Semi
Semi
SEMICONDUCTORS
INTRODUCTION
The term excess carriers is used for electrons and holes that are in excess of their
thermal equilibrium values. Excess carriers can be created in a semiconductor
by a variety of processes such as optical excitation, electron bombardment, or in-
jection from a contact. A study of the behavior of excess carriers is important
because most of the semiconductor devices operate under nonequilibrium condi-
tions in which the electron and hole concentrations are significantly different
from their thermal equilibrium values.
In the case of metals, excess carriers can be introduced only by producing an
electrical charge on the specimen, and the field resulting from the mutual repul-
sion of these charges forces the charge to reside on the surface. Thus, no change
in the bulk properties is observed. In the case of semiconductors, because of the
existence of two types of carriers, an excess carrier density can be created in the
bulk without introducing any significant space charge. Hence, a large concentra-
tion of excess carriers can be maintained in a semiconductor under nonequi-
librium conditions, and this alters the conduction properties of the specimen.
When excess electrons are introduced in a metal, the process of lattice colli-
sions restores equilibrium. In semiconductor crystals, the processes that restore
equilibrium are the diffusion and drift of carriers into or out of the region or the
recombination of carriers inside the region. We discussed the drift and diffusion
of carriers in the last chapter. In this chapter we will discuss the carrier recombi-
nation in detail. Starting with the introduction of excess carriers, we will first
treat the kinetics and then the mechanisms of excess carrier recombination. This
discussion is followed by a brief account of the origin of recombination centers.
Finally, we will discuss the nonequilibrium transport of excess carriers and for-
mulate a set of equations to solve the carrier flow problems in semiconductors.
129
130 EXCESS CARRIERS IN SEMICONDUCTORS
The coefficient a is called the absorption coefficient and has a unit of cm” if
d is measured in centimeters. The typical variation of a with photon energy in a
semiconductor is shown in Fig. 5.2. It is seen that when hv is small compared to
E,, the coefficient a is negligible, and most of the light is transmitted through
the crystal, making it transparent to radiation at these frequencies. The small ab-
sorption observed in this region is caused by free electrons that are excited from
lower energy states to higher energy states in a given band. This free carrier ab-
sorption obviously does not create excess electron-hole pairs. As the photon
energy increases, a begins to rise rapidly when hv approaches E,. A photon with
energy hv = E, is absorbed in the semiconductor because it has enough energy to
break the covalent bond and create an electron-hole pair. For these frequencies,
the crystal becomes opaque to the incident radiation. If a photon has energy con-
siderably in excess of E,, then an electron excited from the valence band to the
conduction band will have energy in excess of the average thermal energy. This
situation is shown in Fig. 5.3. The excited electron now loses energy to the lat-
tice during scattering collisions until its velocity reaches the thermal velocity at
that temperature. The process of creating excess carriers by shining light is known
as photogeneration. Photogenerated carriers cause an increase in the conductivity
of the semiconductor. A plot of conductivity as a function of the photon energy
will have a shape similar to the curve of Fig. 5.2.
Introduction of excess electron-hole pairs in a semiconductor is also known
as injection of excess carriers. When excess carriers are injected in a semicon-
oo
Monochromator Photodetector
Semiconductor sample —
(cm!)
——>
a
E,g
hv(eV) ——>
FIGURE 5.2 The optical absorption coefficient as a function of photon energy in a semiconduc-
tor. (From B.G., Streetman, Solid State Electronic Devices, 2nd ed., © 1980, p. 96. Reprinted by
permission of Prentice Hall, Inc., Englewood Cliffs, New Jersey.)
Photon ~—~_—»
FIGURE 5.3 The optical absorption of a photon with energy hv > E,. (a) An electron-hole pair is
created, and (b) the excited electron loses its energy to the lattice phonons and attains thermal
equilibrium. (From B.G. Streetman, Solid State Electronic Devices, 2nd ed., © 1980, p. 94.
Reprinted by permission of Prentice Hall, Inc., Englewood Cliffs, New Jersey.)
132 EXCESS CARRIERS IN SEMICONDUCTORS
In the present case, the electrons and holes are generated in pairs, and we have
Ne = Pe.
