Thermodynamics and Kinetics of Adsorption

subtitle:
What can we learn about adsorption systems from thermodynamic and kinetic measurements?
IMPRS-Lecture Series 2012: Experimental and Theoretical Methods in Surface Science
Klaus Christmann
Institut fr Chemie und Biochemie, Freie Universitt Berlin

Organization
1. Introduction; terms and definitions
2. Thermodynamics of adsorption

Adsorption as a macroscopic (thermodynamic) phenomenon

Equilibrium thermodynamics and adsorption isotherms: Langmuir and BET isotherm

The adsorption energy: Initial adsorption energy and a-priori heterogeneity

Coverage dependence of the adsorption energy: lateral interactions and a-posteriori heterogeneity

3. Kinetics of adsorption and desorption

The rate of adsorption; trapping and sticking

The rate of desorption; order of desorption and frequency factor

The desorption energy: Binding states and lateral interactions

Experimental procedure and evaluation of TD spectra

4. Summary and conclusions

Practical relevance of adsorption processes

Technical applications of adsorption:

Separation processes of gaseous and liquid mixtures via molecular sieves

Heterogeneous catalysis
Chemical analyses (various types of chromatography)
Flotation processes
Gas mask filters
Lubrication phenomena
Optical coatings
corrosion-resistant coatings (Lotus effect)

As we will see later, one distinguishes

chemical adsorption (chemisorption)
and
physical adsorption (physisorption),
depending of the strengths of the interaction between adsorbent and adsorptive.
While chemisorption is most essential for heterogeneous surface reactions, physisorption plays a
major role during separation and enrichment processes at and near phase boundaries.

1. Terminology: Strength of interaction forces physisorption and chemisorption

Interaction forces responsible for adsorptive bonds
Kind of interaction

Strength of interaction

van-der-Waals

Very weak (usually less than

50 kJ/mole)

ionic

Strong (Coulombic origin)

> 100 kJ/mole

covalent

Strong (quantum-chemical
origin)
> 50 kJ/mole

metallic

Strong (quantum-chemical
origin), E > 50 kJ/mole

1. Terminology: adsorption and desorption

e
e
t

In heterogeneous catalysis, adsorption and desorption are essential elementary processes

because adsorption activates the decisive chemical bonds of the reactants and desorption removes
the reaction products from the surface.
In a typical adsorption process, a foreign material in gaseous or liquid form (the adsorptive) becomes
attached to a solid or liquid surface, the adsorbent, and forms the adsorbate, a complex between
adsorptive and adsorpt.
Consequence: No adsorption without a surface (or, more generally, without a phase boundary
including liquid surfaces). In heterogeneous catalysis we mostly deal with solid surfaces.
Since the adsorptive and the adsorbent often undergo a chemical reaction, the chemical and physical
properties of the adsorbate is not always just the sum of the individual properties of the adsorptive and
the adsorbent, but often represents a phase with novel properties.

1. Terminology: what is a surface?

There exist a lot of definitions among which the most rigorous ones are:
A surface represents the outer or the topmost boundary of a solid or liquid object,
where the bulk properties are no longer sufficient to describe the systems properties.
A surface is the boundary between the inner and outer world of a given phase.
A surface always determines a thermodynamic phase boundary, where the
physical and chemical properties of the adjacent phases change abruptly.

Consequence:
In the surface region, the interaction forces
responsible for the bulk properties become
asymmetric, since the outermost bonds are
unsaturated, resulting in a peculiar chemical
reactivity of the surface. In our context the most
important consequence is that the surface becomes
capable of adsorbing foreign atoms or molecules.

1. Terminology: asymmetry of binding forces

Consequences: Clean surfaces try to rearrange their atoms positions in order to minimize the surface free energy.
One distinguishes:
multilayer relaxation, i.e., in the surface region the distances perpendicular to the surface differ from those of the bulk
reconstruction, i.e., the surface atoms take on new periodic positions parallel to the surface (formation of superstructures)
Both phenomena can (and usually will) be affected by adsorption.

Asymmetry of binding forces and its consequence

Adsorbate-induced surface reconstruction

Free energy as a
function of the
surface
configuration.
This
configuration is
determined by
the adsorbate
coverage. At a
critical value of
this coverage,
the surface flips
to a more stable
equilibrium
configuration.

