed to be the 1,2,3,4-
tetraphenylbut-1,3-ene-1,4-diyl dianion (PhC
= CPhUCPh = C
(A
11g
and 1,4-diphenyl-1,3-
butadiyne
(12
(13
(n = 2)), -propane
(13
(n = 3)), -
butane
(13
(n = 4)), -pentane
(13
(13
(n = 6)) to give
the intramolecular dimer
was
observed immediately after e
decayed
on a time scale of a few Asec with formation of a band at 500 nm with isosbestic points at
470 and 630 nm at r.t. The pseudo-rst-order rate constants of the decay of 450-nm band
and formation of 500-nm band were calculated to be k
d
450
=1.0 10
6
and k
f
500
=1.3 10
6
sec
1
, respectively, at [11a]=10
2
M. It is found that k
decay
450
and k
form
500
increase with
increasing [11a]. This suggests that 11a
with the
formation of 500-nm band. A similar absorption spectral change of 11a
was observed in
the MTHF rigid matrix at 77 K during g-R of 11a (Fig. 8). Formation of (11a)
2
is
completely contrary to previous results of electrolysis, the reduction by Na, photolysis,
and radiolysis of 11a where no dimerization was observed. k
d
=(3.86.6) 10
7
M
1
sec
1
for the dimerization was calculated from the linear plots of k
decay
450
and k
form
500
vs. [11a].
j-Type or diene-type stucture is most likely for (11a)
2
jCPhCPhjC
(11a
11g
, 12
, and 13
(n=26)).
Copyright 2004 by Taylor & Francis Group, LLC
k
d
for dimerization of other 11
and 12
11c
11g
dimerizes with 12 to 12
2
is almost
equivalent with that for 11a
.
Formation of Intramolecular Dimer Radical Anions
The formation of intramolecular dimer
, j-(11a)
2
between 11a
in 13
in 13
.
The yield of j-(11a)
2
in 13
(n = 26) and
comparable with that of intermolecular (11a)
2
in 13
.
No direct evidence for the formation of k-A
2
suggests that A
is stabilized by the
formation of j-A
2
and A.
3.2. Intramolecular Dimer Radical Anions in Diarylmethanols
The formation of an intramolecular dimer
of diaryllmethanol (Ar
2
CHOH, 14ae) was
found as the intermediates in the attachment of e
s
immediately after e
and a
byproduct from 14b
. 14be
(11a)
2
and 11a.
Copyright 2004 by Taylor & Francis Group, LLC
Table 6 Intermolecular Dimerizations of A
Substituents A
A
2
A
2
[A
2
]
max
/[A
]
0
e
max
[M
1
sec
1
] of A
k
d
[M
1
sec
1
] Solvent
11a
4-CH
3
450 510 1.4 0.9 0.66 4.1 10
4
5.7 10
7
HMPA
11c
4-C
6
H
5
500 550 1.6 1.2 0.77 4.7 10
4
4.1 10
7
HMPA
11d
4-NO
2
440 490 0.95 0.09 0.10 2.7 10
4
9.0 10
6
HMPA
11e
4-OCH
3
440 500 1.7 0.15 0.09 5.2 10
4
1.0 10
7
HMPA
11f
(CH
3
)
5
450 540 3.9 0.25 0.06 1.2 10
5
7.3 10
6
HMPA
11a
]
max
/[A
]
0
=(DO.D.
max
at k
max
for A
2
)/(DO.D.
0
at k
max
for A
). e
max
was calculated from DO.D.
0
at k
max
for A
, e
max
=5.4 10
4
M
1
cm
1
for 11a
, l=0.4 cm,
and [11
]0=8.3 10
5
M in HMPA, while e
480
for 12
]
0
=2.8 10
5
M in DMF. k
d
for
the dimerization of A
vs. [A].
Copyright 2004 by Taylor & Francis Group, LLC
Scheme 9 Intramolecular dimerization of 11a
and 11a in 13
.
Scheme 8 j-Type or diene-type dimer
12
2
from dimerization of 12
and 12.
Scheme 10 Intramolecular dimer
in the range
of 8001600 nm were obtained from the decay curve measured at every 100-nm step
(Fig. 10). While the peak of e
s
of 14b,c
and 14c
of 14b,c
.
Although an intramolecular singlet excimer (
1
Ar
2
*
) uorescence of 1,3-bis(4-cyano-
phenyl)propan-1-ol was observed in MTHF at r.t., no CR band of 1,3-bis(4-cyanophenyl)-
propan-1-ol
in 1,3-di-h-naphthylpro-
panes
, no CR band of 1,3-di-a-naphthylpropane
and 1,3-di-1-pyrenyl-propane
was
Figure 9 Transient absorption spectra of 14
and Np
+
of pyrene
+
and Np
+
. Ar
2
of methylterephthalate-
chromophore was assumed from the kinetic study on the anion radical transfer from
methylterephthalate to 1,4-dicyanobenzene in diester of 1,n-alkanediols [116]. These suggest
that Ar
2
of 1b,c
rather than
displaced sandwich-pair structure favored for Ar
2
+
.
4. LIFETIMES OF RADICAL CATIONS AND ANIONS IN THE
EXCITED STATE
Excitation of M
+/
+
*
/
+
*
/
* is usually nonluminescent in
solution, although weak luminescences of thermally stable M
+/
and N-methylphenothiazine
+
have been reported at 470580 nm [120,121] and
at 600 nm [25], respectively. Breslin and Fox [33] have reported the weak luminescence of
two substituted triarylamine
+
with luminescence peaks at 790 and 803 nm and with u
f
=
10
5
, while the other thermally stable M
+/
+/
(D
1
)
because of the small energy between D
1
and D
0
states.
4.1. Lifetime of Stilbene Radical Cation in the Excited State (St
+
*) by the
Hole Transfer Quenching
It is assumed that the photochemical c!t one-way isomerization of c-St
+
to t-St
+
occurs
in rigid matrices at 77 K [40,122] and in solution at r.t. on the basis of LPF of c-St
+
formed by using PR in DCE [83] or secondary ET in acetonitrile [32]. The photochemical
c!t isomerization has been reported to take place in the second excited doublet state (D
2
)
but not in c-St
+
(D
1
) [83]. Because St
+
(D
2
) is nonluminescent, H of St
+
(D
2
) must be
determined by using other experiments such as the excitation energy (E
ex
) quenching or
hole transfer (HT) quenching. Because St
+
(D
2
) has a higher E
ox
than St
+
(D
0
) [1,5], HT
occurs from St
+
(D
2
) to an appropriate quencher (Q). H of St
+
(D
2
), as well as the 1,2-
diphenylcyclo-1-butene
+
(DPCB
+
), as a structurally constrained derivative of c-St, was
estimated from selective HT quenching of St
+
(D
2
) using anisole (AN) as Q through the
PR-LFP combined method [43,83,123125].
Lifetime of t-St
+
*
It is well established that c-St
+
and t-St
+
are selectively generated during the PR of c-St
and t-St in DCEat r.t., and they showbands at 420580 and 700850 nmassigned to D
2
pD
0
and D
1
pD
0
transitions, respectively, and peaks at 515 and 780 nm for c-St
+
and 480 and
760 nm for t-St
+
[42,73,83]. Excitation of St
+
with hv
532
generates St
+
(D
2
) (St
+
*) with
E
ex
=50 and 53 kcal mol
1
for c-St
+
* and t-St
+
*, respectively, that were calculated from
the red edges of bands at 420580 nm.
Irradiation of t-St
+
with hv
532
exhibited no change in the transient absorption
spectra and time proles of DO.D.
480
, where t-St
+
shows a peak (Fig. 12a). Therefore t-
Copyright 2004 by Taylor & Francis Group, LLC
St
+
* does not isomerize to c-St
+
(D
0
) with photoirradiation, but decays to t-St
+
(D
0
)
through internal conversion at k
ic
t
(Scheme 11).
The photoirradiation of t-St
+
in the presence of AN caused a decrease of DO.D.
480
immediately after hv
532
(Fig. 12b). jDO.D.
480
j increased with increasing [AN]. It is obvious
that t-St
+
* is quenched by AN via HT quenching to give t-St and AN
+
with the
bimolecular HT rate constant of k
ht
(Scheme 11). The recovery of DO.D.
480
after hv
532
corresponds to HT fromAN
+
to t-St to yield AN and t-St
+
[73], which occured at k
di
=
7.8 10
9
M
1
sec
1
. The k
ht
= k
di
can be assumed because of the large exothermicity of
Figure 12 Kinetic traces illustrating time proles of DO.D.
480
as a function of time during PR
LFP experiment of t-St (5 10
3
M) in the absence (a) and presence (b) of AN (1 M) in DCE.
Scheme 11 Hole transfer quenching of t-St
+
* by AN.
Copyright 2004 by Taylor & Francis Group, LLC
DG = 25 kcal mol
1
for HT from St
+
* to AN. From the linear SternVolmer plots of
Y
t
1
vs. [AN]
1
, where Y
t
is the chemical yield of [t-St
+
]
disapp
, s
t
, and k
ic
t
for t-St
+
(D
2
)
were calculated to be 240F50 psec and (4.1F1.1) 10
9
sec
1
, respectively.
Lifetime of c-St
+
* and DPCB
+
* (DPCB=1,2-diphenylcylclobutene)
In contrast to t-St
+
*, c!t one-way isomerization of c-St
+
* to t-St
+
was observed
within hv
532
(5 nsec) [32,83]. From the transient spectral changes and time proles of
DO.D.
515
and DO.D.
480
of c-St
+
and t-St
+
, respectively (Fig. 13), the photochemical
conversion of c-St
+
per hv
532
is (80 F 15)% at [c-St
+
]
0
= 1.5 10
5
M before photo-
irradiation, and the chemical yield of t-St
+
from c-St
+
(D
2
) is (75 F 15)%. The remain-
ing (25F15)% of c-St
+
that disappeared is considered to convert to c-St, t-St, or a radical
cation as a product such as dihydrophenanthrene.
The u of photochemical c!t isomerization of c-St
+
to t-St
+
was subsequently
determined to be 0.65F0.15 from c-St
+
that disappeared (u
c
) and 0.49F0.12 from t-St
+
that formed (u
t
). Taking into account u
c
and u
t
, as well as the conversion of (80 F15)%for
c-St
+
per hv
532
, c-St
+
* isomerizes to t-St
+
at (49 F 12)%, converts to other products at
(16F4)%, and deactivates to c-St
+
at (35 F 8)% yield.
The transient phenomena of c-St
+
* that involved the isomerization and HT
quenching of c-St
+
* using AN are shown in Scheme 12. According to Scheme 12, the
linear SternVolmer plots of s
t
/s
t
AN
vs. [AN] aorded s
c
= 120 F 30 psec being one-half
of s
t
at k
ht
= k
di
= 7.8 10
9
M
1
sec
1
. k
i
, k
p
, and k
ic
c
were determined to be (4.1 F 1.0)
10
9
, (1.3 F 0.3) 10
9
, and (2.9 F 0.7) 10
9
sec
1
, respectively.
Because 1,2-diphenylcyclo-1-butene (DPCB) has a rigid planar structure with the c-
St-chromophore structurally constrained by the cyclobutene ring, and because it is stable
for geometrical isomerizations, TPCB
+
is expected to have the same rigid planar
structure. Pulse radiolysis of DPCB in DCE produces DPCB
+
, which shows bands at
Figure 13 Transient absorption spectra recorded before hv
532
(o) and immediately after (.) and
200 nsec (D) and 1 As (E) after hv
532
during PRLFP of c-St (510
3
M) in DCE (a). Kinetic traces
illustrating time proles of DO.D.
480
(b) and DO.D.
515
(c) as a function of time after e
.
Copyright 2004 by Taylor & Francis Group, LLC
420560 and 700850 nm that are assigned to D
2
pD
0
and D
1
pD
0
transitions, respec-
tively [42,73,83]. Excitation of DPCB
+
at 532 nm exhibited no change in the transient
absorption. Therefore, DPCB
+
* is deactivated to DPCB
+
(D
0
) within hv
532
, similarly to
t-St
+
* (Scheme 11). Photoirradiation of DPCB
+
in the presence of p-methylanisole
(MA) caused a decrease of DO.D.
480
immediately after hv
532
with an increasing [MA]. It is
obvious that DPCB
+
* is quenched by MA via HT quenching to give DPCB and MA
+
.
From the linear SternVolmer plots of Y
CB
1
vs. [MA]
1
, s
DPCB
and k
ic
DPCB
of DPCB
+
* were calculated to be 380 F 30 psec and (2.6 F 0.2) 10
9
sec
1
, respectively, at k
ht
=
k
di
= 7.8 10
9
M
1
sec
1
. Consequently, it is concluded that H
DPCB
is longer than s
c
and
s
t
, whereas k
ic
DPCB
is comparable with k
ic
c
and k
ic
t
(Table 7).
Comparison of c-St
+
*, t-St
+
*, and DPCB
+
*
Similar to most M
+
*, c-St
+
*, t-St
+
*, and CB
+
* are nonluminescent. Isomerization of
c-St
+
* to t-St
+
, which proceeds via twisting about the central CjC double bond as a
main process, competes with the internal conversion of c-St
+
*. However, t-St
+
* and
DPCB
+
* are deactivated to the D
0
state without any other reactions. Although t-St
+
*
could slightly twist, it does not isomerize to c-St
+
. The twisting is faster in c-St
+
* than in
t-St
+
* because of the interactions between the Ph groups of c-St. Because the twisting is
severely hindered by structural constraints in DPCB, DPCB is unable to change its initial
planar rigid structure. Both DPCB
+
* and DPCB
+
(D
0
) probably have the rigid planar
structure. The energy levels of c-St
+
*, t-St
+
*, and DPCB
+
* are assumed to be almost
equivalent because of the similar E
p/2
ox
for c-St, t-St, and DPCB, as well as the similar E
ex
of c-St
+
*, t-St
+
*, and DPCB
+
* (Table 7). It is suggested that c-St
+
*, t-St
+
*, and
DPCB
+
* exist in potential minima, and that the order s
c
<s
t
<s
DPCB
is mainly attributed
Table 7 Properties of t-St
+
, c-St
+
, and DPCB
+
in the D
2
State: Lifetime (s), Rate Constants
for Internal Conversion (k
ic
), ct Isomerization (k
i
), and Product Formation (k
p
), and Excitation
Energy (E
ex
)
Radical Cations in D
2
t-St
+
* c-St
+
* CB
+
*
s[psec] 240F50 120 F 30 380 F 30
k
ic
[sec
1
] (4.1 F 1.1) 10
9
(2.9 F 0.7) 10
9
(2.6 F 0.2) 10
9
k
i
[sec
1
] 0 (4.1 F 1.0) 10
9
0
k
p
[sec
1
] 0 (1.3 F 0.3) 10
9
0
E
ex
[kcal mol
1
] 53 50 51
Scheme 12 Reactions of c-St
+
* involving isomerization, ic, and HT quenching.
Copyright 2004 by Taylor & Francis Group, LLC
to the isomerization via twisting only in c-St
+
*. It is clear that the twisting is favorable in
c-St
+
*, slightly favorable in t-St
+
*, but not at all in DPCB
+
*.
Considering u
f
, s, and ct isomerization for St*, it is clearly suggested that the
transient phenomena of St
+
(D
2
) cannot be assumed on the basis of those for St(S
1
).
Judging from s
c
f
120 psec and H
t
f
240 psec, a diabatic process is proposed for the
isomerization of c-St
+
* to t-St
+
. It has been proposed the analogous process in which c-
St
+
* avoidably crosses with c-St
+
(D
1
), followed by an allowed crossing to t-St
+
(D
0
)
at the 40j twisted geometry against a planar c-St
+
*, where the D
2
, D
1
, and D
0
states
closely lie, on the basis of MO calculations [32]. The barrier for the twisting of c-St
+
* is
estimated to be 0.32 kcal mol
1
, if k
i
values are assumed to be 5.3 10
9
sec
1
for the
largest limit and 2.9 10
9
sec
1
for the smallest limit at 300 and 140 K, respectively, on
the basis of the error limits (F30%) for u and k
i
. This suggests that c-St
+
* is stabilized by
twisting at a shallow potential minimum, and that conversion to another product can
compete with isomerization. Consequently, the diabatic isomerization from c-St
+
* to t-
St
+
is strongly expected to take place via an avoidable crossing from c-St
+
(D
2
) to c-St
+
(D
1
), and then to t-St
+
(D
0
) with a small barrier.
In conclusion, selective HT quenching of St
+
* and CB
+
* by AN and MA yielded
s
c
=120, s
t
=240, and s
DPCB
=380 psec, which is attributed to the isomerization via the
twisting only in c-St
+
*. It is also suggested that diabatic isomerization from c-St
+
* to t-
St
+
proceeds via an avoidable crossing of the D
2
, D
1
, and D
0
states. To determine the s of
M
+
* with k
ht
=k
di
can be essentially applied to any M
+
*, although it is necessary to
nd an appropriate molecule as a selective hole quencher. Moreover, the s of M
+
* must
be longer than 50 psec, because [AN] is 2 M at the maximum to keep the solvent property.
4.2. Lifetime of St
**
by the Electron Transfer Quenching
Characterizations of St
(D
2
) are necessary to elucidate the isomerization of St
(D
2
),
which is nonluminescent, similarly to radical anions of aromatic compounds in the excited
states [25,33,118,120,121,126130]. Selective electron transfer (ET) quenching of St
(D
2
)
was performed using biphenyl (Bp) as an electron acceptor and H of St
(D
2
) was
estimated using a PRLFP combined method [43,46,83,123125,131,132].
It is well established that c-St
and t-St
[42,49,83,133]. Excitation of St
with hv
532
generates
St
(D
2
) (St
*) with E
ex
=45 and 48 kcal mol
1
for c-St
* and t-St
*, respectively.
Irradiation of t-St
with hv
532
caused a decrease of DO.D. immediately after hv
532
, while
no spectral shift was observed.
Photoelectron Ejection from t-St
*
The dierences in DO.D. before and after hv
532
(jDO.D.j) decreased with increasing [t-St].
The spectrum of t-St
(D
0
), but decays to t-St
(D
0
) with k
ic
t
of t-St
*. u
e
t
= 0.06 F 0.02 of photoelectron
ejection from t-St
was measured.
The ET quenching of c-St
(D
2
) and t-St
(D
2
) (c-St
* and t-St
*) by an ET
quencher Bp was examined. Transient absorption spectra composed of bands of St
and
Copyright 2004 by Taylor & Francis Group, LLC
Bp
with peaks at 405 and 630 nm at the ratio of [Bp]/[St] = 2100 were observed
immediately after e
on
the time scale of 50200 nsec after e
was
irradiated with hv
532
, DO.D.
500
decreased immediately after hv
532
, and DO.D.
405
increased
with an increase of [Bp] (Fig. 15). This indicates that t-St
within hv
532
. The decay in DO.D.
405
and recovery of DO.D.
500
after hv
532
correspond to ET from Bp
at k
et
= k
di
.
Electronic transfer quenching of t-St
* proceeds at k
et
= k
di
= 7.1 10
9
M
1
sec
1
in DMF. Similar to radical ion pair, (D
+
/A
)
solv
formed during ET between donor
(D) and acceptor (A) molecules in the excited singlet or triplet state, it is suggested that ET
quenching initially gives (St/Bp
)
solv
with competition of the internal conversion of St
*
to St
. (St/Bp
)
solv
then undergoes solvent separation into St and Bp
at k
sep
or returns
to St
formed is
represented by R=k
sep
/(k
et
+k
sep
).
Transient Phenomena of t-St
* and c-St
*
Photoelectron ejection from ArH
*)
z
)
undergoes vibrational relaxation on the time scale of 10
14
sec to yield t-St
*, which has a
long enough lifetime H
t*
= (k
d
t
)
1
to be quenched by Bp via ET involving (t-St/Bp
)
solv
(Scheme 13). u
e
t
=0.06 F0.02 suggests that (t-St
*)
z
undergoes photoelectron ejection and
vibrational relaxation to t-St
was 75 nsec.
Copyright 2004 by Taylor & Francis Group, LLC
Figure 15 Kinetic traces illustrating the time proles of DO.D.
405
and DO.D.
500
as a function of
time after e
was 75 nsec.
Scheme 13 Electron transfer quenching of t-St
* by Bp.
Copyright 2004 by Taylor & Francis Group, LLC
u
Bp
1
vs. [Bp]
1
, s
t
and k
ic
t
were estimated to be 2.5 F 0.7 nsec and (4.0 F 1.2) 10
8
sec
1
,
respectively. Because R=k
sep
/(k
et
+k
sep
) =0.15 F0.4 is calculated fromthe intercept and
u
t
= 0.94, the back ET with DG = 11 kcal mol
1
occurs six times faster than the solvent
separation within (t-St/Bp
)
solv
.
Isomerization of c-St
* to t-St
and t-St
per hv
532
is 49% and that the chemical yield of
[t-St
]
form
from [c-St
]
cons
is 67%. The remaining 33% of c-St
consumed is considered
to give c-St via photoelectron ejection. The us of [c-St
]
cons
and [t-St
]
form
were
measured to be u
c
= 0.21 F 0.05 and u
t
= 0.14 F 0.03, respectively. Photoelectron
ejection of (c-St
*)
z
occurs competitively with vibrational relaxation to c-St
*, and
therefore, u
e
c
was calculated from u
e
c
= u
c
B
t
to be 0.07 F 0.02, which is similar to that
for (t-St
*)
z
, u
e
t
=0.06 F 0.02. Summarizing, (c-St
*)
z
undergoes photoelectron ejection
and vibrational relaxation to c-St
stoichiometrically
at k
di
, c-St
was 70 nsec.
Copyright 2004 by Taylor & Francis Group, LLC
on Scheme 14, the linear plot of u
Bp
1
vs. [Bp]
1
was obtained. s
c
was calculated to be
1.5F0.4 nsec and found to be shorter than s
t
, fromk
et
s
c
, assuming k
et
=k
di
. On the basis of
k
i
:k
ic
c
= 1:5.7 and s
c
, k
i
and k
ic
c
were calculated to be (1.0 F0.3) 10
8
and (5.6 F1.7) 10
8
sec
1
, respectively. R=k
sep
/(k
et
+k
sep
) =0.18 F0.4 is almost equal to Rfor (t-St/Bp
)
solv
because of the nearly equivalent DG = 11 kcal mol
1
for the back ET in (St/Bp
)
solv
.
Comparisons Between St
* and St
+
*
Isomerization of c-St
* to t-St
*, while
only internal conversion occurs in t-St
*. Isomerization of c-St
* to t-St
proceeds via
the twisting which is faster in c-St
* than in t-St
* and t-St
+
* to
t-St
+
* is proposed for the isomerization of c-St
* to t-St
* is
twisted at a potential minimum which undergoes a diabatic crossing to t-St
(D
0
) having a
twisting form without a large barrier. On the other hand, a planar t-St
* exists in a
potential minimum with the relatively large barrier for the twisting and undergoes internal
conversion to t-St
(D
0
) having a planar form.
Photoelectron ejection occurs in (St
*)
z
, while c!t isomerization occurs in c-St
*
and c-St
+
*. However, the k
i
of c-St
+
*, and B
i
of
c-St
+
* (Table 8). k
i
is 5.6 times smaller than k
ic
of
c-St
*, while k
i
is slightly greater than k
ic
of c-St
+
*. The k
ic
of St
* is an order of
magnitude smaller than that of St
+
*. These dierences are responsible for the potential
surfaces of St
(D
2
) and St
(D
0
) and their relations, which are probably dierent from
those of St
+
[138]. It may be considered that the D
2
and D
0
states lie less closely in St
than in St
+
. H =1.0 and 7083 psec for c-St and t-St in the S
1
states have been reported
Figure 17 Transient absorption spectra recorded immediately after e
was 60 nsec.
Copyright 2004 by Taylor & Francis Group, LLC
Figure 18 Kinetic traces illustrating time proles of DO.D.
405
, DO.D.
500
, and DO.D.
520
as a func-
tion of time after e
was 60 nsec.
Copyright 2004 by Taylor & Francis Group, LLC
from S
1
!S
0
uorescence measurement [139145]. It is clearly suggested that the transient
phenomena of M
(D
2
) cannot be assumed on the basis of those of neutral molecules in
the S
1
state or M
+
(D
2
).
In conclusion, Bp acts as a selective electron quencher for St
* via ET quenching.
Assuming that k
et
=k
di
, s
c
and s
t
are estimated to be 1.5 and 2.5 nsec, respectively, which
are longer than s=120 and 240 psec for c-St
+
* and t-St
+
* (Table 8). The order of s
c
<s
t
is attributed to the isomerization via the twisting only in c-St
*, because twisting is
favorable in c-St
* to t-St
* by
the selective ET quenching assuming k
et
=k
di
.
4.3. Lifetime of Radical Anions of Dicyanoanthracene, Phenazine, and
Anthraquinone in the Excited States
The lifetimes of M
* by an
electron quencher (Q) during PRLFP in DMF [47].
Selective Electron Transfer Quenching of Radical Anions in the Excited State
The transient absorption spectra of M
showed three
peaks at 510, 640, and 710 nm for DCA
,
AQ
, and RAQ
. When M
, no
Scheme 14 Reactions of c-St
+
, c-St
+
, t-St
+
, and c-St
+
in the D
2
States: s,
a
k
ic
k
i
, /
i
, and E
ex
Radical Ions (D
2
) H [nsec] k
ic
[sec
1
] k
i
[sec
1
] /
I
E
ex
[kcal/mol]
t-St
+
* 0.24 F 0.05 (4.1 F 1.1) 10
9
0 0 53
c-St
+
* 0.12 F 0.03 (2.9 F 0.7) 10
9
(4.1 F 1.0) 10
9
0.49 F 0.12 50
a
From k
et
s or k
ht
s assuming that k
et
= k
ht
= k
di
with Q = Bp for St
* and AN for St
+
*, respectively.
Copyright 2004 by Taylor & Francis Group, LLC
change occurred in the transient absorption spectra and the time proles. Therefore M
*
converts to M
within hv
532
duration of 5 nsec.
No transient absorption was observed in the range 450740 nm immediately after e
,
while the absorption of DCA
decreased within hv
532
(Fig. 19), as a result of the selective ET quenching of DCA
* by FN
proceeds to yield DCA and FN
at k
et
. Similar ET quenching of PZ
* or AQ
* by FN
and of M
* have dierent
characteristics fromeach other in their structure and electronic properties but an equivalent
E
ex
= 4648 kcal mol
1
. E
ex
might be important for s of M
* have
similar s of 1.01.4 nsec, but they are shorter than s for AQ
* vs. 2-BAQ
* vs. 2-CAQ
* vs. 2-CAQ
* and M
are almost equivalent for AQ and RAQ (Table 9), a shorter s for
RAQ
* than for AQ
* to RAQ
.
s = 1.04.4 nsec of M
* and t-
St
* and St
* occurs with
similar k
ic
as the main process. The structural and electronic characteristics of M
* are
much dierent from those of St
*, while E
ex
= 4648 kcal mol
1
for M
* is similar to
E
ex
= 5053 kcal mol
1
for c-St
* and t-St
*. Therefore, E
ex
is the signicant factor for
the rate of internal conversion of M
* to M
.
Figure 19 Transient absorption spectra recorded at t = 190 nsec after e
(before hv
532
, o) and
immediately after (.) and at 100 nsec (D) after hv
532
during PRLFP of DCA (2.5 10
3
M)FN
(5.0 10
1
M) mixture in DMF. Inset: kinetic trace illustrating time prole of DO.D.
510
as a
function of time after e
.
Copyright 2004 by Taylor & Francis Group, LLC
Table 9 Half-Peak Reduction Potential (E
1/2
red
) of M and Q, E
ex
of M
to M and from M
*
to Q, s of M
*, k
ic
=s
1
of M
*, f
-M
of M
)
solv
) Based
on 1.0 for (PZ/FN
)
solv
a
E
1/2
red
[V vs. SCE]
M
* Q M Q E
ex
[kcal mol
1
] DG [kcal mol
1
] DG* [kcal mol
1
] H [nsec] k
ic
=H
1
[sec
1
] /
M
R
DCA
* BN 0.89 2.49 48 37 11 0
PZ
)+ 0.25 V, when E
1/2
red
(vs. Ag/Ag
and
back electron transfer to M
)
solv
, respectively.
Copyright 2004 by Taylor & Francis Group, LLC
Intermediacy of Radical Anion-Neutral Molecule Pair in the Electron Transfer
Quenching
As shown in Table 9, u
M
depended on the combination of M
* and Q. u
PZ
in PZ
*FN
was the largest among the M
*Q pairs, while u
AQ
in AQ
*FN and u
PZ
in PZ
*FN. u
RAQ
in RAQ
*FN was
comparable with u
AQ
in AQ
*FN, whereas u
DCA
in DCA
)
solv
in the selective ETquenching of M
* by Q. In other words, ET
proceeds with k
et
to initially give (M/A
)
solv
, with competition of the internal conversion of
M
* to M
with k
ic
. (M/Q
)
solv
separates into Mand Q
at k
sep
or returns to M
and Q
via back ET at k
et
. Because u
M
is proportional to the separation factor of (M/Q
)
solv
,
relative separation factors of R = k
sep
/(k
et
+k
sep
) are calculated on the basis of R=1.0 for
(PZ/FN
)
solv
. As k
sep
of (D
/Q
)
solv
is assumed not to depend on the structures of D
and Q
)
solv
. Indeed, R has a correlation with DG but not
completely (Table 9). Although electrostatic interaction are preferential for R in (D
+
/
A
)
solv
, structural relations might also be important for R in (M/A
)
solv
.
In conclusion, the selective ET quenching of M
*.
5. FLUORESCENCE OF RADICAL CATIONS IN THE EXCITED STATE
Fluorescence spectroscopy has sensitivity as high as up to the single-molecular detection
level. However, uorescence detection of M
+/
+
*
/
+/
have been reported but some of them are quite doubtful [148150].
5.1. Fluorescence of 1,3,5-Trimethoxybenzene Radical Cation (TMB
+
)
Ecient uorescence (u
f
= 1.1 10
3
) of 1,3,5-trimethoxybenzene radical cation in the
excited state (TMB
+
generated from
PET of TMB-1,4-dicyanonaphthalene (DCN)Bp mixture in acetonitrile by two-laser,
two-step excitation technique [147153]. Excitation with a XeCl excimer laser (308 nm, 8
nsec) initiated ET between
1
DCN* and Bp giving DCN
and Bp
+
, which led to the
secondary ET with O
2
and TMB to give O
2
and TMB
+
with a peak at 590 nm,
respectively. Excitation of TMB
+
with hv
532
(6 nsec) gave uorescence spectrum around
620 nm symmetry to the absorption spectrum. u
f
= 1.1 10
3
allowed an estimation of s
= 210 psec using the StricklerBerg relationship, giving k
f
= 9.5 10
6
sec
1
for natural
radiative rate constant of TMB
+
*.
Formation Kinetics of [TMB
+
/Cl
+
due to the high ionization potential (IP) of the solvent molecules [146,154156]. However,
Copyright 2004 by Taylor & Francis Group, LLC
the oxidation process is rather complicated because of the generation of Cl
, which is also
an oxidant together with solvent
+
during PR [125,157160]. Cl
forms complexes or an
ion pair directly with M [157]. In addition, such species is also formed from M
+
and Cl
+
with a broad spectrum
peaked at 500 nm during PR of TMB in highly chlorinated solvent such as CCl
4
, CH
2
Cl
2
,
or CHCl
3
, although the absorption spectrum of TMB
+
in acetonitrile showed a sharp
peak at 590 nm during LFP [125,159,160]. The transient absorption spectrum was time-
dependent in DCE (Fig. 20) showing a spectrum similar to that in acetonitrile immediately
after e
+
/Cl
]) formation in DCE. The fast and slow processes for ion pair
formation are considered: trapping of Cl
+
and Cl
at k
n
= 3.0 10
11
M
1
sec
1
, respectively.
Fluorescence of Free TMB
+
and [TMB
+
/Cl
] Ion Pair
Generation of M
+/
+
M
, M
+
anion, etc. in any processes. Absorption
spectroscopy of M
+/
are not suciently sensitive to distinguish ion pair from free ion,
while uorescence spectroscopy is enough sensitive. The u
f
of TMB
+
* can be a measure of
the ratio of free TMB
+
to the ion pair [TMB
+
/Cl
+
in
the ion pair would cause no uorescence or weak uorescence through rapid HT quenching
of TMB
+
* by Cl
+
. The u
f
of TMB
+
as a
function of delay time (101000 nsec) decreased rapidly during PRLFP of TMB in DCE
Figure 20 Time-resolved transient absorption spectra observed at 0 nsec (solid line), 100 nsec
(dashed line), and 1 Asec (dotted line) after e
+
* ob-
served at 300 nsec after e
+
*/
Cl
+
* by Cl
in [TMB
+
*/Cl
]. s
of TMB
+
* in the ion pair in DCE and k
et
for HT quenching of TMB
+
* by Cl
within the
ion pair were estimated to be 3.2 psec and 3.0 10
11
sec
1
, respectively. On the other hand,
free TMB
+
was generated fromHT during PRof TMB in the presence of Bp in DCE. Free
TMB
+
* emits a uorescence with u
f
= 1.1 10
3
.
Free ion fraction in the total ion amount can be estimated from u
f
=u
f
(free ion)+u
f
(ion pair). Pulse radiolysis of TMB (1.0 10
2
M) in DCE gave TMB
+
, whose ion pair
fraction is 50% immediately after e
+
* in the ion pair,
we estimated free ion fraction (R
f
=[free TMB
+
]/([free TMB
+
]+[TMB
+
/Cl
]), R
f
=
0.55 at 0 nsec and R
f
=0.2 at 1 Asec after e
+
* and TMB
+
* in the ion pair is estimated
to be 116 and 3.2 psec, while the k
ic
and k
ht
were also calculated to be 8.5 10
9
and 3.0 10
11
sec
1
, respectively. The transient behaviors of free TMB
+
* and TMB
+
* in the ion pair are
summarized in Scheme 15.
u
f
= 3.6 10
5
and 1.9 10
4
of TMB
+
* in CCl
4
and CH
2
Cl
2
, respectively, at a
delay time of 100 nsec were also estimated in a similar manner. These results can be
explained by the ion pair formation fromTMB and Cl
+
and Cl
as a slow process.
Figure 21 u
f
of TMB
+
as a function of the delay time of hv
532
relative to e
during PRLFT of
TMB (1.0 10
2
M) (open circle) and Bp (5.0 10
2
M)TMB (1.0 10
3
M) mixture (solid circle)
in DCE.
Copyright 2004 by Taylor & Francis Group, LLC
5.2. Fluorescence of TMB
+
Derivatives and the Quenching
Although the uorescence of the TMB
+
* family was studied [125,159,160], uorescence
was observed only in TMB
+
* among mono-, di-, tri-, tetra-, and penta-methoxybenzenes
and among 1,2,3-, 1,2,4-, and 1,3,5-trimethoxybenzenes. This indicates that 1,3,5-trioxyl
substitution is necessary for the uorescence. Pseudo-D
3h
symmetry cause the degenerate of
the singly occupied and second highest occupied molecular orbitals (SOMO and SHOMO,
respectively) in the TMB
+
family bearing 1,3,5-trioxybenzene structure. Therefore, the
TMB
+
family has low D
0
D
1
and high D
0
D
2
energy separations. This makes the D
2
!D
0
transition radiative enough to compete with D
2
!D
1
internal conversion.
Fluorescence was observed for the TMB
+
* family such as 3,5-dimethoxyphenol,
1,3-dihydroxy-5-methoxybenzene (5-methoxyresorcinol), 1-acetoxy-3,5-dimethoxyben-
zene, and 1,3,5-trimethoxy-2-methylbenzene. These results indicated that complete sym-
metry of the substitution on O atoms is not necessary to observe uorescence from the
TMB
+
* family, and that the variation of parent molecules of uorescent radical cation is
possibly performed [153]. Fluorescence was also detected from hexamethxybenzene
+
* as
an example of pseudo-D
6h
molecules. The discussion of the symmetry has been described
here on the uorescence from uorobenzenes
+
* in the vapor-phase or noble gas matrices.
The u
f
of TMB
+
generated during PR in DCE decreased with the introduction of
alkyl or acetyl substituents at O atoms suggesting that they act as quenchers. For example,
u
f
=2 10
3
, 1.1 10
3
, and c0 for 1,3,5-trihydroxybenzene
+
, TMB
+
, and 1,3,5-
triacetoxybenzene
+
(TAB
+
), respectively. However, the mechanisms involved in the
quenching of 1,3,5-trioxybenzenes
+
* are not the same. Internal conversion promoted by
the CUH vibration would operate in TMB
+
* and intramolecular HT would take place in
TAB
+
*.
Excitation of TMB
+
generated by PR in benzonitrile or in DCE in the presence of
benzonitrile gave weak or zero uorescence as a result of intermolecular HT quenching
[153]. This process has an exothermicity of c0.57 eV estimated from the ionization
potentials (IP) of TMB and benzonitrile (IP=8.11 and 9.62 eV, respectively) and E
ex
of
TMB
+
(2.03 eV). Similarly, uorescence from TMB
+
* was not observed in acetone
Scheme 15 Transient behaviors of free TMB
+
* and [TMB
+
*/Cl
].
Copyright 2004 by Taylor & Francis Group, LLC
(IP = 9.67 eV) and cyclohexane (IP=10.32 eV). On the other hand, uorescence of
TMB
+
* was observed in acetonitrile, DCE, 1,1,1,3,3,3-hexauoro-2-propanole, 2,2,2-
triuoroethanol, and other chlorinated solvents (IP>10.4 eV). The absence of the uo-
rescence from TAB
+
* could be a result of intramolecular HT quenching of the TMB
+
*
family by the acetyl group (IP=10 eV).
5.3. Fluorescence of 3,5-Dimethoxyphenol Radical Cation (DMP
+
)
If strong uorescence of M
+
is found, it will be a uorescent M
+
probe to monitor
ET/HT phenomena in homogeneous and heterogeneous environment with a high sensi-
tivity. DMP
+
showed a uorescence stronger than TMB
+
in DCE [150,163].
A transient absorption spectrum with a peak at 575 nm assigned to DMP
+
was
observed immediately after e
+
decreased with time, and most of DMP
+
disappeared at 2 Asec after e
to yield
DMP
+
. It has been shown that bimolecular deproto-
nation of DMP
+
occurs to yield DMP
at k
H
+=5.5 10
10
M
1
sec
1
in the presence of
basic reagents. The basic species for the deprotonation is probably Cl
generated by the
radiation chemical primary processes of DCE. The plausible mechanism involved therein
is summarized in Scheme 16.
The similarity of the absorption spectrum of DMP
+
to that of TMB
+
indicates a
similar electronic structure. The structure of TMB
+
is deviated frompseudo-D
3h
symmetry
because of repulsion of three Me groups, while that of DMP
+
is closer to pseudo-D
3h
symmetry. When D
3h
symmetry is an important factor for higher eciency of the
uorescence and Me group enhances internal conversion, DMP
+
* uoresces stronger
than TMB
+
*. It is interesting to knowwhether unimolecular deprotonation fromDMP
+
*
Figure 22 Time-resolved absorption spectra measured at 30, 220, and 2000 nsec (gate time: 30 nsec)
after e
+
becomes lower than TMB
+
.
Excitation of DMP
+
with hv
532
gave a broad uorescence around 600750 nm
(Fig. 23). The uorescence spectrum showed a mirror image symmetry to the absorption
spectrum of DMP
+
. The integrated uorescence intensity monitored decreased with the
increase of the delay time of hv
532
relative to e
+
*. When DMP
+
was excited with hv
532
,
the absorption of DMP
+
was unchanged and no increase in that of DMP
was observed.
This indicates that DMP
+
* does not undergo deprotonation, unlike its neutral molecule
[164]. Because the acidity of the OH group of DMP
+
in the ground state is enhanced by
the positive charge of DMP
+
and deprotonation takes place even in the presence of weak
bases such as Cl
or H
2
O, and E
ex
= 47 kcal mol
1
and the resonance stabilization of the
Scheme 16 Deprotonation of DMP
+
through [DMP
+
/Cl
].
Figure 23 Fluorescence spectrum of DMP
+
measured at 100 nsec after e
. Energy of hv
532
, 130
mJ. Inset: a plot of the integrated uorescence intensity of DMP
+
as a function of the delay time of
hv
532
relative to e
+
upon excitation. The absence of
the deprotonation upon excitation strongly suggests that MO incorporated in the
excitation of DMP
+
does not correlate with the j* orbital of the OH bond.
The u
f
of DMP
+
in DCE was measured from the uorescence intensity as a function
of the delay time of hv
532
relative to e
+
* by Cl
in [DMP
+
*/Cl
].
u
f
= (2F0.3) 10
3
for DMP
+
* was higher than B
f
= 1.1 10
3
for free TMB
+
* in
DCE. The intense uorescence of DMP
+
* ruled out the demand of complete pseudo-
D
3h
structure in M
+
* and suggested a possibility of substitution of H atom of the OH
group of DMP with an alkyl group as an attachment group for the introduction of DMP
unit to various molecules or surfaces as a uorescent radical cation probe. k
f
= 5.8
10
6
sec
1
was estimated for the uorescence decay of DMP
+
* using the StricklerBerg
relationship. Although k
f
was smaller than that for TMB
+
* (9.5 10
6
sec
1
), u
f
was
larger by a factor of 1.8. It is suggested that internal conversion in DMP
+
* is slower by
a factor of 3.0 as compared to TMB
+
*. Because the deprotonation from DMP
+
* is
not an important process for internal conversion, the Me group would play an important
role in internal conversion of DMP
+
* and TMB
+
* as reported for that of
1
alkylbenzenes*
[165]. sc350 psec of DMP
+
* was estimated from u
f
= k
f
/(k
f
+ k
ic
) = k
f
s c k
f
/k
ic
. It
is strongly suggested that the introduction of Me groups in 1,3,5-trioxybenzenes causes
the deviation from the pseudo-D
3h
structure by their mutual repulsion, reducing u
f
by the
enhancement of internal conversion.
It is found that radical cations of 1,3,5-trixybenzene derivatives in solution uoresce
with moderately high u
f
c10
3
in the rage of 580800 nm.
6. CONCLUDING REMARKS
New aspects of radical ion chemistry in solutions have been described in this chapter.
These studies were performed by PR, g-R, and PRLFP techniques in which M
+
or M
+
withp-MeO
and oxidation of S
+
with p-MeO are explained by chargespin separation in such S
+
.
Unimolecular ct isomerization of such c-S
+
proceeds with a chain mechanism, while
regioselective oxidation occurs in such S
+
because of the spin localization. Cycloreversion
of t,c,t-TPCB
+
occurs to give a j-St
2
+
, while the photochemical cycloreversion of t,c,t-
TPCB
+
and t,t,t-TPCB
+
gives k-St
2
+
and t-St
+
/t-St pair, respectively. Radical cations of
phosphorus compounds (9
+
and 10
+
) form intramolecular k-dimer
+
between two Nps
from which Np
2
+
forms. Formation of intermolecular j-dimer
of aromatic acetylene
(11
and 12
of 13
and diarylmethanol
was observed,
and the n = 3 rule is not eective for intramolecular dimer
.
Anisole and Bp act as selective hole and electron quenchers for St
+
* and St
+
* via
HTand ETquenching, respectively. Assuming that k
ht
=k
et
=k
di
, s
c
and s
t
were estimated
to be 120 and 240 psec for c-St
+
* and t-St
+
*, respectively, which are shorter than H =1.5
and 2.5 nsec for c-St
* and t-St
*. The order of s
c
<s
t
is attributed to the isomerization
via the twisting only in c-St
+
* and c-St
+*/
*
compared with t-St
+*/
*
. Diabatic isomerization from c-St
*/
*
to t-St
+/
is
suggested to proceed via crossing of the D
2
and D
0
states. The selective ET quenching of
Copyright 2004 by Taylor & Francis Group, LLC
M
* (M = DCA, PZ, AQ, and RAQ) by Q (FN and DCB) occurs to give s = 4 nsec of
M
*. Intermediacy of the
M/Q
* by Q.
It is found that radical cations of 1,3,5-trioxybenzene derivatives such as TMB
+
and
DMP
+
in solution uoresce with moderately high u
f
c10
3
in the range of 580800 nm.
Large u
f
was obtained for free TMB
+
, while small u
f
was obtained for the [TMB
+
/Cl
]
ion pair because of fast HT quenching of TMB
+
* by Cl
+
by Cl
. 1,3,5-Trioxyl substitution is
necessary for the ecient uorescence because of the pseudo-D
3h
symmetry causing low
D
0
D
1
and high D
0
D
2
energy separations. Intramolecular and intermolecular HT
quenching of TMB
+
* occurred by substituents of the O atom and hole acceptor molecules
with low IP.
Because M
+
*
/
+
*
/
* is interesting and
still a new subject. The PR method can allow us to study various M
+/
as the starting
molecule. Because we used the 532-nmlaser ash for excitation of M
+/
+/
with absorption at 532 nmcan be used as the starting molecule. When we can use any
k for LFP, various M
+/
+/
. And
because various excited states exist in M
+
*
/
*, selective excitation of M
+/
to M
+/
(D
n
) is of particular interest to elucidate the transient phenomena depending on the
excited state. Although it has been considered that M
+
*
/
* is nonluminescent, we have
found that 1,3,5-trioxybenzenes
+
uoresces with moderately high u
f
c10
3
. Considering
that uorescence was readily observed with conventional photomultiplier as a detector,
we can expect the use of these type of compounds as ET/HT probe for homogeneous
solutions and also for heterogeneous systems such as photocatalysts, and biological
oxidation in future.
ACKNOWLEDGMENTS
The author thanks his collaborators, particularly Mrs. Sachiko Tojo, Drs. Akito Ishida,
Nobuyuki Ichinose, Mitsunobu Nakamura, and his students, as well as the members of
the Radiation Laboratory of ISIR, Osaka University, for running the linear accelerator.
This work was partly supported by a Grant-in-Aid from the Ministry of Education,
Culture, Sports, Science and Technology (MEXT) of Japanese Government.
REFERENCES
1. Fox, M.A. Chem. Rev. 1979, 79, 253.
2. Fox, M.A., Chanon, M., Eds.; Photoinduced Electron Transfer; Elsevier: Amsterdam, 1988.
3. Eberson, L., Ed.; Electron TransferReactions in Organic Chemistry; Springer-Verlag: Berlin,
1987.
4. Mattes, S.L.; Farid, S. In Organic Photochemistry; Padwa, A., Ed.; Marcel Dekker: New
York, 1983; 233 pp.
5. Julliard, M.; Chanon, M. Chem. Rev. 1983, 83, 425.
6. Kavarnos, G.J.; Turro, N.J. Chem. Rev. 1986, 86, 401.
7. Roth, H.D. Acc. Chem. Res. 1987, 20, 343.
8. Yoon, U.C.; Mariano, P.S. Acc. Chem. Res. 1992, 25, 233.
9. Peters, K.S.; Cashin, A. J. Phys. Chem. A 2000, 104, 4833.
Copyright 2004 by Taylor & Francis Group, LLC
10. Lewis, F.D.; Letsinger, R.L.; Wasielewski, M.R. Acc. Chem. Res. 2001, 34, 159.
11. Stevenson, J.P.; Jackson, W.F.; Tanko, J.M. J. Am. Chem. Soc. 2002, 124, 4271.
12. Pascaly, M.; Yoo, J.; Barton, J.K. J. Am. Chem. Soc. 2002, 124, 9083.
13. Nakatani, K.; Dohno, C.; Saito, I. J. Am. Chem. Soc. 2002, 124, 6802.
14. Kanvah, S.; Schuster, G.B. J. Am. Chem. Soc. 2002, 124, 11286.
15. Sinnecker, S.; Koch, W.; Lubitz, W. J. Phys. Chem. B 2002, 106, 5281.
16. Fukuzumi, S.; Ohkubo, K.; Chen, Y.H.; Pandey, R.K.; Zhan, R.Q.; Shao, J.G.; Kadish, K.M.
J. Phys. Chem. A 2002, 106, 5105.
17. Horner, J.H.; Taxil, E.; Newcomb, M. J. Am. Chem. Soc. 2002, 124, 5402.
18. Baciocchi, E.; Bietti, M.; Salamone, M.; Steenken, S. J. Org. Chem. 2002, 67, 2266.
19. Adam, W.; Librera, C.P. J. Org. Chem. 2002, 67, 576.
20. Shukla, S.S.; Rusling, J.F. J. Phys. Chem. 1985, 89, 3353.
21. Galland, B.; Moutet, J.C.; Reverdy, G. Electrochim. Acta 1987, 32, 175.
22. Moutet, J.C.; Reverdy, G. Nouv. J. Chim. 1983, 7, 105.
23. Moutet, J.C.; Reverdy, G. Tetrahedron Lett. 1979, 20, 2389.
24. Labbe, P.; Moutet, J.C.; Paltrier, M.; Reverdy, G. Nouv. J. Chim. 1984, 8, 627.
25. Shine, H.J.; Zhao, D.C. J. Org. Chem. 1990, 55, 4086.
26. Moutet, J.C.; Reverdy, G. J. Chem. Soc., Chem. Commun. 1982, 654.
27. Nelleborg, P.; Lund, H.; Eriksen, J. Tetrahedron Lett. 1985, 26, 1733.
28. Breslin, D.T.; Fox, M.A. J. Org. Chem. 1994, 59, 7557.
29. Fox, M.A.; Dulay, M.T.; Krosley, K. J. Am. Chem. Soc. 1994, 116, 10992.
30. Wang, Z.; Mcgimpsey, W.G. J. Phys. Chem. 1993, 97, 3324.
31. Wang, Z.; Mcgimpsey, W.G. J. Phys. Chem. 1993, 97, 5054.
32. Kuriyama, Y.; Hashimoto, F.; Tsuchiya, M.; Sakuragi, H.; Tokumaru, K. Chem. Lett. 1994,
1371.
33. Breslin, D.T.; Fox, M.A. J. Phys. Chem. 1994, 98, 408.
34. Ichikawa, T.; Moriya, T.; Yoshida, H. J. Phys. Chem. 1976, 80, 1278.
35. Holroyd, R.A. J. Phys. Chem. 1982, 86, 3541.
36. In CRC Handbook of Radiation Chemistry. Tabata, Y., Ed.; CRC Press: Boca Raton, 1991;
63 pp.
37. Hart, E.D.; Boag, J.W. J. Am. Chem. Soc. 1962, 84, 4090.
38. Hunt, J.W.; Thomas, J.K. Radiat. Res. 1967, 32, 149.
39. In CRC Handbook of Radiation Chemistry; Tabata, Y., Ed.; CRC Press: Boca Raton, 1991;
439 pp.
40. Shida, T.; Hamill, W.H. J. Chem. Phys. 1966, 44, 2375.
41. Shida, T.; Haselbach, E.; Bally, T. Acc. Chem. Res. 1984, 17, 180.
42. Tojo, S.; Morishima, K.; Ishida, A.; Majima, T.; Takamuku, S. Bull. Chem. Soc. Jpn. 1995,
68, 958.
43. Ishida, A.; Fukui, M.; Ogawa, H.; Tojo, S.; Majima, T.; Takamuku, S. J. Phys. Chem. 1995,
99, 10808.
44. Majima, T.; Tojo, S.; Ishida, A.; Takamuku, S. J. Org. Chem. 1996, 61, 7793.
45. Majima, T.; Tojo, S.; Ishida, A.; Takamuku, S. J. Phys. Chem. 1996, 100, 13615.
46. Majima, T.; Fukui, M.; Ishida, A.; Takamuku, S. J. Phys. Chem. 1996, 100, 8913.
47. Fujita, M.; Ishida, A.; Majima, T.; Takamuku, S. J. Phys. Chem. 1996, 100, 5382.
48. Majima, T.; Tojo, S.; Takamuku, S. J. Phys. Chem. A 1997, 101, 1048.
49. Yamamoto, Y. Trends. Org. Chem. 1992, 3, 93.
50. Tamai, T.; Mizuno, K.; Hashida, I.; Otsuji, Y.; Ishida, A.; Takamuku, S. Chem. Lett. 1994,
149.
51. Kimura, N.; Takamuku, S. J. Am. Chem. Soc. 1994, 116, 4087.
52. Tsuchida, A.; Ikawa, T.; Yamamoto, M.; Ishida, A.; Takamuku, S. J. Phys. Chem. 1995, 99,
14793.
53. Kimura, N.; Takamuku, S. J. Am. Chem. Soc. 1995, 117, 8023.
54. Kojima, M.; Kakehi, A.; Ishida, A.; Takamuku, S. J. Am. Chem. Soc. 1996, 118, 2612.
Copyright 2004 by Taylor & Francis Group, LLC
55. Shida, T.; Momose, T. J. Mol. Struct. 1985, 126, 159.
56. Suzuki, H.; Koyano, K.; Shida, T.; Kira, A. Bull. Chem. Soc. Jpn. 1982, 55, 3690.
57. Shida, T.; Kato, T.; Nosaka, Y. J. Phys. Chem. 1977, 81, 1095.
58. Shida, T.; Momose, T.; Ono, N. J. Phys. Chem. 1985, 89, 815.
59. Toriyama, K.; Nunome, K.; Iwasaki, M.; Shida, T.; Ushida, K. Chem. Phys. Lett. 1985, 122,
118.
60. Dunkin, I.R.; Andrews, L.; Lurito, J.T.; Kelsall, B.J. J. Phys. Chem. 1985, 89, 1701.
61. Ushida, K.; Shida, T.; Shimokoshi, K. J. Phys. Chem. 1989, 93, 5388.
62. Momose, T.; Suzuki, T.; Shida, T. Chem. Phys. Lett. 1984, 107, 568.
63. Shida, T. J. Phys. Chem. 1978, 82, 991.
64. Lewis, F.D.; Dyksta, R.E.; Gould, I.R.; Farid, S. J. Phys. Chem. 1988, 92, 7042.
65. Lewis, F.D.; Bedell, A.M.; Dykstra, R.E.; Elbert, J.E.; Gould, I.R.; Farid, S. J. Am. Chem.
Soc. 1990, 112, 8055.
66. Kuriyama, Y.; Arai, T.; Sakuragi, H.; Tokumaru, K. Chem. Phys. Lett. 1990, 173, 253.
67. Kuriyama, Y.; Sakuragi, H.; Tokumaru, K.; Yoshida, Y.; Tagawa, S. Bull. Chem. Soc. Jpn.
1993, 66, 1852.
68. Eriksen, J.; Foote, C.S. J. Am. Chem. Soc. 1980, 102, 6083.
69. Lewis, F.D.; Petisce, J.R.; Oxman, J.D.; Nepras, M.J. J. Am. Chem. Soc. 1985, 107, 203.
70. Konuma, S.; Aihara, S.; Kuriyama, Y.; Misawa, H.; Akaba, R.; Sakuragi, H.; Tokumaru, K.
Chem. Lett. 1991, 1897.
71. Tojo, S.; Morishima, K.; Ishida, A.; Majima, T.; Takamuku, S. J. Org. Chem. 1995, 60, 4684.
72. Hamill, W.H. In Radical Ions. Kaiser, E.T., Kevan, L., Eds.; Interscience: New York, 1968;
405 pp.