Let us now consider a silicon sample doped with 10'° donors cm*. At 300K,
we have nn. = 10° cm? and pn. = 2.25 x 10° cm*.We first consider the case of
a low-level injection with p. = n. = 10’ cm’. From Eq. (5.2) we obtain p, =
Pno + De = 10? cm™ and nz = Mao + ne ~ 10° cm~*. Thus, the low-level injec-
tion is essentially a minority carrier injection because the concentration of major-
ity carriers remains practically unchanged. As an example of high-level injection,
we takep. = n, = 10 cm~. Thus, n, = 1.01 « 10” cm andpy»= 1.x 107 cm~.
The majority carrier concentration now has also changed, and both n, and p,
have become nearly equal. Although high-level injection is encountered in a
number of semiconductor devices, to keep the mathematics simple our treatment
in this chapter will be limited mainly to low-level injection.
Obviously, this is the rate at which the electrons and holes in the semiconductor
disappear, so that this relation can also be written as
Eoeae
oe ica [ee 2
ni] (5.6)
which is a general relation describing the balance between generation and recom-
bination processes. As long as the pn product exceeds n2, the right-hand side of
Eq. (5.6) is positive and both dp/dt and dn/dt decay with time, indicating the re-
combination of excess carriers. When the pn product is decreased below n2 by ex-
tracting carriers, the time derivatives in Eq. (5.6) become positive and there is a
net generation of carriers.
Let us consider the recombination of excess carriers just after the light is switched
off. Besides the recombination in the bulk, electron-hole pairs also recombine
at the surface. However, for the sake of simplicity, we ignore surface recombina-
tion and assume the electron-hole concentrations to be constant throughout the
sample. Writing p = (p. + p.) andn = (n, + n.) we can write Eq. (5.5) as
dDn
n= Hae = Oy Nao + Dao)De (5.8a)
Note that a@,(1no + Pno) has the dimension of reciprocal time and can be written as
if
Tp= scl liSoya1 (5.8b)
ita. a7 Dea) a,Na
The constant 7, is known as the minority carrier (hole in this case) lifetime and is
a function of temperature because a, is temperature dependent. From Eqs. (5.8a)
and (5.8b) we have
Or
where p.(0) = [pn(0) — Pro] is the excess carrier concentration at t = 0 when the
light is switched off. It can be shown (see Prob. 5.3) that 7, is the average time
that a hole spends before recombining with an electron. The lifetime 7, is inde-
pendent of injected carrier concentration as long as the low-level injection condi-
tion is maintained.
A similar analysis can be used to study recombination of excess electrons in a
p-type semiconductor to yield
Electron
Hole Ey
G, = [-c'a(v)P(v) du (5.12)
0
=
2 G,
ao( pn —7n;) = ee
2
— nj) (5.13a)
U.= ttn,
nj
+ po + pele a)
n;
eee en (5.14)
Tr
Gita beDps +) Pe)
have
For a strongly n-type semiconductor having a donor concentration Nz, we
is obtaine d by neglecti ng p.
no = Na, Po = 0 and the low-level injection lifetime
in Eq. (5.14), giving
(5.15a)
136 EXCESS CARRIERS IN SEMICONDUCTORS
n? 5.15b
Ty =
Gr De (
TABLE 5.1
Values of the Recombination Coefficient (G,/n7) at 300K
Note: Data in Table 5.1 is taken with permission from M. H. Pilkuhn, Light Emitting Diodes,
in Handbook on Semiconductors, T.S. Moss (ed), Vol. 4, Device Physics edited by C. Hilsum,
North-Holland Amsterdam, 1981, p. 545.
The indirect gap semiconductors Ge, Si, and GaP have a very low value of the
recombination coefficient, and the radiative lifetime in these semiconductors is
several orders of magnitude higher than the measured lifetime. For example, in Si
doped with 10” donors cm”? the lifetime calculated from Eq. (5.15a) is 0.56 sec,
whereas the measured value may range from a fraction of a millisecond to a few
microseconds. Thus, the radiative recombination remains insignificant in these
semiconductors and lifetime is controlled by indirect recombination via localized
states in the band gap.
Note that both G, and n? increase with temperature. However, G, is somewhat
more sensitive to temperature variations than n;. Consequently, the radiative
lifetime in Eq. (5.15a) is a slowly decreasing function of temperature.