1. Terminology: surfaces of interest

We shall subdivide our surfaces in two classes:

Single crystal surfaces are geometrically well-defined. While they are not really used in
heterogeneous catalysis, they have the advantage to provide access to the physics and chemistry
of the elementary steps of a catalytic reaction (model systems). The scientific field of classical
Surface Science - introduced among others, by Gerhard Ertl - can be most successfully applied.
Polycrystalline, heterogeneous surfaces are very important
in practical heterogeneous catalysis, since they have a large
surface area due to micro-, meso- or macropores. Among
these solids are charcoal (active carbon), alumina, silica,
zeolites or carbon nanotubes. Adsorption on these porous
surfaces is often well described by the BET adsorption
isotherm. In addition, the existence of mesopores (pore radius
rp) can modify the thermodynamic phase equilibria due to a
vapor pressure depression or elevation predicted by the Kelvin
equation (curvature effect). The Kelvin equation, which
contains both the size and the surface tension of the dispersed
material also governs growth and nucleation phenomena
(Ostwald ripening etc.)

Pconcave < p0 <

pconvex

p
V
2

= exp

p0
RT rp , Kelvin
Kelvin equation

Terminology: associative and dissociative adsorption

Associative (molecular) adsorption leaves the adsorbing molecule intact. Example: CO adsorption on a
palladium(100) surface. The adsorption energy equals the depth of the potential energy well: Ead = Echem.
Dissociative (atomic) adsorption cleaves the adsorbing molecule either homolytically (H H = H + H) or
heterolytically (H H = H- + H+). Example: H2 adsorption on a nickel (111) surface. The adsorption energy is the
difference between twice the Ni H bond energy and the heat of H2 dissociation (435 kJ/mol): Ead = 2 EMe-H - Ediss

CO-Pd interaction potential

H2-Ni interaction potential

Information on the mechanism of adsorption may be obtained from measurements of adsorption isotherms or from
thermal desorption spectra as will be discussed later.

Terminology: associative and dissociative adsorption

The dissociation of a diatomic molecule can be better characterized by a two-dimensional potential
energy representation. In these so-called elbow-plots the perpendicular distance of the molecule, y,
is plotted against the internuclear distance, x. Whether or not the molecule can successfully dissociate
depends on its translational and/or vibrational excitation.

early barrier

late barrier

Terminology: associative and dissociative adsorption

The dissociation of
molecular hydrogen has
been studied theoretically in
great detail, for example on
Pd(100) surfaces by A. Gro
et al.. Quite important is the
so-called dynamical
steering where the H2
molecule is guided to the
optimum dissociation sites.

Terminology: activated and non-activated (spontaneous) adsorption

In the adsorbed state and in thermal equilibrium the trapped particle resides at the bottom of the
potential well. It is either physically or chemically adsorbed on the solid.
In order to desorb it from the surface, one has to supply it thermally or electronically with the energy
required to return to the gas phase. If activation barriers are involved, these have to be additionally
overcome, and the desorbing particles then possess more energy than they had if they were in thermal
equilibrium with the surface.

Non activated (spontaneous) adsorption: CO on Ni(111)

activated adsorption: hydrogen on a Cu(111) surface

2. Adsorption as a thermodynamic phenomenon

Some general points of interest in an adsorption process:
How many particles can be adsorbed by a given amount of material ?
thermodynamic quantity of interest = surface concentration ad [moles/m2]
How does this amount depend on the state variables (temperature and pressure), i.e.
how can one control the adsorbate concentration by an appropriate choice of P and T?
How can one determine the stationary and maximum adsorbed amount?
Where are the adsorbed particles located at, on, and in the surface, respectively?
How much energy is required to remove these particles from the surface, i.e., how strong
are the adsorptive bonds to the surface?
Relatively straightforward answers to most of these questions are provided by (surface)
thermodynamics.

2. Thermodynamics of adsorption - Fundamental Gibbs equation

Thermodynamics describes the behaviour of matter as a function of state variables (P, T, chemical
composition etc.). However, since it is just a continuum description of the chemical state of systems, it
is unable to describe or predict microscopic details or elementary processes on the atomic scale,
nor is it capable of predicting activation energies and, hence, provides hardly any information about
kinetics (time dependences) of chemical reactions.

Consider the differential change of Free Enthalpy (Gibbs Energy), dG, of a thermodynamic
system during any change of state:

G
G
G
G

+
+
+
dG =
dP
dT
dA
dni

A P ,T
P T , A
T P , A
ni T ,P , A

dG = V dP S dT + dA + i dni
i

= + V;
P T , A
G

= S
T P , A
G

=
A P ,T

with P = pressure, T = temperature, A = surface area, = surface tension, V = system volume, = chemical potential
and S = entropy. The third term becomes decisive, when the surface area is large in relation to the bulk volume (high
degree of dispersion). Example: Raney-Nickel as a catalyst is often utilized in hydrogenation reactions.

The surface tension [N/m] is entirely equivalent to the surface energy [Nm/m2], which is the driving
force for all surface phenomena.