73. Shida, T. Electronic Absorption Spectra of Radical Ions; Elsevier: Amsterdam, 1988.
74. Yamamoto, Y.; Aoyama, T.; Hayashi, K. J. Chem. Soc., Faraday Trans. I 1988, 84, 2209.
75. Kikuchi, O.; Oshiyama, T.; Takahashi, O.; Tokumaru, K. Bull. Chem. Soc. Jpn. 1992, 65, 2267.
76. Hirota, S.; Ogura, T.; Appelman, E.H.; Shinzawaitoh, K.; Yoshikawa, S.; Kitagawa, T. J.
Am. Chem. Soc. 1994, 116, 10564.
77. Tojo, S.; Toki, S.; Takamuku, S. J. Org. Chem. 1991, 56, 6240.
78. Takamuku, S.; Komitsu, S.; Toki, S. Radiat. Phys. Chem. 1989, 34, 553.
79. Badger, B.; Brockleh, B. Trans. Faraday Soc. 1969, 65, 2576.
80. Ichikawa, T.; Ohta, N.; Kajioka, H. J. Phys. Chem. 1979, 83, 284.
81. Brede, O.; Bos, J.; Helmstreit, W.; Mehnert, R. Radiat. Phys. Chem. 1982, 19, 1.
82. Yamamoto, Y.; Chikai, Y.; Hayashi, K. Bull. Chem. Soc. Jpn. 1985, 58, 3369.
83. Ebbesen, T.W. J. Phys. Chem. 1988, 92, 4581.
84. Tsuchida, A.; Yamamoto, M. J. Photochem. Photobiol. A 1992, 65, 53.
85. Sluggett, G.W.; McGarry, P.F.; Koptyug, I.V.; Turro, N.J. J. Am. Chem. Soc. 1996, 118, 7367.
86. Ganapathy, S.; Dockery, K.P.; Sopchik, A.E.; Bentrude, W.G. J. Am. Chem. Soc. 1993, 115,
8863.
87. Barth, A.; Hauser, K.; Mantele, W.; Corrie, J.E.T.; Trentham, D.R. J. Am. Chem. Soc. 1995,
117, 10311.
88. Furuta, T.; Torigai, H.; Sugimoto, M.; Iwamura, M. J. Org. Chem. 1995, 60, 3953.
89. Givens, R.D.; Kueper, L.W. J. Chem. Soc., Chem. Rev. 1993, 93, 55.
90. Nakamura, M.; Majima, T. Chem. Commun. 1997, 1291.
91. Nakamura, M.; Dohno, R.; Majima, T. J. Org. Chem. 1998, 63, 6258.
92. Powell, R.L.; Hall, C.D. J. Am. Chem. Soc. 1969, 91, 5403.
93. Pandey, G.; Hajra, S.; Chorai, M.K.; Kumar, K.R. J. Am. Chem. Soc. 1997, 119, 8777.
94. Pandey, G.; Pooranchand, D.; Bhalerao, U.T. Tetrahedron Lett. 1991, 47, 1745.
95. Yasui, S.; Shioji, K.; Ohno, A.; Yoshihara, M. J. Org. Chem. 1995, 60, 2099.
96. Nakamura, M.; Miki, M.; Majima, T. J. Chem. Soc., Perkin Trans. 2, 2000, 1447.
97. Hamill, W.H. In Radical Ions; Kaiser, E.T., Kevan, L., Eds.; Interscience: New York, 1968;
321 pp.
98. In CRC Handbook of Radiation Chemistry; Tabata, Y., Ed.; CRC Press: Boca Raton, 1991;
395 pp.
Copyright 2004 by Taylor & Francis Group, LLC
99. Arai, S.; Dorfman, L.M. J. Chem. Phys. 1964, 41, 2190.
100. Arai, S.; Grev, D.A.; Dorfman, L.M. J. Chem. Phys. 1967, 46, 2537.
101. Arai, S.; Kira, A.; Imamura, M. J. Phys. Chem. 1977, 81, 110.
102. Shida, T.; Iwata, S. J. Chem. Phys. 1972, 56, 2858.
103. Shida, T.; Hamill, W.H. J. Chem. Phys. 1966, 44, 2369.
104. Shaede, E.A.; Dorfman, L.M.; Flynn, G.F.; Walker, D.C. Can. J. Chem. 1973, 51, 3905.
105. Cserhegy, A.; Chaudhur, J.; Franta, E.; Jagurgro, J.; Szwarc, M. J. Am. Chem. Soc. 1967, 89,
7129.
106. Wang, H.C.; Levin, G.; Szwarc, M. J. Am. Chem. Soc. 1977, 99, 2624.
107. Dadley, D.; Evans, A.G. J. Chem. Soc. B 1967, 418.
108. Levin, G.; Jagurgro, J.; Szwarc, M. J. Am. Chem. Soc. 1970, 92, 2268.
109. Yamamoto, S.; Yamamoto, Y.; Hayashi, K. Bull. Chem. Soc. Jpn. 1991, 64, 346.
110. Aoyama, T.; Yamamoto, Y.; Hayashi, K. J. Chem. Soc. Faraday Trans. I 1989, 85, 3353.
111. Yamamoto, Y.; Nishida, S.; Ma, X.H.; Hayashi, K. J. Phys. Chem. 1986, 90, 1921.
112. Hirayama, F. J. Chem. Phys. 1965, 42, 3163.
113. Ichinose, N.; Hobo, J.; Tojo, S.; Majima, T. Chem. Phys. Lett. 2000, 330, 97.
114. Ishitani, A.; Nagakura, S. Mol. Phys. 1967, 12, 1.
115. Kira, A.; Ishiwata, M.; Imamura, M.; Tabata, Y. Radiat. Phys. Chem. 1979, 15, 663.
116. Tsuchida, A.; Masuda, N.; Yamamoto, M.; Nishijima, Y. Macromolecules 1986, 19, 1299.
117. Lim, B.T.; Lim, E.C. J. Chem. Phys. 1983, 78, 5262.
118. Lund, H.; Carlsson, H.S. Acta Chem. Scand., Ser. B 1978, B32, 505.
119. Nelleborg, P.; Lund, H.; Eriksen, J. Tetrahedron Lett. 1985, 26, 1773.
120. Eriksen, J.; Jerrgensen, K.A.; Linderberg, J.; Lund, H. J. Am. Chem. Soc. 1984, 106, 5083.
121. Eriksen, J.; Lund, H.; Nyvad, A.I. Acta Chem. Scand., Ser. B 1983, B37, 459.
122. Sazhnikov, V.A.; Rakhmatov, M.; Almov, M.V. Chem. Phys. Lett. 1980, 71, 33.
123. Ebbesen, T.W.; Akaba, R.; Tokumaru, K.; Washio, M.; Tagawa, S.; Tabata, Y. J. Am. Chem.
Soc. 1988, 110, 2147.
124. Sumiyoshi, T.; Sakai, H.; Kawasaki, M.; Katayama, M. Chem. Lett. 1992, 617.
125. Sumiyoshi, T.; Kawasaki, M.; Katayama, M. Bull. Chem. Soc. Jpn. 1993, 66, 2510.
126. Carlsson, H.S.; Lund, H. Acta Chem. Scand., Ser. B 1980, B34, 409.
127. Hiratsuka, H.; Yamazaki, T.; Maekawa, Y.; Kajii, Y.; Hikida, T.; Mori, Y. Chem. Phys. Lett.
1987, 139, 187.
128. Ramamurthy, V.; Caspar, J.V.; Corbin, D.R. J. Am. Chem. Soc. 1991, 113, 594.
129. Ikematsu, S.; Hikida, T. J. Photochem. Photobiol. A: Chem. 1990, 52, 193.
130. Oomori, T.; Hikida, T. Chem. Phys. 1993, 178, 477.
131. Bromberg, A.; Schmidt, K.H.; Meisel, D. J. Am. Chem. Soc. 1985, 107, 83.
132. Ishida, A.; Yamamoto, K.; Takamuku, S. Bull. Chem. Soc. Jpn. 1992, 65, 3186.
133. Arai, S.; Grev, D.A.; Dorfman, L.M. J. Chem. Phys. 1967, 46, 2572.
134. Kira, A.; Arai, S.; Imamura, M. J. Chem. Phys. 1971, 54, 4890.
135. Kira, A.; Arai, S.; Imamura, M. J. Phys. Chem. 1972, 76, 1119.
136. Sawada, U.; Holroyd, R.A. J. Phys. Chem. 1981, 85, 541.
137. Vanderde, Gm.; Dousma, J.; Speiser, S.; Kommande, J. Chem. Phys. Lett. 1973, 20, 17.
138. Oshiyama, T.; Takahashi, O.; Morihashi, K.; Kikuchi, O.; Tokumaru, K. Bull. Chem. Soc.
Jpn. 1993, 66, 1622.
139. Saltiel, J.; Waller, A.S.; Sears, D.F. J. Am. Chem. Soc. 1993, 115, 2453.
140. Bartocci, G.; Mazzucato, U. Chem. Phys. Lett. 1977, 47, 541.
141. Saltiel, J.; Waller, A.S.; Sears, D.F.; Garrett, C.Z. J. Phys. Chem. 1993, 97, 2516.
142. Heisel, F.; Miehe, J.A.; Sipp, B. Chem. Phys. Lett. 1979, 61, 115.
143. Courtney, S.H.; Balk, M.W.; Philips, L.A.; Webb, S.P.; Yang, D.; Levy, D.H.; Fleming, G.R.
J. Chem. Phys. 1988, 89, 6697.
144. Lee, M.; Bain, A.J.; Mccarthy, P.J.; Han, C.H.; Haseltine, J.N.; Smith, A.B.; Hochstrasser,
R.M. J. Chem. Phys. 1986, 85, 4341.
145. Todd, D.C.; Jean, J.M.; Rosenthal, S.J.; Ruggiers, A.J.; Yang, D.; Fleming, G.R. J. Chem.
Phys. 1990, 93, 8658.
Copyright 2004 by Taylor & Francis Group, LLC
146. Burrows, H.D.; Kemp, T.J.; Greatore, D. J. Phys. Chem. 1972, 76, 20.
147. Shkrob, I.A.; Sauer, M.C.; Liu, A.D.; Crowell, R.A.; Trifunac, A.D. J. Phys. Chem. A 1998,
102, 4976.
148. Cook, A.R.; Curtiss, L.A.; Miller, J.R. J. Am. Chem. Soc. 1997, 119, 5729.
149. Zimmer, K.; Hoppmeier, M.; Schweig, A. Chem. Phys. Lett. 1998, 293, 366.
150. Ichinose, N.; Tanaka, T.; Kawanishi, S.; Suzuki, T.; Endo, K. J. Phys. Chem. A 1999, 103,
7923.
151. Ichinose, N.; Majima, T. Chem. Phys. Lett. 2000, 322, 15.
152. Ichinose, N.; Tanaka, T.; Kawanishi, S.; Majima, T. Chem. Phys. Lett. 2000, 326, 293.
153. Ichinose, N.; Majima, T. The Spectrum 2001, 13, 14.
154. Grodkowski, J.; Neta, P. J. Phys. Chem. 1984, 88, 1205.
155. Sujdak, R.J.; Jones, R.L.; Dorfman, L.M. J. Am. Chem. Soc. 1976, 98, 4875.
156. Mehnert, R.; Brede, O.; Bos, J.; Naumann, B. Bunsenges. Phys. Chem. 1979, 83, 992.
157. Alfassi, Z.B.; Mosseri, S.; Neta, P. J. Phys. Chem. 1989, 93, 1380.
158. Emmi, S.S.; Beggiato, G.; Casalbore-Miceli, G. Radiat. Phys. Chem. 1989, 33, 29.
159. Chateauneuf, J.E. J. Am. Chem. Soc. 1990, 112, 442.
160. Shoute, L.C.T.; Neta, P. J. Phys. Chem. 1990, 94, 2447.
161. Mah, S.; Yamamoto, Y.; Hayashi, K. J. Phys. Chem. 1983, 87, 297.
162. Yamamoto, Y.; Nishida, S.; Hayashi, K. J. Chem. Soc., Faraday Trans. 1 1987, 83, 1795.
163. Ichinose, N.; Tojo, S.; Majima, T. Chem. Lett 2000, 1126.
164. Gadosy, T.A.; Shukla, D.; Johnston, L.J. J. Phys. Chem. A 1999, 103, 8834.
165. Schloman, W.W.; Morrison, H. J. Am. Chem. Soc. 1977, 99, 3342.
Copyright 2004 by Taylor & Francis Group, LLC
23
Application of Radiation Chemistry
to Nuclear Technology
Yosuke Katsumura
The University of Tokyo, Tokyo, Japan
INTRODUCTION
Nuclear technology is dened as a technology to use the nuclear energy through nuclear
ssion and fusion reactions as nuclear power and to develop useful processing and
application with radiation and radioisotopes. Therefore radiation-induced eect is one of
the important issues when nuclear technology is developed and employed. In the present
chapter, three important applications of radiation chemistry in nuclear technology will be
discussed: coolant water in nuclear power stations, spent nuclear fuel reprocessing, and
radioactive high-level waste repository.
1. RADIOLYSIS OF WATER AT ELEVATED TEMPERATURERADIATION
EFFECT OF COOLANT WATER IN NUCLEAR REACTORS
1.1. Importance for the Understanding of the Radiolysis at Elevated
Temperatures
All over the world, 432 nuclear power reactors are under operation and more than 36 GWof
electricity could be produced as of December 31, 2001. There are several types of reactors
such as boiling water reactor (BWR), pressurized water reactor (PWR), Canada deuterium
uranium(CANDU), and others. In these reactors, light water is normally used not only as a
coolant, but also as a moderator. On the contrary, in CANDU reactors, heavy water is
taken. It is widely known that the quality control of coolant water, the so-called water
chemistry, is inevitably important for keeping the integrity of the plant.
The coolant water at around 300jC is irradiated by both g-rays and fast neutrons at
the core region with a dose rate of several 10 kGy/sec, which is more than 3 orders of
magnitude of the dose higher than those obtained in conventional
60
Co g-ray irradiation
facilities. As a result, O
2
and H
2
O
2
are formed as radiolysis products. It is well known that
the concentrations of O
2
and H
2
O
2
strongly aect the stress corrosion cracking (SCC) and
the electrochemical corrosion potential of the primary circuit. In order to avoid the SCC, it is
necessary to keep the lower level of these concentrations and hydrogen injection (HWC),
and, recently, noble metal chemical addition (NMC) has been widely employed in BWRs. In
Copyright 2004 by Taylor & Francis Group, LLC
the hydrogen injection, both H
2
O
2
and O
2
would be converted into water molecules by
radiolytic processes (see below).
To grasp the chemical condition of water in the pressure vessel, direct measurement is
practically impossible because of high pressure, high temperature, and intense radiation. In
order to predict the concentrations of water decomposition products, a computer simulation
should be applied. This idea was found in 1960s [13]. To performthe simulation, both a set
of G-values for water decomposition products and a set of reactions for transient species are
necessary. For these two decades, much eort has been made in Sweden, Denmark, United
Kingdom, Canada, and Japan to evaluate the G-values and rate constants of the reactions at
elevated temperatures up to 300jC, and nowthere are practically enough accumulated data.
There are several reviews of water radiolysis at elevated temperatures [47] and examples of
practical application of the radiolysis in reactors [8,9].
1.2. Temperature Dependence of Water Decomposition Products
up to 300jC
The coolant water at around 300jC is irradiated mainly at the core of the reactor. At the
initial stage to determine the G-values of water decomposition products at elevated
temperatures, the Fricke dosimeter was chosen [1014] because the mechanism of the
reaction has been established. Since the reactions in neutral solution are of practical interest,
intensive measurement of the G-values of water decomposition products at elevated
temperatures in neutral solutions has been done [1521].
For the measurement, techniques of product analysis and pulse radiolysis were
employed. In product analysis, H
2
yield in degassed 2 mM KBr or 1 mM KNO
2
solutions
were taken to evaluate g(H
2
) and degassed solutions of 1 mM Cd
2+
with 10 mM methanol,
1 mMacetone with 10 mMmethanol, and 1 mMacetone with 10 mM2-rpropanol have been
selected for g(H)+g(H
2
). For the measurement of g(e
aq
)+g(H)+g(H
2
), degassed solution
of 1 mM HClO
4
with methanol was taken. In pulse radiolysis measurement, Fe(CN)
6
3
and
CO
3
were used for g(OH) evaluation and 0.25 mM methyl viologen solution with 10 mM
tert-butanol for g(e
aq
). Many other systems have been used. In the presence of formate, the
yield of G(MV
+
) would be equal to g(e
aq
aq
2:56 3:40 10
3
t 1
gOH 2:64 7:17 10
3
t 2
gH
2
0:43 0:69 10
3
t 3
gH
2
GH 0:97 1:98 10
3
t 4
gH
2
O
2
0:72 1:49 10
3
t 5
Here t stands for temperature in jC.
Fromthe dierence between Eqs. (4) and (3), the temperature dependence of g(H) can
be calculated as
gH 0:54 1:28 10
3
t 6
These lines are also drawn in Fig. 1. It is clear that the agreement is quite nice. Similar tting
equations in heavy water for CANDU reactors were also obtained [19]. Furthermore,
temperature-dependence measurements with fast neutrons and high LET ion beams were
also investigated [20,23].
Common features of water radiolysis at elevated temperatures up to 300jC are
summarized as below.
1. g(e
aq
), g(OH), g(H
2
), and g(H) are all increasing almost linearly with increasing
temperature up to 300jC. About 40% increase was found for g(e
aq
) and more
than 80% for g(OH). This is due to the lower reactivity of OH as compared with
that of e
aq
at elevated temperatures.
2. g(H
2
O
2
) is decreasing with temperature. In spite of the short lifetime of H
2
O
2
due to the thermal decomposition, 50 min at 150jC, 510 min at 150jC, and 1
min at 200jC, this decrease is obvious.
As pointed out before, time scale of scavenging, corresponding to G-values at elevated
temperatures, is not clear yet and further investigation is needed.
1.3. Rate Constants of the Reactions up to 300jC
It is known that more than 30 reactions are needed to reproduce the radiation-induced
reactions occurring in pure water. Intensive measurements with a pulse radiolysis method
have been done at elevated temperature up to 300jC [2542], and the temperature depend-
ence of some reactions does not exhibit a straight line but a curved one in Arrhenius plot.
These examples are the reactions of the hydrated electron with N
2
O, NO
3
, NO
2
, phenol,
SeO
4
2
, S
2
O
3
2
, and Mn
2+
[33,35], and two examples, e
aq
+NO
3
and e
aq
+NO
2
, are shown in
Fig. 2. The rate constant for the reaction of hydrated electron with NO
3
is near
diusion-controlled reaction at roomtemperature and is increasing with increasing temper-
ature. Above 100jC, the rate does not increase and reaches the maximumat 150jC, and then
decreases. Therefore the curve is concave upward in Arrhenius plot.
Copyright 2004 by Taylor & Francis Group, LLC
Figure 1 Variations of G-values (in aprt./100 eV) for the radiolysis of light water as a function of
temperature; (a) reducing species G-values for G(e
aq
), G(H+H
2
), and G(H
2
) and (b) oxidizing
species. Data are taken from Ref. 19 (o5 w q), Ref. 18 (.Ex), and Refs. 20 and 23 (6) for reducing
species and from Ref. 19 (w .o), Ref. 17 (65), and Ref. 24 (Dxj) for oxidizing species. It is noted
that because of the large uncertainties in the G-values determined in Refs. 20 and 23 for both G(OH)
and G(H
2
O
2
) from HClO
4
+10
2
mol kg
1
methanol chemical system at above 100jC, these data are
not shown in (b). Lines are corresponding to Eqs. (1)(5).
Copyright 2004 by Taylor & Francis Group, LLC
Figure 2 The Arrhenius plot rate constants for the reactions of e
aq
with NO
3
(a) and NO
2
(b):
k
obs
(n), k
react
(.), k
di
(
: : :
), calculated t of k
react
(). A
1
/M
1
and A
2
are (5.7F2.1) 10
12
and (4.4
F 6.0) 10
5
, respectively, for (a). A
1
/M
1
and A
2
are (2.0 F 0.8) 10
11
and (2.1 F 0.7) 10
3
,
respectively, for (b). (From Ref. 35.)
Copyright 2004 by Taylor & Francis Group, LLC
In general, observed reaction rate constant is written as below with a diusion term
and a reaction term [43].
1=k
obs
1=k
di
1=k
react
7
The reaction of the hydrated electron would be considered as a charge transfer reaction and
elementary processes can be described as below [44].
D AZ
K
A
D=AZ
k
1
k
2
D=A*Z
m
D
=A
*!
k
3
D
=A
! product 8
Based on the above scheme, an equation for the overall reaction is derived and setting k
3
=k
2
gives a next one [44].
k
react
K
A
k
1
=2 k
2
=m 9
This is rearranged into Eq. (10)
k
react
T A
1
expE
1
=RT=2 A
2
expE
2
=RT 10
Here K
A
k
1
= A
1
exp(E
1
/RT) and k
2
/m = A
2
exp(E
2
/RT).
The temperature dependence of k
di
can be evaluated by the diusion coecients of
the reactants as a function of temperature. The term of k
obs
is experimentally observed.
Then, k
react
can be evaluated by tting procedures. In Fig. 2, k
obs
, k
di
, and k
react
were all
shown [35]. Similar analysis was made for the experimentally obtained reactions of OHwith
Fe(CN)
6
4
, OH, and HCO
2
, H with MnO
4
and O
2
, e
aq
with H
+
, H
2
O
2
, and Cd
2+
, and
so on [34,37,42].
Recent measurement of the hydrated electron at elevated temperature by Shiraishi
et al. [41] shows clearly the existence of an equilibrium between the hydrated electron and
the proton.
e
aq
H
ZH; 11
In order to calculate the reactions in pure water at elevated temperatures, more than 30
reactions as a function of temperature are needed, especially for the reactions in nuclear
reactors. Much work has been done and a reaction set for calculation at 285jCand 300jCis
summarized in Table 1 [4] for light water and in Ref. 5 for heavy water.
1.4. Theoretical Calculations of High-Temperature Radiolysis of Water
In order to reproduce the temporal behavior of water decomposition products, two theo-
retical approaches based on spur diusion model and Monte Carlo calculations have been
developed.
In the spur diusion calculation, initial distribution of the water decomposition after
thermalization is assumed as a spherical one and is set normally as a Gaussian distribution
where hydrated electron has a radius of 2.3 nm, which is larger than the radii of 0.85 nm for
OH, H, H
3
O
+
, H, H
2
, and H
2
O
2
at ambient temperature. Then, the spur reactions start
taking place after 1 psec during the diusion of each species. These calculations have
explained and predicted well the time dependence of the water decomposition products at
roomtemperature. This method was successfully applied to the calculation for the radiolysis
with high LET radiation, where a cylindrical initial distribution is taken. Essence of the idea
has been clearly summarized by Draganic and Draganic [45]. This calculation has been
extended to the water radiolysis at elevated temperatures. Although the initial yields of
water decomposition products at 1 psec are xed as room temperature, the initial dis-
Copyright 2004 by Taylor & Francis Group, LLC
Table 1 Calculated Rate Constants for all Reactions in the Radiolysis of Light
Water at 25jC, 285jC, and 300jC
No. Reaction k
25
k
285
k
300
(1) e
aq
+e
aq
! 6.44E09
a
2.93E11 3.29E11
(2) e
aq
+H
2
O
2
! 1.41E10 2.53E11 2.76E11
(5) e
aq
+O
2
! 1.79E10 2.57E11 2.78E11
(6) e
aq
+O
2
+HO
2
! 1.28E10 1.65E11 1.78E11
(8) H+H! 5.43E09 9.20E10 1.00E11
(9) H+OH! 1.53E10 6.58E10 6.87E10
(10) H+H
2
O
2
! 5.16E07 1.03E09 1.13E09
(11) H+O
2
+O
2
! 3.50E07
Equilibria
(22) H
2
O! 1.95E05 7.57E02 6.90E02
(22) H
+
+OH
+H
2
O! 1.36E06 1.31E08 1.39E08
(25) H! 6.32E00 1.36E05 1.40E05
(25) H
+
+e
aq
+H
2
O! 1.57E01 8.28E03 9.94E03
(26) e
aq
+OH
+H
2
O! 1.36E6 1.31E08 1.39E08
(29) HO
2
! 7.14E05 2.70E05 1.98E05
(29) H
+
+O
2
+H
2
O! 1.36E06 1.31E08 1.39E08
Alkaline solution
(31) O
+H
2
! 1.21E08 1.62E09 1.75E09
(32) O
+H
2
O
2
! 5.53E08 1.04E10 1.13E10
(33) OH+HO
2
! 8.29E09 1.56E11 1.70E11
(34) O
+HO
2
! 3.50E09 6.33E10 6.90E10
(36) e
aq
+O
+O
2
! 3.70E09 3.04E10 3.24E10
(37) O
3
2
[78], carbonate radical (CO
3
) [79], Ag
0
and
Ag
2+
[80], and benzophenone ketyl and anion radicals [81] have been observed from room
temperature to 400jC in supercritical water. The (SCN)
2
radical formation in aqueous
solution has been widely taken as a standard and useful dosimeter in pulse radiolysis study
[82,83]. The lifetime of the (SCN)
2
radical is longer than 10 Asec at room temperature and
becomes shorter with increasing temperature. This dosimeter is not useful anymore at
elevated temperatures. The absorption spectrum of the (SCN)
2
radical again shows a red
shift with increasing temperature, but the degree of the shift is not signicant as compared
with the case of the hydrated electron. It is known that the (SCN)
2
radical is equilibrated
with SCN
aq
! /
2
CO
12
/
2
CO
Z/
2
C
OH 13
Therefore adjustment of pHcan control the chemical form. Both spectra of anion and ketyl
radical were measured at dierent temperatures up to 400jC. The spectrum of the anion
shows red shift, but the ketyl radical indicates blue shift.
Apulse radiolysis of Ag
+
solution was studied, and the behavior of formed silver atom
(Ag
0
) and dimer cation (Ag
2
+
) was measured [80]. The absorption band for dimer shows
a signicant red shift with increasing temperatures, which implies to the CTTS (charge
transfer to solvent) character of the band.
It is common that the absorption bands are normally dependent on temperature, and
most of them show red shift at elevated temperatures. These behaviors are closely related to
the solvation and the formation of hydrogen bonds between transients and water molecules.
Precise theoretical consideration is not made yet.
As mentioned before, the spectroscopic behavior of the transients is strongly depend-
ent on temperature and absorption coecients are not constant with temperature. At
present, it is dicult to evaluate the G-values of water decomposition products at elevated
temperatures. One of the recent attempts for the evaluation with a use of methylviologen
(MV
2+
) is shown in Fig. 6 [87]. The methylviologen has been taken to evaluate the G-values
up to 200jC [16,31]. Similar method was applied to higher temperatures up to 400jC
assuming that the absorption coecients would be estimated by extrapolation of the data
known up to 200jC. As shown in Fig. 6(a), the G-value of hydrated electron is increasing up
to 300jC, which is consistent with reported as described as summarized in Fig. 1. The value
is decreasing above 300jC, and above critical temperature, 374jC, a signicant pressure-
Figure 6 (a) g(e
aq
) as a function of density
at 400jC. (From Ref. 86.)
Copyright 2004 by Taylor & Francis Group, LLC
dependent yield was observed, as shown in Fig. 6(b). When ethanol is taken as a scavenger,
the ethanol radical reduce MV
2+
and the G(MV
) is equal to g(OH)+g(H)+g(e
aq
). It is
evident that not only temperature but also pressure, in other words, density of water, are the
key parameters.
As for the rate constants above 300jC, little experimental results have been reported.
Again, not only temperature, but also pressure dependence was reported. One of the
important issues is the decrease of the dielectric constant of water in supercritical water,
where it is less than 10 and much lower than the value of 79 at room temperature. It was
pointed out that the Coulombic interaction becomes important and the radiolysis of
supercritical water resembles that of the organic liquids with low dielectric constant [88].
In addition, it should be noted that the solubility of the solute is quite low and ion pairing
would have a signicant role [8991], which reects the diculties for the sample prepara-
tion in the actual experiment.
Above observations are the special aspects of the peculiarity of water radiolysis in
supercritical water and thus much further investigation is needed.
2. RADIATION EFFECTS IN SPENT NUCLEAR FUEL REPROCESSING
2.1. Purex Process
In nuclear power reactors, ssion reactions of U-235 are taking place continuously and
energy of 200 MeV per ssion is released. At the electric power plants, ssion energy is
converted into thermal energy at rst and nally to electricity. In commercial reactors, after
the operation for a certain period, new UO
2
fuel is introduced in place of the old one. The
used nuclear fuel is called spent fuel and it is composed of 950 kg U, 9 kg Pu, 75 g Np, 140 g
Am, 47 g Cm, and 31 kg ssion products (FPs) in 1 t of UO
2
after the output of 30 GWday/t
[92]. This composition is a typical one and dependent on both the operation history of the
reactor and the storage time of the fuel after taking out from the reactor. If Pu is separated
from the spent fuel and used as fuel again, high utilization of the fuel is attained. Chemical
processing for the separation of ssile material and removal of the FPs fromthe spent fuel is
called reprocessing. Therefore the main purposes of commercial reprocessing are (1) to
increase the available energy from ssile and fertile atoms and (2) to reduce hazards and
costs for handling the high-level wastes. Two other reasons are sometimes mentioned: (3) to
reduce the cost of thermal reactor fuel cycle and (4) to extract valuable by-products fromthe
high active waste. While several reprocessing methods have been proposed, the Purex
process is the most popular one and has been widely used all over the world. The Purex is an
abbreviation of plutonium uranium rening by extraction, which employs tributyl phos-
phate, TBP, with (C
4
H
9
)
3
PO
4
as an extractant. Since the TBP is rather viscous with a
density of 0.973 g/cm
3
, in an actual process, a 30-vol.% solution in dodecane (n-C
12
H
26
) or
kerosene as a diluent is employed. The mixture solution makes a quick separation of
the organic phase from aqueous phase and exhibits proper viscosity for extraction process.
The spent fuel is dissolved into highly concentrated nitric acid, and UO
2
2+
and Pu
4+
are the
chemical forms for Uand Pu, respectively. After the mixing of organic solvent and aqueous
phase containing dissolved spent fuel, both UO
2
2+
and Pu
4+
in aqueous phase are selec-
tively extracted from aqueous phase to the organic phase and the FPs remain in aqueous
phase. This step is called extraction. Next step is a separation of Pu from organic phase.
After reduction of Pu
4+
to Pu
3+
by chemical or electrochemical method, Pu
3+
goes back to
the aqueous phase selectively, while UO
2
2+
remained in organic phase, which is called
partition. These processes are precisely explained in textbooks [92,93].
Copyright 2004 by Taylor & Francis Group, LLC
The extractant should satisfy several requirements as follows: (1) high distribution
coecient for U and Pu; (2) low distribution coecient for FPs; (3) fast attainment to
equilibrium; (4) stability to water and acid; (5) radiation resistance; (6) low price; (7) low
ammability; (8) big density dierence fromwater; and (9) appropriate viscosity and surface
tension. The mixture of TBP in n-dodecane satises the above requirements and thus has
been widely employed as an extractant in the Purex process. Since the processes are operated
under strong radiation eld not only of g- and h-rays but also of a-rays, degradation of the
solvent takes place and the following detrimental eects appear: (1) loss of Uand Pu; (2) low
decontamination factor (DF) for U and Pu product; (3) lower extraction eciency such as
precipitation, long separation time, the third phase, and emulsion formation; (4) lower
eciency of the process; and (5) instability and danger of the process due to the criticality.
Therefore the understanding of the solvent degradation is inevitable to keep the integrity of
the plant and thus much work has been conducted.
In the eld of the reprocessing, radiation energy released from radionuclides in the
solution is normally expressed in a unit of W hr/1. If all the energy is absorbed in the
medium, the absorbed dose can be calculated as:
1 W hr=1c3600 J=kg 3600 Gy: 14
2.2. Radiolysis of TBP and n-Dodecane
Radiolysis of TBP has been investigated and reported from the 1950s. The main products
are dibutyl phosphate (DBP), monobutyl phosphate (MBP), and gaseous H
2
. Therefore CO
bond scission occurs easier than CC bond scission. The G-value of DBP is higher than
G(H
2
). TBP dimer as a polymeric product is also observed. In addition, other minor
products such as butane, butene, C
1
C
3
hydrocarbons, phosphoric esters, and butanol are
also observed. Reported G-values of TBP radiolysis products are summarized in Table 2
[9498]. The values are scattered probably because of the dierence of the TBPpurity in each
experiment. From ESR measurement of irradiated phosphades alkyl (R), alkylphosphade
[(RO)(R
O)POO
], phosphoranyl [((RO)
2
POO
] radi-
cals have been observed [99101]. Pulse radiolysis study reported the observation of the
solvated electron in TBP liquid at lower temperatures: at 198 K, the peak position is 1580
nm. The position is shifted shorter wavelength with increasing temperature and the peak
position of 1280 nmwas observed at 223 K[102]. It was found fromthe scavenger eect that
the solvated electron reacts with TBP leading to alkyl radical and DBP:
e
sol
C
4
H
9
O
3
P O !
C
4
H
9
C
4
H
9
O
2
POO
15
Geminate recombination process and reactivity and spectroscopic character of the solvated
electron in a variety of phosphates were also investigated [103,104].
It is noted that the decomposition of TBP takes place not only through radiolysis, but
also through chemical degradation. In chemical degradation, pyrolysis, oxidative decom-
position, and hydrolysis play important roles. Above 150jC, the pyrolysis reaction takes
place and the CObond is broken to lead DBP and MBP formation. The POscission is also
occurring but minor. The oxidation reaction proceeds quickly in 5 M HNO
3
at 75jC,
leading to NO
x
. In the presence of uraniumsalt, this reaction is accelerated and UO
2
(NO
3
)
2
is formed, followed by the formation of butane, butene, N
2
, NO
2
, CO, CO
2
, NO, butyne,
and butanol. It is known that the accidents at Savannah River in 1953 and Oak Ridge in
1959 were triggered by this oxidation reaction. The hydrolysis reaction of TBP occurs in
alkaline, acid, and organic phases leading to DBP and MBP. The diluent will react with
Copyright 2004 by Taylor & Francis Group, LLC
HNO
3
leading to the formation of carboxyl acids, nitric acid ester, and nitro and nitroso
compounds. Rate parameters for chemical degradation under dierent conditions have been
investigated [106,107]. The reaction of n-dodecane with HNO
3
is quite slow.
Radiolysis of hydrocarbons has been investigated and there is a review [108]. n-
Dodecane is one of the saturated hydrocarbons, and its radiation degradation is rather
known as compared with that of TBP. Main degradation products are H
2
and dimers, as
summarized in Table 3 [109,110]. Again, the values are dependent on the reports. Recently,
systematic study on the radiolysis of n-parans has been reported [111113]. The formation
process of dimers and double bonds is assumed as follows.
RHoR
16
oRH H
2
17
R
1
R
2
! R
1
H R
2
H 18
! R
1
R
2
19
Here R(H) is corresponding to alkenes.
2.3. Radiolysis of HNO
3
Radiolysis of aqueous nitrate solution has been investigated not only from practical view-
points in nuclear technology, but also from scientic interests in order to understand the
radiolysis of water because nitrate ion is an ecient scavenger for hydrated electron [114]
and even for the precursor of the hydrated electron [115]. In practical process, the con-
centration range of nitrate covers frommillimolars to 10 Mand the radiation eect not only
in diluted nitrate solutions but also in concentrated ones should be considered.
Table 2 G-Values of TBP g-Radiolysis Products
Product Burr [94]
Wilkinson and
Williams [95]
Wagner
et al. [96]
Holland
et al. [97]
Burger and
McClanahan [98]
H
2
1.73 1.11 1.59 2.02
CH
4
0.072 0.05 0.07 0.032
C
2
0.18 0.24 0.13
C
3
and C
4
0.66 0.45 0.75
DBP 2.44 1.52 2.25 3.71 1.9
MBP 0.14 0.12 0.39 0.3
Polymer 2.47
Table 3 G-Values of n-Hexadecane g-Radiolysis Products
Product Holland et al. [97] Dewhurst [109]
Rappoport and
Ga uman [110]
H
2
6.71 4.9 4.9
CH
4
0.05 0.05 0.01
Dimer 1.4
Copyright 2004 by Taylor & Francis Group, LLC
In diluted solutions, the reduction of NO
3
with e
aq
and H generates NO
2
radicals
through the transient intermediates of NO
3
2
, HNO
3
, and H
2
NO
3
which are in equi-
librium with one another. The reactivity of H atoms towards NO
3
.
NO
3
e
aq
! NO
2
3
k 9:7 10
9
M
1
sec
1
114 20
NO
3
H ! HNO
3
k 1:0 10
7
M
1
sec
1
114 21
NO
2
3
ZHNO
3
ZH
2
NO
3
pK
1
4:8; pK
2
7:5 114 22
NO
2
3
H
2
O ! NO
2
2OH
k 1:0 10
3
M
1
sec
1
116;117 23
HNO
3
! NO
2
OH
k 2:0 10
5
M
1
sec
1
116;117 24
H
2
NO
3
! NO
2
H
2
O k 7:0 10
5
M
1
sec
1
116;117 25
The NO
2
radicals tend to recombine with each other and, as a result, N
2
O
4
appears. The
N
2
O
4
decays slowly into HNO
2
and HNO
3
through hydrolysis reaction.
NO
2
NO
2
ZN
2
O
4
k
forward
4:5 10
8
M
1
sec
1
118 26
k
backward
6:0 10
3
M
1
sec
1
118
N
2
O
4
H
2
O ! HNO
2
HNO
3
k 18M
1
sec
1
118 27
In the radiolysis of nitric acid, HNO
2
and H
2
O
2
are formed. They strongly aect the solvent
degradation, oxidation states of metal ions, and corrosion condition of the material.
However, HNO
2
and H
2
O
2
are not coexisting because the next reaction will take place.
HNO
2
H
2
O
2
! HNO
3
H
2
O 28
This reaction is composed of several steps as follows [119,120].
HNO
2
H
ZH
2
NO
2
29
H
2
NO
2
H
2
O
2
! HOONO H
2
O 30
HOONO ! NO
3
H
k 4:6 10
3
H
M
1
sec
1
31
In concentrated solutions, direct action of nitrate ion is also taken into consideration.
In addition, in concentrated nitric acid, nitric acid molecule (HNO
3
) coexists. Although it is
known that NO
3
plays an important role in concentrated nitric acid, precise formation
process was not known. Recently, it was claried that the formation of the NO
3
radical is
formed by two dierent processes: reaction of OHwith molecular HNO
3
and direct action of
radiation to HNO
3
and NO
3
3
oO NO
2
33
g
s2
NO
3
g
s2
NO
2
g
s2
O 34
HNO
3
oO HNO
2
35
g
s2
V HNO
3
g
s2
V HNO
2
g
s2
V O 36
Figure 7 Formation of NO
3
radical observed at 640 nm in 3 M HNO
3
and 6 M NaNO
3
solutions
in the presence and absence of 0.1 M ethanol. (From Ref. 121.)
Figure 8 Dependence of G(NO
3
) as a function of nitrate concentration in HNO
3
and NaNO
3
,
respectively: total (.) and fast formation process (E) yields in HNO
3
solution; fast formation
process yields in LiNO
3
(j) and NaNO
3
(5) solutions. (From Ref. 121.)
Copyright 2004 by Taylor & Francis Group, LLC
The O formed is in a singlet (
1
D) or triplet (
3
P) state. It is known that the O (
1
D) will react
with water to yield H
2
O
2
and this does not contribute to the formation of HNO
2
and O
2
.
Therefore O (
3
P) is needed to explain the formation of O
2
and HNO
2
.
O
1
D H
2
O ! H
2
O
2
37
O
3
P NO
3
! O
2
NO
2
38
The rate constant of the second reaction is estimated to be k[O(
3
P)+NO
3
] =2.210
8
M
1
sec
1
fromthe values of [O(
3
P)+O
2
] =4.010
9
M
1
sec
1
[125] and k[O(
3
P)+NO
3
]/k[O
(
3
P)+O
2
] = 5.610
2
M
1
sec
1
[126]. However, another possible direct path, NO
3
o
O
+NO
2
, does not contribute to the O
2
formation.
The yield of HNO
2
fromwater decomposition, G
W
(HNO
2
), can be calculated by using
the yields of water decomposition products and their material balance relation:
G
W
HNO
2
f
W
0:5g
W
e
aq
H 0:5g
W
OH g
W
H
2
O
2
f
W
g
W
H
2
GH
2
39
Here f
W
stands for the electron fraction of water in the solution. Thus the electron fraction of
the solute is f
S
=1f
W
. In addition, G(HNO
2
), G(H
2
), and G(O
2
) are G-values of HNO
2
, H
2
,
and O
2
formation, respectively, experimentally determined in the solution. Reported G(H
2
)
and G(O
2
) in nitric acid and sodium nitrate solutions are summarized in Fig. 9 [127].
The total yield of HNO
2
in nitric acid, G(HNO
2
), coming from the direct and indirect
eect, G
W
(HNO
2
) and G
S
(HNO
2
), can be expressed as:
GHNO
2
G
W
HNO
2
G
S
HNO
2
GH
2
G
S
HNO
2
40
The values of G(HNO
2
)+G(H
2
) and G(O
2
) could be expressed by using a, the dissociation
constant of molecular nitric acid [128].
GHNO
2
GH
2
2f
S
ag
s2
NO
3
1 ag V
s2
NO
3
41
GO
2
f
S
ag
s2
NO
3
1 ag V
s2
NO
3
42
Figure 9 Reported G(H
2
) and G(O
2
) as a function of nitrate concentration in g-radiolysis of HNO
3
and NaNO
3
solutions. (From Ref. 127.)
Copyright 2004 by Taylor & Francis Group, LLC
From the experimentally determined values of G(HNO
2
), G(H
2
), and G(O
2
), the values of
g
s2
(NO
3
)=1.7 and g V
s2
(NO
3
, and H
2
O
2
have been reviewed [130].
The H
2
evolution is an important safety assessment in the commercial process and
much work has been conducted [131]. The G(H
2
) is decreasing with increasing of the
concentration of nitric acid. Coexisting ions aect the H
2
evolution slightly. Ions such as
Rh
3+
, Sm
3+
, Pr
3+
, Ce
3+
, Cd
2+
, Ce
4+
, and Eu
3+
increase the H
2
evolution in this order
and, contrarily, the presence of Fe
3+
and Cr
3+
reduces the evolution. At higher temper-
ature, the G(H
2
) is higher. It was also found that H
2
evolution is strongly dependent on the
depth of the liquid of nitric acid solution: deeper liquid gives smaller evolution than a
shallower one [131,132]. The eect was observed in both g- and a-radiolysis, and the eect is
more signicant in g-radiolysis [133]. In Fig. 10 [133136], the reported G(H
2
) and G(O
2
) by
a-radiolysis in nitric acid are summarized, and it is clear that G(H
2
) in a-radiolysis is higher
than that in g-radiolysis in diluted solution, but the dierence is not signicant in
concentrated solutions.
2.4. Radiolysis of the Mixture of TBP, n-Hexadecane, and HNO
3
Since, during the actual process, TBP, n-dodecane, and HNO
3
are used in a mixture or
contact condition, the degradation is aected by many factors: composition, concentration
of nitric acid, concomitant metal ions, and irradiation conditions such as aerated or
deaerated, stirring or settling, and high or low dose rate. As a result, a variety of products
are formed as summarized in Fig. 11 [137]. Under aerated condition, organic radicals react
with O
2
quickly to form peroxy-radical and after successive reactions bring variety of
alcohols, ketones, peroxides, and carbonyl compounds. These products are roughly
classied into nitration and oxidation products. The ratio of the nitration products to
oxidation product, g=(nitration/oxidation), is 0.8 under sucient oxygen supply but is
increasing to 8 under insucient oxygen. Recent advancement of the analytical technique
brings more precise data [138].
2.5. Radiation Chemistry of Actinide Ions
Because of the multivalent nature of the actinide ions, understanding the radiation-induced
change of the valence-state of the actinide in solutions under self-irradiation or external
irradiation is a challenge in radiation chemistry. Some of the ions are strong a-emitters. It is
also important from a practical viewpoint that the solution chemistry of actinide ions is
closely related to the storage and the repository of the wastes. Much work combined with
experiment and simulation has been conducted and reviews were summarized [136,140
144].
2.6. Storage of Radioactive Waste
In 50 years of production of electrical power and weapons fromthe nuclear fuel, the United
States has accumulated millions of cubic meters and tens of billions curies of radioactive
Copyright 2004 by Taylor & Francis Group, LLC
Figure 10 Reported (a) G(H
2
) and (b) G(O
2
) as a function of nitrate concentration in a-radiolysis
of HNO
3
solution. In (a), from Ref. 134 (E) and Ref. 133 (5: 9.3 g/L, w : 13 g/L, D: 18.5 g/L, .: 30 g/
L, n: 37 g/L, o: 100 g/L, and x: 160 g/L). In (b), from Ref. 134 (o), Ref. 135 (E), and Ref. 133 (D:
210 g/L, w : 160 g/L, and .: 100 g/L). (From Ref. 133.)
Copyright 2004 by Taylor & Francis Group, LLC
wastes [145]. It is widely recognized that safe treatment and storage of these wastes is an
urgent problem to be solved. At a radioactive waste storage tank, the so-called SY-101 in
Hanford, an intermittent gas evolution was found in 1980s. The released gas, composed of
H
2
, N
2
O, and minor NH
3
, in addition to air was emitted every 813 weeks. The content of
the tank is a mixture of chirate agents and their fragments, carboxylic acids, and alkanes in
addition to the sodiumnitrate and nitrite salts, which were produced during the processes to
separate plutonium from uranium and the FPs in spent metallic uranium fuel. This gas
evolution was considered to be a very important issue to keep safe storage of the waste, and
much radiolysis study was conducted to understand the mechanism of the gas evolution.
Through the work, subjects on (1) direct eect in concentrated solutions, (2) redox reactions
of NO
x
compounds, (3) radiation degradation of chilators, and (4) reactions in grout,
precipitation, and slurry have been pointed out as key issues for understanding. Details have
been reported elsewhere [146148].
3. RADIATION EFFECTS RELEVANT TO RADIOACTIVE HIGH-LEVEL
WASTE REPOSITORY
3.1. Once Through, Reprocessing, and High-Level Waste Repository
After the operation of the nuclear power stations for a certain period, the spent nuclear
fuel is produced. The spent fuel is reprocessed, and both U and Pu are recovered. At the
Figure 11 Radiolysis products in Purex solvents (TBP, n-dodecane or kerosene, and HNO
3
).
(From Ref. 137.)
Copyright 2004 by Taylor & Francis Group, LLC
same time, radioactive high-level liquid waste (HLLW) composed of ssion products and
trace amount of actinide compounds is also formed. HLLW emits strong radiation in-
cluding g-rays, high-energy electrons, a-particles, and fast neurons. Most of the ssion
products have a half-life less than 1000 years, and activity is dropped 3 orders magnitude
after 1000 years but still the activity remains due to the
99
Tc,
93
Zr, and actinide nuclei with
long lifetimes of million years. Therefore isolation of HLW from the activities and envi-
ronment of human being is the most essential requirement in HLW repository. Many
possible methods have been discussed such as disposal into the outer space from the solar
system, into the deep sea, under the ice in polar area, and so on. Actually, a disposal to
deep underground, geological deposition, has been selected as a reasonable and practical
one. In fact, at rst, the high-level liquid wastes are solidied with silicate glass into a
metal container called canister and are packed into metal container called an overpack
made of thick copper, stainless steel, or carbon steel, which depends on the choice. The
assembly would be placed in the bentonite clay mantle, which is built in the tunnel with
the depth up to more than 1000 m from the surface. Solidication, overpack, and ben-
tonite clay act as the multiple barriers which will prevent the invasion of groundwater to
the overpack and reduce the migration of radioactive and hazardous nuclei to biosphere.
Another option is that the spent fuel is directly contained without reprocessing into an
overpack for disposal, which is called once through. The same geological disposal method
will be applied. The precise procedures of the depository have been summarized in
textbooks [92,93].
In order to assess the integrity of the system, we should know what kind of reactions
would take place when the groundwater invades and the overpack is corroded. Conse-
quently, the solidied waste or spent fuel itself will be in contact with groundwater. Since the
waste would still be seriously activated, radiolysis of groundwater will take place and change
the chemical condition, which might aect the dissolution of the solidied waste or UO
2
of
the spent fuel.
3.2. Carbonate Solution
It is known that (bi)carbonate ions are predominant anions in groundwater with a
concentration as high as 50 mM. Therefore it is necessary to know the radiolysis eects
on groundwater by ionizing radiation from the HLW. A computer set of reaction steps and
reliable kinetic data for the radiolysis of (bi)carbonate solutions at ionic strength close to
groundwater is essential for the overall safety evaluation of geological disposal. The re-
actions of water decomposition products toward (bi)carbonate ions have been investigated
and rate constants have been accumulated. However, a little experimental evaluation of the
formation yield for the products such as formate, oxalate, H
2
O
2
, and so on has been
reported [149152]. Two sets of the reaction mechanismin carbonate solution were reported
[152,153]. In Table 4, recent list is summarized [152].
Here it should be mentioned that the chemical structure of carbonate radical is still
under debate. The reaction of the carbonate radical has been intensively investigated by
pulse radiolysis [154156] and laser photolysis [157], and the pK
a
of HCO
3
radical was
determined to be 7.6 [158] and 9.6 [157,159,160] from the change of the reactivity as a
function of pH. However, recent studies of Raman spectroscopy [160] and pulse radiolysis
[161] of carbonate solution proposed that the carbonate radical is a strong acid and keeps a
chemical form of CO
3
, and CO
3
2
(1) H
+
+CO
3
2
=HCO
3
b 5E10
a
(2) CO
2
+H
2
O=H
+
+HCO
3
b 2E4
(3) H
+
+HCO
3
=CO
2
+H
2
O b 5E10
(4) HCO
3
=H
+
+CO
3
2
b 2
(5) CO
2
+e
aq
=CO
2
b 7.7E9
(6) HCO
3
OH CO
3
+H
2
O b 8.5E6
(7) CO
3
2
+
OH CO
3
+OH
b 3.9E8
(8) HCO
3
+H H
2
CO
3
b 4.4E4
(9) CO
3
2
+e
aq
=CO
2
+2OH
H
2
O b 3.9E5
Reactions of CO
3
(10) CO
3
+CO
3
=C
2
O
6
2
b 1.4E7
(11) CO
3
+H
2
O
2
=CO
3
2
+O
2
+2H
+
b 9.8E5
(12) CO
3
+HO
2
=CO
3
2
+O
2
+H
+
b 1E7
(13) CO
3
O
2
=CO
3
2
+O
2
b 4E8
(14) CO
3
CO
=CO
3
2
+CO
2
b (3F1)E8
Reactions of CO
2
(15) CO
2
+e
aq
=HCOO
+OH
H
2
O b 1E9
(16) CO
2
CO
2
=C
2
O
4
2
b 6.5E8
(17) CO
2
+O
2
=CO
2
+O
2
b 2E9
(18) CO
2
+H
2
O
2
=CO
2
+OH
OH b 7.3E5
(19) CO
2
+HCO
3
=HCOO
+CO
3
b (1.0F0.1)E3
Reactions of C
2
O
6
2
and C
2
O
4
2
(20) C
2
O
6
2
=C
2
O
4
2
+O
2
b 1
b
(21) C
2
O
6
2
=HO
2
+OH
+2CO
2
H
2
O b 200F50
b
(22) CO
3
+C
2
O
4
2
=C
2
O
4
+CO
3
2
b (3F1)E3
(23) C
2
O
4
2
+e
aq
=C
2
O
3
4
b 3.1E7
(24) C
2
O
4
2
+
OH C
2
O
4
+OH
b 7.7E6
Reactions of HCOO
(25) CO
3
+HCOO
=HCO
3
+CO
2
b (1.5F0.5)E5
(26) HCOO
OH H
2
O CO
2
b 3.2E9
(27) HCOO
+H=H
2
+CO
2
b 2.1E8
(28) HCOO
+e
aq
=H
2
+CO
2
H
+
b 8E3
a
Read as 5 10
10
.
b
There are relative rate constants. They mean 0.5% of C
2
O
6
2
decomposes via
reaction (20) and 99.5% of C
2
O
6
2
via reaction (21).
Source: Ref. 152.
Copyright 2004 by Taylor & Francis Group, LLC
among oxychlorides and redox reactions induce conversion from one species to another,
which complicates. Recently, it is recognized that oxychlorides play important roles in
atmospheric and environmental chemistry. Oxychlorides are widely used to breach and to
sterilize water. Some oxychlorides exhibit chemical oscillation which is discussed from
viewpoints of chaos and complex system. Much knowledge has been accumulated. While the
reliable reaction sets of oxychloride in aqueous solution are not fully established, a typical
example of the literature survey is shown in Table 5 [162185].
Table 5 Rate Constants of Radiation-Induced Reactions in Aqueous Oxychloride Solution
Reactions Rate constants (M
1
sec
1
) Reference
(1) Cl
+OH!ClOH
4.3E9
a
[162]
(2) ClOH
!OH+Cl
6.1E9 sec
1
[162]
(3) ClOH
+H
+
!
Cl+H
2
O 2.1E10 [162]
(4)
Cl+H
2
O!ClOH
+H
+
1.6E5 sec
1
[163]
(5)
Cl+OH
!ClOH
1.8E10 [163]
(6)
Cl+OH!HClO 1E9 Assumed
(7)
Cl+Cl!Cl
2
8.8E7 [164]
(8)
Cl+Cl
!Cl
2
2.1E10 [162]
(9) Cl
2
Cl+Cl
1.1E5 sec
1
[162]
(10)
Cl+Cl
ZCl
2
K=1.9E5, 4.7E3 M
1
[162,165]
(11) ClOH
*
!
Cl+OH
23 sec
1
[163]
(12) ClOH
+Cl
!Cl
2
+OH
9E4 [166]
(13) Cl
2
+OH
!ClOH
+Cl
4.5E7 [167]
(14) Cl
2
H!2Cl
+H
+
8E9, (7E9) [168,169]
(15) Cl
2
OH!HClO+Cl
1E9 [170]
(16) Cl
2
+H
2
O!