5.3.2 Indirect Recombination via Deep Energy Levels in the Band Gap
Impurity atoms other than donors and acceptors and some types of crystal de-
fects in a semiconductor, introduce localized energy levels deep in the band gap
away from the band edges. These levels act as stepping stones for electrons be-
tween the conduction and valence band, making a substantial enhancement in
the recombination process. Depending on its location in the band gap, a deep
level may act as an electron or a hole trap or a recombination center. An electron
trap has a high probability of capturing a conduction electron and setting it free
after some time. Similarly, a hole trap has a high probability of capturing a hole
that is subsequently released into the valence band. At a recombination center
the probabilities of electron and hole captures are nearly equal. Thus, an electron
capture is followed by a hole capture, and this results in the elimination of an
electron-hole pair. The potential energy of the pair is lowered in two stages. Part
of the energy is released when the electron makes a transition from a state in the
conduction band to the deep level center, and the rest is released when the trapped
electron recombines with a hole. In general, in both steps the energy is dissipated
in the form of phonons, and the recombination is nonradiative.
The mechanism of indirect recombination through the deep level centers has
been investigated by Hall, Shockley, and Read [4]. Many recombination centers
MECHANISMS OF RECOMBINATION PROCESSES [1, 2] 137
have more than one energy level, but in most cases only one level dominates
the recombination. For this reason, we will keep our discussion confined to
single-level recombination. The four steps that occur in the recombination of an
electron-hole pair through a deep-level center are illustrated in Fig. 5.5. The
arrows in this figure indicate the transition of the electron in the process. Pro-
cess (a) depicts the capture of an electron from the conduction band by the cen-
ter located at energy F,. The inverse process of emission of the electron from the
center into the conduction band is process (b). Process (c) represents the capture
of a hole from the valence band by the center. Here the center emits the captured
electron into the valence band, which is equivalent to the transfer of a hole from
the valence band to the center. Finally, in process (d), the center captures an
electron from the valence band leaving behind a hole in the band. This is equiva-
lent to the center’s emission of a hole into the valence band. It should now be
clear that if the level is to act as an efficient recombination center, the electron
capture process (a) must be followed by the hole capture process (c) and both
these processes should have nearly the same probability. If process (a) is followed
by the electron emission process (b), then the center acts as an electron trap.
Similarly, when process (c) is immediately followed by process (d), the center acts
as a hole trap.
Now consider a semiconductor with a density of recombination centers N, lo-
cated at energy E,. The probability that the center is occupied by an electron is
given by
We will next calculate the rates of the individual processes. The rate of the elec-
tron capture process R, should be proportional to the concentration of free
electrons n in the conduction band and should also be proportional to the con-
centration N,(1 — f) of the unoccupied centers. Therefore
Ra = OnVnnN,(l — f) (5.17)
ee
Oo
mies
co
|__HaE—a Ey
SSS!
e Eo
SS SSS See
a SORES i
<
ster we Oana ae
o Ey
Electron Electron Hole Hole
capture emission capture emission
(a) (b) (c) (d)
the deep-
FIGURE 5.5 Schematic representation of electron-hole pair recombination through
level recombina tion center at thermal equilibriu m.
138 EXCESS CARRIERS IN SEMICONDUCTORS
where v,, represents the thermal velocity of the electron and o,, is the capture
cross-section. An electron with velocity v,, must, on the average, come within a
cross-sectional area o,,, of a trap to be captured, and thus sweeps out an effective
volume GenVin per second.
The electron emission rate R, should be proportional to the concentration N, f
of the occupied centers. It should also be proportional to the emission probability
e, that an electron makes a transition from an occupied center into the conduc-
tion band. Thus
where o-p represents the capture cross-section of holes. Finally, the rate of the
hole emission process Rz can be written as
Ra = e,N,1 — f) (5.20)
we obtain
state the rate at which the electrons enter into the conduction band (1.€:5 Gre
R,) must be equal to the rate R, at which the electrons leave the band. Thus
GL, et R, — R, = 0 (5.23)
Similarly, the detailed balance for holes in the valence band leads to
U=R, — R, = R. — Ra (O25)
Onn + OcpP1
f= (5.26)
Oen(n a nN) ar Op(P a Pi)
Finally, substituting this value of fin the rates of the individual processes, we get
where
ny = n; expl((E, — E;)/kT]
Pi = n; exp[(E; — E,)/kT] (5.28)
Note that n; and p; represent the equilibrium electron and hole concentrations
that would result when the Fermi level lies at the trap level E,. After making the
substitutions
1 d 1
ipo = an Tro = (5.29)
Ocp Vth N, OcnV th N,
We note that the driving force for the recombination is the term (pn — n;). To
see the effect of the level E, on the rate of recombination, we assume To = Tro =
tT. for simplicity. Substituting for n, and p;, we can write Eq. (5.30) as
ee
= cr Hi) aay (5.31)
ac + p + 2n; cosh| )]
kT
EXCESS CARRIERS IN SEMICONDUCTORS
140
For given values of p and n, the net rate of recombination U will be maximum
when the third term in the denominator approaches its minimum value. It is clear
that this term will have its minimum value of 2; when E, lies at the intrinsic
level E;. If E, moves away from E; toward the conduction band, then 7, increases.