2. Thermodynamics of adsorption - the heat of adsorption

Thermodynamical derivation of the so-called isosteric enthalpy of adsorption, qst:
Chemical equilibrium between adsorbate and gas phase leads to a constant surface
concentration [moles/m2]. Constant is maintained when the fluxes of adsorbing and
desorbing particles are equal. We now consider the pressure and temperature dependence of
this phase equilibrium. A common procedure is to equate the chemical potentials and their
derivatives of the phases involved. Remember that the chemical potential is the derivative of
the Gibbs energy with respect to the mole number in question, ni :
dG

dn
i P ,T , other mole numbers

i =

In the (dynamical) phase equilibrium, the chemical potentials ad and gas are equal and remain
equal (persisting or ongoing equilibrium):
ad = gas

and

dad = dgas ; = nad/A

ad + OF
dad+OF = -sad+OFdT + vad+OFdP + d
p ,T

(small letters denote partial molar quantities)

= dgas = - SgdT + VgdP.

2. Thermodynamics of adsorption - the heat of adsorption

Rearranging and considering that the term ad +OF d gets zero for constant coverage (d=0) yields
p ,T
the well-known Clausius - Clapeyron
equation for the ongoing phase equilibrium
between gas phase and adsorbate phase(s):

s g sad +OF
S
dP
=
=

dT
V

Vg Vad +OF

Setting

Vad + OF << Vg

s g sad + OF
1 dP

=
P dT
RT

and applying the ideal gas equation, I.e., Vg =

RT
P

yields the relation

. Recalling that the entropy is the reduced heat s =

leads to the expression:

h
finally
T

hg had + OF d ln P
1 dP
qst
=
=
=

2
P dT
RT 2
dT RT

qst is called differential isosteric heat of adsorption and represents the energy difference between
the state of the system before and after the adsorption of a differential amount of particles on that
surface. It can be obtained from equilibrium, i.e., adsorption isotherm measurements. These
considerations apply particularly well for homogeneous single crystal surfaces, for which we shall
provide examples later. Nevertheless, qst is often coverage-dependent due to lateral particle particle
interactions.
Inhomogeneous, porous surfaces that one encounters in practical heterogeneous catalysis require a
modified view and more complicated thermodynamic treatment, since they can provide a whole variety
of energetically and geometrically different adsorption sites.

2. Thermodynamics of adsorption:
Concept of the adsorption isotherm

The adsorption isotherm displays information of how much material will adsorb for a
given set of state variables (pressure P and temperature T when using gases as
adsorptives, concentration c and temperature T when dealing with adsorption from
solutions). The amount of adsorbed material can be introduced as surface coverage ,
i.e., the fraction of occupied/available sites, or by the total volume vm taken up by the
sum of particles adsorbed in the first monolayer.
The adsorption isotherm is one of the most characteristic quantities for an adsorption
process. In practice, the uptake of adsorptive of a given catalyst material is decisive,
since a high uptake usually means a good catalyst efficiency. The respective uptake is
often called sorption capacity and characterised by the sorption (or adsorption)
isotherm. The form of the sorption isotherm provides a lot of first-glance information
about the chemical and physical properties of the catalyst material and about how the
adsorption process proceeds. In principle, various kinds of isotherms can be
distinguished, depending on the nature of the catalyst and the kind of interaction.

2. Thermodynamics of adsorption:
The concept of the adsorption isotherm

Some examples of experimental adsorption isotherms

Langmuir isotherms (type I) are usually observed during adsorption of chemically active gases on
metal surfaces, but also with non-polar gases (methane, nitrogen N2, ethane) on zeolites, whereas
the adsorption of water vapour on -Al2O3 leads to isotherms of type II with condensation. Using an
unpolar surface like charcoal and polar molecules as adsorptive (methanol or H2O) the situation is
characterised by initially rather repulsive interactions leading to a reduced uptake, while the
increasing presence of adsorbate molecules facilitate the ongoing adsorption leading to isotherms of
type III. If porous adsorbents are used, additional capillary condensation effects lead to isotherms of
type IV and V.

2. Thermodynamics: monolayer and multilayer adsorption

Monolayer adsorption on a
homogeneous surface at equilibrium
pressure p/p0 . The heat of adsorption
of the first monolayer is much
stronger than the heat of adsorption
of the second (and all following)
layers (typical chemisorption case)

Multilayer adsorption/condensation
on a homogeneous surface at
equilibrium pressure p/p0 . The heat
of adsorption of the first (blue) layer
is comparable to the heat of
condensation of the subsequent
(red) layers. Often observed during
physisorption.