OH+H
+
+2Cl
!Cl
2
+H
2
O 2.11E11 M
3
sec
1
[170]
(18) Cl
2
+H
2
O
2
!HO
2
+2Cl
+H
+
1.4E5 [167]
(19) Cl
2
+HO
2
!2Cl
+H
+
+O
2
1.0E9 [169]
(20) Cl
2
+Cl
2
!Cl
3
+Cl
Cl!Cl
3
6.3E8 [164]
(22) Cl
2
+HO
2
!Cl
2
+H
+
+O
2
1.0E9 [171]
(23) Cl
2
+Cl
!Cl
3
1.8E5 [172]
(24) Cl
2
+H
2
O!HClO+H
+
+Cl
11 sec
1
[173]
(25) HClO+H
+
+Cl
!Cl
2
+H
2
O 1.8E4 M
2
sec
1
[173]
(26) Cl
2
+OH
!HClO+Cl
3.8E11 [153]
(27) Cl
2
+2ClO
2
!2ClO
2
+2Cl
2.1E1 [Cl
2
(aq)] [ClO
2
] [174]
(28) HClO+2ClO
2
+H
+
!2ClO
2
+Cl
+H
2
O 1.06E6 [HOCl] [ClO
2
] [H
+
] [174]
(29) 2HClO+ClO
2
!ClO
3
+Cl
2
+H
2
O 2.1 E3 M
2
sec
1
[175]
(30) HClO+
Cl!Cl
+ClO
+H
+
3E9 [163]
(31) HClO+H
2
O
2
!H
2
O+O
2
+H
+
+Cl
3.4E3 Assumed
(32) Cl
2
+H
2
O!Cl
2
OH
+H
+
6.12E2 sec
1
[173]
(33) Cl
2
OH
+H
+
Cl
2
+H
2
O 2E10 [173]
(34) Cl
2
OH
+H
+
!Cl
2
+OH
+H
+
2E5 [173]
(35) Cl
2
+OH
+H
+
!Cl
2
OH
+H
+
1E10 [173]
(36) Cl
3
!Cl
2
+Cl
1.6E6 [172]
(37) Cl
3
H!Cl
+Cl
2
+H
+
3E10 [176]
(38) Cl
3
+HO
2
!Cl
2
+Cl
+H
+
+O
2
1.0E9 [171]
(39) HClO+e
aq
!Cl
OH 6.5E8 [177]
(40) ClO
+e
aq
!Cl
+O
OH!Cl
+H
2
O 1.4E8 [177]
Copyright 2004 by Taylor & Francis Group, LLC
In repository of HLW in salt mine, the understanding of radiolysis of concentrated
NaCl up to 5 M is important. In addition, a-radiolysis by actinide ions dissolved is also
taken into account. There are several reports mentioned above [186192].
3.4. Radiation-Induced Dissolution of UO
2
Fuel
In direct geological depository of the spent fuel, dissolution of UO
2
matrix into groundwater
plays a key role for the release of the radionuclides to the biosphere. Therefore mechanismof
Reactions Rate constants (M
1
sec
1
) Reference
(43) ClO
OH!ClO
+OH
+O
(+H
2
O)!ClO
+2OH
2.4E8 [168]
(45) ClO
Cl!ClO+Cl
8.2E9 [163]
(46) ClO
+O
2
!O
+O
2
+Cl
7.7E6 [179]
(47) ClO
+H
2
O
2
!H
2
O+Cl
+O
2
3.4E3 [180]
(48) ClO
+HO
2
!OH
+Cl
+O
2
4.4E7 [181]
(49) HClO!ClO
+H
+
3.0E2 pKa=7.53
(50) ClO
+H
+
!HClO 1E10 [181]
(51) HClO+ClO
2
+H
+
!Cl
2
O
2
+H
2
O 1.12E6 M
2
sec
1
[175]
(52) Cl
2
O
2
+ClO
2
!2ClO
2
+Cl
5.4E4(
f
1E5) [175]
(53) 2Cl
2
O
2
!2ClO
2
+Cl
2
6E10 M
1
sec
1
[175]
(54) ClO
2
+Cl
2
!Cl
2
O
2
+Cl
4.010
4
[175]
(55) Cl
2
O
2
+H
2
O!ClO
3
+Cl
+2H
+
1, 10 sec
1
[175]
(56) ClO
+ClO
+H
2
O!ClO
+ClO
2
+2H
+
2.5E9 [163]
(57) ClO
+ClO
2
!ClO
+ClO
2
9.4E8 [182]
(58) ClO
2
+e
aq
!ClO
+O
OH!ClO
2
+OH
+O
(+H
2
O)!ClO
2
+2OH
!HClO
2
1.0E10 pKa=2.5
(62) HClO
2
!H
+
+ClO
2
3.16E7 sec
1
[174]
(63) ClO
2
+HO
2
!ClO
2
+HO
2
8E4, 1.3E5 [183,184]
(64) ClO
2
+O
2
!ClO
2
+O
2
3.3E9 [183]
(65) ClO
2
+ClO
!Cl
2
O3 1E9 Assumed
(66) Cl
2
O3+H
2
O!ClO
+ClO
3
+2H
+
1E4 Assumed
(67) ClO
2
+
OH!HClO+O
2
1.4E9 [185]
(69) ClO
2
+
OH!ClO
3
+H
+
2.6E9 [185]
(70) ClO
2
+O
!ClO
+O
2
4.9E8 [185]
(72) ClO
2
+O
!ClO
3
2.2E9 [185]
(73) ClO
2
+e
aq
e
aq
!ClO
2
qkq **
(75) ClO
2
+
e
aq
!ClO
(1q)kq **
(76) ClO
2
+
H!overall ks=110
9
Assumed
(77) ClO
2
+
H!H
+
+ClO
2
sks **
(78) ClO
2
+
H!ClO
OH (1s)ks **
*The dissociation rate is calculated after the assumption of the recombination rate constant of H
+
with anion to
be 10
10
M
1
sec
1
when the pK
a
value is known.
**Branching reaction. q (s) and 1q (1s) are the branching fraction.
a
Read as 4.3 10
9
.
Table 5 Continued
Copyright 2004 by Taylor & Francis Group, LLC
the dissolution processes has been claried by using irradiated and unirradiated UO
2
under
various conditions. Inuence of many factors such as radiolysis, pH, temperature, ground-
water composition, and formation of corrosion deposits has been investigated. An excellent
review has been published elsewhere [193]. Among many factors, it was revealed that
radiolysis accelerates the dissolution of UO
2
.
The accelerated dissolution of UO
2
and U
2
O
8
in acid or carbonate solutions under the
irradiation of g-rays was found by Gromov in 1981 [194]. Since then, much experimental
results have been accumulated as summarized in a review of Cristensen and Sunder [195].
The radiolysis products such as OH and H
2
O
2
react with UO
2
, and its surface is oxidized to
form higher oxidizing phase: UO
2+x
, UO
2.33
, UO
2
2+
, and so on. The measurement of the
corrosion potential of the specimen was also done and showed the correlated behavior of the
oxidizing process. There is a time-dependent generation of oxidants and reductants at fuel/
water interface due to a-, h-, and g-radiolysis. Based on the accumulated experimental data,
a calculation code to predict the oxidation and dissolution process was proposed and the
calculated results are in good agreement with the experimental ones [195].
Nuclear spent fuel UO
x
emits not only gamma and beta, but also alpha particles. The
alpha activity lasts over a million years, much longer than the gamma and beta activities
which decay within a few hundred years. In long-term assessment, a-radiolysis should be
also taken into consideration. Recently, a model experiment using He
2+
ion beam from an
accelerator has been investigated and a signicant acceleration of dissolution under beam
irradiation was found, and the formation of uranium peroxide hydrate UO
4
2H
2
O (meta-
studtite) was detected from x-ray diraction at UO
2
surface [196,197].
4. CONCLUSION
It is clear that the development of nuclear technology is impossible without the support of
radiation chemistry. Thus the radiation chemistry is important not only in basic science, but
also in technology.
ACKNOWLEDGMENT
This is supported by the Japan Society for the Promotion of Science under the contract
JSPS-RFTE 98P00901. This is also partly supported by the innovative basic research
program in the eld of high-temperature engineering using HTTR conducted by Japan
Atomic Energy Research Institute.
REFERENCES
1. Jenks, G.H. ORNL-3848, UC-80-Reactor Technology, TID-4500, 1965.
2. Jenks, G.H.; Griess, J.C. ORNL-4173, UC-80-Reactor Technology, 1967.
3. Burns, W.G.; Moore, P.B. Radiat. Eects 1976, 30, 233.
4. Elliot, A.J. AECL-11073, COG-94-167, 1994.
5. Elliot, A.J.; Ouellette, D.C.; Stuart, C.R. AECL 11658, COG-96-390-1, 1996.
6. McCracken, D.R.; Tsang, K.T.; Laughton, P.J. AECL-11895, 1998.
7. Buxton, G.V. Radiation chemistry: present status and future trends. In: Jonah, C.D. Rao, B.S.,
eds. Studies in Physics and Theoretical Chemistry 87. Elsevier Science, 2001, 195 pp.
Copyright 2004 by Taylor & Francis Group, LLC
8. Ruiz, C.P.; Lin, C.C.; Robinson, R.; Burns, W.G.; Curtis, A.R. Water Chem. Nucl. React. Syst.
1989, 4, 131. BNES, London.
9. Ruiz, C.P.; Lin, C.C.; Robinson, R.N.; Burns, W.G.; Henshaw, J.; Pathania, R. Water Chem.
Nucl. React. Syst. 1992, 6, 141. BNES, London.
10. Kabakchi, S.A.; Lebedeva, I.E. High Energy Chem. 1984, 18, 166.
11. Katsumura, Y.; Takeuchi, Y.; Ishigure, K. Radiat. Phys. Chem. 1988, 32, 259.
12. Katsumura, Y.; Takeuchi, Y.; Hiroishi, D.; Ishigure, K. Radiat. Phys. Chem. 1989, 33, 299.
13. Elliot, A.J.; Ouellette, D.C.; Reid, D.; McCracken, D.R. Radiat. Phys. Chem. 1989, 34, 747.
14. Katsumura, Y.; Yamamoto, S.; Hiroishi, D.; Ishigure, K. Radiat. Phys. Chem. 1992, 39, 383.
15. Burns, W.G.; Marsh, W.R. J. Chem. Soc., Faraday Trans. 1 1981, 77, 197.
16. Shiraishi, H.; Buxton, G.V.; Wood, N.D. Radiat. Phys. Chem. 1989, 33, 519.
17. Kent, M.C.; Sims, H.E. AEA-RS-2302, PWR/CTG/(92) 084, 1992.
18. Kent, M.C.; Sims, H.E. Proceedings of the 6th International Conference of Water Chemistry of
Nuclear Reactor Systems; British Nuclear Energy Society: London, 1992, 153 pp.
19. Elliot, A.J.; Chenier, M.P.; Ouellette, D.C. J. Chem. Soc., Faraday Trans. 1993, 89, 1193.
20. Sunaryo, G.R.; Katsumura, Y.; Hiroishi, D.; Ishigure, K. Radiat. Phys. Chem. 1995, 45, 131.
21. Ishigure, K.; Katsumura, Y.; Sunaryo, G.R.; Hiroishi, D. Radiat. Phys. Chem. 1995, 46, 557.
22. Elliot, A.J.; Chenier, M.P.; Ouellette, D.C.; Koslowsky, V.T. J. Phys. Chem. 1996, 100, 9014.
23. Katsumura, Y.; Sunaryo, G.R.; Hiroishi, D.; Ishigure, K. Prog. Nucl. Energy 1998, 32, 113.
24. Stefanic, I.; LaVerne, J.A. J. Phys. Chem. A 2002, 106, 447.
25. Christensen, H.; Sehested, K. Radiat. Phys. Chem. 1980, 16, 183.
26. McCracken, D.R.; Buxton, G.V. Nature 1981, 292, 439.
27. Christensen, H.; Sehested, K. J. Phys. Chem. 1988, 92, 3007.
28. Shiraishi, H.; Katsumura, Y.; Hiroishi, D.; Ishigure, K.; Washio, M. J. Phys. Chem. 1988, 92,
3011.
29. Buxton, G.V.; Wood, N.D.; Dyster, S. J. Chem. Soc., Faraday Trans. 1, 1988, 84, 1113.
30. Elliot, A.J.; McCracken, D.R. Radiat. Phys. Chem. 1989, 33, 69.
31. Buxton, G.V.; Wood, N.D. Radiat. Phys. Chem. 1989, 34, 699.
32. Elliot, A.J. Radiat. Phys. Chem. 1989, 34, 753.
33. Elliot, A.J.; McCracken, D.R.; Buxton, G.V.; Wood, N.D. J. Chem. Soc., Faraday Trans. 1990,
86, 1539.
34. Elliot, A.J.; Buxton, G.V. J. Chem. Soc., Faraday Trans. 1992, 88, 2465.
35. Buxton, G.V.; Mackenzie, S.R. J. Chem. Soc., Faraday Trans. 1992, 88, 2833.
36. Hickel, B.; Sehested, K. Radiat. Phys. Chem. 1992, 39, 355.
37. Buxton, G.V.; Elliot, A.J. J. Chem. Soc., Faraday Trans. 1993, 89, 485.
38. Christensen, H.; Sehested, K.; Lgager, T. Radiat. Phys. Chem. 1993, 41, 575.
39. Christensen, H.; Sehested, K.; Lgager, T. Radiat. Phys. Chem. 1994, 43, 527.
40. Elliot, A.J.; Ouellette, D.C. J. Chem. Soc., Faraday Trans. 1994, 90, 837.
41. Shiraishi, H.; Sunaryo, G.R.; Ishigure, K. J. Phys. Chem. 1994, 98, 5164.
42. Ashton, L.; Buxton, G.V.; Stuart, C.R. J. Chem. Soc., Faraday Trans. 1995, 91, 1631.
43. Noyes, R.M. In: Progress in Reaction Kinetics; Porter, G., Ed.; Pergamon: London, 1961: Vol.
1, 129 pp.
44. Newton, M.D.; Sutin, N. Annu. Rev. Phys. Chem. 1984, 35,437.
45. Draganic, I.; Draganic, Z.D. Radiation Chemistry; Academic Press: New York, 1971, 170 pp.
46. LaVerne, J.A.; Pimblott, S.M. J. Phys. Chem. 1993, 97, 3291.
47. Swiatla-Wojcik, C.; Buxton, G.V. J. Phys. Chem. 1995, 99, 11464.
48. Swiatla-Wojcik, C.; Buxton, G.V. J. Chem. Soc., Faraday Trans. 1998, 94, 2135.
49. Swiatla-Wojcik, C.; Buxton, G.V. Chem. Phys. Phys. Chem. 2000, 2, 5113.
50. Pimblott, S.M.; LaVerne, J.A. J. Phys. Chem. A 1997, 101, 5828.
51. Frongillo, Y.; Goulet, T.; Fraser, M.-J.; Cobut, V.; Patau, J.P.; Jay-Gerin, J.P. Radiat. Phys.
Chem. 1998, 51, 245.
52. Goulet, T.; Fraser, M.-J.; Frongillo, Y.; Jay-Gerin, J.-P. Radiat. Phys. Chem. 1998, 51, 85.
53. Cliord, P.; Green, N.J.B.; Pilling, M.J. J. Phys. Chem. 1982, 86, 13181322.
Copyright 2004 by Taylor & Francis Group, LLC
54. Cliord, P.; Green, N.J.B.; Oldeld, M.J.; Pilling, M.J.; Pimblott, S.M. J. Chem. Soc., Faraday
Trans. 1 1986, 82, 2673.
55. Cliord, P.; Green, N.J.B.; Pilling, M.J.; Pimblott, S.M. J. Phys. Chem. 1987, 91, 4417.
56. Nikjoo, H.; Uehara, S.; Wilson, W.E.; Hoshi, M.; Goodhead, D.T. Int. J. Radiat. Biol. 1998, 73,
355.
57. Herve du Penhoat, M.-A.; Goulet, T.; Frongillo, Y.; Fraser, M.-J.; Bernat, Ph.; Jay-Gerin, J.-P.
J. Phys. Chem. A 2000, 104, 11757.
58. Herve du Penhoat, M.-A.; Meesungnoen, J.; Goulet, T.; Filali-Mouhim, A.; Mankhetkorn, S.;
Jay-Gerin, J.-P. Chem. Phys. Lett. 2001, 341, 135.
59. Meesungnoen, J.; Jay-Gerin, J.-P.; Filali-Mouhim, A.; Mankhetkorn, S. Chem. Phys. Lett.
2001, 335, 458.
60. Meesungnoen, J.; Jay-Gerin, J.-P.; Filali-Mouhim, A.; Mankhetkorn, S. Can. J. Chem. 2002,
80, 68.
61. Meesungnoen, J.; Jay-Gerin, J.-P.; Filali-Mouhim, A.; Mankhetkorn, S. Can. J. Chem. 2002,
80, 767.
62. Savage, P.E. Chem. Rev. 1999, 99, 603.
63. Tester, J.W.; Cline, J.A. Corrosion 1999, 55, 1088.
64. Schmieder, H.; Abeln, J. Chem. Eng. Technol. 1999, 22, 903.
65. Bro ll, D.; Kaul, C.; Kra mer, A.; Krammer, P.; Richter, T.; Jung, M.; Vogel, H.; Zehner, P.
Angew. Chem., Int. Ed. 1999, 38, 2998.
66. Kajimoto, O. Chem. Rev. 1999, 99, 55.
67. Oka, Y. Physics of supercritical-pressure light water cooled reactors. Proc. 1998 Frederic Joliot
Summer School in Reactor Physics, Caderache, France; and references cited herein, 1998; 240
259 pp.
68. Ferry, J.L.; Fox, M.A. J. Phys. Chem. A 1998, 102, 3705.
69. Ferry, J.L.; Fox, M.A. J. Phys. Chem. A 1999, 103, 3438.
70. Wu, G.; Katsumura, Y.; Muroya, Y.; Li, X.; Terada, Y. Chem. Phys. Lett. 2000, 325, 531.
71. Takahashi, K.; Cline, J.A.; Bartels, D.M.; Jonah, C.D. Rev. Sci. Instrum. 2000, 71, 3345.
72. Michael, B.D.; Hart, E.J.; Schmidt, K.H. J. Phys. Chem. 1971, 75, 2798.
73. Dixson, R.S.; Lopata, V.J. Radiat. Phys. Chem. 1978, 11, 135.
74. Christensen, H.; Sehested, K. J. Phys. Chem. 1986, 90, 186.
75. Wu, G.; Katsumura, Y.; Muroya, Y.; Li, X.; Terada, Y. Radiat. Phys. Chem. 2001, 60, 395.
76. Cline, J.; Jonah, C.D.; Bartels, D.M.; Takahashi, K. Proc. of the 1st International Symposium
on Supercritical Water-cooled Reactors, Design and Technology, Nov. 69, 2000, Univ. of
Tokyo; Tokyo; Japan; 2000; 194 pp.
77. Levelt Sengers, J.M.H.; Straub, J.; Watanabe, K.; Hill, P.G. J. Phys. Chem. Ref. Data. 1985, 14,
193.
78. Wu, G.; Katsumura, Y.; Muroya, Y.; Lin, M.; Morioka, T. J. Phys. Chem. A 2002, 106, 2430.
79. Wu, G.; Katsumura, Y.; Muroya, Y.; Lin, M.; Morioka, T. J. Phys. Chem. A. 2001, 105, 4933.
80. Mostafavi, M.; Lin, M.; Wu, G.; Katsumura, Y.; Muroya, Y. J. Phys. Chem. A 2002, 106,
3123.
81. Wu, G.; Katsumura, Y.; Lin, M.; Morioka, T.; Muroya, Y. PCCP 2002, 4, 3980.
82. Schuler, R.H.; Patterson, L.K.; Janata, E. J. Phys. Chem. 1980, 84, 2089.
83. Buxton, G.V.; Stuart, C.R. J. Chem. Soc., Faraday Trans. 1995, 91, 279.
84. Hayon, E.; Ibata, T.; Lichtin, N.N.; Simic, M. J. Phys. Chem. 1972, 76, 2072.
85. Adams, G.E.; Wilson, R.L. J. Chem. Soc., Faraday Trans. 1 1973, 69, 719.
86. Brede, O.; Helmstreit, W.; Mehnert, R. Z. Phys. Chem. (Leipzig) 1975, 256, 513.
87. Lin, M.; Katsumura, Y.; Wu, G.; Muroya, Y.; He, H.; Kudo, H. Proc. of the 45th Annual
Meeting on Radiation Chemistry in Japan, Oct. 911, 2002, Kyushu University; Fukuoka;
Japan; 2002; 193 pp.
88. Takahashi, K.; Bartels, D.M.; Cline, J.A.; Jonah, C.D. Chem. Phys. Lett. 2002, 357, 358.
89. Ho, P.C.; Palmer, D.A.; Wood, R.H. J. Phys. Chem. B 2000, 104, 12084.
90. Ho, P.C.; Palmer, D.A.; Gruszkiewicz, M.S. J. Phys. Chem. B 2001, 105, 1260.
Copyright 2004 by Taylor & Francis Group, LLC
91. Chialvo, A.A.; Ho, P.C.; Palmer, D.A.; Gruszkiewicz, M.S.; Cummings, P.T.; Simonson, J.M.
J. Phys. Chem. B 2002, 106, 2041.
92. Benndict, M.; Pigford, T.H.; Levi, H.W. Nuclear Chemical Engineering. 2nd Ed. McGraw-Hill:
New York, 1981.
93. Choppin, G.; Liljenzin, J.O.; Rydberg, J. Radiochemistry and Nuclear Chemistry. 2nd Ed.
Butterworth-Heinemann: Oxford, 1995.
94. Burr, J.G. Radiat. Res. 1958, 8, 214.
95. Willkinson, R.W.; Williams, T.F. J. Chem. Soc., 4098.
96. Wagner, R.M.; Kinderman, E.M.; Towle, L.H. Ind. Eng. Chem. 1959, 51, 45.
97. Holland, J.P.; Merklin, J.F.; Razvi, J. Nucl. Instr. Methods 1978, 153, 589.
98. Burger, L.L.; McClanahan, E.D. Abstracts of ACS meeting at Miami, 1957.
99. Nelson, D.; Symons, M.C.R. J. Chem. Soc. Perkin II, 286.
100. Zaitsev, V.D.; Karasev, A.L.; Khaikin, G.I.; Egorov, G.F. High Energy Chem. 1988, 22, 351.
101. Zaitsev, V.D.; Khaikin, G.I. High Energy Chem. 1994, 28, 308.
102. Zaitsev, V.D.; Khaikin, G.I. High Energy Chem. 1989, 23, 79.
103. Zaitsev, V.D.; Protasova, E.L.; Khaikin, G.I. High Energy Chem. 1989, 24, 58.
104. Zaitse, V.D.; Protasova, E.L.; Khaikin, G.I. High Energy Chem. 1993, 27, 28.
105. Zaitsev, V.D.; Protasova, E.L.; Khaikin, G.I. High Energy Chem. 1994, 28, 30.
106. von Stieglitz, L.; Becker, R. Atomkernenerg. Kerntech. 1985, 46, 76.
107. Tallent, O.K.; Mailen, J.C.; Dodson, K.E. Nucl. Technol. 1985, 71, 417.
108. Fo ldiak, G. Radiation Chemistry of Hydrocarbons. Elsevier: Amsterdam, 1981.
109. Dewhurst, H.A. J. Phys. Chem. 1957, 61, 1466.
110. Rappoport, S.; Ga uman, T. Helv. Chim. Acta 1973, 56, 531. Helv. Chim. Acta 1973, 57, 2861.
111. Seguchi, T.; Hayakawa, N.; Tamura, N.; Hayashi, N.; Katsumura, Y.; Tabata, Y. Radiat. Phys.
Chem. 1988, 32, 753.
112. Soebianto, Y.S.; Yamaguchi, T.; Katsumura, Y.; Ishigure, K.; Kubo, J.; Koizumi, T. Radiat.
Phys. Chem. 1992, 39, 251.
113. Katsumura, Y. Angew. Makromol. Chem. 1997, 252, 89.
114. Buxton, G.; Greenstock, G.L.; Hermann, W.P.; Ross, A.B. J. Phys. Chem. Ref. Data 1988, 17,
513.
115. Wol, R.K.; Bronskill, M.J.; Hunt, J.W. J. Chem. Phys. 1970, 53, 4211.
116. Gra zel, M.; Henglein, A.; Lilie, J.; Beck, G. Ber. Bunsenges. Phys. Chem. 1970, 74, 292.
117. Lgager, T.; Sehested, K. J. Phys. Chem. 1993, 97, 6664.
118. Gra zel, M.; Henglein, A.; Lilie, J.; Beck, G. Ber. Bunsenges. Phys. Chem. 1969, 73, 646.
119. Bhattacharyya, P.K.; Veeraraghavan, R. Int. J. Chem. Kinet. 1977, 9, 629.
120. Damaschen, D.E.; Martin, L.R. Atmos. Environ. 1983, 17, 2005.
121. Katsumura, Y.; Jiang, P.Y.; Nagaishi, R.; Oishi, T.; Ishigure, K.; Yoshida, Y. J. Phys. Chem.
1991, 95, 4435.
122. Katsumura, Y. Radiation chemistry: present status and future trends. In: Jonah, C.D. Rao,
B.S.M.; Eds.; Studies in physics and theoretical chemistry 87; Elsevier Science: Amsterdam,
2001, p. 163.
123. Daniels, M. Radiation Chemistry I. Gould, F. Ed.; American Chemical Soc., Advances in
Chemical Series; American Chemical Soc., Washington; 1968, 81,153.
124. Pogge, H.B.; Jones, F.T. J. Phys. Chem. 1970, 74, 1700.
125. Kla ning, U.K.; Sehested, K.; Wol, T. J. Chem. Soc., Faraday Trans. 1 1984, 80, 2969.
126. Amichai, O.; Treinin, A. Chem. Phys. Lett. 1969, 3, 611.
127. Mahlman, M.A. J. Chem. Phys. 1961, 35, 936.
128. Redlich, O.; Duerst, R.D.; Merbach, A. J. Chem. Phys. 1968, 49, 2986.
129. Nagaishi, R.; Jiang, P.Y.; Katsumura, Y.; Ishigure, K. J. Chem. Soc., Faraday Trans 1994, 90,
591.
130. Katsumura, Y. In: Alasi, Z.B. Ed.; N-Centered Radicals; John Wiley & Sons, 1998, 393 pp.
131. Nakagiri, N.; Miyata, T. J. Atom. Energ. Soc. Jpn. 1994; 37, 1119; 1995; 1996; 38, 992 pp. in
Japanese.
Copyright 2004 by Taylor & Francis Group, LLC
132. Specht, S. In KfK 2940, 1980; 196 pp.
133. Kuno, Y.; Hina, T.; Masui, J. J. Nucl. Sci. Technol. 1993, 30, 919.
134. Savelev, Yu., I.; Savelev; Ershova, Z.V.; Vladimirova, M.V. Sov. Radiochem. 1967, 9, 221.
135. Becker, R.; Burkhardt, H.G.; Neeb, K.H.; Wu rtz, R. IAEA-SM-245/13, 1979.
136. Bibler, N.E. J. Phys. Chem. 1974, 78, 211.
137. Huggard, A.J.; Warner, B.F. Nucl. Sci. Eng. 1963, 17, 168.
138. Tripathi, S.C.; Bindu, P.; Ramanujam, A. Sep. Sci. Technol. 2001, 36, 1463; Tripathi, S.C.;
Ramanujam, A.; Gupta, K.K.; Bindu, P. Sep. Sci. Technol. 2001, 36, 2863.
139. Miner, F.J.; Seed, J.R. Chem. Rev. 1967, 67, 299.
140. Bibler, N.E. J. Phys. Chem. 1975, 79, 1991.
141. Pikaev, A.K.; Shilov, V.P.; Spitsyn, V.I. Radiolysis of Aqueous Solutions of Lanthanides and
Actinides; Nauka: Moscow, 1983. in Russian.
142. Gordon, S.; Sullivan, J.C.; Ross, A.B. J. Phys. Chem. Ref. Data 1986, 15, 1357.
143. Frolov, A.A.; Andreychuk, N.N.; Rotmanov, K.V.; Frolova, L.M.; Vasiliev, V.Ya. J.
Radioanal. Nucl. Chem. 1990, 143, 433.
144. Bhattacharyya, P.K.; Natarajanin, P.R. In Handbook on the Physical Chemistry of Actinides;
Freeman, A.J. Keller, C., Eds.; Elsevier Science Publishers B.V., Chapt. 13, 97 pp.
145. Ahearne, J.F. Phys. Today, 24 (June).
146. Tank Waste Science Panel, PNL-7595 DE91 009369, 1991.
147. Reynolds, D.A.; D.D. Siemer, D.M. Strachan, R.W. Wallace, PNL-7520 DE91 009680, 1991.
148. Jonah, C.D.; Kapoor, S.; Matheson, M.S.; Mulac, W.A.; Meisel, D. ANL-94/7, 1994, references
cited herein.
149. Draganic, Z.D.; Negron-Mendoza, A.; Navarro-Conzalez, R.; Vujosevic, S.I. Radiat. Phys.
Chem. 1987, 30, 229.
150. Eriksen, T.E.; Dalamba, P.N.; Christensen, H.; Bjergbakke, E. J. Radioanal. Nucl. Chem. 1989,
132, 19.
151. Draganic, Z.D.; Negron-Mendoza, A.; Sehested, K.; Vujosevic, S.I.; Navarro-Gonzalez, R.;
Albarran-Sanches, M.G.; Draganic, I.G. Radiat. Phys. Chem. 1991, 38, 317.
152. Cai, Z.; Li, X.; Katsumura, Y.; Urabe, O. Nucl. Technol. 2001, 136, 231.
153. Sunder, S.; Christensen, H. Nucl. Technol. 1993, 104, 403.
154. Weeks, J.L.; Rabani, J. J. Phys. Chem. 1966, 70, 2100.
155. Behar, D.; Czapski, G.; Duchovny, I. J. Phys. Chem. 1970, 73, 2206.
156. Chawla, O.P.; Fessenden, R.W. J. Phys. Chem. 1975, 79, 2693.
157. Chen, S.-N.; Cope, V.W.; Homan, M.Z. J. Chem. Soc. Chem. Commun., 1970; 991; J. Phys.
Chem. 1970, 77, 1111.
158. Eriksen, T.E.; Lind, J.; Mere` nyi, G. Radiat. Phys. Chem. 1985, 26, 197.
159. Zuo, Z.; Cai, Z.; Katsumura, Y.; Chitose, N.; Muroya, Y. Radiat. Phys. Chem. 1999, 55, 15.
160. Bisby, R.H.; Johnson, S.A.; Parker, A.W.; Tavender, S.M. J. Chem. Soc., Faraday Trans. 1998,
94, 2069.
161. Czapski, G.; Lymar, S.V.; Schwarz, H.A. J. Phys. Chem. A 1999, 103, 3447.
162. Jayson, G.G.; Parsons, B.J.; Swallow, A.J. J. Chem. Soc., Faraday Trans. 1 1973, 69, 1597.
163. Kla ning, U.K.; Wol, T. Ber. Bunsenges. Phys. Chem. 1985, 89, 243.
164. Wu, D.; Wong, D.; DiBartolo, B. J. Photochem. 1980, 14, 303.
165. Adams, D.J.; Barlow, S.; Buxton, G.V.; Malone, T.M.; Salmon, G.A. J. Chem. Soc., Faraday
Trans. 1995, 91, 3303.
166. Bjergbakke, E.; Sehested, K.; Rusmussen, L.O.; Christensen, H. Ris M-2430, 1984.
167. Hasegawa, K.; Neta, P. J. Phys. Chem. 1978, 82, 854.
168. Buxton, G.V.; Subhani, M.S. J. Chem. Soc., Faraday Trans. 1 1972, 68, 947.
169. Navaratnam, S.; Parsons, B.J.; Swallow, A.J. Radiat. Phys. Chem. 1980, 15, 159.
170. Wagner, I.; Kartha user, J.; Strehlow, H. Ber. Bunsenges. Phys. Chem. 1986, 90, 861.
171. Bjergbakke, E.; Navaratnam, S.; Parsons, B.J.; Swallow, A.J. J. Am. Chem. Soc. 1981, 103,
5926.
172. Scott, R.L. J. Am. Chem. Soc. 1953, 75, 1550.
Copyright 2004 by Taylor & Francis Group, LLC
173. Eigen, M.; Kustin, K. J. Am. Chem. Soc. 1962, 84, 1355.
174. Epstein, I.R.; Kustin, K.; Simonyi, R.H. J. Phys. Chem. 1992, 96, 5852.
175. Peintler, G.; Nagypa l, I.; Epstein, I.R. J. Phys. Chem. 1990, 94, 2954.
176. Gogolov, A.Y.; Makarov, I.E.; Pikaev, A.K. High Energy Chem. 1984, 18, 390.
177. Zuo, Z.; Katsumura, Y.; Ueda, K.; Ishigure, K. J. Chem. Soc., Faraday Trans. 1997, 93, 1885.
178. Amber, M.; Hart, E.J. Adv. Chem. Ser. 1968, 81, 79.
179. Long, C.A.; Bielski, B.H.J. J. Phys. Chem. 1980, 84, 555.
180. Held, A.M.; Halko, D.J.; Hurst, J.K. J. Am. Chem. Soc. 1978, 100, 5732.
181. Morris, J.C. J. Phys. Chem. 1966, 70, 3798.
182. Alfassi, Z.B.; Huie, R.E.; Mosseri, S.; Neta, P. Radiat. Phys. Chem. 1988, 32, 85.
183. Eriksen, T.E.; Lind, J.; Merenyi, G. J. Chem. Soc., Faraday Trans. 1 1981, 77, 2115.
184. Hoigne , J.; Bader, H. Water Res. 1994, 28, 45.
185. Kla ning, U.K.; Sehested, K. J. Phys. Chem. 1991, 95, 740.
186. Kim, J.I.; Lierse, Ch.; Bu ppelmann, K.; Magirius, S. Mater. Res. Soc. Symp. Proc. 1986, 84,
603.
187. Bu ppelmann, K.; Magirius, S.; Lierse, Ch.; Kim, J.I. J. Less-Common Met. 1986, 122, 329.
188. Lierse, C.; Sullivan, J.C.; Schmidt, K.H. Inorg. Chem. 1987, 26, 1408.
189. Gregorev, A.E.; Makarov, I.E.; Pikaev, A.K. High Energy Chem. 1987, 21, 99.
190. Bu ppelmann, K.; Kim, J.I.; Lierse, Ch. Radiochim. Acta 1988, 44/45, 65.
191. Kelm, M.; Pashalidis, I.; Kim, J.I. Appl. Radiat. Isotopes 1999, 51, 637.
192. Janata, E.; Kelm, M.; Ershov, B.G. Radiat. Phys. Chem. 2002, 63, 157.
193. Shoesmith, D.W. J. Nucl. Mater. 2000, 282, 1.
194. Gromov, V. Radiat. Phys. Chem. 1981, 18, 135.
195. Cristensen, H.; Sunder, S. Nucl. Technol. 2000, 131, 102, and references cited herein.
196. Corbel, C.; Sattonnay, G.; Lucchini, J.-F.; Ardois, C.; Barte, M.-F.; Huet, F.; Dehaudt, P.;
Hickel, B.; Jegou, C. Nucl. Instr. Methods B 2001, 179, 255.
197. Sattonnay, G.; Ardois, C.; Corbel, C.; Lucchini, J.-F.; Barte, M.-F.; Garrido, F.; Gosset, D. J.
Nucl. Mater. 2001, 288, 11.
Copyright 2004 by Taylor & Francis Group, LLC
24
Electron Beam Applications to Flue
Gas Treatment
Hideki Namba
Japan Atomic Energy Research Institute, Takasaki, Gunma, Japan
1. INTRODUCTION
The emission of toxic gases, sulfur oxides (SO
x
), and nitric oxides (NO
x
) from industrial
plants has become a serious problem in many countries. Fig. 1 shows the sources of SO
x
and
NO
x
emission, which are recognized as the origin of dry and wet deposition known as acid
rain [1].
These toxic components sometimes travel more than a thousand kilometers and cause
problems in other places and, in some cases, in other countries. Therefore the problem of air
pollution attracts worldwide attention. Among these, the coal-red ue gas from thermal
power plants has been recognized as one of the main sources of environmental pollution,
because of its larger amount of S and N components. Now many countries are introducing
stricter emission control regulations to solve the problem.
Fig. 1 also shows the chemical reaction mechanism of SO
x
and NO
x
in air for
producing acid rain [2]. The radicals (O, OH, and HO
2
) produced in air play an important
role in oxidizing SO
x
and NO
x
to produce sulfuric acid and nitric acid, which are the main
components of acid rain. Some of these reactions also occur in a reaction chamber of
electron beam treatment of ue gas.
Electron beam treatment of ue gas was rst performed in Japan [3,4] and is now
regarded as a promising pollution control method with many advantages [5,6]: simultaneous
reduction of SO
x
and NO
x
, being a dry process, yields a by-product which can be used as an
agricultural fertilizer, and so on. Fig. 2 shows a ow diagram of electron beam treatment
process for ue gas treatment. In this method, combustion ue gas containing NO
x
(main
component is NO) and SO
x
(main component is SO
2
) is introduced in a spray cooler to
decrease the temperature. After adding ammonia, the ue gas is irradiated with electron
beam in a process vessel. NO and SO
2
are converted to nitric acid and sulfuric acid and are
nally changed to ammonium nitrate and ammonium sulfate which may be used as
agricultural fertilizers. Cleaned ue gas is emitted from a stack. Radiation chemical
reactions induced by electron beam irradiation and related chemical processes will be
studied in detail in this chapter.
Copyright 2004 by Taylor & Francis Group, LLC
2. INCIDENT ELECTRONS
2.1. ElectronNuclei and ElectronElectron Interactions
Accelerated, or energized, electrons by an accelerator are used in ue gas treatment. The
energy of incident electrons for practical treatment process is usually between 300 and 800
kV. The reason for using this energy region will be discussed in Sec. 2.4. The main energy loss
processes of electrons in the region while passing through matter are (1) the interaction with
nuclei to emit x-ray, so-called bremsstrahlung; and (2) the interaction with electrons of the
materials. The contribution ratio of (1) and (2) is dierent depending on the atomic number
Figure 1 Origin of acid rain and chemical reactions in air. (From Refs. 1 and 2.)
Figure 2 Flow diagram of electron beam treatment process for ue gas treatment.
Copyright 2004 by Taylor & Francis Group, LLC
(Z) of materials. At nearly 1 MeV of incident energy, the ratio is EZ/800, where E is the
energy represented in MeV unit. As ue-gas components consist of light atoms, as shown in
Tables 1 and 2, the contribution of (1) is negligible. That means almost all the energy loss of
incident electrons is through inelastic collisions to generate ionization and excitation.
2.2. Stopping Power
In this condition, Bethe formulated the stopping power for electron according to the Born
approximation. Stopping power is a property of irradiated materials and gives the amount
of energy deposited per unit path length, dE/dx.
dE
dx
2pe
4
m
0
v
2
NZ ln
m
0
v
2
E
2I
2
1 b
2
2
1 b
2
q
1 b
2
ln2 1 b
2
1
8
1
1 b
2
q
2
( )
1
Where, E is the energy, x is the length, m
0
and e are the rest mass and charge of
electrons, respectively, v is the velocity of electron, N and Z are the number of atoms in unit
volume and atomic number of the irradiated material, respectively, and b is the relative
velocity represented by v/c, where c is the velocity of light.
In this equation, I is the mean ionization potential, which corresponds to the average
energy used to ionize all the electrons in the atom, and is obtained experimentally. The
values of mean ionization potential for elements in ue gas are listed in Table 1. The ratio of
I=Zdecreases with increasing Z and becomes almost constant for substances with Z greater
than sulfur to be 10 F 1.
For low energy electron region, namely, below 200 Z
2
eV, this equation will not be
valid because of the limitation of Born approximation; in this energy region, the calculations
will be made using electron-impact ionization and excitation cross sections for gaseous
targets [7].
To get more accurate stopping powers, two corrections are added to the basic stopping
power formula. One is shell correction and the other is density eect correction. Including
these eects, the formula will be written as follows [8,9]:
dE
dx
2pe
4
m
0
v
2
NZ ln
m
0
v
2
E
2I
2
1 b
2
1 b
2
q
1 b
2
ln2 1 b
2
1
8
1
1 b
2
q
2
2
C
Z
d
( )
2
Table 1 Mean Ionization Potentials for Elemental Substances
Composing Flue Gas
Z Element I/eV DI/eV (I/Z)/eV
1 H (gas) 19.2 0.4 19.2
6 C (graphite) 78.0 7 13.0
7 N (gas) 82.0 2 11.7
8 O (gas) 95.0 2 11.9
16 S (solid) (180) 11.3
Source: From Ref. 7.
Copyright 2004 by Taylor & Francis Group, LLC
where C is the total shell correction factor, and d is the density eect correction factor that
can be expressed as
d ln
t
2
x
2
p
I
2
1 b
2
!
1 3
where x
p
= (4pNe
2
Z/m)
1/2
is the so-called plasma frequency.
These corrections, however, are not so signicant in ue gas treatment conditions.
2.3. Range
The range is the distance traveled by an electron from its incident point to the point where it
has lost its energy completely; therefore it can be given by the integration of (dx/dE) over
dE from the initial energy to 0. Fig. 3 shows the relation between range and energy of
electrons in water calculated with continuous slowing-down approximation (csda) [10].
Table 2 Typical Concentration of Components in Coal-Fired
Flue Gas and Their Electron Fraction
Components Concentration
Ratio of total number
of electrons
N
2
72% 0.685
O
2
6% 0.065
H
2
O 10% 0.068
CO
2
12% 0.179
NO 225 ppm 0.0002
SO
2
800 ppm 0.0017
Figure 3 Relation between range and energy of electrons in water calculated with csda. (From
Ref. 7.)
Copyright 2004 by Taylor & Francis Group, LLC
For practical use, it is important to knowthe linear distance in the substances fromthe
incident point, i.e., an accelerator window, to the point at which the energy of electrons has
been exhausted. The denitions of range for theoretical consideration and for practical use
are somewhat dierent, however; the range shown in Fig. 3 gives a good indication for
estimating the distance of electrons traveling in substances, which is called maximum range.
Rough estimation of range is possible for other materials, by simply dividing their density.
For example, the density of air is 0.0013 g/cm
3
at normal pressure and temperature; the
range may be estimated about 770 times larger than in water, while that calculated by csda is
about 860 times in this energy region [11]. As the density of the ue gas is almost the same as
that of air, the range in ue gas will be the same as that in air.
2.4. DepthDose Curve and Energy Loss in Windows
The dose distribution in the materials is given as a depthdose curve. An example of the
curve is illustrated in Fig. 4 obtained with the irradiation of electron from 0.5 to 1.0 MeV
using cellulose triacetate (CTA) lmdosimeter [12]. The existence of the maximumdose is an
important characteristic of the depthdose curve. Irradiation from two opposite sides by
using two accelerators was proposed in order to give better uniformity in water [13]. The
uniform irradiation is also important for ue gas treatment. Better eciency of NO
x
removal was proved with both-side irradiation by using three accelerators for coal-red
ue gas than single-side irradiation at the same dose [14].
Fig. 5 shows a schematic representation of an accelerator and a reaction chamber.
Double windows are usually used to separate the accelerator and the chamber. The primary
window keeps the accelerator in vacuum; therefore the window material must have enough
strength to bear the pressure dierence of more than 1 atm. The secondary windowprevents
the primary windowfromue gas in which acidic chemical components are produced during
Figure 4 Depthdose distribution curves in CTA stack lms with the irradiation of electron.
(From Ref. 11.)
Copyright 2004 by Taylor & Francis Group, LLC
irradiation. That means the windows material might be anticorrosive. In order to minimize
the energy loss in the windows, both windows should be thin. In most cases, the windows are
made of metal or plastic lms with a thickness of a few micrometers to a few tens of mi-
crometers. Titanium or its alloy foils are commonly used for industrial-scale plants. Table 3
shows the power penetration eciency of the windows and air-layer, which are the inverse of
energy loss and beam current loss [15]. At lower energy of incident electrons, the energy loss
is not negligible at the windows region. The eciency increases with increasing energy of
accelerated electron; however, we need a rather large installation to obtain higher energy of
electrons. Thus accelerators with an energy region of 300 to 800 keV are commonly used for
large-scale practical plants.
Figure 5 Schematic representation of accelerator and reaction chamber for ue gas treatment.
Table 3 Power Penetration Eciency of Electron with Dierent Incident
Energy at Windows and Air Layer
Materials Ti (30 Am) Air (5 cm) Ti (17 Am) Total
300 keV 0.56
Energy 0.87 0.945 0.89
Beam 0.86 0.995 0.9
500 keV 0.79
Energy 0.93 0.974 0.95
Beam 0.945 1.0 0.97
300 keV 0.88
Energy 0.96 0.985 0.97
Beam 0.975 1.0 0.98
1 MeV 0.90
Energy 0.97 0.99 0.975
Beam 0.98 1.0 0.98
Source: From Ref. 15.
Copyright 2004 by Taylor & Francis Group, LLC
3. INITIAL PROCESS
3.1. Energy Absorption to Produce Primary Species
The energy, or power, of electron beam induced in the ue gas is divided and absorbed by
their gas components roughly depending on their electron fraction. Therefore almost all
the energy is absorbed by the main components of the ue gas, namely, N
2
, O
2
, CO
2
, and
H
2
O. Table 2 shows a typical concentration of the components in coal-red ue gas in
Japan. The ratio of the total number of electrons in each gas components is also listed in
the same table. The energy absorbed directly by the toxic components (SO
2
and NO) is
negligibly small. For electron beam treatment of ue gas, ammonia gas is added to the
ue gas before the irradiation. The amount of ammonia is usually set as stoichiometrically,
i.e., 2D[SO
2
] + D[NO], where D[SO
2
] and D[NO] are the concentrations of SO
2
and NO
intended to be treated, respectively. The concentration of ammonia is usually higher than
the initial concentration of SO
2
and NO; however, it is still far lower than that of the main
components.
The rst processes produced by high-energy charged particles are ionization and
excitation. Evaluating the amount of energy deposited in molecules is one of the key factors
in radiation chemistry, both theoretically and experimentally. The incident electron pro-
duces other electrons while losing its energy during ionization. The ionized electrons also
produce other electrons till the energy reached is less than ionization potential. These are
called secondary electrons. As the W-value in air is
f
30 eV, a single electron with an
energy of 750 kV will produce 25,000 electrons. That means energy deposition to the mate-
rials is mainly by the secondary electrons.
In order to estimate the contribution of secondary electrons, degradation spectrum,
or in other words slowing down spectrum, is used. Degradation spectrum is dened as the
length of the secondary electron trajectory, where the electron with the initial energy E
o
dissipates its energy between E and E+dE.
Once we can obtain the degradation spectrum and the cross section of each reaction
to produce the initial species, the yields of all the initial species are determined with the
following equation [16];
N
s
N
X
i
n
i
Z
T
max
T
z
yTQ
s
Td T 4
or
N
s
N
X
i
n
i
Z
T
max
T
z
TyTQ
s
Td ln T 5
where N
s
is the number of produced initial species, N is the number density, n
i
is the number
of electron at ith discreet energy level, T is the energy of electron (eV), Q
s
(T) is the cross
section (cm
2
), and y(T) is the degradation spectrum (cm/eV). Fig. 6 shows the electron
degradation spectrum in gaseous nitrogen calculated by binary-encounter collision theory
with an irradiation of 100 keV electron [17]. Examples of contribution of the electron
degradation spectrum to ionization and excitation are also listed in the same gure. The
ionization in the gure shows an ionic process, N
2
oN
2
+
(X
2
Ag
+
) + e
0:69N
2:96e
0:885N
2
D 0:295N
2
P 1:87N
4
S 0:29N
2
*
5:377O
2
o
2:07O
2
1:23O
3:3e
2:25O
1
D 2:8O
3
P 0:18O* 0:077O
2
*
7:33H
2
O
o
1:99H
2
O
0:01H
2
0:57OH
0:67H
0:06O
3:3e
0:46O
3
P
4:25OH4:15H0:51H
2
*
6
7:54CO
2
o
2:24CO
2
0:51CO
0:07C
0:21O
3:03e
5:16O
3
P 4:72CO
where the numerical coecients are the amounts of decomposed molecules or produced
species with 100-eV energy absorption, i.e., the G-values.
Positive ions, electrons, and excited molecules are produced after the irradiation of ue
gas as shown in these equations. These ions and radicals play the most important role in
radiation chemical reactions in ue gas.
Figure 6 Contribution of the electron degradation spectrum to ionization and excitation in gaseous
nitrogen calculated by binary-encounter collision theory with irradiation of 100-keV electron. (From
Ref. 15.)
Copyright 2004 by Taylor & Francis Group, LLC
3.2. Ions and Electrons
As the W-values of the major component of ue gas are nearly 30 eV, the total amount of
positive ions produced will be
f
3 in G-value. The most important reaction of positive ions is
that with H
2
O. For example, N
2
+
reacts with H
2
Oto produce OHradical, decomposing H
2
O
through charge transfer reaction.
N
2
2H
2
O !H
3
O
OHN
2
7
In a real case, part of the positive ion will be produced through more complex pathways,
such as forming N
4
+
, forming ion clusters with H
2
O, and so on. In any case, however, the
nal positive ion will be H
3
O
+
or its clustered ions, H
3
O
+
(H
2
O)
n
. Other positive ions, O
2
+
,
H
2
O
+
, and CO
2
+
, will behave in the same way.
O
2
2H
2
O !H
3
O
OHO
2
H
2
O
2H
2
O !H
3
O
OHH
2
O 8
CO
2
2H
2
O !H
3
O
OHCO
2
On the analogy of these reactions, the radical ions, N
+
, O
+
, and H
+
, will produce
radicals and positive ions in the following events.
N
2H
2
O !H
3
O
OHN
O
2H
2
O !H
3
O
OHO 9
H
2H
2
O !H
3
O
OHH
These positive ions are the main source of the OH radical, which is one of the most
important radicals for electron beam treatment of ue gas as will be discussed in the next
section.
Electrons react mainly with O
2
to produce O
2
O
2
M!O
2
M 10
where M is a third body for the reaction.
These positive and negative ions neutralize to form HO
2
.
H
3
O
O
2
!HO
2
H
2
O 11
Therefore one pair of ions produces one OH and one HO
2
radicals. The total amount of
radicals, which are produced in ue gas by electron beam irradiation, is possible to calculate
using reported G-values. The main radicals produced initially through direct and ionic
decomposition processes are OH, N, HO
2
, O, and H.
4. CHEMICAL REACTIONS
4.1. Radical Reactions
The radicals produced by irradiation will react with NO and SO
2
, which are well-known
good radical scavengers. These radical reactions related to NO
x
and SO
x
removals are very
Copyright 2004 by Taylor & Francis Group, LLC
complicated with more than a few hundreds of reaction pathways [20]. The main radical
reactions committed to the removal of NO
x
and SO
2
are summarized in Table 4.
The total amount of radicals produced after irradiation is estimated at
f
10 molecules/
100 eV ( G-value) from Eq. (6) and from discussions in the previous section. The following
equation will help us to compare the amount of radicals with the concentration of NO and
SO
2
that we are going to treat:
C 0:104 M
w
DG 12
where Cis the concentration (ppm), M
w
is the molecular weight (g/mol), Dis the dose (kGy),
and G is G-value (molecules/100 eV). Because M
w
is almost the same as air (29 g/mol) in ue
gas, the concentration will be 300 ppm at G = 10 and D = 10 kGy.
Some radicals produce other radicals through the reactions shown in Table 4.
Comparing the calculated amount of NO and SO
2
removals through radical reactions to
the experimental results [21], it indicates that NO
x
removals mainly occur through radical
reactions, while SO
2
removals mainly occur through other processes, so-called thermal
reactions, or heterogeneous reactions. More details will be given in the following sections.
Table 4 Main Radical Reactions
Related to the Removal of NO
x
and
SO
2
OH+NO+M ! HNO
2
+M
OH+SO
2
+M ! HSO
3
+M
OH+NO
2
+M ! HNO
3
+M
OH+NO
2
! NO+HO
2
OH+HNO
2
! NO
2
+H
2
O
OH+NH
3
! NH
2
+H
2
O
OH+CO ! CO
2
+H
N(
4
S)+NO ! N
2
+O
N(
4
S)+NO
2
! N
2
O+O
N(
2
D)+O
2
! NO+O
N(
2
D)+NH
3
! NH+NH
2
N(
2
P)+O
2
! NO+O
N(
2
P)+NH
3
! NH+NH
2
HO
2
+NO ! NO
2
+OH
HO
2
+SO
2
! SO
3
+OH
O(
1
D)+H
2
O ! 20H
O(
1
D)+M ! O(
3
P)+M
O(
3
[Pub])+O
2
+M ! O
3
+M
O(
3
[Pub])+NO+M ! NO
2
+M
O(
3
[Pub])+NO
2
! NO+O
2
H+O
2
+M ! HO
2
+M
NH
2
+NO ! N
2
+H
2
O
NH
2
+NO
2
! N
2
O+H
2
O
NH+NO ! N
2
+OH
NH+NO
2
! N
2
O+OH
HSO
3
+O
2
! SO
3
+HO
2
O
3
+NO ! NO
2
+O
2
SO
3
+H
2
O ! H
2
SO
4
M is a third body molecule.
Copyright 2004 by Taylor & Francis Group, LLC
4.2. NO
x
Removal
Fig. 7 shows the main reaction pathways to remove NO
x
. NOwill react with Oradical or O
3
,
in which O
3
is produced with the reaction of O
2
.
NOO
3
P M!NO
2
M
O
3
P O
2
M!O
3
M 13
NOO
3
M!NO
2
O
2
M
Another important reaction to oxidize NO is that with HO
2
.
NOHO
2
!NO
2
OH 14
This NO
2
is converted to nitric acid with the reaction of OH and nally changes to
ammonium nitrate with the chemical reaction of ammonia.
NO
2
OHM!HNO
3
M
15
HNO
3
NH
3
!NH
4
NO
3
This oxidization is the main process of NO
x
removal.
Another important pathway for removing NO
x
is reaction of NO with ground-state
nitrogen radical N(
4
S) or NH
2
radical, in which NH
2
radical is produced with the reaction of
NH
3
and OH. Nitrogen molecules are produced with these reactions [22].
NON
4
S !N
2
O
NH
3
OH !NH
2
H
2
O 16
NONH
2
!N
2
H
2
O
With the same type of reaction, dinitrogen monoxide is produced [23].
NO
2
N !N
2
OO
17
NO
2
NH
2
!N
2
OH
2
O
Nitrous acid is also produced [24] with the reaction of NO and OH; however, this acid
will be oxidized to NO
2
or HNO
3
nally.
NOOHM!HNO
2
M 18
Figure 7 Main reaction pathways for removing NO
x
. Solid line: radical reaction. Dotted line:
thermal reaction.
Copyright 2004 by Taylor & Francis Group, LLC
The ground-state nitrogen radical N(
4
S) decreases NO with reducing reaction;
however, excited-state nitrogen radicals, N(
2
D) and N(
2
P), behave perfectly in another
way to produce NO.
N
2
D;
2
P O
2
!NOO 19
4.3. SO
2
Removal
The main reaction pathways for SO
2
removal are illustrated in Fig. 8. The most important
radical reaction for removing SO
2
is the following reaction to produce HSO
3
.
SO
2
OHM!HSO
3
M 20
This HSO
3
will react with oxygen to form SO
3
and easily react with water to form
sulfuric acid.
HSO
3
O
2
!SO
3
HO
2
21
SO
3
H
2
O !H
2
SO
4
The acid will react with ammonia and become ammonium sulfate.