In such a case, the emission probability e, of the electron increases, and the
trapped electron, instead of recombining with a hole, is emitted back into the
conduction band and the center behaves as an electron trap rather than a recom-
bination center. On the other hand, if E, moves away from E; toward the valence
band, then e, increases and the center becomes a hole trap. Thus, a recombina-
tion center is most effective when its energy is in the middle of the band gap.
Tp= Pn
Se Pao. aye
U U
Now consider the case of a strongly extrinsic n-type semiconductor for which the
recombination centers are situated at the energy level E;. For this case, n; =
Pi = Ni < No, and the above relation simplifies to
1
T P =~ po = —— (5.34)
OcpVin Ni
Physically, this means that the lifetime is determined by the capture cross-section
for holes. This is understandable because the Fermi level is well above E,, and
prior to injection, all the recombination centers are occupied by electrons. Thus,
it is the hole capture that limits the rate of recombination. Hence, 7, is indepen-
dent of electron concentration.
For a strongly p-type semiconductor with level E, lying close to the center of
band gap, it can similarly be shown that
1
LL = (5.35)
OcnVth N,
In silicon and germanium the capture cross-section for holes is larger than that
for electrons, and this implies that ta, > Tpo.
MECHANISMS OF RECOMBINATION PROCESSES [1,2]
141
(a) , (b)
FIGURE 5.6 Band-to-band Auger recombination in a direct gap semiconductor with parabolic
valence and conduction bands; (a) the two electrons and one hole (eeh) process and (b) the one
electron and two holes (ehh) process.
where C, is the Auger coefficient for the ehh process. In thermal equilibrium, the
rate of recombination must be balanced by the Auger generation rate G, giving
and the excess carrier recombination rate U, = (R4 — G4) can be written as
where n and p are given by Eq. (5.2). Under low-level injection, the following ex-
pression for the Auger lifetime +4 = p./U, is obtained from Ea. (5.40):
™m== SH (5.41a)
(5.41b)
MECHANISMS OF RECOMBINATION PROCESSES [1, 2]
143
for a strongly p-type semiconductor. In the case of high-level injection, such that
Pe > n, and p. > po, it can be seen from Eq. (5.40) that
1
TA (5.42)
~ (Cy + C,)p?2
Experimentally determined values of C,, and C, vary over a wide range [2], but
in Ge as well as in Si, both C, and C,, are lower than 3 x 10°“! cm® sec”!. Thus,
band-to-band Auger recombination in these semiconductors becomes important
only when the mobile carrier concentration is higher than about 10'* cm™*. This
situation can occur either in a heavily doped semiconductor or with a very high-
level injection.
Both C, and C, increase with temperature because of an increase in the Auger
generation rate G4. Consequently, the Auger lifetime decreases with temperature.
a p-type semiconductor. In the case of Si and Ge, the last two terms in Eq. (5.43)
are negligibly small unless the carrier concentration 1s very high, so we can write
T = 7,
; AS a example, let us consider an n-type Si sample with Nz = 10'’ cm” and
=2)
G,
— = GyVnN, + —Na + CiNi
Tp nj
5.4.1 Impurities
As mentioned earlier, the impurities that give rise to energy levels near the mid-
dle of the band gap act as efficient recombination centers. Typical examples are
gold and platinum in silicon and copper in germanium.
Properties of gold in silicon have been extensively investigated. A gold atom
in silicon introduces two energy levels — an acceptor level 0.54 eV below Ec and
a donor level 0.29 eV above E,. The acceptor level near the center of the gap is
found to be dominant in controlling the carrier recombination. In the case of
platinum, the level 0.36 eV above the valence band dominates the recombination
process. Lifetime control by gold and platinum doping requires the diffusion of
these impurities into silicon at temperatures in excess of 800°C. The lifetime can
be varied by controlling the concentration of these impurities. For platinum dop-
ing, the lifetime is very sensitive to the resistivity of the silicon. Thus, control of
lifetime using platinum is difficult, and gold doping is commonly used.
High-energy particles such as neutrons, protons, electrons, and y-rays can dis-
place atoms in a semiconductor lattice from their normal positions, creating va-
cancies and interstitials. These vacancies and interstitials can form more complex
lattice defects that behave like deep impurities. The minority carrier lifetime in
ORIGIN OF RECOMBINATION CENTERS
145
OPn
ys Ox x=0
='S(p1(0)' Pps) (5.44)
146 EXCESS CARRIERS IN SEMICONDUCTORS
Penetrating
light
Ti
(a)
Py
P,, (0)
FIGURE 5.8 Excess carrier recombination near a surface having a high density of recombination
centers. (a) Illuminated semiconductor and (b) excess minority carrier distribution.
For low-level injection, the carrier recombination rate is given by Eq. (5.9), which
after using Eq. (5.34) can be written as
= OopVin NA Pn = Dns)
If N,. represents the concentration of centers in the region of thickness x,, then
the total number of centers in this layer is N,.X, = N,, and the total number of
carriers recombining per unit time in the surface layer is given by
In steady state, this rate must be balanced by the flux given by Eq. (5.44). Thus,
S can be expressed as
where N,, represents the number of centers per unit area of the surface. For a
perfectly reflecting surface, § should be zero, whereas for a perfectly absorbing
surface, it should be infinite, which implies that no excess carrier concentration
can be maintained at a perfectly absorbing surface. In many semiconductors, the
rate of recombination at the surface is much larger than in the interior. In such
cases, the effective lifetime is obtained by the reciprocal addition of the two life-
times (see Sec. 20.5).
Surface recombination velocity is a sensitive function of the conditions at
the surface. For sand-blasted surfaces, the typical values of S may be as high as
10° cm/sec, whereas for clean etched surfaces, this value may be as low as 10 to
100 cm/sec.
EXCESS CARRIERS AND QUASI-FERMI LEVELS
147
The concept of the Fermi level used in Eqs. (3.27) and (3.28) is meaningful only
under thermal equilibrium conditions when no excess carriers are present. How-
ever, under nonequilibrium conditions, it is often convenient to write the steady-
state electron and hole concentrations in terms of the quasi-Fermi levels E,, and
Ey defined by the relations
Em a E;
n=n; exp(= (5.47a)
and
LG = En,
Pp = Nn; exp oc. (5.47b)
Since n and p are different from n, and pz, it should be obvious that E,, and Ex,
represent two separate energy levels. In fact, by multiplying the above two rela-
tions, we obtain
2 Lae,
pn = nj exp(=F (5.48)
This relation shows that as long as the pn product is different from n;, Ex,, and
Ey, will be displaced from each other.
The concept of quasi-Fermi levels is important in semiconductors because
their gradients determine the current. For example, consider the case where elec-
tron current is flowing owing to both drift and diffusion processes. The current
density J,(x) can then be written as
dn
Jn(X) = Gunn(x)e(x) + OL
Substituting the value of n(x) from Eq. (5.47a) and expressing the field €(x) =
(dE; /dx)/q, it can be shown (see Prob. 5.15) that
Fn 1 dE rm
J,(X) = pnan(x) ( = otra (") (5.49a)
dE
Jj(X) = Lp P(x) (22) = toe Uige
|Ae 2) (5.49b)
These equations are equivalent to the modified forms of Ohm’s law where the
gradients of the quasi-Fermi potentials (1/q)E, and (1/q)E% determine the total
field. If the two quasi-Fermi levels are constant throughout the semiconductor,
these gradients are zero and current cannot flow.