H
2
SO
4
2NH
3
!NH
4
2
SO
4
22
Sulfur dioxide may react with HO
2
to form SO
3
; however, the reaction (Eq. (14)) will
be overcome when NO simultaneously exists.
SO
2
HO
2
!SO
3
OH 23
In addition to this SO
x
removal process with OH radicals, there exists a so-called
thermal reaction in which SO
2
reacts with NH
3
[2527]. This reaction depends on the
concentrations of NH
3
, SO
2
, and H
2
O, and it has a very large temperature dependence.
5. INDUSTRIAL-SCALE PLANTS AND OTHER FLUE GAS
Large-scale pilot plants have been installed in Indianapolis, USA, in 1984 [28], in Karlsruhe,
Germany, in 1985 [29], in Warsaw, Poland, in 1990 [30], and in Nagoya, Japan, in 1991 [14]
for the treatment of coal-red ue gas with a ow rate of more than 10,000 m
3
/hr at normal
Figure 8 Main reaction pathways for removing SO
2
. Solid line: radical reaction. Dotted line:
thermal reaction.
Copyright 2004 by Taylor & Francis Group, LLC
temperature and pressure. With the success of these pilot-scale tests demonstrating the
feasibility of the process, two large industrial-scale plants were installed in Chengdu, China,
in 1997 and in Pomorzany, Poland, in 2000. The plant in Chengdu is equipped with two
accelerators with a total capacity of 640 kW for treating 300,000 m
3
/hr of ue gas with 1800
ppm SO
2
and 400 ppm NO
x
. The plant is now successfully operating with 80% removal of
SO
2
[31]. The plant in Pomorzany [32] has four accelerators with a total maximum power of
1200 kW to treat 270,000 m
3
/hr of ue gas and has started operation.
The technology for treating ue gas from oil burning boilers is essentially the same as
that from coal-red boilers, because of the same components to treat, i.e., NO
x
and SO
x
.
Containing high concentration of sulfur, heavy oil is regarded as an urgent target; therefore
intensive investigations are going on in many countries. Among them, an industrial-scale
plant for ue gas of 620,000 m
3
/hr will start operation in Nagoya, Japan, in 2003 [33].
The ue gas frommunicipal waste incinerator boilers contains SO
2
, NO
x
, and HCl. To
remove these harmful components simultaneously by dry process, electron beam treatment
method was investigated. The pilot-scale test was conducted in Matsudo, Japan, in 1992
with a ue gas of 1000 m
3
/hr [34]. Recently, dioxins, namely, poly-chlorinated-di-benzo-
paradioxins (PCDDs) and poly-chrorinated-di-benzo-furan (PCDFs), from incinerators
have become a very serious problem because of their high toxicity. Pilot-scale tests to
decompose dioxins by electron beam irradiation were conducted in Karlsruhe, Germany
[35], and in Takahama, Japan [36], using almost the same capacity of ue gas, 1000 m
3
/hr.
Very promising results were obtained with decomposing more than 90% of dioxins.
REFERENCES
1. Okita, T. In Sanseiu (Acid Rain); Hakuyusya, 1996; 25 pp (in Japanese).
2. Mohnen, V.A. Sci Am Aug. 1988, 259 (2), 14.
3. Kawamura, K.; Aoki, S. J. At Energy Soc. Jpn. 1972, 14, 597.
4. Machi, S.; Tokunaga, O.; Nishimura, K.; Hashimoto, S.; Kawakami, W.; Washino, M.;
Kawamura, K.; Aoki, S.; Adachi, K. Radiat. Phys. Chem. 1977, 9, 371.
5. Electron beam processing of combustion ue gases, IAEA-TECDOC-428, 1987.
6. Environmental Applications of Ionizing Radiation; Cooper, W.J., Curry, R.D., OShea,
K.E., Eds.; John Wiley and Sons, Inc.: New York, 1998.
7. Watanabe, T.; Shimamura, I.; Karashima, S. In CRC Handbook of Radiation Chemistry;
Takahashi, T., Tabata, Y., Itoh, Y., Tagawa, S., Eds.; CRC Press: Boca Raton, 1991; 31 pp.
8. Chatterjee, A. Radiation ChemistryPrinciples and Applications; Farhatazis, Rodgers, M.A.J.,
Eds.; VCH Publishers, Inc.: New York, 1987.
9. Mozumder, A. Fundamentals of Radiation Chemistry; Academic Press: San Diego, 1999.
10. Watanabe, T.; Shimamura, I.; Karashima, S.; Takahashi, T. In CRCHandbook of Radiation
Chemistry; Tabata, Y. Itoh, Y., Tagawa, S., Eds.; CRC Press: Boca Raton, 1991; 47 pp.
11. Stopping powers for Electrons and Positrons, ICRU Report 37, International Commission
on Radiation Units and Measurements, Bethesda, 1984.
12. Kanazawa, T.; Haruyama, Y.; Ueno, S.; Yotsumoto, K.; Tanaka, R.; Washino, M.; Yoshida,
K. JAERI-M 86-005, 1986.
13. Woods, J.; Pikaev, K. Applied Radiation Chemistry: Radiation Processing; Wiley-Interscience:
New York, 1994.
14. Namba, H.; Tokunaga, O.; Hashimoto, S.; Tanaka, T.; Ogura, Y.; Doi, Y.; Aoki, S.; Izutsu,
M. Radiat. Phys. Chem. 1995, 46, 1103.
15. Sakamoto, I.; Mizusawa, K.; Kashiwagi, M. In Electron beam processing of combustion ue
gases; IAEA-TECDOC-428, 201, 1987.
16. Rau, A.R.P.; Inokuti, M.; Douthat, D.A. Phys. Rev. 1978, A18, 971.
Copyright 2004 by Taylor & Francis Group, LLC
17. Okazaki, K.; Sato, S. Bull. Chem. Soc. Jpn 1975, 48, 3523.
18. Willis, C.; Boyd, A.W. Radiat. Phys. Chem. 1976, 8, 71.
19. Ma tzing, H. In Adv. Chem. Phys., LXXX; Prigogine, I. Rice, S.A., Eds.; John Wiley and Sons
Inc., 1991; 315 pp.
20. Ma tzing, H. In Adv. Chem. Phys., LXXX; Prigogine, I. Rice, S.A., Eds.; John Wiley and Sons
Inc., 1991; 360 pp.
21. Namba, H.; Tokunaga, O.; Tanaka, T.; Ogura, Y.; Aoki, S.; Suzuki, S. Proceedings of the
International Conference on Evolution in Beam Applications, Takasaki, Japan, 1991; 476 pp.
22. Namba, H.; Aoki, Y.; Tokunaga, O.; Suzuki, R.; Aoki, S. Chem. Lett. 1988, 1465.
23. Namba, H.; Tokunaga, O.; Suzuki, R.; Aoki, S. Appl. Radiat. Isotopes 1990, 41, 569.
24. Ma tzing, H.; Namba, H.; Tokunaga, O. Radiat. Phys. Chem. 1994, 43, 215.
25. Machi, S.; Namba, H.; Suzuki, N. Electron beam processing of combustion ue gases,
IAEA-TECDOC-428, 13, 1987.
26. Paur, H.-R.; Jordan, S. Radiat. Phys. Chem. 1988, 31, 9.
27. Hirota, K.; Niina, T.; Anwar, E.; Namba, H.; Tokunaga, O.; Tabata, Y. Kankyo Kagaku
Kaishi 1993, 6, 143.
28. Frank, N.; Hirano, S.; Kawamura, K. Radiat. Phys. Chem. 1988, 31, 57.
29. Platzer, K.-H.; Willibald, U.; Gottstein, J.; Tremmel, A.; Angele, H.-J.; Zellner, K. Radiat.
Phys. Chem. 1990, 35, 427.
30. Cheimielewski, A.G.; Iller, E.; Zimek, Z.; Licki, J. Radiat. Phys. Chem. 1992, 40, 321.
31. Doi, Y.; Nakanishi, I.; Shi, J.; Fujita, K.; Habe, H.; Suzuki, R.; Hayashi, H.; Ikeda, K.; Konno,
Y.; Amano, T.; Hagiwara, I.; Satoh, K. Ebara Jiho, 1999, 183, 51.
32. Chmielewski, A.G.; Iller, E.; Zimek, Z.; Romanowski, M.; Koperski, K. Radiat. Phys. Chem.
1995, 46, 1063.
33. Tanaka, T. Genshiryoku Eye 1998, 44, 31 (in Japanese).
34. Hirota, K.; Tokunaga, O.; Miyata, T.; Sato, S.; Osada, Y.; Sudo, M.; Doi, T.; Shibuya, E.;
Baba, S.; Hatomi, T.; Komiya, M.; Miyajima, K. Radiat. Phys. Chem. 1995, 46, 1089.
35. Paur, H.-R.; Bauman, W.; Ma tzing, H.; Jay, K. Radiat. Phys. Chem. 1998, 52, 355.
36. Hashimoto, S.; Hirota, K.; Hakoda, T.; Taguchi, M. Proceeding of the 23rd KAIF-JAIF
Seminar on Nuclear Industry, Seoul, Korea, 2001; 251 pp.
Copyright 2004 by Taylor & Francis Group, LLC
25
Ion-Beam Therapy: Rationale, Achievements,
and Expectations
Andre Wambersie and John Gueulette
Universite Catholique de Louvain, Brussels, Belgium
Dan T. L. Jones
iThemba Laboratory for Accelerator Based Sciences, Somerset West, South Africa
Reinhard Gahbauer
Ohio State University, Columbus, Ohio, U.S.A.
ABSTRACT
Radiation therapy is a very important and eective treatment modality for almost all types
of malignancies. Its importance in the local control of primary or bulky cancers will even
increase in the future if more eective chemotherapeutic or other systemic treatments
become available to treat metastatic disease.
Improvements in the physical selectivity, from orthovoltage x-ray to cobalt-60 and
high-energy linear accelerators, combined with more powerful diagnostic tools and radia-
tion delivery methods have continuously improved the results of photon therapy (3-D or
inverse planning, conformal- and intensity-modulated radiation therapy, and stereotactic
methods). The safety and the reliability of photon therapy are well established.
However, newtypes of radiation continue to be explored with the aimof improving the
physical selectivity, the radiobiological dierential eect, or both. In this context, the role of
fast neutrons with their biological selectivity in selected cancer sites and of protons with their
superb physical selectivity will be reviewed.
Fast neutrons were the rst nonconventional radiation used in cancer therapy. Fast
neutrons (a high-LET radiation) were introduced for the following radiobiological reasons:
(1) a reduction of the OER with increasing LET; (2) a reduction in the dierence in
radiosensitivity related to the position of the cells in the mitotic cycle; (3) and less repair and
thus less clinical relevance of the dierent repair mechanisms. The best and clinically proven
indications for fast neutrons are salivary gland tumors, locally advanced prostatic adeno-
carcinomas, and slowly growing, well-dierentiated sarcomas.
Radiobiological issues, mainly related to the RBE of high-LET radiation, are
discussed.
Copyright 2004 by Taylor & Francis Group, LLC
Proton beams bring a signicant improvement in the physical selectivity of the
treatment. The number of proton therapy centers in operation and in the planning stage
increases continuously worldwide. The best clinical results, so far, have been reported for
uveal melanoma, tumors of the base of skull, and some brain tumors in children.
Heavy ions are very promising by combining the advantage of the excellent physical
selectivity of protons with the radiobiological advantages of fast neutrons for some tumor
types. Heavy ions were applied at Berkeley from 1975 to 1992 and at NIRS in Chiba since
1994. A pilot study started at GSI-Darmstadt in 1997.
For historical reasons and for their general principles, pion therapy and the still
experimental Boron Neutron Capture Therapy are also briey discussed.
Besides the technological and radiobiological aspects considered here, other promising
approaches are now being investigated in the dierent therapy centers and research
laboratories worldwide in order to improve the eectiveness of radiation therapy. They
include better, individualized fractionation and time factor, association with drugs to
modulate the radiation sensitivity, combination with gene therapy and immunotherapy,
and protocols combining radiation therapy with surgery and chemotherapy in a more
eective way. These approaches are outside the scope of this chapter.
1. INTRODUCTION: RADIATION THERAPY (AND ION BEAMS)AN
EFFECTIVE APPROACH IN CANCER TREATMENT
1.1. Cancer Incidence and Social Implications
Radiation therapy constitutes, together with surgery and chemotherapy, one of the three
traditional and recognized methods of cancer treatment. In addition, novel approaches,
such as immunotherapy and gene therapy, are developing and appear to be promising.
With increased longevity of the populations resulting from improved control of
epidemic and infectious diseases, the frequency of cancerand the often associated suer-
inghas raised cancer awareness in the 20th century [1].
In developing countries, cancer cases have risen from 2 million in 1985 to 5 million in
2000 and are projected to number 10 million in 2015. By contrast, in developed countries,
there were 5 million cases in 1985 as well as in 2000; no increase is projected to 2015.
For example, in the United States, in 1991, more than 1 million invasive cancers
occurred, i.e., an incidence of about 400 per 100,000 per year. In addition, more than 600,000
nonmelanomatous skin cancers occurred; most of them can now be cured. In developed
countries, the probability of dying fromcancer is 2025%and is expectedto further increase.
1.2. Present Situation in Cancer Cure Rate
The dierent techniques mentioned abovesurgery, radiation therapy, and chemother-
apyare more and more frequently applied in combinations in modern protocols rather
than in competition. Today, about 45% of all cancer patients can be cured (Table 1) [2].
Radiation therapy contributes to the cure of 23%of all cancer patients [alone (12%) or
in combination with surgery (6%) or chemotherapy/immunotherapy (5%)]. Thus about half
of the cancer patients who are cured benet from radiation therapy at least for part of their
treatment: this proportion illustrates the important role of radiation therapy in cancer
management.
Sixty-ve percent of the cancer patients present themselves for the rst consultation
with a localized tumor. About 1/3 of these fail, i.e., nearly 25%of the total number of cancer
Copyright 2004 by Taylor & Francis Group, LLC
patients. This situation has to be improved mainly by improving the eciency of local
treatments, i.e., radiation therapy and/or surgery.
This is the challenge for the coming years for teams involved in development of new
radiation therapy modalities.
The importance of radiation therapy in the local control of primary or bulky cancers
will even increase in the future if more eective chemotherapeutic or systemic therapy (such
as immunology, gene therapy, etc.) becomes available to treat (prevent) metastatic disease.
Besides increasing the cure rate, it is important to improve the tolerance to the
treatment and reduce the long-termsequelae. Also, palliative treatments (in particular, pain
relief) deserve attention for those patients for whom there is no hope anymore of a cure.
1.3. Radiation Therapy in Cancer Treatment
At present, in industrialized countries, about 70% of the cancer patients are referred to a
radiation therapy department for at least part of the treatment. The majority of them are
treated with conventional photon beamtherapy (i.e., the reference treatment modality as
dened in Table 2.1).
Experience accumulated over more than a century indicates that the major improve-
ments in the eectiveness of radiation therapy were always associated with (or made possible
by) signicant progress in technology [35].
The major steps involved were rst the move from200-kVx-rays to cobalt-60 and then
to modern electron linear accelerators providing x-ray beams of about 20 MV(and electrons
when needed for some patients). Improvements in the beam penetration were, in general,
combined with improvements in beam delivery systems, e.g., isocentric gantries, variable
collimators, later on multileaf collimators, etc.
In addition, 3-D dosimetry and treatment planning were introduced and, last but not
the least, dramatic progress was achieved in imaging and related disciplines.
It is recognized that, besides the technological and radiobiological aspects considered
here, other promising approaches are nowbeing investigated in the dierent therapy centers
and research laboratories worldwide in order to improve the eectiveness of radiation
therapy. They include better individualized fractionation and time factor, association with
drugs to modulate the radiation sensitivity, combination with gene therapy and immuno-
therapy, and protocols combining radiation therapy with surgery and chemotherapy in a
more eective way. These approaches are outside the scope of this chapter.
Table 1 Summary of the Present Situation Concerning Cancer Cure Rate
Cure rate (%)
Patients appearing with localized tumor (65%):
Cured by surgery 22
Cured by radiotherapy 12
Cured by combination of surgery and radiotherapy 6
Patients appearing with inoperable or metastatic disease (35%):
Cured by combined complex treatment including, for example,
chemotherapy and immunotherapy
5
Total
a
45
a
Excluding nonmelanoma skin cancers.
Source: Ref. 2.
Copyright 2004 by Taylor & Francis Group, LLC
Before reviewing the results achieved with ion beams in cancer therapy, a short survey
of the general situation in radiation therapy and the main trends are presented in the next
section.
2. GENERAL TRENDS IN RADIATION THERAPY (IN RELATION WITH
TECHNOLOGICAL DEVELOPMENTS)
2.1. Conformal Radiation Therapy and Highly Selective and Novel
Photon Irradiation Techniques
For several decades, regular and continuous technical improvements were made to the linear
accelerators. Similar improvements were achieved in dosimetry and imaging, and, as a
consequence, the quality and eectiveness of the treatments progressively improved.*
Recently, dramatic and novel approaches were introduced in radiation therapy
planning and delivery: conformal radiation therapy, inverse treatment planning, and
intensity-modulated radiation therapy. These improved the optimization of all treatment
parameters in order to achieve the desired and optimal dose distribution, i.e., to match
(conform) as closely as possible the treated volume and the planning target volume (PTV)
and at the same time reduce the dose to the organs at risk (OAR). These techniques should
allow us to bring the conformity index close to unity [6,7]. This goal became possible due to
the contribution of (at least) the three following factors.
1. Patient data. Complete and accurate anatomical information, including the
location of the dierent target volumes and organs at risk, became possible with
the latest developments in modern imaging techniques, including computer to-
mography (CT), magnetic resonance imaging (MRI), positron emission tomog-
raphy (PET), etc. In addition, better knowledge of the natural cancer history
contributes to the selection of the optimum strategic approach (e.g., selection of
CTV, PTV, etc.).
2. Medical physics and dosimetry. Powerful and fast 3-D treatment planning, in-
cluding inverse planning with on-line 3-D dose computation, became avail-
able. Heterogeneities can be easily identied and appropriate corrections can be
applied. In addition, the radiation eld can be made to conform to irregular
contours and shapes of the organs.
3. Accelerator engineering. Full, reliable, and on-line control of the electron ac-
celerator, beam delivery system, and multileaf collimator is the basis of the new
conformal treatment and IMRT approaches. Patient-beam positioning became
more accurate and fully reliable.
Of course, appropriate QAprograms are required not only to check each of the above
three factors, but also to ensure that they are adequately combined.
From a technical point of view, intensity-modulated radiation therapy (IMRT)
optimizes several parameters: selection of multiple beams and, for each beam, optimization
of dose, dose rate, size and shape, etc. Tomotherapy optimizes the dose distribution by
IMRT in successive sections with immediate verication; gamma knife, cyberknife, and
radiosurgery concentrate high radiation doses in well-dened small volumes.
* The brachytherapy applications are not discussed in this chapter.
Copyright 2004 by Taylor & Francis Group, LLC
These new techniques are rapidly expanding and are becoming available in an
increasing proportion of hospitals, at least in industrialized countries. They are commer-
cially available. They constitute the most important trend in radiation therapy today.
2.2. Does Effectiveness of Photon Beam Therapy Reach a Plateau?
The energy of photons with their optimal beam penetration, the reliability of the beam
delivery and collimation systems, and the mechanical stability of the new generation of
accelerators have achieved a nearly optimum level of performance. Actually, with the
modern linear electron accelerators, it is now possible to irradiate at the prescribed dose
(nearly) any target volume of any shape with reduced irradiation of the surrounding organs
at risk (OAR).
However, the impressive development and progress in conformal therapy with
photons raise a dicult issue: to what extent has photon beam therapy reached a kind of
plateau as far as physical selectivity is concerned? This important question is still con-
troversial and heavily debated. If photon therapy had reached a plateau, the search for
improvements would be directed to alternative radiation modalities [8,9].
In particular, it is recognized that when optimizing the treatment parameters in order
to improve, e.g., the conformity index in conformal therapy or IMRT, other indices can
change adversely, for example, the size of the irradiated volume, the integral dose, and the
dose homogeneity throughout the PTV.
2.3. New Types of Radiations
An alternative to further improve or optimize the photon techniques is to replace the con-
ventional photonbeams withnewtypes of radiation. Indeed, since the beginning of radiation
therapy, the radiation oncologists have always been eager to search new types of beams
(dierent from conventional x-rays/photons) in order to improve the ecacy of radiation
therapy. In principle, dierent approaches can be adopted (Table 2).
Improving the physical selectivity. The most straightforward method to improve the
ecacy of radiation therapy is to improve the physical selectivity of the irradiation, i.e., by
using beams with better physical characteristics (penetration, collimation, etc.). This is the
rationale for the introduction of proton beams: no radiobiological advantage is expected.
Improving the radiobiological dierential eect between the cancer and normal cell
populations. This second approach is more complex and involves radiobiological consid-
erations. New and dierent radiation qualities have been considered such as high-LET
radiations. Some of the radiobiological issues involved in the use of high-LET radiation, in
particular the RBE concept, are discussed in Sec. 3.
Table 2 New Types of Beams in Radiation Therapy (Alternative to Conventional Photon
Beam Therapy)
(1) Improving the physical selectivity: proton beams
(2) Improving the radiobiological dierential eect: fast neutrons and other high-LET radiations
(3) Combination of physical selectivity and radiobiological dierential eect: heavy ions
Copyright 2004 by Taylor & Francis Group, LLC
Among high-LET radiations, fast neutrons were rst introduced to exploit a radio-
biological dierential eect; no benet is expected from the physical selectivity.
However, clinical evidence has demonstrated that, from a pure safety point of view, a
physical selectivity at least as good as that obtained with conventional photon beams is
absolutely required for neutrons as well.
The logical combination of these two approaches is the rationale for the application of
heavy ions. The use of heavy ions in cancer therapy is, in principle, easy to justify by several
sets of physical and radiobiological arguments. However, heavy ions are expensive, require
sophisticated quality assurance, and raise complex radiobiological and physical issues as
long as robust clinical data are not yet available.
The specic techniques of brachytherapy and the administration of unsealed sources
of radionuclides, in nuclear medicine [10], are not dealt with in this chapter.
2.4. Reference Treatment Modality
Evaluation of the relative merits of dierent treatment techniques or strategic approaches
existing or newly introducedrequires the denition of a reference treatment modality.
Fractionated photon beam therapy, as dened in Table 3, is generally accepted as the
reference therapy modality. This is due to its broad application, the experience accumulated
over several decades, and its recognized eectiveness [4,5,8].
However, when making these comparisons, a diculty arises from the fact that the
eciency of photon therapy (i.e., the reference) is improving continuously, as clearly
shown above (Sec. 2.1) [9].
3. HIGH-LET RADIATION AND RBE: RADIOBIOLOGICAL CONCEPT
AND CLINICAL APPLICATION
3.1. From Absorbed Dose to Biological Effects
Absorbed dose is a scientically rigorously dened quantity which is used to quantify the
exposure of humans, biological systems, and any type of material to ionizing radiation.
The concept and quantity absorbed dose were introduced by the ICRU in 1951,
with the special unit rad. A new special unit gray (Gy) was introduced in 1972 to be in
correspondence with the SI system of quantities and units (Syste` me International des
Grandeurs et Unite s) [11,12].
Absorbed dose is a fundamental and basic physical quantity which can be used in all
elds where ionizing radiations are used. It is directly related to the physical, chemical, and
biological eects produced by the irradiation. The concept of absorbed dose thus has broad
applications and is indeed widely used. Metrological institutions provide standards and
calibration of instruments in terms of absorbed dose.
Table 3 The Reference Radiation Therapy Modality
External photon beam therapy, involving (multiple) beams of conformally adjusted size and shape,
adequately orientated to cover the PTV(s) in a homogeneous way, with photon energy ranging
from a few megavolts to about 20 MV, using fractionated irradiation with daily fractions of 2 Gy,
ve times a week, over 467 weeks depending on the clinical situation.
Copyright 2004 by Taylor & Francis Group, LLC
Although the biological eect is directly related to the quantity absorbed dose, there is
no unique relationship between absorbed dose and induced biological/clinical eects, and
other factors have to be considered, such as:
1. Radiation quality
2. Dose rate, fractionation, and temporal distribution of the irradiation
3. Level of biological eect/endpoint
4. More generally, the technical conditions in which radiological treatment is de-
livered (resulting in dierent dose distributions and dierent doses to normal
tissues)
Aweighting factor or weighting function, W, thus is applied to correlate and link
absorbed dose and biological eect [13,14]. The factors listed above aect the weighting
function or weighting factors independently. Dierent empirical formulations are used in
radiation therapy to express this biological weighting function. Some agreement has been
reached on the numerical values of the involved quantities, but only for well-established
radiotherapy techniques.
3.2. Energy Distribution at the Microscopic Level, Microdosimetry,
and Biological Efficiency
Radiation quality is dened by the nature, charge, and energy spectrum of the particles and
can be characterized by the linear energy transfer (LET) or, alternatively, by the micro-
dosimetric spectra at the point of interest under the actual irradiation conditions.
With hadrons (i.e., neutrons, protons, and heavy ions), new radiation qualities are
introduced in therapy. The distributions of the ionizations (and energy deposition events)
along the particle tracks are dierent, and, as a result, dierent and increased biological
eects (at equal absorbed dose) may be expected compared with the conventional photon
beams. Fig. 1 illustrates the dierences in dose necessary to produce a given biological eect
as a function of radiation quality [15].
Fig. 2 illustrates schematically the distribution of the ionizations along the particle
tracks after photon and neutron irradiation [1,16]. In a more quantitative way, Fig. 3
presents the microdosimetric lineal energy, y, spectra for two clinical fast neutron beams: the
highest and lowest energy used in cancer therapy (measurements performed at Louvain-la-
Neuve and Essen, respectively) [17].
These large dierences in energy distribution, at the microscopic level, between
radiation qualities at equal absorbed dose, produce dierent biological eects: this leads
to the concept of RBE.
3.3. The RBE Concept
The concept of relative biological eectiveness (RBE) was introduced by the ICRP and the
ICRU [18,19].
The RBE is dened as a dose ratio, between two doses delivered with two radiation
qualities, which produces the same eect on a given biological system, under the same
conditions.*
* If A is the test radiation quality and B is the reference radiation quality, and if D
A
and D
B
are
the doses necessary to produce the eect of interest with radiation A and B, respectively, the RBE
of radiation A, relative to radiation B, is D
B
/D
A
.
Copyright 2004 by Taylor & Francis Group, LLC
Figure 1 Absorbed dose necessary to produce a given biological eect on a given system, in the
present case: 50% inactivation of V79 hamster cells in exponential growth phase. The dose needed
depends on the type of the particles (photons, alpha particles, or uranium ions) and on their energy.
(From Ref. 15.)
Figure 2 Distribution of ionizations in a mediumirradiated by x-rays (a) and neutrons (b). The black
dots represent the ionizations produced along the tracks of electrons set in motion by x-rays (a) and of
protons, a-particles, or heavy ions set in motion by neutrons (b). The distribution of ionizations is very
dierent. After irradiation by neutrons, when a vital sensitive structure (or target, represented by the
circles) is crossed by a track, the deposition of energy (or the damage) is so great that there is a high
probability of cell death whatever the cell line, the position in the cell cycle, the degree of oxygenation,
etc. After irradiation by x-rays, the energy depositions are smaller and more variable. In some cases, a
single particle deposits enough energy to kill the cell; in others, death of the cell requires the
accumulation of damage produced by several tracks. (From Ref. 16, adapted in Ref. 3.)
Copyright 2004 by Taylor & Francis Group, LLC
The RBE is a clear, unambiguous, and well-dened radiobiological concept. A RBE
value is the result of an experiment and is thus associated with an experimental uncertainty.
The biological system, type, and level of eect, the dose, and the experimental conditions in
which a given RBE value has been obtained must be specied.
3.4. RBE Values and Modifying Factors
The RBE of a given radiation quality relative to
60
Co gamma rays is not a unique value but
varies to a large extent with dose, biological system, and eect.
Fast neutrons are selected here as an example to illustrate the RBE variations because
their RBE variations are signicant and are well documented. They can be summarized as
follows (Figs. 4 and 5) [20,21]:
1. RBE values range within large limits (from less than 2 up to about 10).
2. The RBE increases with decreasing dose.
3. The RBE values for late eects are signicantly higher than for early eects
(Figs. 4 and 5).
4. The RBE of the fast neutron beams used in therapy varies with energy, and this
variation is signicant (Fig. 6) [22].
Figure 3 Comparison of energy depositions after irradiation with fast neutrons and x-rays. The
curves indicate distributions of individual energy-deposition events in a simulated volume of tissue 2
Am in diameter; the parameter y (lineal energy) represents the energy deposited by a single charged
particle traversing the sphere, divided by the mean cord length. The maximumwith x-rays is at 0.3 keV
Am
1
and with d(14)+Be neutrons at 20 keV Am
1
. The spectrum for p(65)+Be neutrons shows 4
peaks: the rst is at 8 keV Am
1
and corresponds to high-energy protons, the second at 100 keV Am
1
corresponds to low-energy protons, the third at 300 keVAm
1
is due to a-particles, and the last, at 700
keV Am
1
, is due to recoil nuclei. (From Ref. 17.)
Copyright 2004 by Taylor & Francis Group, LLC
Figure 4 RBE/dose relationships for dierent normal tissues. Summary of the pretherapeutic
radiobiological experiments performed at the Hammersmith cyclotron with d(16)+Be neutrons. The
increase of RBE with decreasing dose is obvious for all tissues investigated. (From Ref. 20.)
Figure 5 RBE/dose relationships for 15-MeV neutrons produced by a (d,T) generator. Dierent
biological endpoints in normal tissues and tumors are investigated. For late tolerance of spinal cord,
the RBE increases from 1.2 to 3.7 when the neutron dose per fraction decreases from 16 to 0.8 Gy.
Higher RBE values were found later on for spinal cord at lower doses. (From Ref. 21.)
Copyright 2004 by Taylor & Francis Group, LLC
Ahigh RBE of a new radiation quality does not per se imply a therapeutic advantage.
The therapeutic gain is actually the ratio of the RBE for the eects on the cancer cell
population and the RBE for the eects on the normal tissues (under the actual irradiation
conditions).
Further information about the RBE and the specic biological eects of the dierent
hadrons used in therapy are found in Sec. 4.
3.5. From Radiobiology to Clinical Application of RBE: The RBE Weighting
Factor, W
RBE
When prescribing the dose in hadron (neutron) therapy, the radiation oncologist has to take
into account the (set of) reported RBE values, and, from them, the radiation oncologist has
to select one RBE value to weight the absorbed dose in order to obtain the desired clinical
eect [13,14].
Figure 6 RBE variation of neutron beams as a function of energy. The p(65)+Be beamof Louvain-
la-Neuve is taken as reference (RBE = 1). The closed squares and circles correspond to six dierent
visited neutron facilities. The open squares and circles correspond to beams produced at the variable-
energy cyclotron of Louvain-la-Neuve. On the abscissa, the eective energy of the neutron beams is
expressed by their half-value thickness (HVT 5/15) measured in specied conditions. Intestinal crypt
regeneration in mice, after a single fraction irradiation, is taken as the biological system. The 95%
condence intervals are shown. A straight line is tted through the points (squares) corresponding to
neutron beams produced by the (p+Be) reaction. For comparison, the neutron beams produced by the
(d+Be) reaction are represented by circles. (From Ref. 22.)
Copyright 2004 by Taylor & Francis Group, LLC
This value is, in a strict sense, no longer a RBE but a weighting factor (selected of
course on the basis of RBEdata) applied to compensate for RBE dierences/variations. It is
the RBE weighting factor, W
RBE
.
It is current medical practice to select the weighting factors as a function of two main
criteria: (1) late tolerance in normal tissues and (2) in relation to a daily fractionation of 2 Gy
(photons or
60
Co-gamma rays). Therefore among the published RBEvalues, those obtained
for late eects on normal tissues, with 2 Gy of photons per fraction as reference, are the most
relevant for the selection of the RBE weighting factor and could be dened as reference
conditions. The reference RBE is then, to some extent, an average or overall RBE for
late tolerance of normal tissues at risk, evaluated vs. 2 Gy per fraction of photon radiation.
The reference RBE provides a rst indication to select the RBE weighting factor
W
RBE
. However, other factors, which are not always clearly identied, may also inuence
the selection of the RBE weighting factor:
1. Past personal clinical experience
2. Clinical experience reported from other centers
3. Some dierences in time factors
4. Some volume eects related to the beam characteristics, geometry, or technical
conditions
Selection of the RBE weighting factor, W
RBE
, implies the (informed) judgment and
clinical experience and thus the responsibility of the radiation oncologist in charge of the
patient.
This RBE weighting factor has been called clinical RBE in the past; this termi-
nology may be confusing.
3.6. The Biologically Weighted Dose for RBE, D
RBE
The product of the neutron absorbed dose D by W
RBE
is the biologically weighted dose
for RBE, D
RBE
:
D
RBE
DW
RBE
where D and D
RBE
are expressed in Gy but a subscript may be useful to avoid confusion
between the (physical) absorbed dose D and the biologically weighted dose D
RBE
.
In collaborative studies involving several neutron therapy centers, the same bio-
logically weighted doses D
RBE
should be prescribed. If the participating centers use neutron
beams with dierent RBEs, dierent W
RBE
values thus have to be introduced and dierent
(physical) neutron doses have to be prescribed.
In addition to D
RBE
, the delivered absorbed doses and the selected RBE weighting
factors W
RBE
should always be reported, in addition to D
RBE
, to allow for eventual later
reevaluation of the results [23].
A similar approach can be followed with protons and heavy ions. There is an
additional diculty with heavy ions (e.g., carbon ions or neon ions) because the RBE varies
signicantly with depth in the tissues (see Sec. 4.4). With proton beams, there are only
discrete RBE variations (less than 1020%) (see Sec. 4.2).
3.7. Discussion
Absorbed dose has proven to be a fundamental and very useful quantity in radiation therapy
which should always be reported. However, absorbed dose alone cannot predict the
Copyright 2004 by Taylor & Francis Group, LLC
biological eects, and dierent weighting factors have to be introduced depending on the
radiation therapy modality.
In external photon beam therapy, weighting factors, W
a/h
, are currently estimated
based on the a/h model to compensate for dierences in fractionation.
In brachytherapy, with the modern equipment now available, the clinically applied
dose rates vary within large limits. The ICRU and the international brachytherapy com-
munity are recommending weighting factor, W
HDR
and W
PDR
, to compensate for the
dierences in dose rate for high dose-rate (HDR) and pulse dose-rate (PDR) brachytherapy,
respectively.
In hadron therapy, a similar approach and symbolism are presented to take into
account the RBE dierences. RBE values for neutrons are well documented, and reliable
data are now becoming available for protons and heavy ions, on which safe weighting
factors W
RBE
can be based. Besides radiobiological determinations, microdosimetry brings
independently additional information on the radiation quality and improves condence in
both sets of data. Microdosimetry provides an objective description of the radiation quality
at the point of interest under the actual irradiation conditions. Correlation between this set
of information and the experimental RBE results is of great scientic and clinical value in
hadron therapy [24,25].
For the dierent therapy modalities, the products of the absorbed doses by the
biological weighting factors (W
a/h
, W
HDR
, W
PDR
, and W
RBE
) are the biologically weighted
doses (D
a/h
, D
HDR
, D
PDR
, and D
RBE
, respectively). They are expressed in Gy like the
absorbed dose, but, to avoid confusion, subscripts (Gy
a/h
, Gy
HDR
, Gy
PDR
, and Gy
RBE
)
must indicate that the absorbed doses have indeed been weighted.
The biologically weighted dose is a quantity intended to be correlated, as closely as
possible, to the relevant biological/clinical eect. However, in all radiotherapy reports, the
(physical) absorbed doses D and the biological weighting factors (W
a/h
, W
HDR
, W
PDR
,
W
RBE
, etc.) should always be indicated separately and not just the biologically weighted
doses (D
a/h
, D
HDR
, D
PDR
, D
RBE
, etc.).
The weighting factors, W, as introduced above, aim to be a pragmatic approach to
harmonize prescription and reporting of radiation treatments. They are based on actual
clinical practice, current techniques, and available radiobiological data. Their use should be
limited to the technical treatment modality for which they were designed. They are not
universally applicable factors, and they should not be multiplied or combined between
each other.
Finally, it must be stressed again that the selected weighting factors W (and thus the
biologically weighted doses) are based on available radiobiological data and clinical
observation, but they always imply clinical judgment and experience.
4. NONCONVENTIONAL RADIATION THERAPY MODALITIES
(RATIONALE, TECHNICAL ASPECTS, AND SHORT SURVEY
OF CLINICAL DATA)
4.1. Fast Neutron Therapy
Rationale: Radiobiological Basis
Fast neutrons were the rst nonconventional radiations to be used in clinical radiation
therapy. The pioneering work in fast neutron therapy was performed by Stone [26] and
his associates at Berkeley between 1938 and 1943. In the late 1960s, after extensive
radiobiological experiments, a neutron therapy program was initiated at the Hammersmith
Copyright 2004 by Taylor & Francis Group, LLC
Hospital in London and, a few years later, in several centers in Europe and in the United
States [27,28].
The radiobiological rationale for introducing high linear energy transfer (LET) ra-
diation in cancer therapy, as proposed in the 1960s, is still valid and has not been con-
tradicted by more recent radiobiological ndings [3].
The radiobiological and clinical data obtained with fast neutrons are reviewed in the
following section; they deserve a detailed and careful analysis as they constitute the rationale
of the clinical application of any high-LET radiation and, in particular, the justication of
the modern heavy-ion therapy programs.
The dierences in the radiobiological eects produced by low- and high-LET ra-
diations are related to the pattern of energy distribution at the level of the particle
tracks. The distributions of energy (or ionizations) at the microscopic level after low- and
high-LET radiation exposure are compared in Figs. 2 and 3 [16,17,29]. For fast neutrons,
at the energies used in therapy, the energy deposited in a 2-Am sphere by one particle
track (proton) is 50 to 100 times higher than after gamma-ray irradiation (secondary
electrons). Such dierences result in relative biological eectiveness (RBE) of neutrons
higher than unity (ranging between 3 and 5) and in dierent shapes of the doseeect
relationships.
The arguments for using high-LETradiation in cancer therapy are based on these well-
established dierences in energy distribution at the microscopic level and can be summarized
as follows:
A reduction in the oxygen enhancement ratio (OER) with increasing LET
A reduction of the dierences in radiosensitivity related to the position of the cell in
the mitotic cycle
Less repair and thus less clinical relevance of the dierent repair mechanisms
Historically, the oxygen eect was the main rationale for introducing fast neutrons in
radiotherapy. Three sets of data support this rationale:
1. Hypoxic cells are present in the malignant tumors; they result from the fast
proliferation of the cancer cells.
2. Hypoxic cells are three times more radio-resistant than well-oxygenated cells
(OER = 3) for low-LET radiation. The presence of a small percentage of hy-
poxic cells (1% or even 0.1%) can thus make the tumor radio-resistant.
3. The OER decreases when LET increases (Fig. 7) [30]. It decreases to about 1.6
for fast neutrons and is close to unity for alpha particles. OER = 1.3 and 1.0 for
alpha particles of 4 and 2.5 MeV (i.e., for 110 and 160 keV/Am), respectively.
Fig. 8 illustrates howincreasing LET reduces the dierences in radiosensitivity related
to the position of the cells in the mitotic cycle [31]. Cells in stationary phase and in S phase
are signicantly more radio-resistant than mitotic cells.
Lastly, because of the large amount of energy deposited in the critical cellular target by
a single high-LET particle track, repair of sublethal damage is reduced (or less relevant).
High-LET radiation is thus particularly ecient against cells which have a high capability
for repair of sublethal damage. Therefore the dose per fraction is less important with the
high-LET than with the low-LET radiation, which gives the radiation oncologist more
freedom in the selection of the fractionation scheme.
The radiobiological arguments discussed above indicate that the high-LET radiation
could bring a benet in the treatment of some cancer types. However, they also imply
the need for the development of predictive tests, allowing the radiation oncologist to
Copyright 2004 by Taylor & Francis Group, LLC
select the best radiation quality (low or high LET) for a given patient group or individual
patient.
The observations of Battermann et al. [32] on lung metastases indicate that slowly
growing tumors (well dierentiated) could benet from high-LET treatment (Fig. 9). The
clinical results accumulated over more than 25 years conrmed these observations.
Technological Aspects
Many centers, including 15 in Europe (former Soviet Union excluded), have applied fast
neutron therapy using dierent types of generators, which can be schematically distributed
into four groups:
Reactors, using fast neutrons in the beam (limited application)
Low-energy cyclotrons, using mainly incident deuterons with energies ranging
from 13 to 16 MeV (d+Be reaction)
(d+T) generators
High-energy cyclotrons or linear accelerators (d/p+Be reaction)
Figure 7 Survival curves of human kidney cells T1 irradiated under hypoxic and aerobic conditions
with dierent qualities of radiation: (a) 250-kV x-rays (LET of about 1.3 keV Am
1
); (b) 14-MeV
neutrons produced by the (d,T) reaction (LET of about 12 keV Am
1
); (c) 4-MeV a-particles (LET =
110 keV Am
1
); (d) 2.5-MeV a-particles (LET = 166 keV Am
1
). (From Ref. 30.)
Copyright 2004 by Taylor & Francis Group, LLC
Figure 8 Dierences in radiosensitivity with the position of the cells in the mitotic cycle. The
dierences are reduced with increasing LET. Single-hit inactivation coecients (a) for homogeneous
populations of mitotic, G1-phase, and stationary phase Chinese hamster cells irradiated with 220-kVx-
rays and various charged particle beams, as a function of median LET (in keV Am
1
). (FromRef. 31.)
Figure 9 Relation between the RBE of neutrons for regression of lung metastases in patients and
their doubling time. The closed circles correspond to measured values of RBE. The open circles
correspond to values estimated fromirradiation with neutrons only. For the 15-MeVneutrons used in
this study [produced by a (d,T) generator], the RBE for the tolerance of the most important normal
tissues is about 3. As a consequence, neutrons are a good indication (RBE>3) for tumors having
doubling times greater that c100 days. Incontrast, they shouldnot be usedfor rapidly growing tumors.
(From Ref. 32.)
Copyright 2004 by Taylor & Francis Group, LLC
Only for the fourth group can the physical selectivity and the technical conditions be
considered to be sucient (or nearly sucient) for adequate treatments, especially in com-
parison with modern linear accelerators.
In some centers, large clinical programs were completed from which important radio-
biological and clinical conclusions could be derived. In some other centers, the facilities were
shut down abruptly:
In most centers, due to technical diculties; for example, all (d+T) generators are
now shut down.
In other centers, due to patient recruitment problems.
In general, due to suboptimal physical selectivity.
Today, in the majority of centers still active in neutron therapy, the technical con-
ditions are becoming progressively comparable to those in modern photon beamtherapy. In
addition, a few new high-energy facilities have been proposed (e.g., in China, Germany,
Poland, Slovakia, and South Africa).
Clinical Aspects (Short Survey)
Salivary gland tumors and locally extended prostatic adenocarcinoma were proven to be an
indication for fast neutrons.
Salivary Gland Tumors. Neutron beam therapy should be considered as the treat-
ment of choice in patients with unresectable malignant salivary gland tumors or in patients
where radical resection would require facial nerve sacrice (Table 4) [33].
For inoperable primary or recurrent tumors, a randomized cooperative study showed,
at 2 years, a signicant advantage for neutrons compared to photons for loco-regional
control (76% vs. 17%, P<0.005) and a trend towards improved survival (62% vs. 25%).
Ten-year analysis continued to show a striking dierence in loco-regional control (56% for
neutrons vs. 17% for photons, P = 0.009), but both groups experienced a high rate of
metastatic failure (Fig. 10) [34].
Prostatic Adenocarcinoma. For prostatic adenocarcinomas, their typical slow
growth rate and lowcycling fraction provide a logical radiobiological rationale for exploring
neutrons (high LET) in the treatment of this disease.
Table 4 Pooled European Data of Local Control in Advanced Salivary
Gland Tumors
Reference No. of patients Local control
Catterall (1987) 65 48 (74%)
Battermann and Mijnheer (1986) 32 21 (66%)
Duncan et al. (1987) 22 12 (55%)
Prott et al.(1996) 64 39 (61%)
Kova cs et al. (1987) 15 13 (87%)
Kru ll et al.(1995) 74 44 (59%)
Skolyszweski et al. (1982) 3 2
Overall 275 179 (65%)
Source: Ref. 33.
Copyright 2004 by Taylor & Francis Group, LLC
The Radiation Therapy Oncology Group (RTOG), 1997, compared mixed beam (a
combination of photons and neutrons) to conventional photons for locally advanced
prostatic cancer. Loco-regional control as well as survival were signicantly superior after
mixed-beam irradiation (Fig.11) [34].
In 1986, the Neutron Therapy Collaborative Working Group (NTCWG) compared
neutrons (alone) and conventional photons. Asignicant dierence ( P<0.01) was observed
in clinical loco-regional failure, with actuarial 5-year failure rates of 11% vs. 32% after
neutrons and photons, respectively (Fig. 12) [34]. Inclusion of routine posttreatment
biopsies resulted in 5-year histological localregional failure rates of 13% and 32%,
respectively ( P = 0.01).
Due to the long natural history of recurrent prostate cancer, longer follow-up is
required to assess the ultimate impact of the improved local control on survival. However,
the prostate specic antigen (PSA) levels could provide an indication: at 5 years, 17% of
the neutron patients showed elevated PSA levels compared to 45% for the photon patients
( P<0.001).
Late sequelae (mainly large bowel complications) were worse in the neutron-treated
group (11% vs. 3%). However, no colostomy was required in 51 patients treated with a
multileaf collimator at the University of Washington, while 6/38 patients, fromother centers
using movable jaw or xed cone collimator, required colostomy.
The data from Louvain-la-Neuve suggest that mixed neutronphoton therapy is
particularly ecient in patients with unfavorable prognostic factors, such as PSA >20
ng/mL [35].
Other Tumor Sites or Types. For other tumor sites or types, such as slowly growing
soft tissue sarcomas, xed lymph nodes in the cervical area, locally extended antrumtumors,
and some bronchus carcinomas, the available clinical results tend to show a benet with
neutrons. However, they need to be conrmed by randomized studies [5,28,36].
Figure 10 Neutron therapy of salivary gland tumors. Probability of localregional failure for
unresectable salivary gland tumors. Starting values of the curves represent initial localregional
failure rates. (From Ref. 34.)
Copyright 2004 by Taylor & Francis Group, LLC
4.2. Proton Beam Therapy
Proton beam therapy was a signicant step towards better physical selectivity, and it is the
most straightforward approach to improve the ecacy of therapeutic irradiations.
The rst cyclotron was built by Ernest O. Lawrence in 1932, and since 1938, cyclotrons
have been used for patient treatment. In Berkeley, in 1954, the rst human target irradiated
with protons was the pituitary gland with the aim to suppress its function for slowing down
the metastatic development of breast cancer.
Figure 11 Clinical loco-regional control in patients treated with mixed (neutron/photon) beams or
photons only (RTOG randomized trial) for locally extended prostatic adenocarcinoma. (From Ref.
34.)
Figure 12 Actuarial clinical loco-regional failure in patients with locally advanced prostate cancer.
Results of NTCWG trial on prostate. (From Ref. 34.)
Copyright 2004 by Taylor & Francis Group, LLC
Today, proton beam therapy is applied in more than 20 centers worldwide and new
facilities are in preparation (or under discussion). So far, about 30 000 patients have been
treated with protons (Tables 5a and 5b). There is a continuously increasing number of
proton therapy centers in operation and an impressive number of projects under construc-
tion or study.
On the other hand, proton beams have no radiobiological advantages and a constant
relative biological eectiveness (RBE) with respect to photons of 1.01.1 is generally
assumed. The vast clinical experience accumulated with photons can thus be transferred
directly to proton therapy.
Rationale: The Bragg Peak
The depthdose curve in a proton beamexhibits a characteristic shape. The absorbed dose is
low at the entrance (initial plateau), then increases in depth more and more steeply, to
Table 5a Worldwide Charged Particle Patients Totals (HTCOGHadron Therapy Cooperative
Group; Janet Sisterson, 2001)
Who Where What
Date
1st RX
Date
last RX
Recent
patient total
Date
of total
Berkeley 184 California, U.S.A. p 1954 1957 30
Berkeley California, U.S.A. He 1957 1992 2054 Jun. 1991
Uppsala Sweden p 1957 1976 73
Harvard Massachusetts,
U.S.A.
p 1961 8747 Jan. 2001
Dubna Russia p 1967 1974 84
Moscow Russia p 1969 3268 Jun. 2000
St. Petersburg Russia p 1975 1029 Jun. 1998
Berkeley California, U.S.A. HI 1975 1992 433 Jun. 1991
Chiba Japan p 1979 133 Apr. 2000
PMRC, Tsukuba Japan p 1983 629 Jul. 1999
PSI (72 MeV) Switzerland p 1984 3253 Dec. 2000
Dubna Russia p 1987 79 Dec. 2000
Uppsala Sweden p 1989 236 Jun. 2000
Clatterbridge England p 1989 999 Jun. 2000
Loma Linda California, U.S.A. p 1990 5638 Dec. 2000
Louvain-la-Neuve Belgium p 1991 1993 21
Nice France p 1991 1590 Jun. 2000
Orsay France p 1991 1894 Jan. 2001
N.A.C. South Africa p 1993 380 Nov. 2000
MPRI Indiana, U.S.A. p 1993 34 Dec. 1999
UCSF-CNL California, U.S.A. p 1994 284 Jun. 2000
HIMAC, Chiba Japan HI 1994 745 Dec. 1999
TRIUMF Canada p 1995 57 Jun. 2000
PSI (200 MeV) Switzerland p 1996 41 Dec. 1999
GSI Darmstadt Germany HI 1997 72 Jun. 2000
HMI, Berlin Germany p 1998 166 Dec. 2000
NCC, Kashiwa Japan p 1998 35 Jun. 2000
3304 Ions
28,700 Protons
Total 33,104 All particles
Copyright 2004 by Taylor & Francis Group, LLC
reach a maximum at the level of the Bragg peak (Fig. 13). The depth of the Bragg peak in
the tissues depends on energy. There is no dose beyond the depth of the Bragg peak and thus
full sparing of the tissues behind the target volume.
The Bragg peak has to be spread over the depth occupied by the planning target
volume [spread-out Bragg peak (SOBP)] (Fig. 14).
Technological Aspects: Beam Delivery
In the past, proton therapy was performed with complex physics machines, adapted to
clinical needs, not always available full time, often expensive to maintain, and dicult to
tune. Today, several commercial companies oer turn key equipment for proton therapy,
adapted to the needs (or to the nancial limitations) of the center. It is likely that this trend
will develop further.
The establishment of the Northeast Proton Therapy Center (NPTC) in Boston was
obviously the trigger of the movement, but such a fast proliferation was certainly not
expected [37].
In general, a new radiotherapy technique may be considered to have gained its place
among the other ones when the equipment can simply be purchased commercially. This is
obviously now the case with proton therapy.
From a technical point of view, two methods can be used to deliver proton beam
therapy: passive scattering and scanning beam (Fig. 15). The use of a scanning beam is a
more complex approach, but it allows close matching of the treated volume with the PTV. In
Table 5b Charged Particle New Facilities (HTCOGHadron Therapy Cooperative Group;
Janet Sisterson, 2001)
Institution Place Type 1st
INFN-LNS, Catania Italy p 2001
NPTC (Harvard) Massachusetts, U.S.A. p 2001
Hyogo Japan p, ion 2001
NAC, Faure South Africa p 2001
Tsukuba Japan 2001
Wakasa Bay Japan 2002
Bratislava Slovakia p, ion 2003
IMP, Lanzhou P.R. China CAr 2003
Shizuoka Cancer Center Japan 2003
Rinecker, Munich Germany p 2003?
CGMH, northern Taiwan Taiwan p 2001?
Erlangen Germany p 2002?
CNAO, Milan and Pavia Italy p, ion 2004?
Heidelberg Germany p, ion 2006
AUSTRON Austria p, ion ?
Beijing China p ?
Central Italy Italy p ?
Clatterbridge England p ?
TOP project ISS Rome Italy p ?
Three projects in Moscow Russia p ?
Krakow Poland p ?
Proton Development N.A. Inc. Illinois, U.S.A. p ?
PTCA, IBA United States p ?
Copyright 2004 by Taylor & Francis Group, LLC
Figure 13 Depthdose curves for proton beams of dierent energy. The position of the Bragg peak
depends on energy and can thus be adjusted according to the clinical requirements. (From PSI,
Villigen, courtesy of Pedroni and Scheib.)
Figure 14 Proton beam irradiation of a deep-seated large tumor. Single Bragg peaks of dierent
energy are combined, in adequate proportions, to obtain a homogeneous dose distribution at the
level of the SOBP. The depthdose curve of a photon beam, shown for comparison, is inferior
compared to the proton curve. However, an optimized multield photon treatment allows to reach
better irradiation conditions. (From Ref. 43.)
Copyright 2004 by Taylor & Francis Group, LLC
Europe, the Paul Scherrer Institute (PSI), Villigen, Switzerland has developed (and is
planning to further improve) a highly sophisticated beamscanning systemin order to exploit
all the physical advantages of the proton beams. The scanning beam technique will be dealt
with in more detail in the section on heavy ions (Sec. 4.4).
Of course, like with photons, several intersecting proton beams can be used in order
to further improve the nal dose distribution. Fixed beams or an isocentric gantry can
be used.
Clinical Aspects (Short Survey)
The clinical results available so far can be summarized as follows [2,5].
Uveal Melanoma. Charged particle beams and, in particular, protons are ideal for
treating intraocular lesions since they can be made to deposit their absorbed dose in the
target volume, while signicantly limiting the irradiation of the noninvolved ocular and
orbital structures (Fig. 16).
Large series of patients with uveal melanoma were treated with protons in several
centers worldwide. The Massachusetts General Hospital/Harvard Cyclotron Laboratory
(MGH/HCL) in Boston played a pioneering role, and 2568 uveal melanoma patients were
treated through September 1998 [38].
Local control of the tumor within the treated eye which was 96%and 95%at 60 and 84
months, respectively, was reported. Eye retention probability after proton therapy depends
on tumor size, being 97%, 93%, and 78% for patients with small, intermediate, and large
tumors, respectively. Survival at 5 years of the patients treated with protons or enucleation is
similar (about 80%).
Tumors of the Base of Skull and Cervical Spine. Proton or other charged particle
beams are a treatment of choice for skull base and cervical spine tumors: irradiation can be
Figure 15 Schematic representation of a purely passive spreading system. A set of scatterers
guarantees a at transverse dose distribution over a large eld. Starting with a single Bragg peak, a
ripple lter (in combination with a at energy degrader as range shifter) produces a SOBP with a
homogeneous dose at each depth. A special compensator conforms the distal surface of the radiation
eld with the shape of the planning target volume. (From Ref. 43.)
Copyright 2004 by Taylor & Francis Group, LLC
focused in the target volume, while achieving signicant sparing of the brain, brain stem,
cervical cord, optical nerves, and chiasma.