148 EXCESS CARRIERS IN SEMICONDUCTORS
Gauss’s law states that the total normal electrical flux coming out of a closed sur-
face equals the charge enclosed by the surface. If p denotes the volume charge
density and D the electrical flux density, then Gauss’s law can be expressed as
$d: ds = | p-dv
Ss Vol
When the surface integral on the left-hand side is converted into a volume inte-
gral using the Divergence theorem, the above equation becomes
Since this relation is true for any arbitrary volume, the integrands on the two
sides of Eq. (5.50) must be equal. Thus
V-D=p (5.51)
which is one of Maxwell's equations for an electromagnetic field. In semiconduc-
tor devices, we are interested in the electric field € rather than the electrical flux
density D. For an isotropic semiconductor of permittivity ¢,, these two quantities
are related by
D =.,8 | (5.52)
and thus
V-€= = (5.53)
Noting that € = —Vd, we obtain
Vb = —pfe, (5.54)
BASIC EQUATIONS FOR SEMICONDUCTOR DEVICE-OPERATION [5] 149
a 1
- = Gy, + Gz — Rin,p) — mm ' Jp (5.55a)
Here G,, represents the rate of thermal generation of holes, and G,, denotes the
rate of generation from other sources. The third term on the right-hand side is the
rate of recombination, and the last term is the divergence term that we have as-
sumed to be negative denoting a net flux of holes entering the volume. A similar
equation can be written for electrons
1
= = + G, — Kp) + ie a (5.55b)
The last term on the right-hand side in Eq. (5.55b) has a positive sign because of
the negative charge of the electron. When a simple concept of lifetime holds, we
can write (R — G,,) = U = p-/t, for holes. In addition, sincep. = (p — Po), we
can write Eq. (5.55a) as
OE or 1
fe avd (5.56a)
Similarly, the rate of change of electron concentration is
One 1
Gee ras as (5.56b)
ot
Jp = Wppé — qD,Vp
and
These equations describe the static and dynamic behavior of mobile carriers
in semiconductors under the influence of an electric field and concentration
gradients.
The transport equations for holes and electrons in a semiconductor are obtained
after substituting the values of J, and J, from Eq. (5.57) into Egs. (5.56a) and
(5.56b). The solution of the resulting equations under appropriate boundary con-
ditions describes the electron and hole distributions as a function of space coor-
dinates and time and thus, furnishes a complete description of carrier transport
under nonequilibrium situations.
A general solution of the transport equations is rather difficult, and several
approximations have to be made. In most cases we will be concerned only with
one-dimensional current flow in which both @ and the concentration gradients
are along the x direction. We can then write
0 0
Vp = P and Wn = we
Ox Ox
and substituting for J, and J, from Eq. (5.57) into (5.56a) and (5.56b), we obtain
OPe op 0 ap
—s= —U — p,é— — —+D,— 5.58
ot Ce ee an ae P ax P ax? (58a)
on on dé an
—s= —-U+u,€— + w,n— + D,-—S 5.58b
ot Gr hae i ax ax? ( )
Here the electric field € consists of the externally applied field and the internal
field that arises owing to the imbalance between the excess electron and hole
concentrations. In the above equations, there are three unknowns p, n and @,
and to obtain a solution for these quantities, a third relation is needed. This rela-
tion is obtained by writing the Poisson equation Eq. (5.53) as
EEL
Ope Be
NIN ppp eaten (5.59)
Analytical solutions of Eqs. (5.58a), (5.58b), and (5.59) are not possible unless
some approximations are made. For simplicity we consider a homogeneously
doped semiconductor. The sum of the first four terms on the right-hand side of
Eq. (5.59) is then zero and
dé “q
ieee es Ge (5.60)
that D, and D,, are different. However, the internal fields that are produced in
practical situations are such that there is only a small imbalance in the electron—
hole concentrations, and charge neutrality is approximately valid. As an ex-
ample, let us consider n-type silicon with n, = 10'° cm™ and p, = 10 cm. If
the electron concentration n, departs 1 percent from pe, then from Eq. (5.60),
this results-in d€/dx ~ 10° V cm‘|. Internal fields of this magnitude are rarely
encountered in a homogeneous semiconductor. Thus, it is clear that in practical
situations, the excess electron and hole concentrations are nearly equal. This
condition is known as the quasi-neutrality condition. Substituting n, = p. and
eliminating d/dx in Eq. (5.58a) and (5.58b) and noting that dp/ax = dp. /dx, we
obtain the following equation:
where
sii D Ds
Do= (n + p)DpDn cite eS (n eee n
= P)tentp (5.62)
nO at ps Wine Di,
D, and wz, are known as the ambipolar diffusion constant and ambipolar mobility,
respectively, and Eq. (5.61) is the so-called ambipolar transport equation. Both D,
and yz, depend on the excess carrier concentration p,, and it is impossible to solve
Eq. (5.61) in a closed form for any arbitrary value of p.. Two special circumstances
in which Eq. (5.61) can be solved are discussed in the following subsections.
2DeD3
D, Sea and pa = (Mo
Po)tenby
= _ 0 (5.63)
DAS De Wn + Mp)
These relations are also true for low-level injection in an intrinsic semicon-
ductor. Under these conditions, the field term in Eq. (5.61) becomes zero and the
transport is dominated by diffusive motion with the diffusion constant given by
Eq. (5.63). The fact that the field term vanishes at high-level injection greatly
simplifies the analysis.
152 EXCESS CARRIERS IN SEMICONDUCTORS
The ratio of the drift to the diffusion current density for holes is given by
a _ __MpPn&
eel
D dp.
where the subscript f is used for the drift field. Substituting the value of € from
Eq. (5.65) and making use of the Einstein relation for holes lead to
For low-level injection, n, > p,, and it is seen that J, is negligible compared to
J,p. Similarly for electrons, the ratio Jig/Inp at low-level injection can be shown to
be given by
ro tierce)iaees
f= (1--#)=(1-—
| 7 z (5.67a)
which obviously is not negligible as long as D, is different from D,,. In fact, from
Eq. (5.67a) it is seen that
D
dh = Ju c= =] = Uap J 5D (5.67b)
SOLUTION OF CARRIER TRANSPORT EQUATIONS
—AN ILLUSTRATION 153
Thus, it is clear that under low-level injection, the minority carrier current (holes
in this case) is dominated by diffusion and is not appreciably influenced by the
field. The majority carrier current, on the other hand, is strongly influenced
by the electric field. These results can be easily understood by referring to the
sketches in Fig. 5.9. In the illuminated bar of Fig. 5.9a, the excess holes and elec-
trons generated near the surface tend to diffuse away from the planex = 0. Since
the diffusion constant for electrons is higher than that for holes, the electrons dif-
fuse away faster than the holes, which raises the hole concentration near x = 0
above the electron concentration. The resulting distribution of excess carriers is
as shown in Fig. 5.9b. The space-charge and electric field distributions are shown
in Fig. 5.9c and Fig. 5.9d, respectively. Since the quasi-neutrality condition is ap-
proximately satisfied at each point, the field established by this charge imbalance
is small. In this field, the holes experience a force along the field direction and a
drift current of holes flows along the positive direction of x. The electrons drift
in the opposite direction, but the conventional drift current of electrons flows in
the same direction as the hole drift current. This current obviously opposes the
electron diffusion current. Under low-level injection, the concentration of holes is
small compared to that of electrons, and the drift component of the hole current
(d)
falling on the
FIGURE 5.9 Steady state injection from the surface of a bar. (a) Ionizing radiation
(d) electric field set up by
surface, (b) excess electron and hole distributions, (c) space charge, and
the charge imbalance.
154 EXCESS CARRIERS IN SEMICONDUCTORS
SI IU (5.68)
where L, = VDyt, is called the diffusion length of holes. We no longer need the
partial derivative since dp./dt = 0 in the steady state. The above equation has
the solution
where A, and A; are constants of integration. The hole concentration p-(x) must
reduce to zero as x approaches infinity. Therefore, we have A; = 0 and
J,(0) == aD,dp.\
iB ls_Seer
qDpp.(0)
In the steady state, the hole flux J,(0)/q must equal the hole generation rate
G_,X,. Thus
Gres
p.(0) = aei ; (5.70)
It is evident from Eq. (5.69) that the excess hole concentration decreases expo-
nentially withx. In fact, Eq. (5.69) has the same form as Eq. (5.10a), and it can be
shown that L, is the average distance a hole diffuses before recombining.
The neglect of minority carrier drift current is not justified at high-level injec-
tion when the majority and the minority carrier concentrations become equal.
Substituting p, = n, in Eq. (5.66), we obtain
Jof = BD (5.71)
REFERENCES
PROBLEMS
51 A 0.5-um thick sample of GaAs at 300K has an area of 1 cm’. The sample
is uniformly illuminated with monochromatic light of hu = 2.1 eV. The
absorption coefficient @ at this wave length is 4 x 10* cm™'. The power in-
cident on the sample is 12 mW. (a) Calculate the power absorbed by the
sample; (b) determine how much power is dissipated by the excess electrons
to the lattice before recombining; and (c) determine the number of photons
per sec falling on the sample and the number of photons per sec emitted
from the electron-hole pair recombination.