At HCL/MGH in Boston, 621 patients with chordomas and low-grade chondrosar-
comas of the skull base and cervical spine were treated with protons between 1975 and
January 1998. For skull base tumors, with follow-up ranging from1 to 254 months (median
of 41 months), local recurrence-free survival is signicantly better for chondrosarcomas
than for chordomas. It is 98%at 5 years and 94%at 10 years for chondrosarcomas and 73%
at 5 years and 54% at 10 years for chordomas (Fig. 17) [39]. Overall survival is also
signicantly better: 91% vs. 80% at 5 years and 88% vs. 54% at 10 years, respectively.
For cervical spine tumors, with follow-up ranging from1 to 172 months (median of 36
months), local recurrence-free survival was not signicantly dierent for chondrosarcomas
and chordomas: 54% vs. 69% at 5 years and 54% vs. 48% at 10 years, respectively. The
overall survival at 5 years for chondrosarcomas and chordomas were 48% and 80%,
respectively, but at 10 years, 48% and 33%, respectively [38].
CNS Tumors in Children. Pediatric tumors located in the CNS are particularly
challenging and deserve highly rened techniques of radiation therapy like proton therapy
(Fig. 18) [4042]. The preliminary results from Loma Linda, MGH/HCL, and Centre de
Proton The rapie dOrsay are promising and showan excellent immediate and late tolerance.
The reduced integral dose may be of particular importance in the pediatric population.
Other Localizations. Results of proton beam therapy for other tumor types and
localizations have been reported, in particular, for retinoblastoma and age-related macular
degeneration [2]. In the pioneering centers, patient selection, special care applied to each
treatment, as well as expertise of the multidisciplinary clinical teams make it dicult to
conclude whether the excellent results were due to the protons themselves or reect the
expertise of the dierent teams.
Figure 16 Proton therapy for uveal melanoma. Dose distribution obtained with a beam of 60-
MeV protons with an appropriate spread-out Bragg peak (energy modulated from 1460 MeV).
Transverse section through the center of the eye. The position of the tumor [gross tumor volume
(GTV)] is indicated by the posterior hatched area. Protons allow to obtain a homogeneous dose over
the whole GTV with eective sparing of the normal structures. This implies a great precision in
patient-beam positioning. (Courtesy from PSI, cited in Ref. 3.)
Copyright 2004 by Taylor & Francis Group, LLC
4.3. Pion Therapy
Between 1974 and 1994, 1100 patients were treated with negative pions at three centers.
These particles, which provide a mixture of high- and low-LET components and some
physical selectivity advantage, were very expensive to generate and the clinical results were
unconvincing. All three facilities have now been closed down [5].
4.4. Heavy-Ion Therapy
Introduction: Terminology
Strictly speaking, hadrons include protons, neutrons, and mesons, but, in the context of
hadron therapy, it is common practice to include (charged) ions heavier than protons and
often to exclude neutrons. As seen above (Sec. 4.3), k-mesons are no longer used in therapy.
In the radiation oncology (or radiobiology) community, hadron therapy currently
includes the application of protons, helium, carbon, neon, and argon ions. Carbon, neon, or
argon ions are called heavy ions, while protons and heliumions are called light ions. In
contrast, in the physics community, all ions mentioned above are called light ions
compared to much heavier ions such as, for example, uranium [43].
In the following section, heavy ions include ions heavier than protons (and helium
ions). They combine the advantages of an excellent physical selectivity comparable to that of
protons with the radiobiological advantages of high-LET radiations for some types of
tumors (as discussed for fast neutrons in Sec. 4.1.1). However, as carbon ions are, at the
moment, the only type of heavy ions used in therapy, the next section will deal mainly with
carbon ions.
Rationale: The Bragg Peak and High-LET
Four advantages can be identied for heavy ions in cancer therapy.
Figure 17 Proton beam therapy of tumors of the base of skull. Probability of local control in
patients with chordomas (n = 169) and low-grade chondrosarcomas (n = 165). (From Ref. 39.)
Copyright 2004 by Taylor & Francis Group, LLC
Copyright 2004 by Taylor & Francis Group, LLC
Bragg Peak and Physical Selectivity. The physical selectivity of carbon ions, other
heavy ions, and protons is quite similar. Fig. 19 shows the Bragg peak for carbon ions as a
function of depth [44]. The Bragg peak has to be spread out, as for protons, to fully cover the
planning target volume (PTV). The ratio between the dose at the level of spread-out Bragg
peak (SOBP) and of the initial plateau is compared for carbon ions and protons and for
dierent ions in Figs. 20 and 21 [44,45].
Nuclear fragmentation observed with carbon and other heavy ions could be a
disadvantage as some energy is deposited beyond the primary Bragg peak. This issue is
probably not clinically relevant as the involved absorbed dose level is lowand the fragments
are low-LET particles. The penumbra is somewhat narrower with carbon ions than with
protons as the particles are heavier.
High LET and RBE. More important is the fact that the LETthus also the RBE
increases with depth in the hadron beam (Figs. 22 and 23) [43]. This further increases the
Figure 19 Bragg curves for carbon beams with dierent energies. When increasing the energy, the
range of the particles increases and the Bragg peak occurs at larger depths. Concurrently, the height
of the Bragg peak decreases. (From Ref. 44.)
Figure 18 Comparison of photon and proton dose distributions for the treatment of a cra-
niopharyngioma in a child. Typical planning sections for a large suprasellar craniopharyngioma, in a
3-year-old child, treated with photons (A) or protons (B). The planning target volume (PTV) is
indicated by the hatched area. For photons and protons, four equally weighted beams are used and
the normalization point was chosen at the intersection of the beam axes. For the four proton beams,
the Bragg peak was spread over 4 cm. (From Ref. 41.)
Copyright 2004 by Taylor & Francis Group, LLC
ratio of the biological equivalent doses between the SOBP and the initial plateau (Fig. 24)
[44]. The clinical benet of a higher RBE at the level of the distal part of the SOBP is
illustrated in Fig. 25 [45].
High-LET Advantage. Thirdly, high-LET irradiation is delivered at the level of the
SOBP where the PTV(thus the cancer cell population) is located. Heavy-ion beams are thus
specically eective against some tumor types as discussed for fast neutrons (see Secs. 4.1.1
and 4.1.3).
Repair Capacity. Finally, when fractionated treatment is applied, the high-LET
radiation at the level of the SOBP (i.e., PTV) partly prevents or reduces cell repair. In
contrast, the normal tissues irradiated at the level of the initial plateau are exposed to
Figure 20 Comparison of calculated extended Bragg peaks of protons and carbon ions (
12
C) at
dierent penetration depths. (From Ref. 45.)
Copyright 2004 by Taylor & Francis Group, LLC
Figure 21 Comparison of the biological dose proles for He, C, and Ne ions. For carbon ions,
the ratio of the biological dose at the level of the SOBP and in the initial plateau is maximized and
the fragmentation tail behind the Bragg peak remains relatively low. These arguments lead to select
carbon ions as most appropriate for heavy-ion beam therapy. (From Ref. 44.)
Figure 22 Comparison of interactions of 1-MeV protons and 1 MeV/u C
6+
ions with water. (From
Ref. 43.)
Copyright 2004 by Taylor & Francis Group, LLC
low(er)-LET radiation and can thus fully benet from repair mechanisms if fractionated
irradiation is used.
Technological AspectsBeam Delivery
The technical problems related to beam delivery are, to some extent, similar to those
encountered with protons [46]. They are, however, more complex with carbon (heavier) ions
and involve more cumbersome and more expensive equipment.
The scanning beam allows a better shaping of the treated volume. This is especially
important when the PTVis close to a critical normal structure. This advantage holds for any
shape of the PTV. The principle of the scanning beam technique is presented in Fig. 26 [45].
A rotating isocentric gantry is, or will be, installed in many of the carbon therapy
facilities. The possibility of orientating the beam in any suitable direction is obviously an
advantage for accurate patient-beam positioning and for allowing the selection of the
optimum treatment plan. Isocentric gantries are used in all modern photon linear accel-
erators. For making any comparison relevant and reliable, it is thus important that carbon-
ion therapy could be delivered with the same geometrical possibilities and with the same
accuracy as the classical photon treatments. Mounting an isocentric gantry for carbon-ion
therapy raises complex mechanical problems and requires huge and expensive magnets (Fig.
27) [43].
Clinical Results (Short Survey)
The Berkeley Program. The rst heavy-ion therapy program was initiated at
Berkeley, and 433 patients were treated between 1975 and 1992 [47]. The program was
limited by the availability of the machine and its complexity (which resulted in many un-
scheduled down times), and, as a consequence, there was a patient recruitment problem.
Figure 23 Variation of RBE as a function of depth in the carbon-ion beam used for clinical
applications at HIMAC, Chiba, Japan (carbon-12, 290 MeV/u, SOBP 60 mm). The biological system
is the well-codied intestinal crypt regeneration in mice. The selected criterion is 20 regenerating
crypts per circumference after a single fraction irradiation. RBE determinations were performed at
the beginning, middle, and end of the SOBP and at the level of the initial plateau. The doseeect
relationship for cobalt-60 is indicated for comparison. An estimation of the LET is presented for
each depth where biological determinations were made. (From Gueulette, unpublished.)
Copyright 2004 by Taylor & Francis Group, LLC
Nevertheless, a great deal of valuable radiobiological and clinical information was obtained
(Table 6).
A summary of the clinical results obtained with neon ions in Berkeley is presented in
Table 7 [4,47]. Some fast neutron therapy results are also presented. Although the recruit-
ments are not comparable, it should be pointed out that tumor types or sites for which an
advantage was found with neon ions are those for which an advantage was also found with
fast neutrons. This suggests a specic high-LET eect.
Figure 24 (a) Comparison of the physical and biological depthdose curves for a carbon beam
with a 10-cm SOBP. The biological dose is obtained by multiplying, at each depth, the absorbed dose
by a weighting factor W
RBE
which takes into account the RBE increase. The highest RBE is obtained
in the deepest part of the PTV where the dose is delivered by a single Bragg peak (thus with high
LET and RBE). In contrast, the dose in the proximal part of the PTV is due to many initial-
plateau contributions (thus with a lower LET and RBE). (b) In order to cover the whole PTV with a
homogeneous biological dose, the shape of the physical dose has to be adapted taking into account the
RBE variation over the PTV. (From Ref. 44.)
Copyright 2004 by Taylor & Francis Group, LLC
Figure 25 Treatment of a tumor of the base of skull with a carbon-ion beam. Comparison of the
(physical) absorbed dose and of the biological dose (i.e., the dose weighted for the RBE variation
in the carbon beam). The treatment is carefully planned in order to obtain a homogeneous biological
dose to the PTV. For the same biological dose to the PTV, the normal tissues in the initial plateau
receive a lower dose. (From Ref. 45.)
Figure 26 Principle of the irradiation of a PTV of any shape using the scanning-beam technique.
The PTV is stratied into a series of layers, perpendicular to the incident beam. The dierent layers
are irradiated successively with narrow pencil beams of equal range, scanned over the surface
(dierent shape at each depth) by two deecting magnets. (From Ref. 45.)
Copyright 2004 by Taylor & Francis Group, LLC
The HIMAC-NIRS Program. After the shutdown of the therapy facility at Berkeley,
the National Institute for Radiological Sciences (NIRS) in Chiba, Japan initiated a therapy
program with carbon ions. In 1993, construction and installation of the Heavy Ion Medical
Accelerator in Chiba (HIMAC) was completed. This was the worlds rst carbon-ion
accelerator fully dedicated to medical use in a hospital environment. The HIMAC project
was part of the Japanese governments 10-year plan against cancer.
There was no limitation on machine time, and it was thus possible to study common
tumors on a large scale. A total number of 1297 patients were treated between 1994 and
August 2002.
Some of the clinical results are summarized in Table 8. More detailed information on
lung cancers [nonsmall cell lung cancer (NSCLC)] is given in Table 9, in particular, about
patient selection and treatment conditions [48]. The possibility of hypofractionation has
been investigated especially for lung and liver cancers with no increase in toxicity.
The clinical experience previously gained in the same center, during the last decades,
with fast neutrons was reported to be useful to identify the patients suitable for high-LET
Figure 27 Isocentric gantry for irradiation of the PTV from any direction. The use of an isocentric
gantry for carbon-ion therapy requires huge and expensive magnets. (From Ref. 43.)
Table 6 Summary of the Clinical Results Obtained with Helium Ions and Neon Ions at Berkeley
Tumor site
Local control rate with
Helium ions Neon ions
Conventional
treatment
Salivary gland 80% (10 patients) 28% (188 patients)
a
Nasopharynx paranasal sinus 53% (13 patients) 63% (21 patients) 21% (97 patients)
(UCSF)
Sarcoma 65% (17 patients) 45% (24 patients) 28%
a
Prostate 100% (9 patients) 6070%
a
Lung 39% (18 patients) 2240% (UCSF)
Brain/glioblastoma
(median survival)
17 months
(13 patients)
912 months (UCSF,
RTOG, NCOG)
a
Literature review.
Source: Ref. 47.
Copyright 2004 by Taylor & Francis Group, LLC
therapy and to select the best weighting factor W
RBE
to take into account the RBE eects
(see Secs. 3.5 and 3.6).
A second medical facility dedicated to cancer treatment has been opened at Hyogo,
Japan in 2001.
The European Experience. In Europe, the rst patients were treated with carbon ions
at the Gesellschaft fu r Schwerionenforschung (GSI), Darmstadt, Germany at the end of
Table 7 Comparison of Some Clinical Results Obtained with Neon Ions and Fast Neutrons
(i.e., External Beam and High-LET Radiation)
Tumor site or type
Local control rates after treatment with
a
Fast neutrons
(pooled data)
Neon ions
(Berkeley)
Salivary gland tumors 67% (24%) 80% (28%)
Paranasal sinuses 67% 63% (21%)
Fixed cervical lymph nodes 69% (55%)
Sarcomas 53% (38%) 45% (28%)
Prostatic adenocarcinoma 77% (31%) 100% (6070%)
a
For comparison, the best estimates of local control rates currently obtained with conventional photon beam
therapy are given in parentheses.
Source: Ref. 2.
Table 8 Results of Carbon-Ion Beam Therapy at NIRS (Treatments: June 1994August 2001)
Protocol Phase Material
Treatment
(fractions/week)
No. of
patients
Response
rate (%)
a
Two-year
local
control (%)
b
Three-year
survival (%)
H&N-1 I/II Locally advanced 18/6 17 73 80 44
H&N-2 I/II Locally advanced 16/4 19 68 71 44
H&N-3 II Locally advanced 16/4 134 52 61 42
Lung-1 I/II Stage I (peripheral) 18/6 47(+1) 54 62 88
Lung-2 I/II Stage I (peripheral) 9/3 34 85 86 65
Lung-3 I/II Stage I (Hillar) 9/3 10 90 100
Lung-4 II Stage I (peripheral) 9/3 50(+1) 65 100 73
Lung-6 I/II Stage I (peripheral) 4/1 18 67
Liver-1 I/II T2
f
4 MONO 15/5 24(+1) 75 79 50
Liver-2 I/II T2
f
4 MONO 4
f
12/1
f
3 82(+4) 72 83 45
Liver-3 II T2
f
4 MONO 4/1 11 55
Prostate-1 I/II B2
f
C C ion+hormone 35 100 94
Prostate-2 I/II A2
f
C C ion+hormone 61 100 97
Prostate-3 II TIC
f
C C ion+hormone 47
Uterus-1 I/II III = Iva(ACC) 24/6 30 100 50 40
Uterus-2 I/II Iib-Iva(SCC) 24/6 14 100 67 36
Uterus-3 I/II Iib-Iva(SCC) 20/5 11 100
Uterus I/II Advanced (Adenoca) 20/5 12 100 38 39
Bone/Soft-1 1/II unresectable 16/4 57(+7) 36 77 50
Bone/Soft-2 II unresectable 16/4 30(+1) 57
a
Response rate: percent of tumors with >50% reduction in size.
b
Local control rate: percent of tumors with no evidence of local recurrence or relapse.
Source: Ref. 48.
Copyright 2004 by Taylor & Francis Group, LLC
1997. Because of the strict limitation in the availability of the machine time, patients with
dicult tumor localizations were selected. For these patients, full advantage could be taken
from the scanning beam and the energy modulation system [45].
At the end of 2002, 156 patients were treated, a large proportion of themfor a tumor of
the base of skull. At the end of 2001, for chondrosarcomas (23 pts), no local recurrence was
observed, while, for chordomas (44 pts), the local control rate was 87% (Kraft, personal
communication). The same dierence in prognosis between the two histologies is found as
with protons (Sec. 4.2.2).
Taking into account the clinical experience reported fromBerkeley and Chiba and the
clinical results of the pilot study at the GSI in Darmstadt, both conrming the radio-
biological and physical arguments in favor of heavy ions, the German Cancer Research
Center decided to build a carbon-ion therapy facility at the University Hospital in
Heidelberg. The technical and medical experience gained at the GSI, mainly by Kraft and
his team, will help to select the optimumconguration. The possibility to treat with protons
at the same facility is also planned: this would allowthe medical teamto evaluate the benet
of high-LET vs. low-LET radiations delivered with the best possible physical selectivity. It
would of course be of great interest to have the possibility to treat either with low- or high-
LET radiations depending on the tumor characteristics and patient conditions, in the same
center, under the same technical conditions and with the same medical team.
Four other hadron-therapy centers are planned in Europe: MED-AUSTRON in
Wiener Neustadt in Austria, Espace de Traitement Oncologique par Ions Le gers dans le
Cadre Europe en (ETOILE), TERAin north of Italy, and the Swedish project in Stockholm.
AEuropean network ENLIGHT has been initiated in 2002 by the European Society
for Therapeutic Radiology and Oncology (ESTRO) and supported by the European Com-
mission (EC). The goal of the European Network on Light Ion Therapy (ENLIGHT) is to
coordinate the ve European programs and, in particular, facilitate the mobility of scientists
and exchange information. The scientic coordination of ENLIGHT is the responsibility of
A. Wambersie (ESTRO-Brussels), J.-P. Ge rard (Lyon), and R. Po tter (Vienna).
4.5. Boron Neutron Capture Therapy
Only a short review of BNCT is presented here. The rationale for BNCT is to reach a
physical selectivity at the cellular level [5,49,50]. BNCT is a bimodal, binary therapy where
Table 9 Results of Carbon-Ion Beam Therapy at NIRS: Clinical Studies for Stage I Nonsmall
Cell Lung Cancer (NSCLC)
Protocol no.
Lung-1
(9303)
Lung-2
(9701)
Lung-3
(9801)
Lung-4
(9802)
Lung-6
(0001)
Phase I/II I/II I/II II I/II
Period of the study Oct. 1994 to
Sep. 1997
Sep. 1997 to
Feb. 1999
Apr. 1998 to
present
Apr. 1999 to
Dec. 2000
Dec. 2000 to
present
Tumor type All type
a
Peripheral Central Peripheral Peripheral
Total Dose (GyE) 59.4
f
95.4 68.4
f
79.2 57.6
f
64.8 72 54 or 60
Fraction/weeks (xed) 18f/6w 9f/3w 9f/3w 9f/3w 4f/1w
# Pats (# Tumors) 47(48) 34(34) 15(15) 50(51) 35(35)
Adenoca/SCC/Large cellca 26/22/0 18/15/1 13/2/0 32/19/0 23/11/1
a
All type includes both peripheral and central type of tumor.
Source: Ref. 48.
Copyright 2004 by Taylor & Francis Group, LLC
an alpha particle is produced at the cellular target by the capture reaction of boron
compounds and thermal/epithermal neutrons.
BNCT using thermal neutron beams was started in the United States in 1951 at the
Massachusetts Institute of Technology (MIT) and at the Brookhaven National Laboratory
(BNL). The clinical results were very poor. The technique was introduced in Japan by
Hatanaka in 1968, and some promising results were obtained.
BNCT was restarted in the United States in September 1994 at Brookhaven National
Laboratory and shortly thereafter at MIT using epithermal neutron beams (BNL trials
ended in 1999 after the treatment of 53 patients but continued at MIT); these programs are
supported by the Department of Energy. Forty patients were treated by the end of 1997. In
Europe, the European Commission supports a BNCT programin Petten, The Netherlands.
The three rst patients were treated in 1997. The thermal neutron beam program continues
in Japan.
So far, the only available sources of epithermal neutrons, with sucient output, are
nuclear reactors. However, construction of compact proton accelerators, producing epi-
thermal neutrons of adequate energy (e.g., 2.5-MeV protons on a lithium target), has been
envisaged.
The main expected advantage is that such an accelerator could be hospital-based, and
thus fractionation (of the irradiation and drug administration) could be optimized since the
machine would be available 24 hr a day. In addition, patient positioning could be made
easier and more accurate.
Lastly, BNCT is used today in combination with fast neutron therapy in some centers
such as Seattle, Essen, and Orle ans. Boron is incorporated in the tumor cells; it captures
thermal neutrons produced in the body by the fast neutron beam. Combination of BNCT
and external photon beam therapy has also been suggested [51].
In conclusion, boron neutron capture therapy (BNCT) is still in an experimental
phase. Although the rationale is particularly attractive (selectivity at the cellular level), it is
dicult to draw conclusions from the available clinical data.
5. SUMMARY AND CONCLUSIONS: PRESENT TRENDS
AND EXPECTATIONS IN RADIATION ONCOLOGY
5.1. Radiation Oncology Over the Past Decades
Historically, the progress in radiation therapy has been linked mainly to technological
developments. The physical selectivity of the irradiations was signicantly increased when
200-kV x-rays were progressively replaced by cobalt-60, betatrons, and linear accelerators.
As a consequence, the clinical results were dramatically improved.
More recent technical developments of the electron linear accelerators are impressive.
The reliability, mechanical stability, beam delivery, and collimation systems of the new
generation of accelerators have reached a high level of quality [52]. These machines now
allow irradiation of nearly any target volume with reduced irradiation of the surrounding
organs at risk (OAR).
This progress in the treatment machines could be fully exploited because of the
dramatic and impressive developments in imaging (such as CT, MRI, and PET). In addition,
our better knowledge of the natural history of cancer helps in selecting the optimum
treatment strategy.
For the future, (at least) three main approaches can be identied for improving the
eectiveness of radiation therapy.
Copyright 2004 by Taylor & Francis Group, LLC
5.2. Optimizing Photon Techniques
Powerful irradiation techniques are now available or under development for photons:
intensity-modulated radiation therapy (IMRT), tomotherapy, gamma knife, cyber-
knife, radiosurgery, etc. These techniques make it possible to irradiate any volume of any
complex shape with minimum irradiation of critical normal structures.
This technical progress has been made possible by:
1. Advances in modern imaging techniques, which have made available complete
and accurate anatomical information, including the location of the dierent
target volumes and organs at risk.
2. Developments in medical physics and dosimetry that provide powerful and fast
3-D treatment planning, including inverse planning, and take into account
heterogeneities. Fast computation is absolutely required for IMRT.
3. Engineering improvements that allow full and reliable control of the machine
and beam delivery system and patient immobilization.
These new techniques are becoming available in more and more hospitals.
However, challenges remain.
If one matches the treated volume (TV) and the planning target volume (PTV) as
closely as possible, the conformity index will approach unity [6,7,53]. Dierent technical
approaches are used to reach this goal; among them is IMRT. However, when optimizing
the treatment parameters to improve the conformity index in IMRT, other indices can
change adversely. For example, the irradiated volume increases (integral dose) and the
homogeneity throughout the PTV may become worse. The compromises to be made vary
with the treatment modality, and the true therapeutic gain of the newer modalities still needs
to be evaluated.
This brings us to an important question: has the ecacy of photon beam therapy now
reached a plateau (at least as far as physical selectivity is concerned)? If this is the case, little
additional clinical benet can be expected from further technical improvements with
photons, and one has to search for other beam or radiation qualities to further improve
the eectiveness of radiation therapy.
This is an important point to consider; however, there is no general consensus as yet.
5.3. Proton Beam Therapy
The introduction of proton beams aims at further improving the physical selectivity of the
irradiation. The clinical results obtained by the pioneers in proton therapy, with machines in
physics laboratories, were suciently convincing to justify building and buying dedicated
hospital-based proton machines.
The challenge for the next few years will be to conrm, on a larger scale, the results of
the pioneers and to conrm that their excellent results were actually due to the protons
themselves and not only to the expertise and the motivation of the teams.
The benet expected from the better physical selectivity of protons will have to be
evaluated in comparison with conventional photon beam therapy, which remains the
reference radiation therapy modality.
When making this comparison, one must remain aware of the modern developments in
photon therapy such as conformal therapy (3DCRT), IMRT, tomotherapy, gamma knife,
stereotactic x-ray techniques, iodine-125 (and palladium-103) seeds and plaques, etc.
There are, however, some (rare) tumor localizations for which nothing can compete
with the excellent physical selectivity of protons: for example, tumors adjacent to the spinal
Copyright 2004 by Taylor & Francis Group, LLC
cord, adjacent or invading the brain stem, tumors of the base of skull, and certain (especially
pediatric) brain tumors (Fig. 18).
Another advantage of protons, relative to the best photon techniques, is a reduction of
the integral dose and of the irradiated volumes. This factor could inuence the risk of radio-
induced cancers, although, so far, epidemiological evidence is still lacking.
Proton therapy equipment is now commercially available from several companies.
This is a sign that protons have gained their place in the radiation therapy arsenal. Indeed,
the number of proton therapy facilities increases worldwide, both the hospital-based,
therapy-dedicated facilities and the facilities in physics laboratories adapted for medical
applications.
5.4. High-LET Radiations, Fast Neutrons, and Hadrons
The third approach is the introduction of another type of radiation quality: high-LET
radiation. Clinical experience with neutrons has demonstrated that high-LETradiations are
superior to low-LETradiations for some tumor types or sites. Fast neutrons were indeed the
rst high-LET radiations to be applied clinically (see Sec. 4.1). Although in the rst studies
they were applied in suboptimal conditions froma technical or dose distributions point of
view, their advantage for some types of tumors is well established, particularly for slowly
growing, well-dierentiated tumors. Randomized trials have indeed shown their superiority
over conventional photons for salivary gland tumors and prostatic adenocarcinomas.
Unfortunately, several neutron therapy centers were closed for technical reasons or
patient recruitment diculties. However, the centers applying neutron therapy today benet
from the same conditions of safety, reliability, and physical selectivity as photon therapy
centers.
Heavy ions combine the benet of the high physical selectivity of proton beams and the
biological advantage of high-LET radiation for some tumor types. They appear today to be
one of the most promising radiation therapy modalities when the clinical indication is
correctly selected (some tumor types and/or localizations).
When comparing the relative merits of the dierent types of radiation used in therapy,
two criteria have to be considered: the physical selectivity and the radiation quality (actually
the LET). They are schematically presented in Fig. 28 [54]. Improving the physical selectivity
is per se always an advantage. In contrast, selection between low and high LET depends on
the tumor characteristics and should be based on the histology, grade, doubling time, etc. It
is a pure radiobiological and medical issue.
Reliable criteria to identify the (individual) patients suitable for high-LET radiation
therapy need to be developed. At present, the available criteria are derived (mainly) fromthe
clinical fast neutron experience. It can be expected that novel approaches based on modern
techniques involving molecular biology or gene identication may provide appropriate and
still missing information. They may also provide information on the susceptibility or risk for
secondary radio-induced cancer.
5.5. Reporting the Treatments
It is important for the future that protocols and results of hadron therapy be reported and
analyzed using the terminology, denitions, concepts, and approaches currently in use for
the other radiation therapy modalities [7,53].
As fractionated photon beamtherapy is the reference radiation therapy modality, used
for more than 80%of the patients, the terminology in use for photons should be applied for
Copyright 2004 by Taylor & Francis Group, LLC
the other techniques. Recommendations for reporting are contained in several ICRU
reports [6,7,53]. It is, however, recognized that each technique has its specicity and may
sometimes require introduction of specic concepts or denitions.
5.6. Global Approach and Collaboration in Cancer Treatment Strategy
The subject of this chapter is to report the rationale, achievements, and expectation of ions in
cancer therapy. To illustrate their potential, the proven advantages of low- and high-LET
ions are discussed in detail.
In the eld of radiation therapy, it is recognized that there are other promising
approaches to improve the eciency of the treatment such as better individual adaptation of
fractionation or association with drugs to modulate the radiation sensitivity. Combination
of radiotherapy with the two other classical methods of cancer treatment, surgery, and
chemotherapy may also be optimized.
Figure 28 Schematic presentation of the relative situation of the dierent types of radiations used
in therapy. Two criteria are considered: the physical selectivity and the LET (or radiobiological
properties). For the low-LET radiations, the physical selectivity was improved from the historical
200-kV x-rays to cobalt-60 gamma rays and the modern linacs. Even with the linacs today, sig-
nicant improvement is continuously achieved (IMRT, etc.). Among the low-LET radiation, the
proton beams have the best physical characteristics, but one of the issues is the proportion of patients
who will benet from proton irradiation. A similar scale can be drawn for high-LET radiation: the
heavy-ion beams have a physical selectivity similar to protons. Selection between low- and high-LET
radiation is a biological/medical problem; it depends on the tumor characteristics, and reliable criteria
still need to be established (see text). (From Ref. 54.)
Copyright 2004 by Taylor & Francis Group, LLC
Last but not the least, the progress in gene therapy, immunotherapy, and other
approaches based on novel molecular biology techniques has to be kept in mind as well.
The recent impressive developments in ion beam therapy should not be regarded as
being in competition with the other approaches mentioned above. In contrast, the
advantages of combinations should be considered and a potential enhancement of the
respective eectiveness may reasonably be expected [5].
ACKNOWLEDGMENT
This work was supported by the European Commission, Directorate General Research-
Quality of Life and Management of Living Resources, contract: QLGI-CT-2002-01574.
REFERENCES
1. Levin, V.; Meghzifene, A.; Izewska, J.; Tatsuzaki, H. IAEA, International Atomic Energy
Agency, Wagramerstrasse 5, A-1400: Vienna, Austria, 2001; 25 pp.
2. Wambersie, A.; Chauvel, P.; Gademann, G.; Ge rard, J.-P.; Sealy, R. EULIMA, Final Report,
European Commission, Rue de la Loi, 200, B-1049: Brussels, Belgium, 1992; 2 pp.
3. Tubiana, M.; Dutreix, J.; Wambersie, A. Introduction to Radiobiology; Taylor & Francis:
London, 1990.
4. R., Po tter, T., Auberger, A., Wambersie., Eds.; HadronsA challenge for high-precision ra-
diotherapy. Strahlenther. Onkol. 1999, 175 (Suppl. II),1128.
5. Wambersie A.; Gahbauer R. Radiochim. Acta 2001, 89, 245.
6. International Commission on Radiation Units and Measurements (ICRU). Prescribing, Re-
cording and Reporting Photon Beam Therapy, ICRU Report 50; Bethesda, Maryland, 1993.
7. International Commission on Radiation Units and Measurements (ICRU). Prescribing, Re-
cording and Reporting Photon Beam Therapy (Supplement to ICRU Report 50), ICRU Report
62; Bethesda, Maryland, 1999.
8. Wambersie A.; T. Auberger; Gahbauer R.; Jones D.; Po tter R. Strahlenther. Onkol. 1999, 175,
122.
9. Gahbauer R. Strahlenther. Onkol. 1999, 175, 121.
10. Wambersie A. International Atomic Energy Agency (IAEA), International Nuclear Data
Committee, Nuclear Data for Production of Therapeutic Radioisotopes, INDC (NDS)-432,
IAEA: Vienna, 2002.
11. International Commission on Radiation Units and Measurements (ICRU). Quantities and
Units in Radiation Protection Dosimetry, ICRU Report 51; Bethesda, Maryland, 1993.
12. International Commission on Radiation Units and Measurements (ICRU). Fundamental
Quantities and Units for Ionizing Radiation, ICRU Report 60, Bethesda, Maryland, 1998.
13. Wambersie A.; Menzel H.G.; Gahbauer R.A.; Jones D.T.L.; Michael B.D.; Paretzke H. Radiat.
Prot. Dosim. 2002, 99, 445.
14. Wambersie A.; Menzel H.G.; Gahbauer R.A.; DeLuca P.; Whitmore G. In Progress in Radio-
Oncology VII; Kogelnik, H.D., Lucas P., Seldmayer, F., Eds.; Monduzzi Editore: Bologna,
Italy, 2002, 361 pp.
15. Goodhead D.T. Nuclear and Atomic Data for Radiotherapy and Related Radiobiology; IAEA:
Vienna, 1987; 37 pp.
16. Barendsen G.W. Radiat. Res. 1994, 139, 257.
17. Pihet, P. Etude microdosimetrique de faisceaux de neutrons de haute energie. Applications dos-
imetriques et radiobiologiques, Thesis; Universite Catholique de Louvain: Louvain-la-Neuve,
Belgium, 1989.
Copyright 2004 by Taylor & Francis Group, LLC
18. International Commission on Radiological Protection (ICRP) Report of the RBESubcommittee
to the International Commission on Radiological Protection and the International Commission
on Radiation Units and Measurements. Health Phys. 1963, 9, 357.
19. International Commission on Radiation Units and Measurements (ICRU). Quantitative Con-
cepts and Dosimetry in Radiobiology, ICRU Report 30, Bethesda, Maryland, 1979.
20. Field S.B.; Hornsey S. In High-LET Radiations in Clinical Radiotherapy; Barendsen, G.W.,
Broerse, J.J., Breur, K. Eds.; Pergamon Press: Oxford, 1979; 181 pp.
21. van der Kogel, A.J. Late eects of radiation on the spinal cord: dose eect relationships and
pathogenesis (Thesis); University of Amsterdam, Publication of the Radiobiological Institute
TNO: Rijswijk, The Netherlands, 1979.
22. Gueulette J.; Menzel H.G.; Pihet P.; Wambersie A. Fast neutrons and high-LET particles
in cancer therapy. In Recent Results in Cancer Research. Engenhart-Cabillic, R., Wambersie, A.
Eds.; Springer-Verlag: Heidelberg, 1998; Vol. 150, 31.
23. Wambersie A.; Menzel H.G. Strahlenther. Onkol. 1993, 169,57.
24. Menzel H.G.; Pihet P.; Wambersie A. Int. J. Radiat. Biol. 1990, 57, 865.
25. Pihet P.; Menzel H.G.; Schmidt R.; Beauduin M.; Wambersie A. Radiat. Prot. Dosim. 1990,
31, 437.
26. Stone R.S. Am. J. Roentgenol. 1948, 59, 771.
27. Wambersie A.; Barendsen G.W.; Breteau N. J. Eur. Radiothe r. 1984, 5, 248.
28. Engenhart-Cabillic, R.,Wambersie, A. Eds.; Fast Neutrons and High-LET Particles in Cancer
Therapy, Recent Results in Cancer Research; Springer: Heidelberg, 1998; 1209.
29. Pihet P.; Norman C.; Gueulette J.; Menzel H.G.; Wambersie A. Radiophysique; XXIV Congre` s
de la Socie te Franc aise des Physiciens dHo pital. Tours, Socie te Franc aise des Physiciens
dHo pital, Fondation Curie, 26 Rue dUlm, F-75248: Paris, France, 1985; 269 pp.
30. Barendsen G.W. Curr. Top. Radiat. Res. Q. 1968, 4, 293 (North Holland, Amsterdam).
31. Chapman J.D. In Radiation Biology in Cancer Research; Meyn, R.E.,Withers, H.R. Eds.;Raven
Press: New York, 1981; 21 pp.
32. Battermann J.J.; Breur K.; Hart G.A.M.; van Peperzeel H.A. Eur. J. Cancer 1981, 17, 539.
33. Kru ll A.; Schwarz R.; Brackrock S.; Engenhart-Cabillic R.; Huber P.; Prott F.J.; Breteau N.;
Favre A.; Lessel A.; Koppe H.; Auberger T. In Fast Neutrons and High-LET Particles in Cancer
Therapy; Engenhart-Cabillic, R.,Wambersie, A. Eds.; Springer: Heidelberg, 1998; 88 pp.
34. Lindsley K.L.; Cho P.; Stelzer K.J.; Koh W.J.; Austin-Seymour M.; Russel K.J.; Laramore
T.W.; Grin T.W. Bull. Cancer, Radiothe r. 1996, 83 (Suppl. 1), 78s.
35. Scalliet, P.; Remouchamps, V.; Lhoas, F.; Van Glabbeke, M.; Curran, D.; Ledent, T.; Wam-
bersie, A.; Richard, F.; Van Cangh, P. Proceedings of the XXIX PTCOG Meeting, DKFZ,
Heidelberg and GSI, Darmstadt, Germany, 1998.
36. Breteau, N., Le Bourgeois, J.-P., Barendsen, G.W., Stannard, C.E., Rosenwald, J.-C., Wam-
bersie, A. Eds.; Hadrons in radiation therapy. Bull. Cancer, Radiothe r. 1996, 83 (Suppl. 1),
1s230s.
37. Suit H.D. Int. J. Radiat. Oncol. Biol. Phys. 2002, 53, 798.
38. Munzenrider J.E. Strahlenther. Onkol. 1999, 175 (Suppl. II), 68.
39. Munzenrider J.E.; Liebsch N.J. Strahlenther. Onkol. 1999, 175 (Suppl. II), 57.
40. Habrand J.L.; Mammar H.; Ferrand R.; Pontvert D.; Bondiau P.-Y.; Kalifa C.; Zucker J.-M.
Strahlenther. Onkol. 1999, 175 (Suppl. II), 91.
41. Wambersie A.; Gre goire A.; Brucher J.-M. Int. J. Radiat. Oncol. Biol. Phys. 1992, 22, 275.
42. Jones D.T.L.; Schreuder A.N.; Symons J.E.; de Kock E.A.; Vernimmen F.J.A.; Stannard C.E.;
Wilson J.; Schmitt G. Strahlenther. Onkol. 1999, 175 (Suppl. I), 30.
43. Regler, M.; Benedikt, M.; Poljanc, K. CERN Accelerator School, Seville, Spain, October 2001,
Hephy-PUB-757/023, 2002.
44. Grundinger, U., Ed.; Nachrichten GSI, Gesellschaft fu r Scherionenforschung mbH, D-64291
Darmstadt, Germany, 1993; 1193.
45. Kraft G. The Physics of Highly and Multiple Charged Ions; Currel, F.J., Ed., Kluwer Academic
Publisher, 2002 Chapter 10, pages 145192.
Copyright 2004 by Taylor & Francis Group, LLC
46. Jones D.T.L. Radiochim. Acta 2001, 89, 235.
47. Castro J.R.; Chen G.T.Y.; Blakeley E.A. Radiat. Res. 1985, 104 (Suppl. 8), S263.
48. Tsujii H.; Morita S.; Miyamoto T.; Mizoe J.-E.; Kamada T.; Kato H.; Tsuji H.; Yamada S.;
Yamamoto N.; Murata H. Annual Report NIRS; National Institute of Radiological Sciences:
Chiba-shi, Japan, 2002; 61 pp.
49. Gahbauer R.; Gupta N.; Blue T.; Goodman J.; Barth R.; Grecula J.; Soloway A.H.; Sauerwein
W.; Wambersie A. Fast Neutrons and High-LETParticles in Cancer Therapy; Engenhart-Cabillic,
A., Wambersie, A. Eds.; Springer: Heidelberg, 1998; 183 pp.
50. Rorer, R.; Wambersie, A.; Whitmore, G.; Zamenhof, R. Current status of neutron capture
therapy, International Atomic Energy Agency, IAEA-TECDOC-12233, 2001.
51. Barth, R.F.; Grecula, J.C.; Yang, W.; Rotaru, J.H.; Nawrocky, M.; Gupta, N.; Albertson, B.J.;
Ferketich, A.M.; Moeschberger, M.L.; Coderre, J.; Rofstad, E.K. Int. J. Radiat. Oncol. Biol.
Phys. in press.
52. Wambersie A. Radiochim. Acta 2001, 89, 255.
53. International Commission on Radiation Units and Measurements (ICRU). Prescribing, Re-
cording and Reporting Electron Beam Therapy, ICRU Report 71, Bethesda, Maryland, 2004.
54. Wambersie A. Progress in Radio-Oncology V; Kogelnik, H.D. Ed.; Monduzzi, Eds: Bologna,
Italy, 1995; 685 pp.
Copyright 2004 by Taylor & Francis Group, LLC
26
Food Irradiation
Jo zsef Farkas
Szent Istvan University, Budapest, Hungary
1. INTRODUCTION
Food irradiation is the process of exposing food, either prepackaged or in bulk, to
controlled levels of certain types of ionizing radiation to increase storage life of food,
reduce postharvest food losses, and inactivate specic food-borne pathogenic organisms.
Food irradiation is one of the most thoroughly and intensively investigated methods of
food preservation.
The ionizing radiation applied in food irradiation is limited to high-energy electro-
magnetic radiation (gamma rays or x-rays) with energies up to 5 MeV or high-energy
accelerated electrons with energies up to 10 MeV. These radiations are chosen because
1. They produce the desired eects with respect to the food.
2. They do not induce radioactivity in foods or their packaging materials.
3. They are available in quantities and at costs that allow practical uses of the
process.
Other kinds of ionizing radiation, in some respect, do not suit the requirements of
food irradiation. Except for dierence in penetrability, electromagnetic ionizing radiations
and accelerated electrons are equivalent in food irradiation and can be interchangeably
used.
Radiation treatment can be considered as one type of nonthermal processing of
food because even at the largest absorbed radiation dose (see Sec. 4.11) to be applied,
which is about 50 kGy, the amount of energy is equivalent to 50 J. At this dose level, if all
the absorbed radiation energy would be degraded to heat, the temperature of a high-
moisture food, thermodynamically roughly equivalent to water, would rise about 12jC.
Because of the negligible heating eect, irradiation treatment is able, e.g., to kill the cells of
microbes contaminating frozen foods without thawing them up, or, those in fresh
commodities without changing the original physical state of the product.
Irradiation can be applied through any packaging materials including those that
cannot withstand heat. This also means that radiation treatment can be performed after
packaging, thus avoiding recontamination or reinfestation of the product.
The useful eects of ionizing radiation as a food processing means are summarized in
Table 1. In addition to the preservative eects listed in Table 1, in some cases even
Copyright 2004 by Taylor & Francis Group, LLC
improvement of certain functional or sensory quality characteristics of food can be
achieved with irradiation.
The basic process is the application of a prescribed amount of ionizing radiation to
foods, plus the eventual use of certain other procedures that may be needed to accomplish
the purpose of the processing.
Food irradiation is a very complex topic and has an enormous literature. Apart from
thousands of journal articles and proceedings of large number of international conferences
and panel meetings, its state of the art has been extensively reviewed during the course of
decades by a number of noteworthy books [15]. The present brief chapter mainly focuses
on the principles and some potential applications of food irradiation and refers to some
most recent research and developments in these regards.
1.1. Principles of Radiation Sources
Two basic types of radiation sources can satisfy the requirement of industrial use of food
irradiation:
1. Machine sources such as electron accelerators and those converting acceler-
ated electron beams to x-ray photons. Accelerated electrons have low penetra-
bility. Thus they cannot meet all the goals of food irradiation. The 10 MeV elec-
trons, the highest energy level of electron irradiation presently recommended
by the Codex Alimentarius [6], can penetrate food with typically about 4-cm
thickness.
2. Long half-life, man-made radionuclides emitting gamma rays. The most readily
available is cobalt-60. Much more limited is the use of cesium-137. Gamma rays
and x-rays have highly penetrating characteristics; thus, they can be used to treat
food in large containers.
Typical irradiation facilities consist of a process chamber containing the radiation
source, some sort of conveyor systems to transport products inside and outside the
shielding walls, and sophisticated control and safety systems. Irradiation facilities are built
with several layers of redundant protection to detect equipment malfunctions and protect
employees from accidental exposure. Technical details depend on the type of irradiation.
Typical processing parameters are compared in Table 2 [7].
Table 1 Useful Eects of Irradiation as a Food Processing Treatment
Eects Results
Inhibition of sprouting of tubers
and bulbs
Increased storability
Decrease of after-ripening and delaying
senescence of some fruits and vegetables
Increased shelf life
Killing or sterilizing stored
product insects
Insect disinfestation of food
Inactivation of parasites transmissible
by food
Prevention of food-borne
parasitic diseases
Inactivation of food-borne
microorganisms
Microbial decontamination of food: Increased
shelf life and/or prevention of food poisoning
Copyright 2004 by Taylor & Francis Group, LLC
1.2. Typical Dose Requirements
The technological feasibility of a food irradiation treatment depends on how much
irradiation the food withstands without adversely changing its qualities, i.e., how much
useful eect can be achieved without signicant change to the chemical composition,
nutritional value, and sensory properties of the product. Generally, there is a minimum
dose requirement. Whether every mass element of a food requires irradiation will depend
Table 2 Comparison of Typical Processing Parameters
Gamma X-ray E-beam
Typical source
power
3.5 MCi 25 kW 35 kW
Typical processing
speed
12 tonnes/
hr at 4 kGy
10 tonnes/
hr at 4 kGy
10 tonnes/
hr at 4 kGy
Source energy 1.33 MeV 5 MeV 510 MeV
Penetration depth 80100 cm 80100 cm 810 cm
Dose homogeneity High High Low
Dose rate Low High Higher
Best application Bulk processing of large
boxes or palletized
product in shipping
cartons in a warehouse
environment
Bulk processing large
boxes or palletized
product in shipping
cartons in a warehouse
environment
Sequential processing of
primary or secondary
packaged product
in-line or at-line
Source: Ref. 7.
Table 3 Dose Requirements of Various Applications
of Food Irradiation
Application Dose Requirement (kGy)
Inhibition of sprouting of
potatoes and onions
0.030.12
Insect disinfestation of seed
products, ours, fresh and
dried fruits, etc.
0.20.8
Parasite disinfestation of
meat and other foods
0.13.0
Radurization of perishable
food items (fruits, vegetables,
meat, poultry, sh)
0.510
Radicidation of frozen meat,
poultry, eggs and other
foods and feeds
3.010
Reduction or elimination of
microbial population in dry
food ingredients (spices, starch,
enzyme preparations, etc.)
3.010
Radappertization of meat, poultry,
and shery products
2560
Copyright 2004 by Taylor & Francis Group, LLC
on the purpose of the treatment. In some cases, irradiation of the surface will suce; in
others, the entire food must receive the minimum dose.
Ranges of dose requirement of various applications are listed in Table 3. The details
of the requirements for each food must be specically considered.
2. BIOLOGICAL EFFECTS OF IONIZING RADIATION
Chemical and biological eects of ionizing radiation are thought to occur through two
main mechanisms: direct interaction of the radiation with food components and living
cells in materials exposed to it, and indirect action from radiolytic products, such as the
radicals formed from water molecules (see Chap. 12).
The primary target of biological eects appears to be the DNA (see Chap. 15),
although eect on the cytoplasmic membrane may also play a role [8]. Ionizing radiation
aects DNA in a number of ways. It causes chemical changes in specic nucleotide bases;
it causes single-strand breaks; and it causes double-strand breaks in a dose-dependent
manner.
2.1. Radiation Sensitivity of Stored Product Insects and Food-Borne
Parasites
The radiation dose required to kill an insect depends on the species and a number of other
factors such as age, sex, and stage of development [9]. In general, radiation sensitivity is
highest at the egg stage and the lowest at the adult stage of development. Fruit ies are the
most radiation-sensitive insect pests while the moths are the most resistant ones.
Radiation eects on parasitic protozoa and helminths are associated with loss of
infectivity, loss of pathogenicity, interruption or prevention of completion of life cycle,
and death of parasites [10].
2.2. Radiation Resistance of Food-Borne Microorganisms
The actual number of cells or percentage of microbial population that will be killed by an
absorbed radiation dose depends also on various factors such as the inherent resistance of
the particular organism, the growth stage, as well as environmental factors such as tem-
Table 4 D
10
Values (kGy) of Some Nonsporeforming Bacteria
Bacteria Nonfrozen Food Frozen Food
Vibrio spp. 0.020.14 0.040.44
Yersinia enterocolitica 0.040.21 0.200.39
Campylobacter jejuni 0.080.20 0.180.32
Aeromonas hydrophila 0.110.19 0.210.34
Shigella spp. 0.220.40 0.220.41
Escherichia coli (incl. O157:H7) 0.240.43 0.300.98
Staphylococcus aureus 0.260.57 0.290.45
Salmonella spp. 0.180.92 0.371.28
Listeria monocytogenes 0.201.0 0.521.4
Copyright 2004 by Taylor & Francis Group, LLC
perature during irradiation, oxygen presence, and water content. Vegetative cells of path-
ogenic bacteria, the main concern in many food-borne infections and intoxication, are
relatively radiation sensitive, including both long-time recognized pathogens as well as
emerging or new ones.
Table 4 shows the ranges of decimal reduction doses (D
10
values) of the most
important nonsporeforming pathogens determined in various atmospheres and foods,
summarized from publications by a large number of laboratories [11,12]. Because of the
eect of irradiation temperature on the radiation resistance of microorganisms, a detailed
knowledge of product temperature prole is critical for eective ionizing radiation
pasteurization of foods, particularly meat products [12,13]. Some nonpathogenic micro-
organisms and the bacterial spores in general are more resistant to radiation than those
listed in Table 4. Viruses, mycotoxins produced by certain types of molds, and prion
particles thought to be responsible for bovine spongiform encephalpathy (BSE) in cattle
are highly resistant to irradiation [11].
3. RADIATION-INDUCED CHEMICAL CHANGES IN MAIN CONSTITUENTS
OF FOODS
More or less, water is present in almost all foods. Therefore the radiolysis of water is of
particular interest in food irradiation. However, this subject is amply dealt with in Chap.
12. The end products of water radiolysis,
.
OH, e
aq
, and
.
H, are very reactive transient
species. The hydroxyl radical is a powerful oxidizing agent whereas the hydrated electron
is a strong reducing agent. The stable end-products such as hydrogen and hydrogen
peroxide are of less signicance because they are largely consumed with their respective
reactions with e
aq
and
.
OH [14].
The presence or absence of oxygen during irradiation can have an important
inuence on the course of radiation-induced changes of food components. When foods
are irradiated, oxygen can add to some of the radicals produced to form peroxy radicals,
.
RO
2
. Through this reaction, the small amount of oxygen normally present in a food can
be quickly consumed during irradiation. Because diusion of oxygen from the atmosphere
is slow, electron irradiation, due to its high dose rate, can create an anoxic condition in the
food, whereas gamma sources have much lower dose rates than electron accelerators. Thus
anoxic conditions are not necessarily created unless the food is gamma irradiated in an
oxygen-free package. This might create a dose rate eect that is actually an oxygen eect.
The radiolysis of water is pH-dependent, too. Thus pH inuences the result of a radiation
treatment.
The temperature during irradiation also inuences the chemical changes. Freezing
can have a strong protective eect because reactive intermediates of water radiolysis are
trapped in the ice of frozen foods and are thus kept from reacting with other food
components. During thawing of frozen, irradiated food, the radiolytic products of water
apparently react preferentially with each other rather than with other food components.
Thus freezing of food has a certain protective eect on some radiation-sensitive vitamins
[15] and decreases the associated chemical changes, e.g., with the formation of volatiles
producing o-avor in some irradiated foods [16].
Because in multicomponent systems such as foodstus, a mutual protection of
dierent components is exerted, irradiation does not cause much chemical change in
foods.
Copyright 2004 by Taylor & Francis Group, LLC
Minerals and trace elements cannot be aected under the process conditions of food
irradiation.
3.1. Effects on Main Organic Components
Apart from water, the major constituents of most foods are carbohydrates, proteins, and
lipids. These organic components of foods are less sensitive to irradiation than they are in
their pure solutions as a single-component system.
Irradiation of sugars results in the formation of low levels of radiolytic products
mostly derived from the reaction with hydroxyl radicals by hydrogen abstraction from the
CH bonds, resulting in products such as sugar acids, keto sugars, and deoxy compounds
[17]. When polysaccharides are irradiated, the reactions observed with monosaccharides
can occur, and, additionally, the glycosidic bonds that connect the monosaccharide units
can be broken. This reduces the degree of polymerization. Especially in case of starch and
pectin, this causes changes in the physical properties of the foods that contain them.
Changes in properties such as viscosity, mechanical strength, swelling, and solubility are
likely to reduce their functionality in food; however, sometimes, the changes improve a
particular function.
Major in vitro reactions of irradiated amino acids involve decarboxylation and
oxidative (if oxygen is present) or reductive (if anoxia exists) deamination [14]. The
sulfur amino acids such as cysteine, cystine, and methionine may act as free radical scav-
engers, thus ameliorating the degradative eects on other components of food. However,
their breakdown also generates end products such as hydrogen sulde, which has
undesirable sensory eects. Radiation damage to constituent amino acids in irradiated
food is very limited. With some 20 amino acids, the total range of possible reaction
products is great, but the quantitative eect on proteins in foods during irradiation is
small. Consequently, enzyme inactivation is also insignicant. Actually, radiation-
sterilized foods destined for long-term storage must receive a heat treatment (blanching)
in addition to the radiation treatment to prevent enzymatic spoilage (see Sec. 4.11). In
some irradiated plant tissues, the in vivo enzyme activities may increase as a result of
release, or diusion, through leaky membranes more easily reaching their hitherto
unavailable substrates. When meat is irradiated at high doses (1050 kGy), radiation-
induced aggregation of some proteins may occur, which leads to decreased protein
solubility [14].
Upon irradiation of fats, the formation of a multitude of products is possible af-
ter primary ionization and excitation, and deprotonation followed by various dimer-
ization, disproportionation reactions, dissociations, or decarboxylation. It is generally
assumed that irradiation in the presence of oxygen leads to accelerated autooxidation
of lipids, and that the pathways are the same as in light-induced or metal-catalyzed
autooxidation.
The irradiation of unsaturated fatty acids in foods predominantly results in the
formation of a hydroperoxyl radical and then the formation of a hydroperoxide. The
hydroperoxides are generally unstable in foods and break down to form mainly carbonyl
compounds, many of which have low odor threshold, and contribute to the rancid notes
often detected when fat-rich foods are irradiated [18]. In the absence of air, their formation
is limited.
In studies on radiation eects on cholesterol in meat [19] and egg powder [20],
elevated levels of the same oxides were found that are known to result from
autooxidation during storage of unirradiated foods. Vacuum packaging or addition
Copyright 2004 by Taylor & Francis Group, LLC
of antioxidants can largely prevent such formation of cholesterol oxidation products
[21,22].
Radiation eects on vitamins will be discussed in Sec. 5.
4. PRACTICAL APPLICATIONS OF FOOD IRRADIATION
4.1. Control of Sprouting and Germination of Vegetable Crops
Inhibition of sprouting of various vegetable crops at low doses was one of the earliest
extensively studied application possibility of food irradiation. Most important from these
opportunities is the control of sprouting to extend the storability of tubers and bulbs [23].
These studies included the response of varieties/cultivars to radiation treatment, optimal
dose, dose-rate eect, time interval between harvest and irradiation treatment for ecacy,
susceptibility of irradiated batches to storage rots, biochemical mechanisms underlying
sprout inhibition, inuence of storage conditions, and the eect of irradiation on
technological properties inuencing the utilization of irradiated crops for processed
products.
The dose required to inhibit sprouting of onions, shallots, and garlics is 0.030.12 kGy.
For good sprout control of tubers such as potatoes and yams, somewhat higher doses, 0.08
0.14 kGy, are required. Because of decreased wound-healing ability after irradiation, doses
in excess of 0.150.2 kGy may induce increased microbial rot in storage [24].
An important factor determining the ecacy of radiation treatment of tubers and
bulbs is the time delay between harvest and irradiation. The sprout inhibition is most
pronounced if the irradiation of tubers and bulbs is applied shortly after harvest, when they
are still in their dormancy stage. However, the dormancy period may vary among cultivars
and cropping season, and is also dependent on the postharvest storage temperature.
In onion bulbs, if some growth of inner buds takes place already before irradiation,
the treatment causes the death and discoloration of the inner buds. The area of this
discoloration depends on the size of the inner buds at the time of irradiation. Pilot scale
studies in Hungary have shown that the yield of unirradiated onions prepared for drying
after storage was 29% of the original compared to 55% in irradiated onions. The
dehydrated onions prepared from irradiated bulbs had better quality than did the controls.
No discoloration of the inner buds occurred in bulbs of cultivar Also go di irradiated at
0.05 kGy and stored up to 8 months [25]. Similar results were reported earlier from Egypt
[26] whereas some studies in the United States have indicated that inner-bud discoloration
of irradiated onions may lower the quality of dehydrated onion slices or powder prepared
from them [27].
In irradiated potatoes, especially in some varieties and as a function of cultivating
conditions of the raw material, after-cooking darkening may occur. This discoloration is
attributed to formation of ferric-phenolic complexes. This phenomenon depends on the
iron content, and is related to increased polyphenol formation and reduced citric acid
levels, which are inuenced by agronomic and climatic factors. Various technological
measures have been developed to prevent this after-cooking darkening [23].
Other types of radiation-induced discoloration in potatoes have also been reported
in some cases and measures to avoid them have been studied [28,29].
Because irradiation interferes with the natural wound-healing process, it is an
important prerequisite of successful application of irradiation that mature tubers with
fully developed periderm are suitable for radiation processing and tubers must be properly
cured from harvesting and handling injuries before irradiation.
Copyright 2004 by Taylor & Francis Group, LLC
Because potatoes are good source of vitamin C, it is important to point out that
irradiation does not adversely aect the vitamin C levels [23,30]. Although some ascorbic
acid is converted into dehydroascorbic acid on irradiation, the latter is also biologically
active.
In Japan, where the use of chemical sprout suppressants is not permitted, commer-
cial irradiation of potatoes has been introduced in 1973 in Shihoro, Hokkaido, where an
industrial scale irradiator has been processing about 15,00020,000 tons of potatoes
annually [31,32]. The success of this system is due in large measure to careful handling
of the product before and after treatment.
Since the late 1990s, commercial irradiation of bulb crops, particularly garlic, has
shown a steady increase using
60
Co irradiators in the major garlic-producing provinces of
China [23].
A commercial demonstration facility for irradiation of onions having a processing
capacity of 10 tons/hr was reported in 1999 to be under construction in Nasik District,
Maharashtra State, India [23].
Pilot-scale irradiation and consumer acceptance studies with tubers and bulbs have
been performed with positive results already in the 1970s in a number of other countries,
e.g., Argentina, Bangladesh, Chile, German Democratic Republic, Hungary, Israel, the
Philippines, Thailand, and Uruguay, showing the techno-economic feasibility of irradi-
ation of these crops. However, in countries under temperate climates, cool storage by
circulating cold outside air, together with the use of chemical sprout inhibitors, provide
inexpensive and satisfactory sprout control with reasonably good product quality for both
industrial processing and household consumption. Thus industrial interest in commercial
implementation in Western countries for sprout inhibition is still very low, basically
because it involves high capital investment for relatively low-price crops, perceived
consumer opposition to irradiated foods, regulation and trade limitations, and labeling
requirements. However, the use of chemicals is coming under increasing scrutiny from the
viewpoints of environmental pollution and health risk from residues left in the products.
Therefore it is likely that their continued use will be restricted.
In yams, an important crop in tropical climates, neither chemical sprout inhibitors
nor cool storage are eective for long-term storage. On the other hand, irradiation would
provide an eective alternative for postharvest treatment.
Controlling the germination of malting barley is an interesting potential application
of low-dose irradiation. Doses of 0.250.5 kGy applied to air-dried barley do not prevent
the emergence of shoot tips and tendrils during malting but markedly retard the root
growth. In this way, high-quality malt can be obtained while the losses resulting from root
growth are reduced [33]. Because this eect of radiation processing persist for at least 7
months, the above radiation treatment applied before the barley is put into storage has the
added benet of destroying insect pests that may be present in the grain.
4.2. Insect Control in Stored Foods
Radiation disinfestation of stored food oers a viable alternative to chemical disinfestation
without adverse eects on the product quality. The use of methyl bromide, the only broad-
spectrum fumigant used for the disinfestation of stored products, was anticipated to be
phased out by around 2001 [34], which highlights the urgency of an alternative treatment.
Disinfestation of stored food by irradiation was extensively studied already in the
1960s and 1970s [9,35], and it has shown that radiation disinfestation can be eciently
applied to almost all dried foods.
Copyright 2004 by Taylor & Francis Group, LLC
The Codex Alimentarius Commission recommends a dose of 1 kGy for killing insects
in all food and agricultural products [6]. However, if Good Manufacturing Practices
(GMP) and Good Irradiation Practices (GIP) are followed, according to subsequent
recommendations published by the International Consultative Group on Food Irradiation
(ICGFI), radiation disinfestation of stored product commodities should be achievable at
doses up to 0.5 kGy [34]. The radiation doses used to control various stored product
insects are summarized in Table 5.
Presently, many commodities have to be fumigated more than once to control
insects. Irradiation by a single treatment sterilizes or kill, depending on the dose and the
time interval allowed after irradiation, all developmental stages of the common insect
pests, including eggs deposited inside grains or even the weevil that may lodge deep inside
the seed of the mango.
With the availability of inexpensive, convenient, and easily applied pest control
methods based on application of pesticides, irradiation disinfestation has not yet received
priority. Nevertheless, radiation disinfestation was performed on an industrial scale in the
Table 5 Radiation Doses Used to Control Stored Product Insects
Species Stage Dose (kGy)
Coleoptera
Sitophilus oryzae All 0.16
S. granarius All 0.16
S. zeamais All 0.16
Tribolium castaneum All 0.20
T. confusum All 0.20
T. destructor All 0.20
T. madeus All 0.20
Rhyzopertha dominica Larvae 0.25
Latheticus oryzae Adults 0.20
Oryzaephilus surinamensis All 0.20
O. mercator All 0.20
Callosobruchus analis All 0.20
C. chinensis All 0.20
C. maculatus All 0.20
Bruchus rumanus All 0.40
Bruchidius incarnatus All 0.40
Trogoderma granarium All 0.25
Dermestes maculatus All 0.50
Lasioderma serricorne All 0.50
Nerobia rupes All 0.30
Araecerus faciculatus All 0.75
Lepidoptera
Anagastus kuehniella Larvae, pupae 0.60
Plodia interpunctella Larvae 0.45
Cadra cautella Larvae, pupae 0.45
Sitotroga cerealella All 0.60
Nemapogon granellus All 0.60
Source: Ref. 34.
Copyright 2004 by Taylor & Francis Group, LLC
Soviet Union, where an electron irradiation plant to treat imported grains went into
operation in 1980 at Port Odessa and some 400,000 tonnes/year of grain were successfully
treated by two electron accelerators. This facility is not currently in use in the Ukraine,
after the collapse of the Soviet Union.
The doses applied for the widely used microbial decontamination by irradiation of
spices, dried herbs, and dry vegetable seasonings (see Sec. 4.9) are much higher than the
disinfestation doses. Thus radiation decontamination of these commodities is more than
enough to kill also any insects eventually infesting them.
Dried fruits, vegetables, and nuts, as well as dried sh, an important source of
protein in many developing countries, are also good candidates for radiation disinfesta-
tion. Application of 0.20.7 kGy doses, if they have been suitably packaged to prevent
reinfestation, can eliminate the insect problem from these products that cannot be
eectively disinfested by either chemical or physical means other than irradiation.
As a residue-free physical treatment, irradiation does not provide any protection
from insects that might reinfest the product after postirradiation storage. Therefore it is of
paramount importance that proper care is needed to store, transport, and market
irradiated products in insect-proof containers or packages [34].
4.3. Irradiation as a Quarantine Treatment
Radiation treatment can signicantly contribute to the variety of means for quarantine
disinfestation in the international trade to prevent the importation of invasive nonnative
insects with food and agricultural commodities that can harbor them. The use of
irradiation disinfestation in quarantine treatment has great potential especially against
fruit ies. A number of past quarantine treatments have been recently prohibited, e.g.,
fumigation with ethylene dibromide.
Irradiation is the fastest among quarantine treatments available, although it cannot
be used to treat so large loads at once as can be carried out with fumigation or by cold
storage. Irradiation treatment is an eective alternative for many types of fresh produce
because it can be used on riper fruit and on fruit that cannot tolerate, e.g., heat treatment.
However, one characteristic of irradiation quarantine treatment that needs specic con-
sideration by regulatory/inspecting agencies is that irradiation does not provide signicant
acute mortality (within 48 h) at doses tolerated by fresh agricultural produce; thus
identication methods for irradiated commodities and insects, or acceptance of certi-
cation as proof of adequate irradiation, are needed [36].
Quarantine treatment doses of several pests (mainly fruit ies) supported by adequate
research vary between 0.07 and 0.225 kGy [36]. Such low doses are tolerated by most fruits.
However, because of the diculties in evaluating ecacy by acute mortality and uncer-
tainties of dose measurements, radiation quarantine treatment research shows some
inconsistencies, and thus further work is required. Nevertheless, motivated by banning
fumigation with ethylene dibromide (EDB), several semicommercial trial shipments of
tropical fruits have been irradiated and marketed in the continental United States since
1986, and a commercial linear accelerator e-beam/converted x-ray facility was built in
Hawaii to treat fruit and began operating in August 2000 [36].
4.4. Parasite Disinfection
Various animal parasites are associated with certain raw and partially processed food, not
only in the developing part of the world but also in developed countries. In the United
Copyright 2004 by Taylor & Francis Group, LLC
States, a recent survey by the Centers for Disease Control and Prevention estimated that
there are 2.5 million cases annually due to food- and beverage-borne parasites [37].
There are a number of traditional and new control measures to protect consumers
from food-borne parasites. Ionizing radiation is one of the new technologies to control or
limit the impact of a number of food-borne parasites on public health [10].
Although relatively high doses (46 kGy) are required to kill food-borne parasites,
much lower doses are adequate to prevent their reproduction and maturation resulting in
loss of infectivity. Table 6 summarizes the eect of ionizing radiation on the most
important parasites that may be associated with shery products and meats [38]. Thus
the parasitic roundworm Trichinella spiralis, which causes trichinosis, the pork and beef
tapeworms, the protozoon in pork responsible for toxoplasmois, and various ukes that
infest sh, can be rendered noninfective by low-dose radiation treatment. Doses of
Table 6 The Eect of Irradiation on Parasites
Parasite Mode of Infection Dose (kGy) Eect of Irradiation
Clonorchis spp. Chinese liver uke, occurs
in raw sh
0.15 In vitro minimum
eective dose
Opistorchis viverrini Liver uke found in
contaminated raw,
pickled or smoked sh
0.1 In vitro minimum
eective dose
Paragonimus spp. Parasitic worm found
in crabs and
craysh in Asia
0.1 In vitro minimum
eective dose
Gnathostoma
spinigerum
Parasitic worm found in
raw, undercooked
or fermented sh
7 Reduces worm recovery
rate in mice
Angiostrongylus
cantonensis
Parasitic worm found in
uncooked molluscs, shellsh
2 Minimum eective dose
Anisakis spp. Nematode is ingested if
sh is eaten raw or
slightly salted
210 Reduces infectivity
of larvae
Trichinella spiralis Nematode occurs in raw or
inadequately cooked pork
0.3 Minimum eective dose
031 FDA permitted dose
to control trichina
in pork
Toxoplasma gondii Consumption of
undercooked
meat or poultry;
or contact with
infected animals
0.7 Minimum eective
dose for
fresh pork
Cysticercus bovis
(Taenia saginata,
in man)
Tapeworm found
in uncooked
or undercooked beef;
causes taeniasis
0.3 Preliminary minimum
eective dose
Cysticercus cellulosae
(Taenia solium,
in man)
Tapeworm found in pork 0.3 Preliminary minimum
eective dose
Source: Ref. 38.
Copyright 2004 by Taylor & Francis Group, LLC
irradiation applied to reduce microbial load (to be discussed in further chapters) are
suciently high to control also many parasites.
4.5. Extension of Shelf Life of Fresh Fruits and Vegetables
Both ripening and reducing fungal decay in many species of harvested fruits and vege-
tables have been studied in relation to food irradiation alone or in combination with other
treatments [39].
Appreciable delay of ripening and consequent enhancement of shelf life have been
noted in some tropical fruits such as bananas, plantain, and mangoes after low-dose
irradiation (0.20.7 kGy) [40,41]. Maximum delay of ripening has been observed with
fruits of lower maturity. However, these feasible levels of doses are close to those that
induce phytotoxicity, and many factors (varietal dierences, the fruits stage of maturity at
the time of irradiation, etc.) make the outcome of the treatment uncertain.
More promising is the delay of maturation in cultivated mushroom (Agaricus sp.)
where a dose of 12 kGy when applied soon after picking at the closed button stage can
extend the shelf life by oppressing cap opening and stalk elongation. At optimal doses,
darkening of the gills, cap and stalk, shriveling, and surface mold development are also
inhibited [42,43].
The feasibility of irradiation of temperate fruits, particularly pome and stone fruits, is
generally limited by adverse eects of texture (softening of the tissue) due to the radiation
sensitivity of pectin. Among temperate fruits, strawberries seem to be the most amenable
to radiation processing, which extends the market life of this perishable fruit by control-
ling molds that otherwise quickly cause decay, particularly Botrytis and Rhizopus rots. The
dose requirement for this decay control is 1.52 kGy, with an upper limit of 3 kGy,
without unduly aecting quality [40]. Thereby, strawberries can be harvested when fully
ripened and can be transported and displayed for longer periods while maintaining
desirable sensory qualities longer than nonirradiated strawberries.
From intact vegetable crops, apart from tuber, bulb, and root crops, tomatoes and
asparagus show potential for radiation treatment. The development of fungal decay in
tomatoes caused by Alternaria, Botrytis, or Rhizopus spp. can be controlled using doses of
approx. 3 kGy. However, softening and the loss of characteristic avor may occur. Doses
of 0.1 kGy and above result in a delay of ripening of tomatoes [43].
Irradiation (12 kGy) of fresh asparagus in combination with wrapping with
polyvinyl chloride (PVC) lm can double the shelf life under refrigeration with no eect
on color and avor [44].
4.6. Irradiation of Fresh Meat and Poultry
Increase of shelf life under refrigeration and control of pathogenic nonsporeforming
bacteria in fresh meat and poultry can be achieved by a 13 kGy dose. Doses for
irradiation are selected under the consideration of threshold dose levels for sensory
changes (o-odor), which depends on the type of animal meat (Table 7) [45]. O-odor
is due to the generation of volatile compounds from lipids and nitrogenous compounds
formed by the reaction of these constituents with the reactive species produced by the
radiolysis of water.
The considerable extension of the shelf life by doses as low as 1 kGy under aerobic
conditions is due to the radiation sensitivity of Pseudomonas spp. [46] and other Gram-
negative bacteria mainly associated with spoilage of fresh meat and poultry. Also, a
Copyright 2004 by Taylor & Francis Group, LLC
signicant reduction in the number of pathogens including Salmonella, Campylobacter,
Yersinia, Staphylococcus, Escherichia coli, and Listeria can be achieved in fresh or frozen
meat and poultry by radiation pasteurization [47,48]. The coliforms, including E. coli
O157:H7, are quite sensitive to ionizing radiation; thus a standard irradiation pasteuriza-
tion process with 1.52.5 kGy causes its 56 logs reduction in ground beef, and it would
reduce also Salmonella and Listeria monocytogenes by 24 logs [49]. In other recent studies
in Hungary, in comparison of gamma irradiation and high hydrostatic pressure-induced
eects on minced beef, radiation pasteurization proved to be superior in maintaining
quality [50]. Ground meat poses particular food safety concerns because the grinding
process can spread pathogens present on the meats surface throughout the product.
Recent results in the United States [51] showed that irradiation of pork trim with a
processing dose of min. 1.25 kGy and max. 1.9 kGy prior to production of dry fermented
sausage such as pepperoni can yield a sausage product with quality indicators closely
resembling those of the traditional dry sausage while still providing the 5-log reduction of E.
coli O157:H7 as mandated by the U.S. Department of Agriculture (USDA). A heat
treatment of the pepperoni sticks performed on the nished product to an internal temper-
ature of 60jC as decontamination alternative signicantly altered the texture and color.
Because freezing confers greater radiation resistance of microorganisms, higher
doses are needed under freezing [52] (see also Table 4) but the sensory results can be better.
One of the major concerns in irradiating meats is its eect of meat quality. Minced
beef is susceptible to lipid oxidation. However, our own studies in cooperation with the
University College Cork, Ireland [53], demonstrated that increased endogenous alpha-
tocopherol concentration by dietary vitamin E supplementation prior to slaughter of the
animals, and combination of this feeding supplementation with a rosemary extract as
natural antioxidant additive to the ground beef, resulted in a color-stabilizing and lipid-
protecting eect. Studies by Giroux et al. [54] also showed that incorporation of ascorbic
acid into the beef patties before irradiation (0.54 kGy) stabilize the color during
refrigerated storage, and its use in combination with irradiation increased the shelf life
Table 7 Threshold Doses for
Some Foods of Animal Origin for
an Organoleptically Detectable
O-Flavor
Food
Threshold Dose
a
(kGy)
Turkey 1.5
Pork 1.75
Beef 2.5
Chicken 2.5
Lobster 2.5
Shrimp 2.5
Rabbit 3.5
Frog 4.0
Trout 4.5
Halibut 5.0
Lamb 6.25
Horse 6.5
a
Irradiated at 510jC.
Copyright 2004 by Taylor & Francis Group, LLC
of ground meat without detrimental eects on taste and odor. Dark rm dry (DFD) pork,
highly susceptible to microbial spoilage, as a result of a higher pH than normal, was
resistant to oxidative changes; thus the microbial safety and the shelf life of DFD meat can
be signicantly improved by irradiation, thereby improving the utilization of raw DFD
pork [55].
Pink color formation in irradiated broiler breast llets after cooking, and the char-
acteristic irradiation odor, can be signicantly reduced through shelf display of raw llets
under aerobic conditions [56].
Radiation decontamination of meat was rst commercially implemented in Brittany,
France, when e-beam irradiation treatment was established for frozen slabs of mechan-
ically separated chicken meat [57,58].
Serious outbreaks due to contamination with and survival of enterohemorrhagic E.
coli in improperly cooked hamburger patties in the west coast of the United States in 1993
prompted a petition to the Food and Drug Administration (FDA) for approval of meat
irradiation. This approval was granted in 1997 in view of new outbreaks caused by this
pathogen and the subsequent recall of some 10,000 tons of ground beef because of
contamination with E. coli O157:H7 [59]. The nal rule authorizing irradiation of
refrigerated or frozen raw meat products permits the use of a dose of 4.5 kGy for
refrigerated meats and a dose of 7 kGy for frozen products [60].
4.7. Irradiation of Fish and Shellfish Products
Fish and many other shery products are perishable commodities with short shelf lives
even under refrigeration. Fish and shellsh can be contaminated by pathogenic micro-
organisms, either because of polluted water or handling after catch [61]. Irradiation could
eectively extend the storage life and reduce the levels of pathogenic organisms. The role
of radiation processing in the improvement of the hygienic quality of shery products has
been reviewed, e.g., by Venugopal et al. [62].
In a recent literature, Kamat and Thomas [63] reported that radiosensitivity of four
food-borne pathogens, L. monocytogenes, Yersinia enterocolitica, Salmonella typhimurium,
and Bacillus cereus was not inuenced by the fat content of Indian sh varieties and
application of 3 kGy dose at refrigeration temperature would eectively inactivate approx.
10
5
CFU/g of all the organisms tested, except spores of B. cereus. Savvaidis et al. [64]
reported recently from Greece that a shelf life of 28 days at 4jC was recorded for salted,
vacuum-packaged freshwater trout irradiated at 2 kGy, compared with a shelf life of 7
days for the unirradiated sample based on sensory odor scores. Under the same
conditions, the growth of L. monocytogenes inoculated into the samples was suppressed
by 2 log cycles after 2 kGy irradiation and storage for up to 18 days at 4jC.
The increased incidence of food-borne diseases from shellsh contaminated with
Vibrio species have brought forward renewed interest in irradiation of shellsh and other
seafood. Results are very promising for clams and oysters without killing the molluscs [65].
The radiation decimal reduction dose (D
10
) determined for Vibrio cholerae O1 biotype El
Tor inoculated into various molluscs was 0.14 kGy [66]. Similar radiation sensitivity of
this organism was found in inoculated sh llets and shrimp tails [67]. The radiation dose
to eliminate as high as 10
7
CFU/g Vibrio spp. in oysters was 1.2 kGy [68]. The radiation
D
10
value for V. cholerae O1 biotype El Tor in marine snails was 0.11 kGy [69].
Petitions by the National Fisheries Institute for approval of irradiation treatment to
control food-borne pathogens in raw or processed crustaceans are pending at the
regulatory authorities in the United States [70].
Copyright 2004 by Taylor & Francis Group, LLC
In France, most, if not all, frog legs marketed are being treated by irradiation to
ensure their hygienic quality [71].
4.8. Irradiation of Minimally Processed or Ready-To-Eat Foods
Minimally processed is an equivocal term that is applied to such dierent types of
products as precut, prepackaged fresh produce, and mildly cooked or pasteurized foods
(meals or meal components) that can be stored under refrigeration for more than 1 week.
Some conventional products such as cured meats can be considered as minimally
processed but more frequently a new generation of partially processed, refrigerated foods
are described with this term.
There is an increasing consumer trend and interest by the catering industry toward
less extensively processed, convenient, or ready-to-eat foods. In response to the demands,
minimally processed foods are gaining importance, and consumers perceive these foods to
be superior because they are chilled rather than canned, dried, or frozen [72].
Many fresh-cut produce takes the advantage of the internal development of a
modied atmosphere packaging (MAP), which improves the retention of quality under
delicate interaction between respiratory increase of CO
2
, gas transmission properties of
packaging materials, and storage temperature.
Various mildly cooked chilled product are also marketed under MAP, packaged with
elevated concentrations of CO
2
over the product. CO
2
suppresses the growth of aerobic
spoilage bacteria. Special versions of cook-chill foods are the sous-vides. Sous-vide is a
process where the food is cooked under controlled conditions of temperature and time
under 100jC inside heat stable, vacuumed plastic pouches [73].
The safety of all these foods largely depends on the temperature of refrigerated
storage and eventually on other hurdles to microbial growth (reduced pH, reduced water
activity, or antimicrobials). With the extended shelf life, there are opportunities for growth
of surviving psychrotrophic microorganisms, and even mesophilic species in case of
abusive temperature.
In products prepared without pasteurization, nonsporeforming as well as sporeform-
ing pathogens should be considered as potential hazard. With mildly cooked minimally
processed food, the pathogens of greatest concern are the psychrotrophic nonproteolytic
types of Clostridium botulinum and certain strains of B. cereus [74].
Considering the generally quality-friendly character of treatment with ionizing
radiation, and by adopting a hazard analysis and critical control point (HACCP)-
based approach to risk management, irradiation processing oers a physical critical
control point for improving microbiological safety of minimally processed foods [75,76].
(A critical control point is a point, step, or procedure at which control can be applied and
a food safety hazard can be prevented or reduced to acceptable level.)
Regarding irradiation of minimally processed fresh produce, irradiation treatment at 1
kGy reduced the viable cell count of L. monocytogenes inoculated onto precut packaged
vegetables such as sliced bell pepper, carrot cubes, shredded white cabbage, and sliced radish
[77]. The irradiation also drastically reduced the viable load of spoilage bacteria thereby
improving the microbiological shelf life and extending the sensorial keeping quality of the
precut vegetables. Loss of vitamin C content as an eect of irradiation did not exceed that in
untreated samples throughout their useful life. Similarly, the benecial eects of irradiation
at 1.0 kGy of diced celery were reported by Prakash et al. [78].
Howard et al. [79] reported promising results on the eects of gamma radiation
processing (1 kGy) for extending the shelf life of refrigerated pico de gallo, a Mexican-
Copyright 2004 by Taylor & Francis Group, LLC
style cold salad prepared by chopping and mixing fresh tomatoes, onions, and jalepeno
peppers. Microbiological shelf life and microbiological safety of diced tomatoes and
cantaloupes could be improved at 0.51.0 kGy without signicant loss of consumer quality
[80,81]. Similarly, low doses of 0.35 and 0.55 kGy can improve the safety and micro-
biological shelf life of cut romaine lettuce packaged under modied atmosphere and
shredded iceberg lettuce, respectively [82,83].
Ionizing radiation eectively inactivates nonsporeforming pathogenic bacteria also
from cured or cooked meats [84]. Pathogen reduction and shelf life extension in cooked
pork chops and cured hams inoculated with L. monocytogenes and S. typhimurium were
studied by Fu et al. [85]. Low-dose irradiation (0.750.9 kGy) reduced L. monocytogenes
by more than 2 logs and S. typhimurium by 1 to 3 logs. Alur et al. [86] described
eradication of Salmonella and Staphylococcus from several processed pork products in
India by gamma radiation doses of 2.54 kGy. When Sommers and Thayer [87] surface-
inoculated several brands or types of commercially available frankfurters with a cock-
tail of four L. monocytogenes strains and vacuum packaged, the gamma radiation D
values ranged from 0.49 to 0.71 kGy depending on the individual product formulation.
The eect of irradiation (2 and 3 kGy) and storage at 23jC on the sensory quality
of cook-chill ready meals consisting of roast beef and gravy, and cauliower together with
roast and mashed potatoes were assessed by McAteer et al. [88]. The authors noted that
growth of Pseudomonas spp. caused obvious spoilage of the nonirradiated meals after 15
days of storage, whereas irradiation reduced the number of microorganisms in the meal to
less than 100 CFU/g initially, and signicant microbial growth did not occur during the
entire storage. However, using sensory proling techniques, a trained panel found that the
organoleptic eects of irradiation and chilled storage were most apparent in the cauli-
ower and mashed potato components, and mostly occurred in the color, appearance, and
textural attributes. In another study, the same research group performed a consumer trial
with 107 consumers to assess the acceptability of an irradiated (2 kGy) chilled meal
consisting of beef and gravy, Yorkshire pudding, carrot, broccoli, and roast potato. The
testing after 4 days after radiation treatment indicated that untrained consumers found
the irradiated meal moderately to very acceptable, and not signicantly dierent from the
nonirradiated meal [89]. Only for the Yorkshire pudding was there a signicant dierence
between the nonirradiated and irradiated samples.
In a later series of experiments on other types of ready-to-eat meals (roast pork and
gravy and mixed vegetables), 2 kGy was selected as the most appropriate irradiation dose
[90]. Microbial numbers in control meals reached unacceptable levels by 6- and 4-day at
3jC and 10jC, respectively, while samples irradiated at 2 kGy remained acceptable
throughout the 14-day storage at 3jC, but irradiated samples stored at 10jC had
unacceptably high numbers by day 6. The 2 kGy radiation dose reduced thiamin content
in the pork component of the ready-to-eat meal by 22%. After storage at 3jC for 14 days,
a further 26% thiamin loss was measured, and a further 14% decrease occurred when the
meals were reheated as per the manufacturers instruction at 160jC for 25 min. Irradiation
caused 51% loss in total vitamin C content of potatoes in the ready meals. The vitamin C
content of potatoes greatly decreased in both nonirradiated and irradiated potatoes during
storage. Reheating had a highly detrimental eect on the levels of vitamin C.
In a recent study in the United States, irradiation of a prepared meal consisting of
Salisbury steak, gravy, and mashed potatoes at 5.7 kGy eectively eliminated the back-
ground microbial population and high concentrations of L. monocytogenes contamination
without causing adverse eects on quality [91].
Copyright 2004 by Taylor & Francis Group, LLC
Sous-vide cooking is thought to be quality friendlier than regular cooking due to the
mild heat treatment and vacuum packaging. The combined eect of irradiation and sous-
vide cooking of chicken breast meat was investigated with respect to survival and growth of
L. monocycotenes, shelf life, thiamin content, and sensory quality by Shamsuzzaman et al.
[92]. Combining electron irradiation of 2.9 kGy and cooking in vacuum packaging to an
internal temperature of 65.6jC, reduction of more than 5.5 logs of CFUs of L. mono-
cytogenes was achieved and the pathogen remained undetectable in the irradiated product
during an 8-week storage at 2jC. Regarding the general microbiological quality, the product
that received the combined treatment had at least 8 weeks shelf life under refrigeration. The
electron beam treatment had little eect on odor and avor, and the thiamin content was
reduced by only 5%. In a second experiment [93], the same product inoculated with L.
monocytogenes to approx. 10
6
CFU/g was cooked to 71jC. This sous-vide cooking alone
reduced Listeria counts by only 1.5 log cycles; the survivors multiplied quickly during the
8jC storage and nonirradiated samples spoiled within 2 weeks. The combination of 3.1 kGy
and sous-vide cooking reduced Listeria counts to undetectable levels without adversely
aecting the sensory quality and prevented microbial spoilage for at least 8 weeks. The loss
of thiamin content due to combined treatment varied from 23% to 46%.
In studies of Farkas et al. [94], the survival and growth of spores of a psychrotrophic
B. cereus strain inoculated in a meal of smoked-cured pork in boiled beans sauce were
investigated as aected by combination of 5 kGy and sous-vide cooking and storage time
at 10jC. The microbiological analyses demonstrated that this medium dose irradiation
prior to sous-vide cooking sensitized surviving spores to the subsequent heat treatment
and provided microbiological stability for at least 2 months.
4.9. Radiation Decontamination of Dry Food Ingredients
The microbiological action of ionizing radiation is well proven to decrease the viable cell
counts in dry food ingredients, thus improving the microbiological safety of such products
and enhancing the storage stabilities of foods prepared with them. Spices, dry vegetable
seasonings, herbs, protein preparations, and commercial enzyme preparations used in the
food industry can be suciently decontaminated with doses of 310 kGy without altering
their avor, texture, or other important technological and sensory properties. It is also of
practical signicance that the surviving microora of radiation-pasteurized ingredients,
consisting mainly of bacterial spores, becomes more sensitive by the radiation treatment to
subsequent food processing treatments than the microora of untreated ingredients
[95,96]. Heat processing techniques have a much more limited feasibility of these
commodities because of the heat sensitivity of many of these ingredients. The irradiation
process is a viable alternative to the formerly used fumigation with ethylene oxide, which
was prohibited by a European Union (EU) directive in 1991 and has been banned in a
number of other countries, too, for health, environmental, or occupational safety reasons,
due to its carcinogenicity [97].
A detailed monograph on irradiation of dry food ingredients has been published in
1988 [98], and an updated shorter summary appeared recently [99]. A Code of Good
Irradiation Practice for the Control of Pathogens and Other Microora in Spices, Herbs,
and Other Vegetable Seasonings has been issued by the International Consultative Group
on Food Irradiation [100].
Presently, irradiation of spices is the most widely utilized application of food
irradiation that is practiced in more than 20 countries, including Argentina, Belgium,
Copyright 2004 by Taylor & Francis Group, LLC
France, Hungary, Mexico, The Netherlands, Norway, South Africa, and the United
States, and global production of irradiated spices has increased from about 5000 tons in
1990 to over 60,000 tons in 1997. In the United States alone, over 30,000 tons of spices,
herbs, and dry vegetable seasonings were irradiated in 1997, as compared to only 4500
tons in 1993 [101]. In 1999, about 95 million pounds of these products were irradiated
accounting for about 10% of their total consumption [102].
In addition to reducing microbial contamination, irradiation is an eective method
for increasing of extraction yield of medicinal herbs [103].
4.10. Combination Processes in Food Irradiation
When irradiation is used with other preservative or antimicrobial factors, the global
eciency is reinforced through additive or synergistic action. The combination of irradi-
ation with mild heat treatment has a number of advantageous eects [104]. The possibility of
using heat in combination with irradiation was rst suggested in the 1950s when synergistic
eects were observed with a variety of biological systems including bacteria. More recently,
Thayer et al. [105] investigated the eects of heat and ionizing radiation on S. typhimuriumin
mechanically deboned chicken meat and reported that irradiation to a dose of 0.9 kGy
caused heat-sensitization of S. typhimurium.
The synergy between irradiation and heat, coupled with the persistence of the
radiation-induced heat sensitization, could be used to advantage in foods that are cooked
before consumption, and cook-chill ready meals that are reheated prior to consumption
[106] (see also Sec. 4.8).
The synergistic eect of irradiation plus heat on vegetative organisms may be due to
the inability of cells to repair radiation damage because heating might inactivate repair
enzymes. In bacterial spores, the synergism [107] may be related to the partial rehydration
of the spore core due to the radiation-induced degradation of the cortex peptidoglycan
that is providing otherwise an osmoregulative eect maintaining the dehydrated state of
the core [108]. A recent paper [109] demonstrated a method to quantify the radiation-
induced sensitivity of bacterial spores to heat in conduction heating foods and illustrated
the associated savings in the heating cycle of the ultimate thermal process.
Mild heat treatment such as a hot water dip in combination with low-dose
irradiation found to be ecient to decrease rot and without adverse eect on nonmicrobial
qualities of certain fruits such as mangoes [110,111] and clementines [112].
In the eld of muscle foods, the use of marination before irradiation reduced the
dose necessary to eliminate Salmonella in poultry [113]. Some antimicrobial additives,
especially the natural ones (e.g., bacteriocins) [94] and GRAS (generally recognized as
safe) preservatives [114] can be usefully combined with irradiation to reduce dose
requirements. Some antioxidants have also been used to prevent the undesirable oxidative
eects in irradiated foods.
4.11. Radiation Sterilization of Food (Radappertization)
The purpose of this process is the production of foodstus that are shelf stable at ambient
temperatures and that have better quality characteristics than the corresponding heat-
sterilized products. Radappertization dose of low-acid foods must ensure elimination of
the spores of the most resistant bacterial pathogen, C. botulinum. The dose selected for the
purpose is around 50 kGy, 12 times of the D
10
value. As the product is intended to be
shelf-stable at ambient temperature, while autolytic enzymes have high radiation resist-
Copyright 2004 by Taylor & Francis Group, LLC
ance, the radiation treatment is supplemented with a mild heat treatment to inactivate the
enzymes. This high-dose treatment must be delivered to a vacuum-packaged and deeply
frozen (30jC, or less) product to avoid avor changes [115]. The whole process thus
involves the following steps:
1. Heating to an internal temperature of 6575jC.
2. Packaging under vacuum in a sealed container impermeable to moisture, air,
light, and microorganisms.
3. Cooling/freezing to irradiation temperature.
4. Irradiation.
The technology has been developed mainly for meats, poultry, and certain seafoods
in the United States, and radappertized foods were successfully used in space missions
[116]. Similar radiation-sterilized products were manufactured by the Atomic Energy
Corporation of South Africa for hikers [117].
Radiation sterilization of deep frozen meals prepared for hospital patients whose
immune systems have been suppressed is approved in several countries such as the United
Kingdom, The Netherlands, and the Federal Republic of Germany.
5. WHOLESOMENESS OF IRRADIATED FOODS
Wholesomeness (toxicological innocuity, nutritional adequacy, and microbiological
safety) of irradiated food has been carefully evaluated by an unprecedented width of
research and testing over more than 50 years. All scientically acceptable evidence
resulted from these studies supports the safety of irradiated foods for consumption
[14,118121].
At low and medium doses, it is well established that the nutritional value of proteins,
carbohydrates, and fats as macronutrients are not signicantly impaired by irradiation,
and neither the mineral bioavailability is impacted. Like all other energy depositing
process, the application of ionizing radiation treatment can reduce the levels of certain
sensitive vitamins. Nutrient loss can be minimized by irradiating food in a cold or frozen
state and under reduced levels of oxygen. Thiamin and ascorbic acid are the most
radiation sensitive, water-soluble vitamins, whereas the most sensitive, fat-soluble vitamin
is vitamin E. In chilled pork cuts at the 3 kGy maximum at 010jC, one may expect about
3540% loss of thiamin; in frozen, uncooked pork meat irradiated at a 7 kGy maximum at
20jC approx., 35 % loss of it can be expected [122].
The Joint Food and Agriculture Organization (FAO)/International Atomic Energy
Agency (IAEA)/World Health Organization (WHO) Expert Committee on Food Irradi-
ation (JECFI) has evaluated the very extensive literature on wholesomeness of irradiated
food and concluded already in 1980 that ...the irradiation of any food commodity up to
an overall average dose of 10 kGy presents no toxicological hazard, hence, toxicological
testing of foods so treated is no longer required, and . . .irradiation of foods up to an
overall average dose of 10 kGy introduces no special nutritional or microbiological
problems [119]. Independent evaluations by experts, e.g., in Denmark, France, Japan,
The Netherlands, the United Kingdom, and the United States found neither toxic eects
as a result of consuming irradiated food. The FAO/WHO Food Standards Program,
Codex Alimentarius, accepted the JECFIs recommendations and established a Codex
General Standard for Irradiated Food and a Recommended International Code of
Practice for the Operation of Radiation Facilities used for the Treatment of Foods [6].
Copyright 2004 by Taylor & Francis Group, LLC
There is no evidence or reason to expect that irradiation produces more virulent pathogens
among those that survive irradiation treatment [123,124].
After publishing an up-to-date review on safety and nutritional adequacy of
irradiated food [120], the FAO/IAEA and the WHO jointly conveyed a Study Group
on High Dose Irradiation in 1997, which reviewed all relevant data related to the
toxicological, microbiological, nutritional, radiation chemical, and physical aspects of
food exposed to doses greater than 10 kGy. The Study Group came to the conclusion that
foods treated also with doses of radiation sterilization can be considered safe and
nutritionally adequate when produced under established Good Manufacturing Practices,
and recommended approval of food irradiation without any dose maximum limit [121].
The recommendation has been submitted to the Codex Alimentarius Commission for
possible amendment of the 1984 Codex General Standard on Irradiated Foods.
6. LEGISLATION OF FOOD IRRADIATION
The WHO closely collaborates with its Member States and the other international
organizations, particularly through the International Consultative Group on Food
Irradiation (ICGFI) established in 1984 under the aegis of FAO, IAEA, and WHO.
Through the eorts of the ICGFI, which had a membership of 46 governments in 1998,
harmonized regulations on food irradiation are being promulgated in developing countries
in Asia, the Pacic, Africa, Latin America, and the Middle East [125]. The ICGFI issues
numerous publications relating to food irradiation, including codes of good irradiation
practice for various classes of foods [126] and compilations of technical data for author-
ization and control of food irradiation.
All these international documents and activities of the specialized agencies of the
UN facilitated legislation of food irradiation in many countries. According to the
database developed by the ICGFIs Secretariat at the Food and Environmental Pro-
tection Section of the Joint FAO/IAEA Division in Vienna, currently some 50 countries
granted national clearances of irradiation of at least one or more food items of food
classes. The itemized ICGFI database on these clearances can be visited on the website
http://www.iaea.org/cgi-bin/rifa-ste.1. Legislatory authorities require that irradiated food
products be labeled. In general, the international food irradiation symbol, the so-called
Radura logo:
is required with a statement that the product has been intentionally subjected to
radiation. The severity of labeling requirements for multi-ingredient products to identify
irradiated components on the list of ingredients is not the same with each authority.
After 10 years of debate, the European Parliament and the Council of the European
Union issued on February 22, 1999, a framework Directive 1999/2/EC on approxima-
tion of the laws of the Member States concerning foods and food ingredients treated with
Copyright 2004 by Taylor & Francis Group, LLC
ionizing radiation [127], and a Directive 1999/3/EC on the establishment of a Community
List of foods and food ingredients treated with ionizing radiation [128]. The present
category of foodstus authorized by this Directive for irradiation treatment is dried
aromatic herbs, spices, and vegetable seasonings, and the permitted maximum overall
average absorbed dose is 10 kGy. The Directive enforces strict labeling. Unfortunately, all
requisite labeling of irradiated foods appears much more in the form of a warning than an
item for neutral information.
Working on the subject between 1986 and 1998, the EU Scientic Committee of
Food (SCF), an independent expert body advising the European Commission on health
matters, concluded that the food irradiation process posed no problem for health over a
wide ranged of uses, if used under prescribed conditions and endorsed for authorization of
17 irradiated foods/food classes [129]. The Commission proposal in December 1998 for an
EU directive contained eight food categories and three products.
By the published Directive 1999/2/EC, the European commission was charged to
develop a nal positive list of permitted items until the end of 2000, but the contents of
this list are not yet published. There was even a consultation on this problem on the
Internet and industry and consumer organizations responded [130]. As long as the
positive list is not adopted, Member States can maintain the existing regulations except
for spices [131].
Regarding packaging materials for irradiated foods, most commonly used food
packaging material are suitable for the purpose. However, irradiation of prepackaged
food requires approval of its packaging material [132].
7. FOOD IRRADIATION PROCESS CONTROL AND DETECTION OF
IRRADIATED FOODS
Proper control of food irradiation applications should fulll the requirements for both
food technologies and radiation technologies. Application of well-established methods for
measurement of absorbed radiation dose and the dose distribution helps to provide
assurance that the radiation treatment is both eective and legally correct [133]. Computer
tomography (CT) can provide detailed, high-resolution, and accurate dose maps for any
arbitrary product and package congurations [134]. Such dose maps are an essential part
of process validation.
Signicant progress has been made in the eld of analytical detection of irradiated
food to improve consumers condence and to assist international trade of irradiated food
[135,136]. Due to national and international programs and activities of the European
Committee for Standardization (CEN), ve validated and standardized detection methods
are now available. The CEN is also considering the adoption of further ve detection
methods: three will be screening methods (positive results from a screening method must
be conrmed using a standardized method) [136].
One of the standardized methods, electron spin resonance (ESR) technique, permits
identication of food that contains a hard, dry matrix, e.g., bone. When food containing
bone is irradiated, free radicals are produced and trapped in the crystal lattice of the bone,
which can be detected by ESR spectroscopy [137]. Thermoluminescence of contaminating
minerals for detection of radiation treatment of, e.g., spices and dried fruits can be
successfully applied [138, 139]. Another standardized method that has been developed for
identication of irradiated fat-containing foods is the mass-spectrometric detection of
radiation-induced 2-alkylcyclobutanones after gas-chromatographic separation [140]. The
Copyright 2004 by Taylor & Francis Group, LLC
DNA comet assay (analysis of DNA fragmentation by irradiation) can be applied as a
fast and simple screening method [141]. For simple screening of irradiation decontami-
nation of dry ingredients of high starch contents, viscosity measurements of their heat-
gelied suspensions is suitable because of their reduced viscosity due to the radiation-
induced starch damage [142144].
8. ECONOMIC FEASIBILITY OF FOOD IRRADIATION
The costs of irradiation are inuenced by a large number of factors depending on the actual
use and local conditions. To optimize processing costs, one must fully understand the
processing scenario and then design an industrial facility that best matches it [145]. The
many dierent potential applications of food irradiation create many dierent requirements
in irradiator design [146]. The throughput requirement can range from a few to hundreds
of tons per hour. With such a variety of process specications discussed in the previous
chapters, the cost of irradiating food also substantially varies from one application to
another. As any other processing technologies, due to varying conditions, cost of the same
type of treatment may dier from place to place, and a cost estimate in one country is not
necessarily representative of other countries.
The capital costs to build a commercial food irradiation plant is about a fewmillions of
U.S. dollars but within the range of plant costs for other physical methods of food pre-
servation. Total annual operating costs are close to $1 million. The total processing cost is
the sumof the total operating costs plus the cost of money, and the depreciation of capital
(amortization) [146]. The unit cost of irradiation equals the total annual processing costs
divided by the annual throughput. This unit processing cost in all irradiators rapidly
decreases with the initial increase of throughput because the xed costs are spread over a
larger number of units [147]. The economic throughput is a function of many factors and
these economics on scale are very pronounced at smaller sizes [148]. The eect of increasing
dose on unit processing costs is linear because unit processing costs inversely relates to
annual throughput.
The ICGFI [149] estimates that irradiation cost range from $10 to $15 per tonne for
a low-dose application (e.g., inhibition of sprouting of potatoes or onions), and $100 to
$250 per tonne for a high-dose application (e.g., to ensure hygienic quality of spices).
These unit costs are considered to be competitive with alternative treatments.
In addition to economic feasibility, radiation processing is less energy consumptive
than other decontamination techniquesan increasingly encouraging feature of radiation
technology [150].
9. RECENT DEVELOPMENT OF COMMERCIAL IMPLEMENTATION OF
FOOD IRRADIATION
Acceptance of irradiated food was greatly inuenced by the perception of food irradiation
as a nuclear technology. As Professor J.F. Diehl, one of the world authorities in food
irradiation, rightly points out [151]: Much more than other modern methods of food
processing, food irradiation has to overcome barriers created by prejudice, misleading
information, restrictive legal and regulatory measures, and a resulting reluctance of food
manufacturers and food trade to make use of the new technology. A signicant barrier to
Copyright 2004 by Taylor & Francis Group, LLC
adoption of the irradiation technology was the wait and see attitude of the food
industry and trade because they either perceived signicant consumer resistance or because
they had concerns about being seen as the leader in a technology perceived as controversial
[125]. Still much needs to be carried out for the wider commercial application of the
technology and public understanding and acceptance of irradiated food. However, in spite
of certain anti-irradiation groups that have been eecting slow take-o of food irradiation
at work with the news media, the notion of overwhelming consumer resistance to the
technology is a myth and a changing attitude can be noted on market research studies.
Numerous consumer acceptance studies and market tests in several countries during the
past decades indicate that majority of the consumers are willing to purchase irradiated
food once they understand the safety and benet of the process [152155]. A group of
rejectors is that segment that generally rejects any new product and no amount of in-
formation would convince this group [156].
Besides the pioneering implementations of specic low-dose applications as well as
the widely utilized irradiation of spices now, mentioned in Secs. 4.1, 4.2, and 4.9, small-
scale commercial application of irradiation to ensure hygienic quality of food, especially
those of animal origin, has been carried out in Chile, China, Indonesia, and Thailand in
the past two decades. In the recent years, new commercial irradiators including some that
are dedicated to food irradiation have been commissioned in Brazil (which plans to add up
to 10 facilities in the coming years), China, India, Republic of Korea, Mexico, and
Thailand [157].
Very important recent progress is noted in the United States. Following the approval
in December 1999 of irradiation of refrigerated or frozen raw meat and meat products to
eliminate or signicantly reduce E. coli O147:H7 and other hazardous bacteria, in May
2000, commercial production and marketing of irradiated frozen beef patties started,
electron-beam treated by Sure-Beam Corp., Sioux City, Iowa. This irradiation plant can
process yearly more than 100 thousand tons of beef [158]. The other irradiation company,
Food Technology Service, Inc., Mulberry, Florida, which has been irradiating poultry
since 1996 with its MDS Nordion-designed cobalt-60 plant, also began treating fresh
ground beef and frozen beef patties. The Ion Beam Applications (IBA) company was
expected to complete in mid-2001 a 135 kW electron accelerator to be used in x-ray mode
for food. Because several other irradiation plants are under construction in various states
in the United States [159], a nationwide distribution of these irradiated products becomes
possible. Due to low conversion rate of production of x-rays, the x-ray technology is
probably not as economical in food irradiation as gamma rays or electron beam
technology. Nevertheless, an x-ray plant has been recently constructed in Hawaii to
disinfest tropical fruit [160]. An important step toward wider future applications is a
petition in 1999 to the FDA, for approval of irradiation of many types of ready-to-eat
food by the Coalition of Food Irradiation, which involves approx. 25 dierent associa-
tions of food producers.
Due to legislative stumbling blocks (described in Sec. 6), after early pioneering eorts
in several countries, e.g., Netherlands, France, and Belgium, where considerable amounts
of frozen seafood, frog legs, and dry food ingredients have irradiated already been in the
1980s, progress in Europe is now lagging behind the developments in the United States.
However, increasing activities on food irradiation in the United States, which is one of the
largest markets in the world, and also similar activities in some countries of Asia and Latin
America, will likely inuence other countries and regions to implement the use of
irradiation for commercial purposes.
Copyright 2004 by Taylor & Francis Group, LLC
REFERENCES
1. Josephson, E.S.; Peterson, M.S. Preservation of Food by Ionizing Radiation. Vol. 13. CRC
Press, Inc.: Boca Raton, FL, 19821983.
2. Urbain, W.M. Food Irradiation; Academic Press, Inc.: Orlando, 1986.
3. Food Irradiation. A Technique for Preserving and Improving the Safety of Food; World Health
Organization: Geneva, 1988.
4. Satin, M. Food Irradiation: A Guidebook; Technomic Publishing Co., Inc.: Lancaster-Basel,
1996.
5. Food Irradiation: Principles and Applications; Molins, R., Ed.; Wiley-Interscience: New York,
2001.
6. Codex General Standard for Irradiated Foods and Recommended International Code of Practice
for the Operation of Radiation Facilities Used for the Treatment of Food, CAC/Vol. XV-Ed. 1,
Food and Agriculture Organization of the United Nations: Rome, 1984.
7. Understanding the Key Radiation Processing ParametersA Guide for Food Producers, Leaet
published by PURIDEC Irradiation Technologies, 2001.
8. Kim, A.; Thayer, D.W. Appl. Environ. Microbiol. 1996, 62, 1759.
9. Tilton, E.W.; Brower, J.H. In Preservation of Food by Ionizing Radiation; Josephson, E.S.,
Peterson, M.S., Eds.; CRC Press, Inc.: Boca Raton, FL, 1983; Vol. 2, 269.
10. Farkas, J. Int. J. Food Microbiol. 1998, 44,189.
11. High-Dose Irradiation: Wholesomeness of Food Irradiated with Doses Above 10 kGy, World
Health Organization: Geneva, 1999.
12. Thayer, D.W.; Boyd, G. J. Food Prot. 2001, 64, 1624.
13. Sommers, C.H.; Niemira, B.A.; Tunick, M.; Boyd, G. Meat Sci. 2002, 61, 323.
14. Diehl, J.F. Safety of Irradiated Foods, 2nd Ed.; Marcel Dekker, Inc.: New York, 1995.
15. Proctor, B.E.; OMeara, J.P. Ind. Eng. Chem. 1951, 43, 718.
16. Merritt, L.C. Jr.; Angelini, P.; Wierbicki, E.; Shults, G.W. J. Agric. Food Chem. 1975, 23, 1037.
17. von Sonntag, C. Adv. Carbohydr. Chem. Biochem. 1980, 37, 7.
18. Hammer, C.T.; Wills, E.D. Int. J. Radiat. Biol. 1979, 35, 323.
19. Zabielski, J. Radiat. Phys. Chem. 1989, 34, 1023.
20. Lebovics, V.K.; Gaa l, O
. Reduction of
Cholesterol and Lipid Oxidation in Radiation Decontaminated Mechanically Deboned Turkey
Meat, Poster presented at the 48th ICoMST, Rome, 2530 August 2002.
23. Thomas, P. In Food Irradiation: Principles and Applications; Molins, R., Ed.; Wiley-
Interscience: New York, 2001; 241 pp.
24. Matsuyama, A.; Umeda, K. In Preservation of Food by Ionizing Radiation; Josephson, E.S.,
Peterson, M.S. Eds.; CRC Press, Inc.: Boca Raton, FL, 1983; 159 pp.
25. Ka lma n, B.; Kiss, I.; Farkas, J. In Food Preservation by Irradiation; International Atomic
Energy Agency: Vienna, 1978; Vol. 1, 113.
26. Salems, A. J. Sci. Food Agric. 1974, 25,257.
27. Dallyn, S.I.; Sawyer, R.L. Proc. Am. Soc. Hortic. Sci. 1954, 73, 398.
28. Ogawa, M.; Uritani, I. Agric. Biol. Chem. 1970, 34, 870.
29. Mondy, N.I.; Gosselin, B. J. Food Sci. 1989, 54, 982.
30. Thomas, P. CRC Crit. Rev. Food Sci. Nutr. 1984, 21, 95.
31. Umeda, K. Food Irradiat. Inf. 1978, 8, 31.
32. Umeda, K. Food Irradiat. Newsl. 1983, 7, 19.
33. Farkas, J.; Kiss, I.; Ra zga, Z.; Vas, K. KE
and AlO
+
, has
attracted a great deal of interest in atomic physics [8,9] and in materials science [10]. The
cluster ion beam has distinctive properties of vicinity eects caused in substances, which is
Table 1 Ion Beam Irradiation Parameters and Their Related Ion Beam Technologies
Parameter Requirement Ion beam technology
Ion species Variety Ion source
Cluster ion source
Multiplex ions Dual/triple beam
Energy Wide range from keV to GeV Accelerator
Quick change Cocktail beam acceleration
Intensity Wide range from a single
ion to mA
Ion source
Accelerator
Single-ion hit technique
Irradiation area and
particle density
uence
Uniformity Beam defocusing and a scatterer
Beam scanning: raster scan,
beam wobbling, spiral scan
Beam spot and
targeting resolution
Minimization
Precision
Microbeam and nanobeam
Micro-PIXE analysis
Time structure Flexibility Pulsated beam
Copyright 2004 by Taylor & Francis Group, LLC
rather dierent from those of a single ion. By irradiation of the high-energy cluster ion
beam, a number of atoms are implanted simultaneously into a very small area, bringing
high implantation density and large energy transfer. It is anticipated that the cluster ion
applications will result in a breakthrough in the invention of new functional materials.
Cluster ion sources with dierent production techniques have been devised [10]. The
cluster ions are produced, for instance, by vaporizing the materials using a heated oven, a
laser ablation apparatus, or sputtering. Essential features of the cluster ion source are
increased intensity and stability of the beam, and control of the mass distribution of the
particles. Further development of cluster ion sources is in progress at several facilities [7,11].
3.3. Acceleration of Ions
Ion Accelerator
Ion beams are powerful tools for modication of specic material characteristics, inves-
tigation of material properties, and processing and creation of materials, bringing great
benet to industrial applications. Ion beams in the energy region of
f
keV to several MeV
have the advantage of enhancing the eectiveness of ion interaction with matter without
signicant damage. On the other hand, the ion beams of energy greater than several tens of
MeV cause great damage to the substances along their trajectory. With larger range and
higher LET in substances, the high-energy ion beams are typically utilized for investigating
the radiation tolerance of materials, for creating new materials, and for generating new
plant resources.