156 EXCESS CARRIERS IN SEMICONDUCTORS
5.2 Let the sample in Prob. 5.1 be n-type with n, = 10'° cm™. (a) Assuming
that each of the absorbed photons produce one electron-hole pair in the
sample, calculate the excess electron and hole concentrations in the steady-
state. (b) Calculate the photoconductivity of the sample and determine
whether it is the case of low- or high-level injection. Use the data from
Table 4.2 for GaAs.
5.3 Using the result of Eq. (5.10a), show that the average time a hole spends be-
fore recombining with an electron is 7p.
5.4 A homogeneous semiconductor bar is illuminated uniformly by a penetrat-
ing light that generates electron-hole pairs at a constant rate G, Cm) sec
Assuming low-level injection, (a) calculate the excess carrier concentration
as a function of time if the light is switched on at t = 0; and (b) determine
the steady state values of electron and hole concentrations and show that
the photo conductivity Ac of the sample is given by Aa = q(un + bp)GLTp.-
5.5 A p-type Ge sample with a resistivity of 40 Q-cm at 300K is uniformly il-
luminated with light that generates 10° excess electron-hole pairs per cm?
per sec. In the steady state, calculate the change in resistivity of the sample
caused by the light. If the light is switched off at ¢ = 0, calculate the time
required for the excess conductivity to drop to 10 percent of its value at
f= 0. Assume 7, = 10 ° sec.
5.6 A Si sample is doped with 10'° donors cm°. Calculate the excess electron
and hole concentrations required to increase the sample conductivity by
15 percent. What carrier generation rate is required to maintain these ex-
cess concentrations? Assume wp = 0.3 bn, Tp = 10°° sec and T = 300K.
5.7 A sample of n-type Si has a dark resistivity of 1 K Q-cm at 300 K. The sample
is illuminated uniformly to generate 10” electron-hole pairs per cm* per
sec. The hole lifetime in the sample is 1 x sec. Calculate the sample resis-
tivity and the percentage change in the conductivity after illumination ow-
ing to majority and minority carriers.
5.8 An n-type bar of GaAs has a length / and an area of cross-section A. The
bar is illuminated with light that generates G, electron-hole pairs cm~°
sec"'uniformly. Assuming the hole mobility to be negligible in comparison
to electron mobility, show that the steady state photocurrent in the bar can
be written as I,, = gAlG_7,/7,, where 7, is the average transit time of elec-
tron through the bar.
5.9 For a semiconductor with indirect recombination characterized by a single
trap level at FE, and 7, = To, Show that under low-level injection the maxi-
mum possible lifetime occurs when E£; lies at E;, and that this maximum is
given by
Tey. Sy E; :
Tp = |
Tpo| 1 + cos cosh( iT )
5.11 Consider an infinitely long, square but thin bar of an n-type semiconductor
illuminated by light that creates electron-hole pairs uniformly at a rate
G, pairs cm’ sec‘. The surface at x = 0 has the surface recombination
velocity S cm sec’. Set up the continuity equation for holes and deter-
mine their concentration p,(x) as a function x. Determine Pn(x) when S ap-
proaches infinity.
5.12 Excess carriers are injected in a region of a uniformly doped n-type semi-
conductor with n, = 10'* cm-*. The excess carrier concentration is main-
tained at 2 x 10° cm throughout the region. (a) Calculate the Auger
recombination lifetime assuming C, = 2.7 x 10°! cm® sec’! and C, =
1.1 x 10~*' cm® sec™'. (b) If the measured lifetime in the above sample is
4 x 10 sec, determine the lifetime in the absence of Auger recombination.
5.13 A Si sample is doped uniformly with 10° donors cm and has 7, = 1 psec.
(a) Determine the photogeneration rate that will produce 2 x 10” excess
pairs cm * in the steady state. (b) Calculate the conductivity of the sample
and the position of the electron and hole quasi-Fermi levels in the steady
state at 300K.
5.14 Show that the nonequilibrium pn product in a semiconductor with band
gap FE, is the same as the equilibrium product p,n, with band gap E£, —
(Em as Ep).