Figure 3 A schematic drawing of a typical ECR ion source. The ions and electrons are conned in
the plasma by the magnetic eld generated by the solenoid coils and the hexapole magnet.
Copyright 2004 by Taylor & Francis Group, LLC
Ion accelerators can cover a wide range of energy from
f
keV to
f
GeV, depending
entirely on the requirement of the application. A single accelerator can hardly provide
various ion beams in such a broad dynamic range of energy due to limitations of the
acceleration voltage, the magnetic eld, beam quality, and other reasons. Practically, the
required energy range may be covered by a combination of dierent accelerators like the
accelerator complex at the TIARA facility of JAERI, Takasaki, as shown in Fig. 4 [12].
In principle, a charged particle is accelerated by traversing an electric eld formed
between electrodes. Acceleration methods are classied into two types: potential acceler-
ation using static elds and sequential acceleration using alternating elds [13]. Electrostatic
accelerators such as the Van de Graa accelerator and a tandem accelerator, belonging to
the former type, provide ion beams with energy of
f
keV to
f
MeV. The latter type of
accelerators, such as the cyclotron, the linac, and the synchrotron, cover higher energy
region from
f
MeV to
f
GeV.
Flexibility in ionic species, energy, intensity, time structure, beam spot size, etc., are
the most essential characteristics common to all the accelerators used for ion beam ap-
plications. The TIARA facility is unique in its design specialized for the ion beam appli-
cations, especially to materials science and to biotechnology. Agreat variety of ionic species
from proton through bismuth is available through four types of accelerators: a K110 AVF
cyclotron, a 3-MV tandem accelerator, a 3-MV single-ended accelerator, and a 400-kV ion
implanter [14,15]. The energy range covered by the four accelerators is shown in Fig. 5. A
cyclotron [16] has the advantages of the diversity of accelerated particles from protons to
heavy ions with mass number of greater than 200, a broad energy range from
f
MeV to
f
GeV, the ability to increase beam intensity up to mA, and compactness for saving space.
An AVF cyclotron [17,18] has a radially increasing magnetic eld to compensate for the
Figure 4 A birds-eye view of the accelerator complex at the TIARA facility of JAERI, Takasaki.
Copyright 2004 by Taylor & Francis Group, LLC
relativistic mass increase for isochronism, and an azimuthally varying eld to provide the
additional vertical focusing. A schematic drawing of the AVF cyclotron is shown in Fig. 6.
Ions generated by the external ion source are axially injected at the center of the AVF
cyclotron. The ions are led to the medium plane of the cyclotron magnet by the electric eld
generated between inector electrodes in the shape of a spiral. The ions make revolutions
periodically in the magnetic eld generated by the cyclotron magnet. The revolution period
of the particles is kept constant to accelerate charged particles using a xed-rf voltage
supplied to the electrodes in the frequency range of MHz. After the ions reach an extraction
radius, the beam trajectory is shifted outward by using the electric eld of a deector,
Figure 6 Schematic drawing of the ion acceleration in the AVF cyclotron.
Figure 5 Energy range of the ion beams available at the TIARA facility of JAERI, Takasaki.
Copyright 2004 by Taylor & Francis Group, LLC
followed by a magnetic channel for weakening the magnetic eld, to extract the ions from
the cyclotron.
Accelerator technologies for the cyclotron have evolved to meet such requirements of
the ion beam as the increases in energy and beam intensity. Precise experiments demand a
cyclotron beamof high quality, namely, high energy and time resolutions. Development of a
at-top acceleration system [19,20] and a beam cooling system [21] resulted in the
minimization of the energy spread in the cyclotron beam. The at-top acceleration
technique is indispensable for the production of a cyclotron microbeam [22]. A highly
stable beam required for the at-top acceleration can be provided by a precise and exible
control of the temperature of the cyclotron magnet yoke.
Quick Change of Ion Species and Energy
Most of beamtimes allocated for ion beamapplications are completed within 24 hr, which is
comparatively shorter than that for nuclear physics. Ordinarily, ion species and energy have
to be changed to collect the necessary data for investigation of dependence on LET in
substances, for example. Such researches require several beam times for data acquisition,
taking users almost a year or more. In case of a cyclotron, more than 2 hr of overall time is
needed for reexcitation of the cyclotron magnet and optimization of the parameters of the
whole systems. The change of the ion beam condition during the individual beam time is
inecient for increasing the rate of cyclotron operation. Quick change of the beam without
loss of time is essential for ecient use of the beam time.
Acocktail beamacceleration method [23] is a smart technique, extremely useful for
quickly changing ion species and energy in the cyclotron. Cocktail ions, consisting of
dierent ion species, having almost identical mass-to-charge ratio (M/Q) produced by the
ECR ion source, are simultaneously injected into the cyclotron. The cyclotron parameters
are optimized for one of the cocktail ions. The selected ion is fully accelerated under the
isochronous condition given by 2pf
rf
/h = ( Q/M)(B
0
/m
0
c
2
), where f
rf
is the rf frequency, h is
an acceleration harmonic number, B
0
is the magnetic eld for isochronism, m
0
is the unied
atomic mass unit, and c is the speed of light. Other ions with slightly dierent M/Q values
are gradually shifted an additional amount in rf phase, and get out of the accelerating phase
of the rf cycle. Other ion species can be extracted fromthe cyclotron by slightly changing the
frequency or the magnetic eld to match the dierence of the M/Q. Changing of the
frequency or the magnetic eld can be completed within a few minutes.
Ion beams are useful to simulate the environment in space, where semiconductor
devices are exposed to high-energy heavy-ion impact. Incorrect operation of semiconductor
devices such as single-event upset results from the heavy-ion irradiation. The cocktail ion
families of M/Q = 4 and 5, available at the JAERI AVF cyclotron facility [24], are
frequently utilized to investigate the tolerance of the semiconductor devices to the radiation,
and to survey highly radiation-tolerant semiconductor devices appearing in the market.
Eciency of the radiation-tolerance testing for thousands of kinds of semiconductor devices
has been totally improved by the cocktail acceleration technique.
3.4. Irradiation of Ions
Simultaneous Irradiation of Dierent Ion Beams
Structural materials for ssion reactors (as well as for future fusion reactors) are being
exposed to intense neutron ux for many years. In the case of fusion reactor, 14-MeV
neutrons, produced by the fusion reaction of d+t !
4
He+n, induce nuclear reactions of
Copyright 2004 by Taylor & Francis Group, LLC
(n, a) and (n, p), while helium and hydrogen gases are accumulated in the reactor mate-
rials. Constituent atoms of the reactor materials are recoiled by the neutrons. The neutron
irradiation causes a degradation of the material properties through displacement of the
constituent atoms, accumulation of the hydrogen and helium gases, and other nuclear
transmutation products.
Development of advanced structural materials for blankets of the fusion reactor is
particularly signicant, because the eciency and life expectancy of the blankets are limited,
respectively, by the operating temperature and the radiation damage of the materials. Prop-
erties of candidate materials such as low-activation ferritic steel and silicon carbide ber-
reinforced silicon carbide (SiC/SiC) composite are being investigated under dual or triple
ion beam irradiation at TIARA facility. A variety of the neutron irradiation conditions in
the reactors can be simulated in a relatively short period by simultaneous irradiation of mul-
tiple ion beams. Especially, triple ion beam irradiation is extremely eective in simulating
the synergistic eect of displacement damage and the heliumand hydrogen accumulation in
the reactor materials. It helps to develop new reactor materials.
Simulation of the neutron-induced damages using triple ion beams is schematically
shown in Fig. 7. Aproton and a heliumion are provided by the ion implanter and the single-
ended accelerator, respectively. Heavy ions, such as iron or silicon, accelerated by the
tandem accelerator, are injected into the target simultaneously. For example, the SiC/SiC
composite was tested under triple ion beam irradiation consisting of a 380-keV proton, a
1.2-MeVheliumion, and a 7.8-MeVSi
3+
ion. The triple irradiation systemis equipped with
an energy degrader and a beam scanner for uniform three-dimensional (3-D) irradiation.
Uniform Irradiation by Beam Scanning
A uniform-irradiation technique is commonly used for ion beam applications in materials
science and biotechnology, and for biomedical application such as cancer therapy. Uni-
formity of the irradiated-particle density distribution is essential to bring about the same
Figure 7 Schematic drawing of the damage in reactor materials induced by neutrons, and its
simulation by the triple-ion-beam irradiation.
Copyright 2004 by Taylor & Francis Group, LLC
irradiation eects in a substance with a large area or with a number of target samples aligned
in several lines. A relative dose variation of at most F10% is maximally tolerable for the
uniform irradiation. Several uniform-irradiation systems have been developed so far to
meet individual demands of ion beam applications. The most economical method for
uniform irradiation is a double-scattering beam delivery using two sets of scatterers [25].
In this system, we should pay attention to the energy loss in the scatterers, a change of a
charge state of the ions, activation of the scatterer material, and production of nuclear
fragments.
Formation of a beam spot with a homogeneous prole by using multipole magnetic
elds is a good way to atten the particle distribution [26]. Multipole elds, such as octupole,
dodecapole, and 16-pole, generate a stronger focusing force for the particles traveling away
from the axis of the multipole, while the focusing force acting on the particles close to the
multipole axis is comparatively weak. The nal particle distribution is highly dependent on
initial beam size and angular spread. A very careful designing of the multipole magnet and
the beam transport system is required to obtain a uniform particle distribution with a hard
edge at an irradiation point.
A 2-D beam scanning system equipped with a set of magnets or electrostatic plates to
deect the beamhorizontally and vertically is commonly used for uniformirradiation over a
large area, e.g., >10 10 cm
2
. A schematic drawing of the 2-D beam scanning method is
shown in Fig. 8. The scanning area may be varied by changing the excitation currents of the
Figure 8 Schematic drawing of the two-dimensional beam scanning system using a set of magnets
to deect the beam horizontally and vertically.
Copyright 2004 by Taylor & Francis Group, LLC
deection magnet or the voltages supplied to the electrostatic plates. By using a raster beam
scanning system [27], a large rectangular area can be irradiated with continuously delivered
particles. The beam-sweeping speed and the spacing between neighboring trajectories are
kept constant. Uniformity of the particle distribution is determined by beam spot size and
trajectory spacing. The raster beam scanning system is utilized for estimating damage in
solar cells from exposure to high-energy protons and for inducing mutation in plants by
heavy ions.
A wobbler beam delivery system [28], another 2-D beam scanning system, is used
typically for biomedical applications. Abeamis circularly wobbled at several dierent radii
to form a uniformdose distribution in an area up to 30 cm in diameter. The beam-sweeping
speed at the outer wobble radii increases, since the horizontal and vertical deection
frequencies are constant. When beam intensity remains constant, longer irradiation time
makes up for a deciency of particle density in outer wobbles to achieve a uniformradiation
eld. A continuous uniform irradiation over a round area can be realized by a spiral beam
scanning method [29]. The beam-sweeping speed is kept constant even for larger radii, while
the trajectory spacing in a radial direction is invariable. Precise control of the irradiation
time is unnecessary for the spiral beam scanning system.
3.5. Microbeam
Superiority of the Microbeam
Spatial andtargeting resolutions of the ion irradiation in a nite area are greatly enhanced by
using a microbeamwith a beamspot size of 1 Amin diameter. More precise microanalysis of
elements included in substances is feasible by combining the nuclear microprobe with
analytical techniques like PIXE, NRA, and ion channeling [30]. In recent years, the micro-
analysis technique has become widespread in the eld of life sciences [31]. The microbeam
has also been applied to material processing and to elucidation of radiation eects caused by
high-LET irradiation such as single-event upset of semiconductor devices used in space.
Ahigh-energy ion has higher LETin substances than do electrons or gamma rays with
the same energy. A heavy-ion microbeam with energy of hundreds of MeV is extremely
useful for research in biology and biotechnology as a ne probe for investigation of cell
response and as a specic tool for cell surgery. Asingle ion can cause great damage locally to
a part of the DNA without destroying the whole cell. The microbeam technology is
indispensable for research in the state-of-the-art bioscience, for example, in the investigation
of cell-to-cell communications like bystander eects, the analysis of cellular spatial sen-
sitivity, the interaction of damages caused by individual irradiation, the cellular repair dy-
namics, and in such an intracellular process as apoptosis.
Production of the Microbeam
The rst microbeam was generated by using a Russian quadruplet lens conguration in the
early 1970s at Harwell in Britain [32]. In recent years, microbeam technology has made
remarkable progress to fulll the requirements for several kinds of microprobe analyses.
Nowadays, more than 200 facilities, at which microbeams with energy greater than several
hundreds of keV/nucleon can be provided, are available worldwide for a variety of research
elds.
Beamcollimation using a microaperture foil with a hole of 5 to 10 Amin diameter or a
precisely manufactured slit is an easy method for the microbeam production [33,34]. The
minimum beam spot size was limited to around 5 Am due to the size of the microaperture
and contamination of particles scattered at the edge of the microaperture.
Copyright 2004 by Taylor & Francis Group, LLC
Beam focusing using a multiplet of quadrupole lenses [30,31] is a sophisticated
technique to reduce the beam size to less than 5 Am. A schematic diagram of the focusing
system for microbeam production is shown in Fig. 9. The object of the primary beam is
limited by the rst slit, followed by a series of second slits used for divergence dening and
for beam collimation. An image of these slits can be projected at a focal plane by means of
the focusing lenses. The microbeam size is determined by the demagnication factor of the
lens system. Spherical and chromatic aberrations in the lens system are also taken into
account to minimize the beam size on the basis of the classical optical theory for charged
particles [35].
Microbeam Systems at the TIARA Facility
Three microbeamsystems were developed at the TIARAfacility for application to materials
science and biotechnology. Aheavy-ion microbeamsysteminstalled on a beamline of the 3-
MV tandem accelerator is the rst one developed to study single-event upset (SEU) of
semiconductor devices used for space [36]. The microbeam system can focus heavy-ion
beams such as a 15-MeVnickel ion with a spot size of less than 1 Am. In order to observe the
SEUphenomena at a specic position of the microdevice, the microbeamsystemis equipped
with a single-ion hit system, consisting of single-ion detectors and a fast beam switcher.
A light-ion microbeam system connected with the 3-MV single-ended accelerator was
developed for high-resolution ion beam microanalysis [37]. The highest spatial resolution
of 0.25 Am was achieved for 2-MeV proton and helium ions. The beam spot size was
estimated from the intensity distribution of the secondary electrons emitted from a silicon
relief pattern irradiated with the 2-MeV helium ion microbeam as shown in Fig. 10.
PIXE analysis using the microbeam has an overwhelming advantage in visualizing
very small quantities of elements with a very high sensitivity. For example, 2-D distribution
of small amounts of elements included in cells can be obtained by the micro-PIXE tech-
nique. An in-air micro-PIXE system using a light-ion microbeam was developed at the 3-
MV single-ended accelerator facility [38]. A schematic diagram of the in-air micro-PIXE
system is shown in Fig. 11. A proton microbeam with spatial and targeting resolutions of
1 Ampenetrates a 4-Am-thick Mylar lm, used as a sample backing and a vacuumpartition,
Figure 9 Schematic diagram of the focusing system for the microbeam formation.
Copyright 2004 by Taylor & Francis Group, LLC
Figure 11 Schematic diagram of the in-air micro-PIXE system.
Figure 10 Estimation of the spot size of the 2-MeV helium ion microbeam from the image of a
silicon relief pattern obtained by detecting secondary electrons.
Copyright 2004 by Taylor & Francis Group, LLC
and it is delivered to an experimental sample mounted on an annular sample holder in the
atmosphere. The X-rays emitted from the elements bombarded by the incident proton are
measured by a Ge or Si(Li) detector placed near the sample. The microbeam is scanned
horizontally and vertically to obtain 2-D element distribution. The elemental species is
identied by the X-ray energy. The in-air micro-PIXE system enables multielemental
mapping of samples in atmospheric environment, which avoids drying the samples.
Formation of the submicron microbeam contributes enormously to the enhancement of
spatial resolution in the micro-PIXE analysis. The micro-PIXE analysis technique is widely
applied to various research elds such as biomedical research, dentistry, environmental
science, and geology.
For living-cell irradiation using high-energy heavy ions, a vertical microbeamline was
developed at the JAERI AVF cyclotron facility. Asingle-ion hit technique was also applied
to the microbeam system to irradiate the specic part of individual cells with a single ion
[33,34]. Heavy ions accelerated up to hundreds of MeV are transferred to the end of the
beam line and delivered to the atmosphere through the microaperture with an inner
diameter of 10 Am. A sample stage is placed just after the microaperture. The positioning
accuracy of the sample stage relative to the microaperture is better than 1 Am. Single ions
are detected with a plastic scintillation counter placed downstream of the sample stage.
Every time a single ion is detected, the ion injection into the cyclotron is inhibited by kick-
ing out the ion beam with a beam switcher installed on the injection line until the single-
ion detection system and a data processing system become ready for the next single-ion
event.
Commissioning of a newmicrobeamsystemis in progress to improve the resolution of
the high-energy heavy-ion microbeam. The new microbeam system has been installed on
another vertical beam line of the JAERI AVF cyclotron. A schematic diagram of the
microbeamsystemis shown in Fig. 12. The spot size of hundreds of MeVheavy-ion beams is
expected to be reduced to less than 1 Am by using a quadruplet of quadrupole lenses in
combination with a series of slits. The microbeam system is also equipped with a single-ion
hit system.
3.6. Future Prospects of the Ion Beam Technologies
The research and technology elds of the ion beam applications will continue to expand
hereafter. Further development of the ion beam technologies is indispensable for the prog-
ress of the state-of-the-art research and technology. For example, ion beams with nano-
meter or subnanometer size are expected to be applied to the processing of materials with
hyperne structure and 3-D nanoanalysis. Pinpoint targeting of ions at a base of DNA on
nanometer scale will bring a breakthrough in biotechnology, especially for plant breeding.
A remarkable progress in biotechnology and in materials science is envisaged by the
applications of heavy ions with energy of more than 100 MeV/nucleon. Relative biological
eectiveness (RBE) in plant cells is enhanced in the LETrange from200 to 300 keV/Am. The
GeVheavy ions provide suitable range and LET for plant breeding to generate useful plants
such as UV-resistant crops, disease-resistant crops, insect-resistant crops, and environment-
remediable plants. The GeV heavy ions can make an almost straight hole along their paths
in an organic lm with a very high aspect ratio. The lm will be used to develop newly
functioned devices such as a precise lter for selecting atoms and molecules and an organic
semiconductor device with which ultrafast data processing may be possible. The heavy-ion
beam will also be useful for a radiation test of fabricated semiconductor devices in an
atmospheric condition to simulate single-event phenomena in space.
Copyright 2004 by Taylor & Francis Group, LLC
4. APPLICATIONS TO SPACE AND MATERIAL SCIENCE
AND ENGINEERING
Using ion beam technologies denoted in Section 27.3, research and development of
advanced materials and devices have been made for their space, nuclear, and industrial
applications. As typical examples of such researches, evaluation and development of semi-
conductor devices for space application, characterization of radiation degradation of nu-
clear materials, beamanalysis and modication of inorganic materials, and beamprocessing
of organic materials are described in this section.
4.1. Evaluation and Development of Semiconductors Used in Space
Articial satellites, which are now used for communication, broadcast, weather forecast,
etc., are equipped with a variety of semiconductor devices, which are often exposed to the
high levels of radiation found in space. Such energetic particles, called cosmic rays, cause
the degradation and malfunction of semiconductor devices, which lowers both the mission
lifetime and reliability of satellites. Using ion beam irradiation facilities at TIARA, which
have been uniquely adapted for simulating the radiation environment of space, we have
Figure 12 Schematic diagram of the high-energy heavy-ion microbeam system using a quadruplet
of quadrupole lenses and slits. The focusing-type microbeam apparatus is combined with the single-
ion hit system, similar to the existing single-ion hit system installed in the collimation-type micro-
beam apparatus.
Copyright 2004 by Taylor & Francis Group, LLC
been examining and characterizing the radiation resistance of semiconductor devices
intended for space applications. Radiation eects on semiconductor devices have been
investigated to clarify the mechanisms of radiation-induced degradation, and as a result, to
propose a way for developing better radiation-hard devices.
Radiation Degradation of Solar Cells
For an example of evaluating radiation degradation of semiconductor devices, the irradi-
ation test results of silicon (Si) and gallium arsenide (GaAs) solar cells using protons and
electrons are shown here. Solar cells, which are the main power sources for articial satel-
lites, are degraded predominantly by high-energy protons and electrons in space because
the existence probability of such particles is high in their mission trajectories such as the
geostationary orbit. Their degradation results from the introduction of displacement
damage in semiconductors, and thus, control of irradiation parameters of charged particles
such as uniformity of beam ux is important for detailed analysis of radiation tolerance of
solar cells. In the tests using a proton beamaccelerated at 10 MeVwith AVFcyclotron, solar
cell samples were irradiated uniformly in an area of 10 10 cm
2
with a magnetic beam
scanner. In irradiation of protons at energies from 1 to 6 MeV using a tandem accelerator,
the beamscan area was 4 4 cm
2
. In both cases, the nonuniformity of the irradiation uence
was kept within 3%. Electron irradiation of 1 MeV to solar cells was also performed by
using Cockcraft-Walton-type or Dynamitron Cascade-type accelerator at JAERI. The
performance of solar cells was examined under illumination of pseudosunlight at zero air
mass (AM0) condition before and after irradiation, or during irradiation by using simulta-
neous irradiation technique of charged particles and pseudosunlight [39].
Figure 13 shows typical results of the degradation of crystalline Si solar cells having
a back surface eld and reector structure (Si-BSFR), which were qualied by National
Space Development Agency of Japan (NASDA) for space usage, when irradiated by 10-
MeVprotons and 1-MeVelectrons. The pn junction of the cell samples, with a size of 2 cm
2 cm 50 Am, was fabricated by phosphorus (P) doping to a depth of 0.15 Am into boron
Figure 13 Dependence on 10-MeV-proton and 1-MeV-electron uence of the remaining factor of
short circuit current (I
sc
) for Si-BSFR solar cells. Circles and squares represent the results for
samples irradiated with 10-MeV protons and 1-MeV electrons, respectively. Broken lines represent
the tting results based on a model mentioned in the text.
Copyright 2004 by Taylor & Francis Group, LLC
(B)-doped, p-type substrate with a resistivity of 10 V cm. As shown in Fig. 13, which is a
gure of the electrical performance of solar cells, normalized short-circuit current I
sc
decreased with increasing uence up to 1 10
13
/cm
2
for proton irradiation and 1 10
16
/
cm
2
for electron irradiation. The logarithmic degradation observed in such uence ranges is
explained by the conventional model, i.e., a decrease in the minority carrier lifetime [40].
Over these uences, we found an anomalous change in I
sc
, i.e., I
sc
recovered about 5% to
10%, and abruptly dropped to nearly zero at proton uences around 1 10
14
/cm
2
and
electron uences around 1 10
17
/cm
2
. The conventional model cannot account for this
anomalous phenomenon. Based on an analysis of all the results obtained, the following
model was proposed: The hole concentration in the p-type substrate decreases owing to the
introduction of lattice defects, which act as hole traps. It raises the Fermi level of the
substrate corresponding to a change in the electrical properties of the substrate from p-type
to intrinsic. Thus, the depletion layer extends toward the backside of the solar cell, which
enhances the eciency of carrier collection and consequently the I
sc
recovers. When the cells
are irradiated further, minority-carrier mobility in the depletion region decreases and thus a
rapid decrease in drift length takes place. As the drift length becomes shorter than the width
of the depletion layer, electrons generated by light illumination do not reach the pn junction,
leading to I
sc
=0. Simulation results using the model are also shown by broken lines in Fig.
13, indicating that the experimental data are well described by the proposed model [39].
Based on the model described above, the degradation of Si solar cells equipped in a
Japanese articial satellite ETS-6 has been predicted. The result is illustrated in Fig. 14.
Actual ight data are also plotted in the gure. It should be noted that the obtained data are
in good agreement with prediction, indicating high reliability of the method. The results
obtained in this research have been summarized in Handbook of Si Solar Cells for Space
Application [41], which is used as an important database on space Si solar cells developed in
Japan. In addition, both junction-type Si solar cells with high radiation resistance have been
developed from these results [42].
GaAs solar cells have also been evaluated at TIARA because in 1998 the articial
satellite COMETS, which was equipped with similar cells, orbited across a strong ionization
Figure 14 Prediction (solid line) and actual ight data (squares) of the degradation of Si solar cells
equipped in ETS-6.
Copyright 2004 by Taylor & Francis Group, LLC
radiation environment called the van Allen belt due to its incorrect positioning. The char-
acterization of proton and electron irradiated cells has been used to predict the total lifetime
of the COMETS satellite. This has contributed to a revision of the COMETS initial mission
program. On the basis of those experiences, at present we have been studying radiation
eects on newtypes of solar cells, such as InGaP/GaAs/Ge triple-junction solar cells. It was
found from those investigations that the improvement of radiation resistance of GaAs
middle cells is a key issue for developing higher radiation resistant InGaP/GaAs/Ge solar
cells [43].
Radiation-Induced Malfunction in Electronic Devices
In contrast to the radiation degradation of solar cells, malfunction called SEU is caused by
irradiation of energetic particles into large-scale integrated circuits (LSIs) such as highly
integrated memory devices. Because a malfunction of electronic devices is regarded as one of
the most serious problems in space, the evaluation of single-event tolerance of LSIs is
indispensable for their space applications. It is well known that SEUs are triggered by the
generation of a high concentration of charge (highly dense electronhole plasma) in
semiconductors due to impingement of a heavy ion. The probability of SEU depends on
the LET of an ion in a device substrate, typically Si, because the value of charge generated in
semiconductors is proportional to LET. For accurate examination of SEUs taking place in
an LSI by a single ion, a special apparatus has been designed and installed in TIARA. By
using the apparatus, high-energy heavy ions accelerated by AVF cyclotron irradiate sample
devices at low uence rates, e.g., several hundreds of ions per second. We also used cocktail
beams developed in TIARA to make SEU tests eciently. Details of cocktail ion beams
have been described in the previous section.
Typical results obtained for the 256-kbit static random access memories (SRAMs),
which were developed for space application by NASDA, are shown in Fig. 15. The SEU
cross section increased with LET in a range above a threshold value and reached a constant
Figure 15 Single-event upset cross section for 256-kbit SRAM as a function of LET. Closed circles
and solid line represent experimental and tting results, respectively.
Copyright 2004 by Taylor & Francis Group, LLC
value called the saturated cross section. The threshold LET and the saturated cross sec-
tion can be related to the critical charge necessary to cause SEU and to the size of memory
cells, respectively. It was determined from those tests that the threshold LET and the sat-
urated cross section were 6.4 MeV/mg/cm
2
and 3.2 10
8
cm
2
, respectively, for the 256-kbit
SRAM. Here the threshold LET is dened as the LET value corresponding to one hun-
dredth of the saturated cross section. Using these values, the SEU tolerance for the devices
in space have been estimated. As a result, it was deduced that SEU occurs approximately
once a week in the geostationary orbit, showing its strong radiation resistance sucient for
space application. Based on results obtained at TIARA, radiation-hard devices like 1Mgate
arrays have been newly developed for components used in articial satellites [44].
Because SEU occurs when the amount of electric charge collected in a memory cell
exceeds a critical value, an understanding of the transportation of charges induced in mem-
ory devices by ion irradiation is indispensable to clarify the mechanism of SEU. For this
purpose, a microbeam irradiation system was designed and installed in a beam line con-
nected tothe tandemaccelerator. Details of the microbeamsystemhave beendescribedinthe
previous section. To measure transient currents in several nanoseconds, a wide-bandwidth
measurement systemwith a charge sensitivity of 20 fCand a temporal resolution of 10 ps was
developed. The developed system is schematically illustrated in Fig. 16. Using the system,
charge collection behaviors for several kinds of devices such as Si pn junction diodes, GaAs
Schottky diodes, etc., have been examined. Typical results obtained for silicon-on-insulator
(SOI) devices are described below [45].
For the development of SEU-hardened memory devices, it is expedient to reduce
charge collected in a memory cell. For this purpose, the formation of buried oxide in device
structures, i.e., the fabrication of SOI structure, is considered a useful method because such
a buried oxide layer can be expected to suppress the charge collection due to the drift and
funneling processes. However, no experimental approach had been made for the charge
collection in SOI devices. To investigate the charge collection in the SOI structure, transient
currents induced in SOI pn junctions by heavy ions such as 15-MeV carbon (C) or oxygen
(O) ions have been measured.
The samples were pn junction diodes formed on an SOI wafer fabricated by wafer
bonding technique. The thicknesses of the n-type top Si layer with resistivity of 24 V cm,
the oxide layer, and the n-type Si substrate with resistivity of 150 V cm are 5.7, 0.48, and
630 Am, respectively. The p region, which is 50 Am in diameter and 0.5 Am in depth, was
Figure 16 Schematic drawing of the system developed for transient current measurement using
heavy-ion microbeam.
Copyright 2004 by Taylor & Francis Group, LLC
formed by diusion of B. The sample diodes were mounted on a chip carrier with 50-V
double-ended microstriplines. They were irradiated with 15-MeVCor 15-MeVOions using
a heavy-ion microbeam, and transient currents induced in the sample diodes by ion
irradiation were monitored using a 50-GHz wideband sampling oscilloscope and a super-
conducting delay line, which were connected with the p contacts.
Figure 17 shows the transient currents induced in the SOI pn junction diode irradiated
with 15-MeV C ions when the reverse bias applied to the junction was 10 V. The result
obtained for the bulk Si pn junction diode when the reverse bias was 10 Vis also displayed in
Fig. 17 for comparison. The values of currents obtained for the SOI sample were shown to be
lower than those for the bulk Si diode. The amount of charge collected in the p region can be
derived fromtime integration of the obtained transient currents. For the SOI diode, the total
amount of collectedcharges was estimatedto be 210 fC, which was approximately one half of
that obtained for the bulk Si diode (470 fC). The lowvalues of collected charges in the SOI pn
junction can be interpreted as the only charges generated in the top Si layer reach to the p
region, i.e., the buried oxide suspends the charge conduction arising from the tunneling
eect.
The amount of positive charge created in a depletion region can be estimated
theoretically by using the length of a depletion layer, the value of LET for incident ions,
and the ion energy necessary to produce an electronhole pair in Si (3.6 eV). Assuming that
all positive charges generated in the top Si layer of the SOI device by ion irradiation were
transferred to the p contact, the amount of collected charge was estimated to be 260 fC. The
total amount of collected charge obtained from transient currents in the SOI device was
comparable with the estimated value. It indicates that the charge collection in the SOI pn
junction is limited in the top Si layer, whereas the penetration depth of 15-MeV C ions
(about 14 Am) is much deeper than the top Si layer. Similar results were also obtained for 15-
MeV O ion irradiation. All the above ndings demonstrate that the SOI structure has a
strong SEU resistance, and therefore they are suitable for the application to electronic
components used in space.
Figure 17 Transient current induced in SOI pn junction diode by 15-MeV C-ion irradiation (solid
line). The result obtained for bulk Si pn diode is also shown as a dotted line in the gure for
comparison. The reverse bias of 10 V was applied to pn diodes during measurements.
Copyright 2004 by Taylor & Francis Group, LLC
Development of Radiation-Resistant SiC Devices
Silicon carbide is regarded as a promising material for electronic devices used in space due to
its outstanding electrical properties [46] and excellent radiation resistance [47]. For its
application to highly integrated devices, it is indispensable to establish a selective doping
technique of electrically active (donor and acceptor) impurities. Because diusion technique
cannot be employed due to the extremely low diusion coecients of impurities in SiC, ion
implantation is the most promising way to dope impurities in selective regions of SiC.
For the electrical activation of implanted impurities, removal of defects by post-
implantation annealing is required. Whereas doping of high concentration of donors or
acceptors is necessary to produce low resistive regions, amorphous SiC layers subjected to
high-dose implantation at room temperature (RT) can hardly be recovered by postim-
plantation annealing. In this case, hot implantation is expected to be eective for avoiding
amorphization of ion-implanted SiC, because induced defects are partially removed during
implantation. The inuence of hot implantation on the electrical activation of P donors in
SiC have been investigated [48]. Fourfold implantation of P ions at 80, 100, 150, and 200
keVwas performed for p-type SiCwith a net acceptor concentration of 4 10
15
/cm
3
to form
a box prole of P atoms with a mean concentration of 1 10
18
to 5 10
19
/cm
3
and a depth
0.25 Am. Implantation temperature was altered in the range from RT to 1200jC. After
implantation, the samples were annealed in an Ar ambient up to 1500jC for 20 min. Ohmic
contacts were formed by A1 deposition on the P-implanted layers. The electrical properties
of the samples were examined by Hall measurements using van der Pauw method.
Figure 18 shows the carrier (electron) concentration in SiC samples implanted with
P at 5 10
19
/cm
3
at RT or 1200jC as a function of annealing temperature. The carrier
concentrations measured at RT are plotted in this gure. The carrier concentration for both
samples increased with annealing temperature. This indicates an increase of electrical
activation of P and a decrease of residual defects due to annealing. As for the comparison
between RT and 1200jC implantations, samples implanted at RT showed lower carrier
concentration as compared to samples implanted at 1200jCat every annealing temperature.
Since, in this sample, an amorphous layer was produced by implantation, such a highly
defective layer is thought to remain even after 1500jC annealing, which causes poor elec-
Figure 18 Free carrier (electron) concentration in SiCsamples implanted with P ions at RT (circles)
and 1200jC (squares) as a function of annealing temperature. Annealing was performed for 20 min in
Ar atmosphere. The electron concentration was obtained from Hall eect measurement at RT.
Copyright 2004 by Taylor & Francis Group, LLC
trical activation of implanted P atoms. The relationship between implanted P concentration
and implantation temperature has been investigated to clarify optimum condition of P
doping. It was found that the eects on hot implantation enhanced with increasing im-
planted P concentration, and the optimumimplantation temperature depends on implanted
P concentration [48]. For example, for high-dose P implantation, the highest carrier concen-
tration was obtained in samples implanted at 800jC. In addition, from positron annihila-
tion spectroscopy (PAS) study, this result can be explained in terms of the migration and
formation of vacancy defects due to annealing [49].
In fabrication of p-type SiC, doping of acceptors such as B and aluminum (Al) is one
of the key issues because their electrical activation ratio (atoms activated electrically/
implanted atoms) was not high enough even when hot-implantation technique was adopted.
It was reported that B and Al atoms located at Si sublattice sites act as shallow acceptors in
SiC[50]. In addition, the incorporation of Band Al in SiCcould be controlled by the change
in Si/C ratio during epitaxial growth of SiC [51]. From these ndings, the introduction of
additional C atoms by implantation, called C coimplantation, is expected to enhance the
electrical activation of acceptors. The eects of Ccoimplantation on the electrical activation
of Al and B acceptors in SiC have been investigated. It was found that the electrical prop-
erties of B- and A-implanted SiC are improved by C co-implantation [52,53]. In this study,
n-type 4H-SiC samples with a net donor concentration of 5.3 10
15
/cm
3
were used. The
samples were implanted with B ions at RT. Abox prole of B with a mean concentration of
5 10
18
/cm
3
and a depth of 0.5 Am was achieved by conducting a vefold implantation.
Prior to the implantation of B, a vefold implantation of C or Si ions was performed at RT
or 800jC to form a similar box prole to that of B. After implantation, samples were an-
nealed at 1630jC for 30 min in a SiC crucible in pure Ar ambient. Ohmic contacts were
formed using the deposition of Al and subsequent sintering at 950jC for 3 min in vacuum.
Hall measurements were carried out in a temperature range from 180 to 700 K using a van
der Pauw arrangement.
Figure 19 shows the free hole concentration as a function of reciprocal temperature for
the samples implanted with B, co-implanted with C and B (C/B) and co-implanted with Si
Figure 19 Free hole concentration as a function of reciprocal temperature for B- (triangles), C/B-
(circles), and C/Si-implanted (squares) SiC. Solid lines represent the tting results.
Copyright 2004 by Taylor & Francis Group, LLC
and B (Si/B). The ionization energy was determined from the slope of these curves (solid
lines in the gure) to be equal to 285 meV, which agrees with the B ionization energy in 4H-
SiC [54]. This indicates that implanted B species are activated electrically in all these
samples. The free hole concentration was found to increase by the C coimplantation and to
decrease by the Si coimplantaion. The hole concentration at RT in samples coimplanted
with C at 800jC and B was approximately 7.1 times as high as that in the sample implanted
with only B. Considering that hole concentration at RT in samples coimplanted with C at
RT and B was approximately 4.1 times as high as that in the sample implanted with only B
[52]; the results obtained indicate that hot coimplantation of C is quite eective for im-
proving the electrical properties of B acceptors. A similar increase in the free hole con-
centration due to Ccoimplantation was also obtained for Al acceptors in SiC [53,55]. It was
also observed that the intensity of the PL peak at 383.9 nm, which arises from shallow B
acceptors, increased by C coimplantation. In addition, the value of the D
II
center, which is
related to C interstitials, increased by coimplantation of C. These results are interpreted in
terms of a site competition eect and/or an alternation of the degree of compensation. In
conclusion, the C coimplantation technique is quite eective to form p-type layers in SiC.
Oxidation for forming insulator on SiC is also one of the key issues for the fabrication
process of SiC devices. The relationship between the oxidation process and the electrical
characteristics of SiC metaloxidesemiconductor eld eect transistors (MOSFETs) have
been studied [56,57]. A photograph of the enhancement-type SiC MOSFETs fabricated in
this study is shown in Fig. 20. The source and drain regions of the MOSFETs were formed
using P hot implantation mentioned above. In this study, oxide on SiC was formed by
pyrogenic oxidation (a wet oxidation process using direct reaction between hydrogen and
oxygen gases) and subsequent thermal treatment in steam or H
2
atmosphere. Details of the
fabrication process of the MOSFETare described in Ref. [57]. Fromthis study, it was found
that the postoxidation annealing methods can improve the channel mobility of SiC
MOSFETs, which is one of the most important characteristics, although the obtained
value was only 30% of the ideal value. Furthermore, the radiation response of the SiC
MOSFETs has been examined. It was found that SiCMOSFETs showed approximately 100
times higher radiation resistance than Si MOSFETs. [58,59].
Figure 20 Photograph of the enhancement-type SiC MOSFET.
Copyright 2004 by Taylor & Francis Group, LLC
4.2. Characterization of Radiation Degradation of Nuclear Materials
The investigation of radiation degradation of materials used in nuclear fusion reactors as
well as ssion reactors is indispensable for their safety and long-term operation with a high
reliability. In fusion reactors, structural materials, solid tritium breeder materials, and
insulating ceramic materials will be exposed to fast (14 MeV) neutrons generated via DT
reactions in high-density plasma. When reactor materials are irradiated with such fast
neutrons, heavy displacement damage as well as He and H atoms, which are produced by
transmutation reactions like (n, p) and (n) are introduced into the materials.
The Development of Triple-Beam Irradiation Apparatus
In order to develop materials for fusion reactors, it is important to investigate synergistic
eects of displacement damage and transmutation-induced gas products on their physical
and mechanical properties. For this purpose, a triple-beam irradiation apparatus has been
designed and installed at TIARA. Aphotograph of the triple-beam irradiation apparatus is
shown in Fig. 21. Using the apparatus connected with a 3-MV tandem accelerator, a 3-MV
single-end accelerator, anda 400-kVion implanter, three species of ions, e.g., heavy ions, He,
and H can be simultaneously irradiated into samples for simulating the radiation environ-
ment of a fusion reactor. Each beam line has an angle of 15j from the other beam line, and
three kinds of beams are focused on the target sample in a vacuum chamber. Besides, for
wide-area irradiation with high uniformity, a beam scanning system is also installed in each
line. Since Al foils are installed for energy degradation in the irradiation chamber, the
penetration depth of irradiated ions can be adjusted. The temperature of samples is also
controllable from 80 to 1300 K. Details of the apparatus were described elsewhere [60].
Austenitic Steel, Ferritic/Martensitic Steel, and Vanadium Alloy
Using the triple-ion beam irradiation apparatus, the microstructural evolution of austenitic
stainless steel, which is considered as a structural material for water-cooled fusion reactors
Figure 21 Photograph of the triple-beam irradiation apparatus.
Copyright 2004 by Taylor & Francis Group, LLC
such as the international thermonuclear experimental reactor (ITER) has been investigated
[61,62]. The austenitic stainless steel samples with a composition of Fe16Ni14Cr0.25Ti
2.5Mo in wt.%were irradiated simultaneously with 12-MeVNi, 1-MeVHe, and 350-keVH
ions in a temperature range between 300j and 400jC. The irradiated samples were observed
by transmission electron microscopy (TEM), and the number density of dislocation loops in
the irradiated region was studied.
Figure 22 shows the depth distribution of the number density of dislocation loops for
the sample irradiated with 12-MeV Ni, 1-MeV He, and 350-keV H triple-ion beams. The
number density of dislocation loops decreases in a region between 1.0 and 1.5 Am. It was
reported that hydrogen interacts with dislocation in steels at a relatively low temperature
[63]. Considering that the estimated projection range of He and Husing TRIMis around 1.5
Am, the results obtained in this study suggest that the generation of dislocation loops is
suppressed by the interaction of defects and He and/or H. In addition, very small cavities
whose average diameter and density were 1.4 nmand 1.5 10
24
/cm
3
respectively were found
in the triple-beam-irradiated region. Since such high-density small cavities were not
observed in a region irradiated with only 12-MeV Ni ions, the creation of high-density
small cavities is thought to be an eect of simultaneous irradiation of Ni, He, and H.
Sekimura et al. [64] reported that high-density small cavities (diameter: 10 nm, density: 10
21
/
cm
3
) were created in austenitic stainless steel samples implanted with dual beams of Al and
He ions. Therefore, the creation of small cavities is one of the synergistic eects of
displacement damage and transmutation-induced gas products. The growth of the cavities
in F82 was also obtained in the region irradiated with triple beams. The growth strongly
depends on heat and mechanical treatment process [65]. Since such a growth relates to
microhardness and leads to swelling of the steel, this nding is considered very important for
the predicted lifetime of steels. Besides, the remarkable swelling for pure vanadium (V) and
V alloys, which are also candidates for fusion materials due to triple-beam irradiation, was
found [66]. Because this swelling could not be attributed to single-or/and dual-beam
irradiation, this also can be concluded to be synergistic eects.
SiC/SiC Composites and Other Materials
Ceramic matrix composites are expected to be applied to structural materials for fusion
reactors because of their excellent mechanical properties at high temperature and their
Figure 22 Depth distribution of the number density of dislocation loops for austenitic stainless
steels irradiated with 12-MeV Ni, 1-MeV He, and 350-keV H triple beams.
Copyright 2004 by Taylor & Francis Group, LLC
noncatastrophic failure behavior. SiC/SiC composite materials are regarded as one of the
attractive candidates for fusion materials. In addition to the excellent mechanical properties
as ceramics, SiC/SiC materials have an advantage of low residual radioactivity. The eects
of triple-beam irradiation on SiC/SiC materials are described below. SiC-ber-reinforced
SiC-matrix (SiCf/SiC) composites and a monolithic SiC(h-SiC) were used in this study [67].
The samples were irradiated with triple beams (6-MeV Si
2+
, 1.2-MeV He
+
and 250-keV
H
+
ions) at 1000jC. As a result of TEM observation, He bubbles and cracks were found in
SiC matrix, and the number of He bubbles and cracks strongly depended on the fabrication
process of SiC/SiC composites. The size of the bubbles also depended on the fabrication
process of SiC/SiC composites. The turbostratic carbon structure was formed in the carbon
interphase layer due to irradiation, although the interphase layer was the nongraphitic
carbon structure before irradiation. Furthermore, a new crystalline SiCphase was observed
in the carbon interphase irradiated with Si ions. Details of irradiation eects on SiC/SiC
composites are described in Refs. [6769].
In addition to SiC/SiC, irradiation eects on insulators such as Al
2
O
3
were also
examined [70]. As a result, synergistic eects such as the diusion of H and the growth of
cavities were observed in samples irradiated with triple beams. As for carbon ber com-
posites, the microhardness increased by triple-beam irradiation [71], which was a dierent
result from that obtained from 14-MeV neutron irradiation (8 10
19
n/cm
2
) experiment.
4.3. Beam Analysis and Modification of Inorganic Materials
Applying ion beams, surface-sensitive analysis and modication in atomic and electronic
structures of inorganic materials have been developed. Ion beam modication of titanium
dioxide (TiO
2
), carbon-based materials, and the analysis of Nb/Cu multilayers and VO
2
using ion beam are described as follows.
Modication of the Characteristics of TiO
2
Titanium dioxide has many attractive characteristics for photoactive applications, such as
photocatalytic coatings and dye-sensitized solar cells. However, its large optical bandgap
(3.03.2 eV) is a disadvantage taking into account the utilization of sunlight. Doping TiO
2
with dierent impurities using ion implantation is known to provide an eective modica-
tion of its electronic structures, thereby improving the separation of the photogenerated
electronhole pairs and/or extending the wavelength range of the TiO
2
photoresponse into
the visible region [72]. Therefore, the eects of uorine (F) and chromium (Cr) implanta-
tions on the characteristics of TiO
2
have been taken up [7375].
First, the results of F implantation into TiO
2
will be described. Single-crystalline TiO
2
rutile with a h001i crystallographic axis was implanted with 200-keV F ions at a uence
range from1 10
16
to 1 10
17
/cm
2
and subsequently annealed in air. The radiation damage
and its annealing behavior were investigated by Rutherford backscattering spectrometry in
channeling geometry (RBS/C) and PAS. X-ray photoelectron spectroscopy measurements
conrmed the occupation of F atoms on O-lattice sites in the outermost region of the
implanted surface layer. Secondary ion mass spectrometry (SIMS) was employed
for probing F distribution in the as-implanted and post-annealed samples. The electronic
structures of the F-doped TiO
2
were also evaluated by theoretical calculations to predict the
doping eect on the spectral response of TiO
2
[74].
The disorder peaks in RBS/C spectra lessened during isochronal thermal treatment.
This can be attributed to a columnar regrowth within the total thickness of the damaged
Copyright 2004 by Taylor & Francis Group, LLC
layer [76]. The pronounced recovery in the Ti and O sublattices was achieved by such a
recrystallization process due to annealing up to 1200jC. PAS study revealed that the crystal
structure was recovered by the migration of vacancy-type defects to the surface. According
to the SIMS results shown in Fig. 23, the F depth prole was shifted to a shallower region
along with the damage recovery and this resulted in the formation of the F-doped layer
where the impurity concentration steadily increased toward the surface. Importantly, this
method enabled F atoms to be introduced in a controlled manner at specic locations to
realize impurity concentration gradients in TiO
2
. From the ab initio band calculations, it
was found that the F doping caused a modication of the electronic structure around the
lower edge of the conduction band. This probably leads to a reduction in the eective
bandgap energy, in other words, to an improved optical response in the visible-light region.
The uorination of TiO
2
surfaces would enhance their chemical and optical stabilities and
open avenues toward photoelectronic materials with various applications. In this sense, F
ions have the possibility of acting as promising dopants in a TiO
2
photocatalyst.
The surface charge separation eciency was characterized using photo-induced
transient charge separation (PITCS) measurement proposed by JAERI [77]. By this
technique, the total number of photocarriers transported toward the surface can be
estimated as a function of illuminated photon energy without disturbing the spontaneous
surface band bending, because of contactless observation under no-applied bias. The PITCS
measurements were carried out for studying hole generation in TiO
2
. The samples used in
this study were TiO
2
rutile lms. The samples were implanted with Cr ions at 150 keV at a
dose of 5.0 10
15
ions/cm
2
and subsequently annealed at 600jC for 5 hr in air. For
comparison, uniformly Cr-doped TiO
2
rutile lms, highly oriented undoped TiO
2
anatase,
and TiO
2
rutile lms were also examined. Figure 24 shows the incident-photon-energy
dependence of the surface charge separation eciency of the samples [75]. The vertical axis
represents the ratios of the number of surface transported holes (h
+
) to the number of
induced photons. An increase in charge separation eciency for samples implanted with Cr
ions, and subsequently annealed, was observed, especially around 350 nm, as compared to
other samples. The total number of h
+
for Cr implanted lm was estimated to be
approximately 15 times higher than that for TiO
2
anatase. The obtained result suggests
that the charge separation properties are improved by making a gradient of Cr distribution.
Figure 23 Depth prole of F in TiO
2
rutile single crystals implanted with 200-keV F ions. (a) As-
implanted, (b) 600jC-annealed, and (c) 1200jC-annealed samples.
Copyright 2004 by Taylor & Francis Group, LLC
Carbon-Based Materials
In the research and development of inorganic materials, the formation mechanism and
thermal stability of carbon-based materials have been intensively investigated. The for-
mation processes of carbon onions (graphitic spheres) and nanocapsules in copper and gold
have been elucidated through simultaneous TEM observation under C-ion irradiation [78].
This apparatus consists of two sets of accelerators and a TEM. One of the accelerators is a
400-kV ion implanter at TIARA, which can accelerate more than 30 kinds of elements
including high-melting-point metals. The other one is a duoplasmatron ion source with
accelerating voltage of 40 kV. In this apparatus, the following techniques are available as
well as conventional transmission electron microscopy: high-resolution electron micro-
scopy, scanning electron microscopy, scanning transmission electron microscopy, CCD
camera and video-recording system and electron energy loss spectroscopy. For the
observation of C onions, copper (Cu) plates were implanted with C ions at temperatures
from300j to 700jC. The high-magnication micrograph of Conions is shown in Fig. 25. As
a result of TEMobservation during Cion irradiation, it was found that the accumulation of
C atoms above 1020 at.% (i.e., >1 10
17
C/cm
2
) and high enough temperature for the
migration of C atoms are needed for creating a nucleus of C onions in Cu plates. The
production rate of C onions using ion implantation was 6 to 8 times higher than that using
electron irradiation. Furthermore, the creation of C onions and nanocapsules was found to
depend on the quality of substrate materials. Details of this study are described in Ref. [79].
In addition to C onions, C atoms condense into various kinds of chemically bonded
forms, and they are known to have excellent physical properties depending on the bonding
nature. This means that research and applications not only in the materials science but also
in other scientic elds are expected. At JAERI, the optimum growth conditions have been
successfully obtained for the preparation of high-quality C
60
, diamondlike carbon, and
nanocrystalline diamond by means of ion-beam-assisted deposition [8082]. The suscept-
ibility of Ni/C
60
thin lms to thermal treatment, the formation of nanocrystalline diamond
and nanotubes due to codeposition of Co and C
60
, and the surface modication of glassy
Figure 24 Incident photon energy dependence of the surface charge separation eciency (h
+
/
photon) measured by PITCS. Squares, circles, triangles, and diamonds represent the results obtained
for Cr-implanted rutile, Cr-doped rutile, undoped rutile, and anatase TiO
2
lms, respectively.
Copyright 2004 by Taylor & Francis Group, LLC
carbon by hydrogen doping have also been characterized [8386]. In the following, materials
synthesized from C and C
60
are described as examples.
Ion beam deposition (IBD) was carried out by employing an acceleratordecelerator
system under ultrahigh vacuum. In this study,
12
C
+
ions were used as source, and Si(111)
and Ir(100)/MgO(100) single crystalline were used as substrates [87]. The typical ion beam
conditions were in the energy range of 50200 eVand in the beamcurrent around 1 mA. On
the other hand, as for the ion-beam-assisted deposition (IBAD) experiments, a special
chamber was designed. In the chamber, three dierent sources (C
60
, Ne ions, and transition
metals) can be installed. In the case of evaporating C
60
molecules, energetic Ne
+
ions were
also introduced in an energy range up to 5 keV. Ne
+
ion beamwas obliquely incident at 60j
relative to the surface normal. In the case of using ions with high Z element, the sputtering
eect should be considered during irradiation. The chamber can be evacuated up to 1.2
10
6
Pa, and the vacuum was kept at less than 5 10
5
Pa during IBAD experiments.
As a result of IBD and IBAD experiments [80], nanodiamonds immersed in the
dominant sp
3
amorphous carbon lms were found. Patterns were formed via the dynamical
process between the sputtering and the deposition eects [82]. Hexagonal nanosized
diamonds were also prepared from the C
60
vapor with the simultaneous irradiation of
1.5-keV Ne
+
ions at a temperature of 700jC. Furthermore, although C and iridium (Ir) are
immiscible, defects introduced by Cimplantation could favor the supersaturated Catoms in
the subsurface region. Fig. 26 shows an atomic force microscope (AFM) micrograph of
diamondlike carbon (DLC) lm on Ir(100) prepared by IBD method (100-eV
12
C
+
)
followed by thermal annealing at 600jC under He gas environment. After annealing,
DLC lm suered from blistering because of immiscibility between C and Ir. One can
recognize a symmetric, owerlike pattern of blistering with a needlelike dot at the center.
Figure 25 High-magnication micrograph of C onions.
Copyright 2004 by Taylor & Francis Group, LLC
The petals correspond to the bumps in blistering and the centered dot has been conrmed
to be nanodiamond by micro-Raman spectroscopic analysis. The growth from the
nucleated embryos was enhanced by a heat treatment. These results suggest that the
nucleation sites can be controlled by using ion beam technique. Therefore, it is concluded
that nanopattern as well as materials with interesting features may be synthesized by using
ion beam irradiation technique with changing ion energies and substrate temperature.
Study of Materials by Ion Beam Analysis
Ion beamanalysis performed on materials synthesized at TIARAis described in this section.
Multilayers in which dierent thin metal lms are stacked alternatively are considered as a
mirror material reecting short wavelength lights such as soft X-rays. To form high-quality
mirror materials, it is indispensable to measure the thickness of lms and the atness of the
lminterfaces. Niobium(Nb) lms deposited on single-crystal a-Al
2
O
3
, MgOsubstrates by
electron beamevaporation technique has been investigated by RBS/Cand X-ray diraction
analyses. Figure 27 shows 2.7-MeV
4
He
+
RBS spectra from Nb on a-Al
2
O
3
substrate
measured under random conditions and the h110i aligned conditions. As a result, it was
found that some amount of disorder exists around the interface region between Nb layer
and a-Al
2
O
3
substrate [88]. From the fact that the ratio of the h110i to random spectrum
yield from Nb lm is 0.044 at just behind surface peak, the excellent crystallinity of the lm
was conrmed. Multilayers consisting of Nd and Cu lms have also been examined to
understand the relationship between crystallinity and the fabrication process [89].
15
N
nuclear reaction analysis (
15
N NRA) was conducted using
15
N ions accelerated by 3-MeV
tandem accelerator in this study. Since
15
N NRA can detect small amounts of residual
hydrogen existing at the interface between Nb and Cu layers, the thickness of each layer can
be estimated. The results obtained from
15
N NRA are shown in Fig. 28. The calculated
result is also shown as a solid line in the gure. From the comparison between experimental
and calculated results, the optimum condition of the fabrication of the multilayers was
determined. Details are described in Refs. [88,89].
Figure 26 AFM micrograph of DLC lm on Ir(100) prepared by IBD method (100 eV
12
C
+
),
followed by thermal annealing at 600jC under He gas environment. After annealing, DLC lm
suered from blistering because of immiscibility between C and Ir. Size of the micrograph is 9.85
9.85 Am
2
.
Copyright 2004 by Taylor & Francis Group, LLC
The fabrication process of vanadium oxide (VO
2
) has also been studied using RBS/C.
Since optical and electrical properties of VO
2
are dramatically changed at 68jCdue to phase
transition, VO
2
is regarded as one of the candidates for thermally activated electronic or
optical switching devices for optical bers or sensors. To obtain the desired properties, the
development of the fabrication process for very thin lms, without crystalline defects on
various substrates, is required. Single-crystalline VO
2
thin lms on (0001) plane of a
sapphire substrate have been synthesized by a laser ablation method. The quality of VO
2
was examined by X-ray diraction and RBS/C method. The electrical resistance and the
optical transmittance of the VO
2
lm were measured under increasing and decreasing
temperatures. At a temperature of 68jC, an abrupt transition of resistance from metal to
Figure 27 RBS/C spectra of 2.7-MeV-
4
He
+
from Nb(110) epitaxial lm on a-Al
2
O
3
substrate [Nb
(140 nm)/Cu (42 nm)/Nb (48 nm)/a-Al
2
O
3
]. The 100-nm-thick Nb lms were fabricated at 750jC by
electron beam evaporation.
Figure 28 Hydrogen distribution observed in Nb/Cu multilayers by
15
N NRA (circles). The tting
result (solid line) is also shown in the gure. The structure of Nb/Cu multilayers is Nb (2 nm)/Cu/Nb
(5 nm)/Cu/Nb (10 nm)/Cu/Nb (20 nm)/a-Al
2
O
3
, where the thickness of Cu layer is 20 nm.
Copyright 2004 by Taylor & Francis Group, LLC
insulator was observed [90]. In addition, as a result of the study of Mo doping eects,
transition temperature was found to decrease with increasing Mo concentration [91].
4.4. Beam Processing of Organic Materials
Irradiation of heavy ions like Au and Kr deposits a heavy damage along the ion track in
polymer lms. When some polymer lms are irradiated with heavy ions and dipped in an
alkaline aqueous solution, micropores are produced along the ion track. Since the shape and
size of pores of polymers can be controlled by etching conditions and by the species of heavy
ions, these polymer lms are expected to be used for high-performance lters and ion
detectors [92]. The development of the polymer lters and ion detectors are described as
follows.
Thermoresponsive Polymer Films
Polymer lms that have useful functions such as selective transportation and absorption of
specic molecules are expected to be applied for high-performance lters used in medicine
and biology. The fabrication process of polymer lms with pores using heavy-ion irradi-
ation consists of the following four steps: (1) the preparation of a thin core lm, (2)
irradiation of heavy ions into the core lm, (3) the treatment of the core lm in an alkaline
solution, and (4) grafting of a polymer, which changes the shape responsive to temperature.
In this study [92], a copolymer consisting of diethylene glycol diallyl carbonate (DEBA) and
acryloyl-L-proline methyl ester (A-ProOMe) was used as the core lm. DEBA acts as a
comonomer with high sensitivity to ion beam irradiation, and A-ProOMe enhances the
anity to the polymers that are subsequently grafted. The core lms were irradiated with
84
Kr ions at 6.19 MeV/n at a uence of 10
7
/cm
2
. After irradiation, the lms were treated in
an aqueous 6 M NaOH solution at 60jC for 1 hr to make pores 1.3 Am in diameter. Then,
the lms were immersed in an aqueous A-ProOMe solution, and irradiated with gamma
rays at a dose of 30 kGy for grafting. A-ProOMe is a cross-linkable polymer to give a
temperature-responsive hydrogel. Thus, its volume increases by absorption of water belowa
critical temperature (14jC). The volume of swelled A-ProOMe is about 400 times larger
than nonswelled one. The diameter of cylindrical pores was 0.7 Amat 30jC. Figure 29 shows
the temperature dependence of the permeability of p-nitrophenol for A-ProOMe grafted
lms. No signicant change of the permeability was observed in temperature ranges below
12jC. However, a remarkable increase in permeability was found above 14jC, which
corresponds to the lower critical solution temperature. It is noteworthy that molecules can
be selectively separated by changing the volume of the grafted polymers in the cylindrical
pore, which are regulated by the change in temperature. Furthermore, when N-isopropy-
lacrylamide (NIPAAm) was used as a responsive material grafted on lms, polyethylene
glycol molecules with the same mass could be separated from polyethylene glycol with a
dierent mass by controlling temperatures, as shown in Fig. 30.
Ion Detectors Based on CR39
Polyallyl diglycol carbonate (CR39) is well known as a plastic track detector [93]. However,
the sensitivity of a plastic detector to ion irradiation should be greater compared to that of
CR39 for detection of lighter ion particles with higher energy, when the ion detectors are
applied to the eld of cosmic ray physics, nuclear physics, space radiation, and neutron
dosimetries. Therefore, The development of a newplastic track detector based on CR39 that
shows high sensitivity to proton beams are described next.
Copyright 2004 by Taylor & Francis Group, LLC
Several types of CR39 copolymers containing 13 wt.% of N-isopropylacrylamide
(NIPAAm) and 0.01 wt.% of Naugard 445 were prepared by cast polymerization. The
copolymerizations were performed with 3% diisopropyl peroxydicarbonate (IPP) as an
initiator at 70jC for 24 hr to give copolymer lms (CR39/NIPAAm). Details of the
fabrication process of CR39/NIPAAm are described in Ref. [94]. Protons at an energy of
10 MeV were incident onto CR39/NIPAAm. The eects of NaOH concentration and
temperature of an etchant to ion beam sensitivity of CR39/NIPAAm were investigated.
Results for CR39/NIPAAm show a broad maximum of sensitivity at 5 N NaOH at 70jC.
Figure 29 Temperature dependence of the permeability of p-nitrophenol for A-ProOMe grafted
lms. The permeability (P) was calculated from the equation, P=jV/AC, where j, V, A, and C are
the slope of the permeationtime curve, the volume of the receiver chamber, the surface area of the
membrane, and the concentration of p-nitropenol in the donor chamber, respectively.
Figure 30 Temperature dependence of the permeability (P) of polyethylene glycol molecules for
N-isopropylacrylamide (NIPAAm) grafted lms. Polyethylene glycol with four kinds of masses (1
10
2
, 110
3
, 210
3
, 710
4
) are contained in the solution.
Copyright 2004 by Taylor & Francis Group, LLC
For the comparison between CR39 and CR39/NIPAAm, the sensitivity of CR/NIPAm to
ion particles was much higher than that for CR39 in LET ranges below10 keV/Am[94]. This
indicates that CR/NIPAAm is very useful for an ion particle detector.
5. APPLICATIONS TO BIOLOGICAL SCIENCE
5.1. Mutagenesis in Plants
Introduction and Irradiation Methods
Since the discovery of mutation induction by X-rays in maize [95], numerous studies on
plant mutagenesis by ionizing radiation have been carried out. The biological eect of ion
beams was also investigated and it was found that ion beams show high RBE of lethality,
mutation, and so on compared with low-LET radiation such as gamma rays, X-rays, and
electrons [96,97]. Because ion beams deposit high energy on a target densely and locally, as
opposed to low-LET radiation, it is suggested that ion beams predominantly induce single-
or double-strand DNA breaks with damaged end groups whose reparability would be low
[98]. Therefore, it seems plausible that ion beams can frequently produce large DNA
alteration such as inversion, translocation, and large deletion rather than point mutation,
resulting in producing characteristic mutants induced by ion beams. However, the charac-
teristics of ion beams on the mutation induction have not yet been clearly elucidated.
As the rst facility of ion beams for biological and material use, TIARA was
established in 1991. The characteristics of ion beams for the biological eects and the
induction of mutation have been investigated using model plants, Arabidopsis, chrysanthe-
mum, tobacco, wheat, and so on. Several kinds of ion beams such as C, He, Ne, and Ar were
experimentedon in this study and mostly 220-MeVCions were used. All ions were generated
by the AVFcyclotron in TIARA. The physical properties of the 220-MeVcarbon ions are as
follows: Incident energy at the target surface was 17.4 MeV/u, mean LET in a target (0.25-
mm thickness) was estimated to be 110 keV/Am as water equivalent, and the range of ions
was ca. 1.0 mm. These physical properties were calculated by ELOSS code program, a
modied OSCAR code program [99]. Particle uences of the ions were determined using a
diethyleneglycol-bis-allylcarbonate (CR39) lm track detector. Electron beams (2 MeV;
JAERI, Takasaki, Japan) were also used as a low-LET ionizing radiation control.
In general, ion beams were scanned at more than 50 50 mm, and exited the vacuum
chamber through the beam window made of a 30-Am-thick titanium foil. The irradiation
sample was placed in the air at a distance of 10 cm from the beam window. In the case of
Arabidopsis or tobacco seeds, for example, 1003000 seeds were sandwiched between
kapton lms (7.5-Am thickness) to make a monolayer of the seeds for homogeneous ir-
radiation. In the case of rice or barley seeds, the embryo side was kept facing toward ion
beams, whereas for calli or explants in a petri dish, the lid of the petri dish was replaced with
a kapton-lm cover to decrease the loss of the energy of ion beams. Samples were irradiated
for less than 3 min for all doses.
Biological Eects of Ion Beams
Determination of biological eects of ion beams is an inevitable problemfor the application
of ion beams to mutagenesis in plants. In the previous investigations, it was shown that RBE
for growth inhibition and so on reached a peak at around 72174 [100] or 190 keV/Am[101],
suggesting LET dependence of RBE of higher plants is similar to that of mammalian cells.
However, in these experiments, the Bragg peak was adjusted on the target and/or only a few
Copyright 2004 by Taylor & Francis Group, LLC
LETs were investigated. To acquire more detailed data on biological eects by using
penetrating ions with various kinds of LET, several investigations have been carried out in
TIARA [102106]. Most eorts were focused on LET dependence of survival and other
endpoints of ion beams in the case of plant seeds. In general, all the experiments showed that
the RBE peak of survival and other endpoints was around ca. 230 keV/Am and more (Fig.
31). By using carbon ions with dierent LET, Hase et al. [106] indicated that the RBE for
survival and chromosome aberration in tobacco increased with increasing LET and showed
the highest value at 230 keV/Am. Shikazono et al. [105] found that the highest LET of the
peak RBE for lethality of Arabidopsis was around 221 keV/Am for carbon ions and around
350 keV/Am and more for neon and argon ions. Therefore, it is almost certain that the LET
having a maximum of RBE for lethality is higher in plant seeds than in mammalian cells.
Mutation Rate
In order to compare mutation frequency induced by 220-MeV Cions with that by low-LET
radiation (electrons), Arabidopsis visible phenotype loci were chosen as follows: transparent
testa (tt) whose seed coat is transparent because of lack of pigment; glaborous (gl), which
have no hair on their leaves and stems; and long hypocotyl (hy) whose hypocotyl is longer
than that of the wild type in the light condition. Mutation frequencies of tt, gl, and hy
induced by carbon ions were 8- to 34-fold higher than those by electrons (Table 2). In this
study, irradiation doses for the induction of mutation were determined from the RBE of
carbon ions compared with that of electrons on the survival of plants, which was
approximately 5. Both doses are at three-quarters of the shoulder dose of each survival
curve [104].
Mutation Spectrum
It is necessary to compare the mutation spectruminduced by ion beams with that by the low-
LETradiation to gure out a special feature of ion beams as an eective mutagen. Mutation
Figure 31 The relative biological eectiveness (RBE) of lethality as a function of linear energy
transfer (LET). (From Refs. 105 and 106.)
Copyright 2004 by Taylor & Francis Group, LLC
spectrumwas rst investigated on the ower color of chrysanthemumcv. Taihei (pink color)
by Nagatomi et al. [107]. The explants of leaf and oral petals incubated in agar medium
were irradiated with carbon ions of 220 MeV. After irradiation, the culture was transferred
to a new medium to induce callus. The mutation induction of the regenerated plants from
the callus was investigated, comparing the results with those with gamma-ray irradiation.
The mutation rates of ower color induced by ion beams were approximately half of those
induced by gamma rays in both oral petal and leaf. Most ower color mutants induced by
gamma rays were light pink, and a few were dark pink (Table 3). On the other hand, ower
mutants induced by ion beams showed complex and stripe types other than single color
(Fig. 32). The complex-type ower color mutants increased as the dose of ion beams rose.
The color spectrum of the ion-beam-induced mutants shifted from pink to yellow. Those
specic mutants such as complex and striped color types have never been obtained by gam-
ma-ray irradiation in the cultivar. Thus, mutation spectrum of ower color induced by ion
beams is wide and novel mutation phenotypes can be induced.
Recently, mutation spectrum of ower color and ower shape of carnation was also
investigated by Okamura et al. [108]. In the carnation variety Vital (spray-type, cherry pink
owers with frilly petals) tested, ower color mutants such as pink, white, and red were
obtained by X-ray irradiation, whereas the color spectrum of the mutants obtained by
carbon ion irradiation was wide such as pink, light pink, salmon, red, yellow, and complex
and striped types. Furthermore, many kinds of round shape of petals were induced in
addition to ower colors. These indicate that ion beams can induce novel ower color and
shape with high frequency.
Novel Mutants
Until now, several new mutants were induced by ion beams. Representative mutants will be
described below.
Table 2 Mutation Frequency Induced by Carbon Ions and Electrons
Mutagen (dose) Locus
Mutation frequency/locus/diploid
cell/dose (Gy) ( 10
6
)
Carbon ions (150 Gy) tt 2.3 (34-fold)
a
gl 1.7 (8)
Hy 2.0 (16)
Electrons (750 Gy) Tt 0.068
gl 0.21
hy 0.12
a
Compared to the frequency of the electrons.
Table 3 Mutation Spectrum of Flower Color in Chrysanthemum
Mutagen
Mutation frequency (%)
White Light pink Dark pink Orange Yellow Complex/Stripe
Unirradiated 0 0.3 0 0 0 0
Gamma rays 0 27.7 2.1 0 0 0
Carbon ions 0.3 4.6 0.3 0.3 0.2 10.2
Original variety Taihei with pink owers was used.
Copyright 2004 by Taylor & Francis Group, LLC
Arabidopsis as a Model Plant. In an attempt to isolate novel mutants by means of
ion beams, Arabidopsis [Arabidopsis thaliana (L.) Heynh.] was used as a model plant. This is
because thousands of mutants were already induced in Arabidopsis by using chemical
mutagens, X-rays, T-DNA mutagemesis, etc. Therefore, if a novel mutant phenotype is
discovered by using ion beams, it would be well grounded that ion beams are a powerful tool
as a new mutagen. As one of such a mutant phenotype, UV-B-resistant plants were chosen,
because many UV-B-sensitive mutants have been isolated in Arabidopsis but there has been
no report of mutants in any higher plants that are resistant to UV-B light. Four UV-B-
resistant mutants [ultraviolet-light insensitive (uvi) 14] have been isolated in 5100 M
2
fam-
ilies derived from carbon-ion-irradiated 1280 M
1
seeds. The fresh weight of uvi1, uvi2 and
uvi4, after 1 month under UV-B radiation, was 2-fold higher than that of wild type, and 1.5-
fold higher in the uvi3 (Fig. 32). Especially, uvi1 showed very great ability of both photo-
reactivation and dark repair. The reduction of CPD and (6-4) photoproducts that are two
major DNA damages caused by UV-B are both faster in uvi1 than in wild type, indicating
that uvi1 mutation would gain the DNA reparability [109].
New anthocyanin-accumulated or -defective mutants were also isolated. A new
mutant, anthocyanin spotted testa (ast), which has the spotted pigmentation on the seed
coat, has been isolated from 11,960 M
2
plants derived from 1488 self-pollinated M
1
.
Anthocyanin content was about 6 times higher at 6 days after owering than that of the
wild-type seeds [110]. Two new loci of anthocyanin-defective tt phenotype (tentatively ttA,
ttB) have been found along with the research of mutation frequency experiment (unpub-
lished data). Both genes seem to encode enzymes of anthocyanin biosynthetic pathway. On
the other hand, a novel ower mutant, frl1, has serrated petals and sepals, but the other
oral and vegetative organs appear to be normal (Fig. 32). FRL1 gene should act in petal
and sepal development in an organ-identity-specic manner [111].
Other Plants. As already described in Mutation Spectrum, complex and striped
types of ower color have been obtained in chrysanthemum. A higher mutation frequency
Figure 32 Novel mutants induced by ion beams. Top left: 1-month-old plants of Arabidopsis wild
type (top) and UV-B-resistant mutants (bottom) under high UV-B (1113 kJ/m
2
/day) condition. Top
center and right: complex (center) and striped (right) ower-color varieties of chrysanthemum.
Bottom left: ower of Arabidopsis frl1 mutant. Bottom center and right: new carnation varieties with
round petals (center) and Dianthus type petals (right).
Copyright 2004 by Taylor & Francis Group, LLC
of complex ower color mutants were derived from oral petal irradiation than that from
leaf irradiation [107]. Also, many kinds of ower color such as pink, light pink, salmon, red,
yellow, complex and striped types, and various kinds of round petals were induced fromthe
carnation variety Vital [108] (Fig. 32).
In order to develop an ecient procedure for obtaining a desired mutation, ion beams
were applied to tobacco anthers, and potato virus Y (PVY)-resistant mutants have been
selected. A high frequency (2.93.9%) of resistant mutants was obtained by the irradiation
of C and He ions with a dose of 510 Gy [112,113].
Two mutant lines of yellow mosaic virus resistant barley were found in ca. 50,000 M
2
families. The character of the resistance in a eld was not changed over three generations
[114].
High eciency of getting blast-resistant mutants of rice was also obtained using ion
beams, although resistant mutants were already obtained by gamma rays and by thermal
neutrons [115].
Molecular Mechanisms of Mutation
In order to investigate the DNA alteration of mutations induced by ion beams in plants,
polymerase chain reaction (PCR) and sequencing analysis was performed to compare DNA
fragments amplied from carbon-ion- and electron-induced Arabidopsis mutants [116,117].
Fourteen out of 30 loci possessed intragenic mutation (small mutation) such as point
mutation or deletion of several to hundreds of bases. For comparison, 16 out of 30 loci
possessed intergenic mutation (large mutation), such as chromosomal inversion, trans-
location, and total deletion covering their own loci. These results imply that in the case of
mutation induced by ion beams, half of the mutants have intragenic small mutation and the
other half have large DNA alteration such as inversion, translocation, and large deletion
(Table 4). In such an alteration, a common feature was that all the DNA strand breaks
induced by carbon ions were found to be rejoined using short homologies [117]. These
results suggest that the nonhomologous end joining pathway operates after plant cells are
exposed to ion beams.
Conclusion
The uvi1uvi4 mutants were induced from 1280 M
1
, and the ast and frl1 mutants were
obtained from the ospring of 1488 M
1
seeds. As ion beams showed high induction of
mutation of the known loci such as tt, gl, hy, it would be true that ion beams can highly
induce not only known mutants but also novel mutants. In chrysanthemum and/or
carnation, complex and striped ower color and new ower-shape mutants that have never
been induced by low-LET radiation have been produced, indicating that ion beams could
induce various kinds of mutants on the similar phenotypes. In conclusion, the character-
Table 4 Classication of Mutation Induced by Carbon Ions
Intragenic mutation
(point mutation, small deletion, etc.)
Large DNA structural alteration
Inversion
translocation
Total
deletion
47% 40% 13%
Copyright 2004 by Taylor & Francis Group, LLC
istics of ion beams for the mutation induction are (1) to induce mutants with high frequency,
(2) to show broad mutation spectrum, and (3) to produce novel mutants. For these reasons,
chemical mutagens such as EMS and low-LET ionizing radiation such as gamma rays and
electrons will predominantly induce many but small modications or DNA damages on the
DNA strands, producing several pointlike mutations on the genome. On the contrary, ion
beams as a high-LET ionizing radiation will cause few but large and irreparable DNA
damages locally, producing a limited number of null mutations (Fig. 33).
5.2. Cross Incompatibility
In plant breeding, obtaining interspecic hybrid is important for introducing desirable genes
from wild species to cultivated species. However, it is very dicult to obtain a viable hybrid
plant between widely related species. These problems would predominantly come from
sexual reproduction such as cross incompatibility, hybrid inviability, etc [118,119].
The wild species of tobacco, Nicotiana gossei, has been reported to be resistant to more
diseases and insects than other Nicotiana species, several attempts to hybridize it with N.
tabacum (cultivar) have been made. In N. gossei crossed with N. tabacum, it was easy to get
hybrid seeds with conventional cross, but the hybrid plants cannot survive. To overcome the
cross incompatibility, mature pollen from N. tabacum cv. Bright Yellow 4 were irradiated
with helium ions from a tandem accelerator (6 MeV). Two viable hybrid plants were
obtained at the rate of 1.1 10
3
. Both the hybrids were resistant to tobacco mosaic virus
and aphid although the degree of resistance varied. With varying energy and kinds of ions,
hybrid production rate ranged from 10
3
to 10
2
[120122]. The highest rate, 3.6 10
2
,
was obtained with 10 Gy of 100-MeV He ion beams. In the case of gamma irradiation, the
rate of producing hybrids was 3.7 10
5
. Thus, ion beams are powerful tools to overcome
the hybrid incompatibility.
Figure 33 Characteristics of energy transfer of ion beams for mutation induction.
Copyright 2004 by Taylor & Francis Group, LLC
5.3. Gene Transfer
Agrobacterium-mediated DNA transfer technique and direct gene transfer into protoplasts
on the basis of the electroporation technique have frequently been used for the gene transfer
methods in plants. However, these methods were hardly adapted to monocotyledonous
plants or to varieties in which regeneration system was not established. These deciencies
found in gene transfer techniques will be largely overcome if pollen could be used as a DNA
vector.
Penetration-controlled irradiation with ion beams for biological study has been
established [123]. When dry pollens were exposed to He or C ions with shallow penetration
depth (e.g., 4-Am depth in the case of He ions), leaky pollen, which was thought to be a
physical lesion in the pollen envelop induced by ion beams, was eciently produced. Using
this penetration-controlled irradiation system, the transient exogenous GUS expression was
observed at a high frequency (55.1%) when tobacco pollen was incubated in a DNA
solution, following ion beam irradiation of pollen at the penetration depth of 4 Am [124].
The GUS expression frequency increased with the irradiation dose, indicating that the
foreign DNA can be eciently introduced into pollen [125]. However, any transformant
cannot be found in more than 10,000 seeds obtained from the crosses using the DNA
introduced pollen. It is thought that the pollen transformation is dicult only by raising the
transient GUS expression frequency and that the nature of the pollen, such as bimembrane
structure and DNA incorporation capability, may be a fundamental problem in a pollen
transformation system [126].
5.4. Application of Microbeams
Aheavy-ion microbeamprovides a unique way to control the number of particles traversing
individual cells and localization of dose within the cell. A collimated heavy-ion microbeam
apparatus has been installed in a beam line from the AVF cyclotron to develop a novel cell
surgery technique [127] (Fig. 34) (Table 5).
Functional Analysis of Animal Development
As a functional analysis of animal development, an accurate fate map was carried out using
the silkworm, Bombyx mori [128]. Among the beamsizes tested (60- to 250-Amf), 250-Amf
was adopted for the fate mapping of egg at the cellular blastderm stage. When eggs were ir-
radiated near the anterior, posterior, or dorsal periphery, no defective larvae were observed.
A close correlation was observed between the site of irradiation and the site of the defect
induced. Head defects were induced only by irradiation at the anterior half of the egg, where-
as defects in thoracic and abdominal segments were induced by irradiation at the middle and
the posterior part of the egg, respectively. Based on the correlations, a fate map of the B.
mori egg was established [128].
Functional Analysis of Plant Physiology
Microbeamsystemis also very useful for analyzing plant physiology. Root gravitropismis a
simple but wondrously complex phenomenon. Perception of gravity in plant roots has been
well studied, but its signaling fromgravity sensing to physical change of root bending is still
unclear. When primary root apical tissues of Arabidopsis were irradiated by microbeams
with 120-Am diameter, strong inhibition of root elongation and curvature were observed at
root tip. Irradiation of cells that become the lower part of the root cap after gravistimulation
Copyright 2004 by Taylor & Francis Group, LLC
showed strong inhibition of root curvature, whereas irradiation of cells that become the
upper part of the root cap after gravistimulation did not show severe damage in either root
curvature or root growth [129]. These results indicate not only that the root tip cells are the
most sensitive sites for root gravity, but also that the signaling of root gravity would go
through the lower part of the root tip cells after the perception.
5.5. Application of Positron-Emitting Radioisotopes
In plant researches, a real-time imaging system to visualize the movement of various
molecules and ions has been desired for understanding the regulation of plant functions.
The positron-emitting tracer imaging system (PETIS) makes use of positron-emitting nu-
clides and enables the investigator to trace the radioactivity in a plant in real time without
touching the sample plant [130] (Fig. 35). This has been rst realized at TIARA by use of
Figure 34 Schematic diagram of microbeam system for biological study. An ion beam from an
AVF cyclotron is collimated by collimators and a tantalum disk with microaperture with a diameter
of 5250 Am. The collimated ion beam passes through the room air to a target that is set on the
micropositioning stage of an optical microscopic system.
Table 5 Physical Parameters of Heavy-Ion Microbeams
Ions
Specic energy
(MeV/u)
Beam size
(f, Am)
Projectile range
(in water, mm)
LET
a
(keV/Am)
4
He
2+
12.5 40
f
250 1.8 14
f
50
12
C
5+
18.3 10
f
250 1.2 100
f
300
20
Ne
7+
13.0 5
f
250 0.7 400
f
800
40
Ar
13+
11.5 5
f
250 0.24 1200
f
2500
a
LET value on the surface of the target.
Copyright 2004 by Taylor & Francis Group, LLC
cyclotron-produced positron-emitting radioisotopes from which various labeled com-
pounds needed for plant research can be produced. As most positron-emitting radioisotopes
are very short-lived, except for
22
Na, they are produced at the site for experiments using the
PETIS and labeled compounds of nutrients and amino acids (Table 6).
Using the PETIS, real-time [
11
C]methionine translocation was studied for barley. For
the mechanismof Fe uptake in an Fe-decient barley, it was found that leaf methionine does
not participate in the reaction of mugineic acid synthesis, but the methionine produced in
barley roots is used in the biosynthesis of mugineic acid phytosiderophores [131,132].
13
NH
4
+
and H
2
15
O translocation in rice plants were also investigated [133,134]. On the
study of root-nodule functions in common bean, the root site of nitrate uptake and
assimilation was claried by using
13
N-nitrate or ammonium [135]. Furthermore, use of
PETIS and positron-emitting nuclides has given us deep insights into water movement in
living plants [136] and nitrogen (N) distribution in soybean using N-decient and/or N-
sucient plants [137].
Figure 35 Schematic of positron-emitting tracer imaging system (PETIS) and an example of
experimental setup.
Table 6 Production and Application of Positron Emitters in TIARA
Nuclide Half-life Reaction Tracer for experiments on plants
11
C 20 min
14
N(p, a)
11
C
11
CO
2
, [
11
C]methionine
13
N 10 min
16
O(p, a)
13
N
13
NO
2
/
13
NO
3
,
13
NH
4
+
, [
13
N]N
2
18
F 110 min
16
O(a, pn)
18
F
18
O(p, n)
18
F
Aqueous
18
F-, [
18
F]-2-deoxy-2-uoro-
D-glucose,
18
F-proline
48
V 16 days
45
Sc(a, p)
48
V [
48
V]H
2
VO
4
solution
52
Mn 5.6 days
nat
Cr(p, xn)
52
Mn
52
Mn
2+
solution
52
Fe 8.3 hr
nat
Cr(a, xn)
52
Fe
52
Fe
3+
solution
62
Zn 9.2 hr
63
Cr(p, 2n)
62
Zn
62
Zn
2+
solution
Copyright 2004 by Taylor & Francis Group, LLC
5.6. Future Prospects
Ion beams as new mutagen showed breakthrough for the next step of biotechnology by
making new mutants that cannot be obtained by other means. One of the most important
objectives of the postgenome age is to elucidate the function of genes. Unfortunately, it is
very dicult in the plant kingdom to apply the gene-knockout system that was used in
mammals or bacteria. Therefore, ion beams will give us thousands of novel mutants that
inform the function of the mutated gene. On the other hand, microbeam irradiation system
is very useful for research on a cell or cell-to-cell function. If the size of the microbeams are
on the order of several microns or submicrons, accurate and eective irradiation for cell or
chromosome inactivation will be carried out. In researches using a whole plant, behavior of
molecules and ions in a living plant can be analyzed with PETIS. As the problems of
elevated CO
2
concentrations and environmental pollution increase, real-time imaging for
physiological understanding of plants is necessary for the security of food resources and
conservation of the environment.
REFERENCES
1a. Tanaka, R. Present Status and Future Development of Research for Advanced Radiation
Application Using Fast Ions, Oyo Buturi 2000, 69, 369 (in Japanese) (a monthly publication of
The Japan Society of Applied Physics).
1b. Tanaka, R. Application of Ion Beams, Radioisotopes 2001, 50, 27 (in Japanese) (A monthly
publication of Japan Radioisotope Association).
2. See the proceedings of the following International Conferences appeared in special issues of
Nuclear Instruments and Methods in Physics Research Section B: Nucl. Instrum. Methods
Phys. Res. B 2002, 193, Proc. of the 19th Intern. Conf. on Atomic Collisions in Solids; Nucl.
Instrum. Methods Phys. Res. B 2002, 190, Proc. of the 15th Intern. Conf. on Ion Beam Analy-
sis; Nucl. Instrum. Methods Phys. Res. B 2001, 175177, Proc. of the 12th Intern. Conf. on
Ion Beam Modication of Materials; Nucl. Instrum. Methods Phys. Res. B 2001, 181, Proc. of
the 7th Intern. Conf. on Nuclear Microprobe Technology and Applications.
3. Brown, I.G. The Physics and Technology of Ion Sources; Wiley-Interscience: New York, 1989.
4. Geller, R. Electron Cyclotron Resonance Ion Sources and ECR Plasmas; Institute of Physics
Publishing: Bristol, 1996.
5. Melin, G.; Bourg, F.; Briand, P.; Delaunay, M.; Gaudart, G.; Girard, A.; Hitz, D.; Klein, J.P.;
Ludwig, P.; Nguyen, T.K.; Pontonnier, M.; Su, Y. Rev. Sci. Instrum. 1994, 65, 1051.
6. Gammino, S.; Ciavola, G.; Celona, L.; Romano, P.; Torrisi, L.; Hitz, D.; Girard, A.; Melin, G.
In Proceedings of the 16th International Conference on Cyclotrons and Their Applications 2001;
Marti, F., Ed.; American Institute of Physics: New York, 2001; 223 pp.
7. Della-Negra, S.; Brunelle, A.; Le Beyec, Y.; Curaudeau, J.M.; Mouron, J.P.; Waast, B.;
Hakansson, P.; Sundqvist, B.U.R.; Parilis, E. Nucl. Instrum. Methods 1993, B74, 453.
8. Tombrello, T.A. Nucl. Instrum. Methods 1995, B99, 225.
9. In Proceedings of an International Workshop on the Structure of Small Molecules and Ions;
Naaman, R., Vager, Z., Eds.; Plenum Press: New York, 1988.
10. Milani, P.; Iannotta, S. Cluster Beam Synthesis of Nanostructured Materials; Springer: Berlin,
1999.
11. Saitoh, Y.; Mizuhashi, K.; Tajima, S. Nucl. Instrum. Methods 2000, A452, 61.
12. Tanaka, R.; Arakawa, K.; Yokota, W.; Nakamura, Y.; Kamiya, T.; Fukuda, M.; Agematsu, T.;
Watanabe, H.; Akiyama, N.; Tanaka, S.; Nara, T.; Hagiwara, M.; Okada, S.; Maruyama, M.
In Proceedings of the 12th International Conference on Cyclotrons and Their Applications;
Martin, B., Ziegler, K., Eds.; World Scientic: Singapore, 1989; 566 pp.
13. In Advances of Accelerator Physics and Technologies, Advanced Series on Directions in High
Energy Physics; Schopper, H., Ed. World Scientic: Singapore, 1993; Vol. 12.
Copyright 2004 by Taylor & Francis Group, LLC
14. Arakawa, K.; Nakamura, Y.; Yokota, W.; Fukuda, M.; Nara, T.; Agematsu, T.; Okumura, S.;
Ishibori, I.; Karasawa, T.; Tanaka, R.; Shimizu, A.; Tachikawa, T.; Hayashi, Y.; Ishii, K.;
Satoh, T. In Proceedings of the 13th International Conference on Cyclotrons and Their
Applications; Dutto, G., Craddock, M.K., Eds.; World Scientic: Singapore, 1992; 119 pp.
15. Saitoh, Y.; Tajima, S.; Takada, I.; Mizuhashi, K.; Uno, S.; Ohkoshi, K.; Ishii, Y.; Kamiya, T.;
Yotumoto, K.; Tanaka, R.; Iwamoto, E. Nucl. Instrum. Methods 1994, B89, 23.
16. Livingood, J.J. Principles of Cyclic Particle Accelerators; D. Van Nostrand: Princeton, 1961.
17. Thomas, L.H. Phys. Rev. 1938, 54, 580.
18. Richardson, J.R. Sector Focusing Cyclotrons, Progress in Nuclear Techniques and Instrumenta-
tion; North-Holland Publishing Co.: Amsterdam, 1965.
19. Gordon, M.M. Part. Accel. 1971, 2, 203.
20. Joho, W. In Proceedings of the 11th International Conference on Cyclotrons and Their
Applications; Sekiguchi, M., Yano, Y., Hatanaka, K., Eds.; IONICS: Tokyo, 1986; 31 pp.
21. Handbook of Accelerator Physics and Engineering; Chao, A.W., Tigner, M., Eds.; World
Scientic: Singapore, 1999.
22. Kurashima, S.; Fukuda, M.; Nakamura, Y.; Nara, T.; Agematsu, T.; Ishibori, I.; Tamura, H.;
Yokota, W.; Okumura, S.; Arakawa, K.; Kumata, Y.; Fukumoto, Y. In Proceedings of the 16th
International Conference on Cyclotrons and Their Applications 2001; Marti, F., Ed.; American
Institute of Physics: New York, 2001; 303 pp.
23. McMahan, M.A.; Wozniak, G.J.; Lyneis, C.M.; Bowman, D.R.; Liu, Charity.Z.H.; Moretto,
L.G.; Kehoe, W.L.; Mignerey, A.C.; Namboodiri, M.N. Nucl. Instrum. Methods 1986, A253, 1.
24. Fukuda, M.; Arakawa, K.; Okumura, S.; Nara, T.; Ishibori, I.; Nakamura, Y.; Yokota, W.;
Agematsu, T.; Tamura, H. In Proceedings of the 1999 Particle Accelerator Conference; Luccio,
A., MacKay, W., Eds. http://accelconf.web.cern.ch/AccelConf/, 1999; 2259 pp.
25. Koehler, A.M.; Schneider, R.J.; Sisterson, J.M. Med. Phys. 1977, 4, 297.
26. Sherrill, B.; Bailey, J.; Kashy, E.; Leakeas, C. Nucl. Instrum. Methods 1989, B40/41, 1004.
27. Rogers, E.J. Nucl. Instrum. Methods 1981, 189, 305.
28. Renner, T.R.; Chu, W.T. Med. Phys. 1987, 14, 825.
29. Fukuda, M.; Okumura, S.; Arakawa, K. Nucl. Instrum. Methods 1997, A396, 45.
30. Breese, M.B.H.; Jamieson, D.N.; King, P.J.C. Materials Analysis Using A Nuclear Microprobe;
Wiley-Interscience: New York, 1996.
31. Llabador, Y.; Moretto, P. Applications of Nuclear Microprobes in the Life SciencesAn Ecient
Analytical Technique for Research in Biology and Medicine; World Scientic: Singapore, 1998.
32. Cookson, J.A.; Ferguson, A.T.G.; Pilling, F. J. Radioanal. Chem. 1972, 12, 39.
33. Kobayashi, Y.; Taguchi, M.; Shimizu, T.; Okumura, S.; Watanabe, H. J. Radiat. Res. 1996,
36(4), 290.
34. Kamiya, T.; Yokota, W.; Kobayashi, Y.; Cholewa, M.; Krochmal, M.S.; Laken, G.; Larsen,
I.D.; Fiddes, L.; Parkhill, G.; Dowsey, K. Nucl. Instrum. Methods 2001, B181, 27.
35. Grime, G.W.; Watt, F. Beam Optics of Quadrupole Probe-Forming Systems; Adam Hilger:
Bristol, 1984.
36. Kamiya, T.; Utsunomiya, N.; Minehara, E.; Tanaka, R. Nucl. Instrum. Methods 1992, B64,
362.
37. Kamiya, T.; Suda, T.; Tanaka, R. Nucl. Instrum. Methods 1995, B104, 43.
38. Sakai, T.; Kamiya, T.; Oikawa, M.; Sato, T.; Tanaka, A.; Ishii, K. Nucl. Instrum. Methods
2002, B190, 271.
39. Ohshima, T.; Morita, Y.; Nashiyama, I.; Kawasaki, O.; Hisamatsu, T.; Nakao, T.; Wakow, Y.;
Matsuda, S. IEEE Trans. Nucl. Sci 1996, 43, 2990.
40. Tada, H.Y.; Carter, J.R., Jr.; Anspaugh, B.E.; Downing, R.G. Solar Cell Radiation Handbook;
3rd Ed. JPL Publication, 1982; 8286.
41. Handbook of Si Solar Cells for Space Application. Private communication by JAERI and NAS-
DA, 1999.
42. Anzawa, O.; Aoyama, K.; Imaizumi, M.; Matsuda, S.; Ohshima, T.; Hirao, T.; Itoh, H.; Saito
M.; Matsumoto, Y. JAERI-Review 2000, 2000-024, 8.
43. Ohshima, T.; Imaizumi, M.; Takamoto, T.; Sumita, T.; Ohi, A.; Kawakita, S.; Itoh, H.; Mat-
Copyright 2004 by Taylor & Francis Group, LLC
suda, S. Proceedings of the 5th International Workshop on Radiation Eects on Semicon-
ductor Devices for Space Application, Takasaki; Japan Atomic Energy Research Institute
(Japan), Oct 911, 2002; 113 pp.
44. Nemoto, N.; Shindou, H.; Makihara, A.; Kuboyama, S.; Matsuda, S.; Sugimoto, K.; Oikawa,
K.; Itoh, H. Proceedings of European Space Components Conference (ESCCON2000); Euro-
pean Space Agency (ESA): Noordwijk, 2000; 325 pp.
45. Hirao, T.; Hamano, T.; Sakai, T.; Nashiyama, I. Nucl. Instrum. Methods, 1999, B158, 260.
46. For example, Properties of Silicon Carbide; Harris, G.L., Ed.; EMIS Datareviews Series No. 13;
INSPEC Publication, IEE: London, 1995.
47. Itoh, H.; Yoshikawa, M.; Nashiyama, I.; Misawa, S.; Okumura, H.; Yoshida, S. IEEE Trans.
Nucl. Sci. 1990, NS-37, 1732.
48. Ohshima, T.; Abe, K.; Itoh, H.; Yoshikawa, M.; Kojima, K.; Nashiyama, I.; Okada, S. Appl.
Phys. 2000, A71, 141.
49. Ohshima, T.; Uedono, A.; Itoh, H.; Yoshikawa, M.; Kojima, K.; Okada, S.; Nashiyama, I.;
Abe, K.; Tanigawa, S.; Frank, T.; Pensl, G. Mater. Sci. Forum 2000, 338342, 857.
50. Fukumoto, A. Phys. Rev. B 1996, 53, 4458.
51. Larkin, D.J. Inst. Phys. Conf. Ser. 1996, 142, 23.
52. Itoh, H.; Toroer, T.; Peppermuller, C.; Pensl, G. Appl. Phys. Lett. 1998, 73, 1427.
53. Itoh, H.; Troer, T.; Pensl, G. Mater. Sci. Forum 1998, 264268, 685.
54. Troer, T.; Schadt, M.; Frank, T.; Itoh, H.; Pensl, G.; Heindl, J.; Strunk, H.P.; Maier, M. Phys.
Status Solidi A 1997, 162, 277.
55. Ohshima, T.; Itoh, H.; Yoshikawa, M. Mater. Sci. Forum 2001, 353356, 575.
56. Ohshima, T.; Yoshikawa, M.; Itoh, H.; Aoki, Y.; Nashiyama, I. Mater. Sci. Eng. B1999, 6162,
480.
57. Ohshima, T.; Yoshikawa, M.; Itoh, H.; Kojima, K.; Okada, S.; Nashiyama, I. Mater. Sci.
Forum 2000, 338342, 1299.
58. Ohshima, T.; Itoh, H.; Yoshikawa, M. J. Appl. Phys. 2001, 90, 3038.
59. Lee, K.; Ohshima, T.; Itoh, H. Mater. Sci. Forum 2002, 389393, 1097.
60. Hamada, S.; Miwa, Y.; Yamaki, D.; Katano, Y.; Nakazawa, T.; Noda, K. J. Nucl. Mater. 1998,
258263, 383.
61. Hamada, S.; Zhang, Y.C.; Miwa, Y.; Yamaki, D. Radiat. Phys. Chem. 1997, 50, 555.
62. Ioka, I.; Naito, A.; Shiba, K.; Robertson, J.P.; Jitsukawa, S.; Hishimura, A. J. Nucl. Mater.
1998, 258263, 1664.
63. Hirsh, J.P. Metall. Trans. A 1980, 11A, 861.
64. Sekimura, N.; Kawanishi, H.; Nodaka, M.; Ishino, S. J. Nucl. Mater. 1984, 122, 322.
65. Sawai, T.; Wakai, E.; Tomia, T.; Naito, A.; Jitsukawa, S. J. Nucl. Mater. 2002, 307311, 312.
66. Sekimura, N.; Iwai, T.; Arai, Y.; Yonamine, S.; Naito, A.; Miwa, Y.; Hamada, S. J. Nucl.
Mater. 2000, 283287, 224.
67. Taguchi, T.; Wakai, E.; Igawa, N.; Nogami, S.; Snead, L.L.; Hasegawa, A.; Jitsukawa, S. J.
Nucl. Mater. 2002, 307311, 1135.
68. Nogami, S.; Hasegawa, A.; Abe, K.; Taguchi, T.; Yamada, R. J. Nucl. Mater. 2000, 283287,
268.
69. Nogami, S.; Hasegawa, A.; Taguchi, T.; Abe, K.; Yamada, R. Mater. Trans. 2001, 42, 171.
70. Katano, Y.; Aruga, T.; Yamamoto, S.; Narumi, K.; Nakazawa, T.; Yamaki, D.; Noda, K. J.
Nucl. Mater. 2000, 283287, 942.
71. Eto, M.; Baba, S.; Ishihara, M.; Ugachi, H. J. Nucl. Mater. 1998, 258263, 843.
72. Anpo, M.; Ichihashi, Y.; Takeuchi, M.; Yamashita, H. Res. Chem. Intermed. 1998, 24, 143.
73. Yamaki, T.; Sumita, T.; Yamamoto, S. J. Mater. Sci. Lett. 2002, 21, 33.
74. Umebayashi, T.; Yamaki, T.; Itoh, H.; Asai, K. Appl. Phys. Lett. 2002, 81, 454.
75. Sumita, T.; Yamaki, T.; Yamamoto, S.; Miyashita, A. Thin Solid Films. 2002, 416, 80.
76. Fromknecht, R.; Khubeis, I.; Massing, S.; Meyer, O. Nucl. Instrum. Methods 1999, B147, 191.
77. Sumita, T.; Yamaki, T.; Yamamoto, T.; Miyashita, A. Jpn. J. Appl. Phys. 2001, 40, 4007.
78. Abe, H. Diam. Relat. Mater. 2001, 10, 1201.
79. Abe, H.; Naramoto, H.; Iwase, A.; Kinoshita, C. Nucl. Instrum. Methods 1997, B127128, 681.
Copyright 2004 by Taylor & Francis Group, LLC
80. Naramoto, H.; Zhu, X.; Xu, Y.; Narumi, K.; Vacik, J.; Yamamoto, S.; Miyashita, K. Phys.
Solid State 2002, 44, 668.
81. Zhu, X.D.; Naramoto, H.; Xu, Y.; Narumi, K. J. Chem. Phys. 2002, 116, 10458.
82. Zhu, X.D.; Naramoto, H.; Xu, Y.; Narumi, K.; Miyashita, K. Phys Rev. B 2002, 60, 165426.
83. Vacik, J.; Naramoto, H.; Narumi, K.; Yamamoto, S.; Miyashita, K. Proc. Mater. Res. Soc.
Symp. 2001, 648, P3.50.1.
84. Zhang, Z.J.; Narumi, K.; Naramoto, H. Appl. Phys. Lett. 2001, 79, 2934.
85. Vacik, J.; Naramoto, H.; Narumi, K.; Yamamoto, S.; Miyashita, K. J. Chem. Phys. 2001, 114,
9115.
86. Zhu, X.D.; Xu, Y.H.; Naramoto, H.; Narumi, K.; Miyashita, K. J. Phys.: Condens. Matter
2002, 14, 5083.
87. Naramoto, H.; Xu, Y.H.; Zhu, X.D.; Narumi, K.; Vacik, J.; Yamamoto, S.; Miyashita, K. Proc.
Mater. Res. Soc. Symp. 2001, 647, O5.18.1.
88. Yamamoto, S.; Naramoto, H.; Narumi, K.; Tsuchiya, B.; Aoki, Y.; Kudo, H. Nucl. Instrum.
Methods, B 1998, 134, 400.
89. Yamamoto, S.; Naramoto, H. Nucl. Instrum. Methods 2000, B161163, 605.
90. Zhu, P.R.; Yamamoto, S.; Miyashita, A.; Wu, Z.P.; Narumi, K.; Naramoto, H. Philos. Mag.
Lett. 1999, 79, 603.
91. Wu, Z.P.; Miyashita, A.; Yamamoto, S.; Abe, H.; Nashiyama, I.; Narumi, K.; Naramoto, H. J.
Appl. Phys. 1999, 86, 5311.
92. Yoshida, M.; Asano, M.; Safranj, A.; Omichi, H.; Spohr, R.; Vetter, J.; Katakai, R. Macro-
molecules 1996, 29 8987.
93. Cartwright, B.G.; Shirk, E.K.; Price, P.B. Nucl. Instrum. Methods 1978, 153, 457.
94. Ogura, K.; Hattori, T.; Asano, M.; Yoshida, M.; Omichi, H.; Nagaoka, N.; Kubota, H.;
Katakai, R.; Hasegawa, H. Radiat. Meas. 1997, 28, 197.
95. Stadler, L.J. Science 1928, 68, 186.
96. Blakely, E.A. Radiat. Environ. Biophys. 1992, 31, 181.
97. Lett, J.T. Radiat. Environ. Biophys. 1992, 31, 257.
98. Goodhead, D.T. Radiat. Environ. Biophys. 1995, 34, 67.
99. Hata, K.; Baba, H.H. JAERI-M 1988, 88184.
100. Hirono, Y.; Smith, H.H.; Lyman, J.T.; Thompson, K.H.; Baum, J.W. Radiat. Res. 1970, 44,
204.
101. Mei, M.; Deng, H.; Lu, Y.; Zhuang, C.; Liu, C.; Qiu, Q.; Qiu, Y.; Yang, T.C. Adv. Space Res.
1994, 10, 363.
102. Hase, Y.; Shimono, K.; Inoue, M.; Tanaka, A.; Watanabe, H. Radiat. Environ. Biophys. 1999,
38, 111.
103. Shimono, K.; Shikazono, N.; Inoue, M.; Tanaka, A.; Watanabe, H. Radiat. Environ. Biophys.
2001, 40, 221.
104. Tanaka, A.; Shikazono, N.; Yokota, Y.; Watanabe, H.; Tano, S. Int. J. Radiat. Biol. 1997, 72,
121.
105. Shikazono, N.; Tanaka, A.; Kitayama, S.; Watanabe, H.; Tano, S. Radiat. Environ. Biophys.
2002, 41, 159.
106. Hase, Y.; Yamaguchi, M.; Inoue, M.; Tanaka, A. Int. J. Radiat. Biol. 2002, 78, 799.
107. Nagatomi, S.; Tanaka, A.; Kato, A.; Watanabe, H.; Tano, S. JAERI-Review 1997, 96-017, 50.
108. Okamura, M.; Ohtsuka, M.; Yasuno, N.; Hirosawa, T.; Tanaka, A.; Shikazono, N.; Hase, Y.;
Tanase, M. JAERI-Review 2001, 2001-039, 52.
109. Tanaka, A.; Sakamoto, A.; Ishigaki, Y.; Nikaido, O.; Sun, G.; Hase, Y.; Shikazono, N.; Tano,
S.; Watanabe, H. Plant Physiol. 2002, 129, 64.
110. Tanaka, A.; Tano, S.; Chantes, T.; Yokota, Y.; Shikazono, N.; Watanabe, H. Genes Genet.
Syst. 1997, 72, 141.
111. Hase, Y.; Tanaka, A.; Baba, T.; Watanabe, H. Plant J. 2000, 24, 21.
112. Hamada, K.; Inoue, M.; Tanaka, A.; Watanabe, H. Plant Biotechnol. 1999, 16, 285.
113. Hamada, K.; Inoue, M.; Tanaka, A.; Watanabe, H. Plant Biotechnol. 2001, 18, 251.
Copyright 2004 by Taylor & Francis Group, LLC
114. Kishinami, I.; Tanaka, A.; Watanabe, H. Abstracts of the Fifth TIARA Research Review
Meeting; Japan Atomic Energy Research Institute, Tokai-mura, Japan, 1996; 165166 (in
Japanese).
115. Nakai, H.; Watanabe, H.; Kitayama, S.; Tanaka, A.; Kobayashi, Y.; Takahashi, T.; Asai, T.;
Imada, T. JAERI-Review 1995, 95-019, 34.
116. Shikazono, N.; Yokota, Y.; Tanaka, A.; Watanabe, H.; Tano, S. Genes Genet. Syst. 1998, 73,
173.
117. Shikazono, N.; Tanaka, A.; Watanabe, H.; Tano, S. Genetics 2001, 157, 379.
118. Goodspeed, T.H. The Genus Nicotiana; Chronica Botanica Company: Waltham, MA, 1954.
119. Takenaka, Y. Bot. Mag. Tokyo 1962, 75, 237.
120. Inoue, M.; Watanabe, H.; Tanaka, A.; Nakamura, A. JAERI TIARA Annual Report 1993;
Japan Atomic Energy Research Institute, Tokai-mura, Japan, 1994; 44 pp.
121. Yamashita, T.; Inoue, M.; Watanabe, H.; Tanaka, A.; Tano, S. JAERI-Review 1995, 95-019,
37.
122. Yamashita, T.; Inoue, M.; Watanabe, H.; Tanaka, A.; Tano, S. JAERI-Review 1997, 96-017,
44.
123. Tanaka, A.; Watanabe, H.; Shimizu, T.; Inoue, M.; Kikuchi, M.; Kobayashi, Y.; Tano, S. Nucl.
Instrum. Methods Phys. Res. B. 1997, 129, 42.
124. Tanaka, A.; Watanabe, H.; Hase, Y.; Inoue, M.; Kikuchi, M.; Kobayashi, Y.; Tano, S. JAERI-
Review 1997, 97-015, 21.
125. Inoue, M.; Kitamura, S.; Toda, Y.; Watanabe, H.; Tanaka, A.; Hase, Y. JAERI-Review 1999,
99-025, 80.
126. Hase, Y.; Sakamoto, A.; Wada, S.; Kitamura, S.; Tanaka, A. JAERI-Review 2001, 2001-039,
87.
127. Kobayashi, Y.; Taguchi, M.; Okumura, S.; Watanabe, H. JAERI-Review 1997, 96-017, 38.
128. Kiguchi, K.; Kinjoh, Y.; Masahashi, K.; Tu, Z.T.; Tamura, H.; Shirai, K.; Kanetatsu, K.;
Kobayashi, Y.; Taguchi, M.; Watanabe, H. JAERI-Review 2000, 2000-024, 51.
129. Tanaka, A.; Kobayashi, Y.; Hase, Y.; Watanabe, H. J. Exp. Bot. 2002, 53, 683.
130. Keutgen, N.; Matsuhashi, S.; Mizuniwa, C.; Ito, T.; Fujimura, T.; Ishioka, N.S.; Watanabe, S.;
Sekine, T.; Uchida, H.; Hashimoto, S. Appl. Radiat. Isotopes 2002, 57, 225.
131. Nakanishi, H.; Bughio, N.; Matsuhashi, S.; Ishioka, N.S.; Uchida, H.; Tsuji, A.; Osa, A.;
Sekine, T.; Kume, T.; Mori, S. J. Exp. Bot. 1999, 50, 637.
132. Bughio, N.; Nakanishi, H.; Kiyomia, S.; Marsuhashi, S.; Ishioka, N.S.; Watanabe, S.; Uchida,
H.; Tsuji, A.; Osa, A.; Kume, T.; Hashimoto, S.; Sekine, T.; Mori, S. Planta 2001, 213, 708.
133. Kiyomiya, S.; Nakanishi, H.; Uchida, H.; Tsuji, A.; Nishiyama, S.; Futatsubashi, M.; Tsukada,
H.; Ishioka, N.S.; Watanabe, S.; Ito, T.; Mizuniwa, C.; Osa, A.; Matsuhashi, S.; Hashimoto, S.;
Sekine, T.; Mori, S. Plant Physiol. 2001, 125, 1743.
134. Mori, S.; Kiyomiya, S.; Nakanishi, H.; Ishioka, N.S.; Watanabe, S.; Osa, A.; Matsuhashi, S.;
Hashimoto, S.; Sekine, T.; Uchida, H.; Nishiyama, S.; Tsukada, H.; Tsuji, A. Soil Sci. Plant
Nutr. 2000, 46, 975.
135. Matsunami, H.; Arima, Y.; Watanabe, K.; Ishioka, N.S.; Watanabe, S.; Osa, A.; Sekine, T.;
Uchida, H.; Tsuji, A.; Matsuhashi, S.; Ito, T.; Kume, T. Soil Sci. Plant Nutr. 1999, 45, 955.
136. Nakanishi, T.M.; Furukawa, J.; Tanoi, K.; Yokota, H.; Ueoka, S.; Ishioka, N.S.; Watanabe, S.;
Osa, A.; Sekine, T.; Itoh, T.; Mizuniwa, T.; Matsuhashi, S.; Hashimoto, S.; Uchida, H.; Tsuji,
A. J. Radioanal. Nucl. Chem. 2001, 249, 503.
137. Ohtake, N.; Sato, T.; Fujikake, H.; Sueyoshi, K.; Ohyama, T.; Ishioka, N.S.; Watanabe, S.; Osa,
A.; Sekine, T.; Matsuhashi, S.; Ito, T.; Mizuniwa, C.; Kume, T.; Hashimoto, S.; Uchida, H.;
Tsuji, A. J. Exp. Bot. 2001, 52, 277.
Copyright 2004 by Taylor & Francis Group, LLC