2. Thermodynamics of adsorption: the Langmuir adsorption isotherm

A way of getting to the adsorption isotherm is to consider the adsorption - desorption equilibrium.
The adsorption process between gas phase molecules, A, vacant surface
sites, S, and occupied surface sites, SA, can be described by the equation

S + A SA

assuming that there are a fixed number of surface sites present on the
surface. An equilibrium constant, K, can be written:

K=

[SA]
[A] [S ]

If we introduce the symbol for the fraction of surface sites occupied (0 < < 1) and consider that [SA] is proportional to
the surface coverage of adsorbed molecules, i.e., proportional to , whilst [S] is proportional to the number of vacant
sites, 1 - , and [A] is proportional to the collision number, i.e., the gas pressure P, we arrive at
an expression:

K P
K=
=
or, rearranged for ,

(1 )P

1+ K P

which has been derived as early as 1916 by Irving Langmuir. Note that K is only a constant if the enthalpy of adsorption
is independent of coverage. As with all equilibrium constants, K and, hence, the position of adsorption - desorption
equilibrium, will depend on (i) the relative stabilities of the adsorbed and gas phase species involved, (ii) on the
temperature of the system, and (iii) on the pressure of the gas above the surface. Factors (ii) and (iii) exert opposite
effects on the concentration of adsorbed species - that is to say that the surface coverage may be increased by raising
the gas pressure but will be reduced if - at constant pressure - the surface temperature is raised.

2. Thermodynamics of adsorption The Langmuir adsorption isotherm

A somewhat more revealing derivation of the Langmuir adsorption isotherm is based on the fact that in
adsorption - desorption equilibrium the rates of adsorption and desorption are equal. While the rate of
adsorption is determined by the collision frequency of the gas phase particles with the surface
according to:

Rad =

particles
P
(1)
S 0 = k ad
P S0

2
2 mkT
m s

(S0 stands for the uncovered fraction of the surface ( (1 - )), P for the gas pressure), the desorption
rate can easily be formulated as being proportional to the fraction of the surface covered by adsorbed
particles (here denoted as S1, i.e., ), a frequency factor , and an exponential term containing the
depth of the potential well, i.e., the adsorption energy (Arrhenius equation):

particles
E
Rdes = S (1) exp des = S (1) k des (T )

2
kT
m s
Equating Rad and Rdes and introducing the coverage by noting that S1 and S0 (1 - ) yields
immediately the Langmuir expression derived before.

2. Thermodynamics of adsorption The Langmuir adsorption isotherm

A closer inspection of the adsorption/desorption equilibrium helps to interpret the constant
K(T) which comes out as :

s0
Edes
12
(2 mkT ) exp ( )
K (T ) =
kT
N max
with s0 representing the initial sticking coefficient, the frequency factor of the desorption reaction
and Nmax the maximum number of adsorbed particles: For non-activated adsorption, the
desorption energy Edes is equal to the energy of adsorption, Ead. In the derivation given above we
have assumed first-order processes both for the adsorption and the desorption reaction. For
second-order processes, e.g. adsorption/desorption reactions involving rate-limiting dissociation
and recombination steps, a similar expression can be derived

( p )T =

K ' (T ) P

1 + K ' (T )P

From the initial slope of a log - log plot of a Langmuir adsorption isotherm (K(T)P << 1), the order of
adsorption can easily be determined (slope 1 = 1st order adsorption; slope 0,5 = 2nd order adsorption).

2. Thermodynamics of adsorption The Langmuir adsorption isotherm

For single crystal

surfaces, this well-known
Langmuir isotherm
describes the adsorption
processes often quite well,
especially at low and
medium coverages.
However, it fails if porous
polycrystalline surfaces
are used as adsorbents.
Then, better the BET
isotherm is applied.

If we plot the surface coverage against the concentration or the pressure of particles of the adsorbing phase, we arrive at
typical saturation curves: Near the monolayer capacity all surface sites are occupied, and higher pressures cannot
increase the coverage or surface concentration beyond = 1 anymore.

2. Thermodynamics of adsorption The Langmuir adsorption isotherm, contd

What are the prerequisites for applying the Langmuir adsorption model?
All adsorption sites must be equivalent regardless of whether we look at
the very first or the final adsorbing particle: All have the same binding energy
to the surface, and consequently, there is no coverage dependence of the
heat of adsorption.
The adsorption capacity is strictly limited to one and only one monolayer.
Adsorption into multilayers is ruled out.
Apparently, the Langmuir model is quite limited and seldom applicable for
practical purposes - in fact, the coverage dependence of the heat of
adsorption must be taken into account as well as the possibility that more
than just a single monolayer will adsorb. Various more sophisticated
isotherms have been developed, among others the Temkin isotherm or the
BET isotherm. The Temkin isotherm resembles somewhat the Langmuir
isotherm except that a (linear) decrease of qst with is included, but still the
adsorption is restricted to a single monolayer.

2. Thermodynamics of adsorption The BET adsorption isotherm

The famous BET Isotherm first introduced in the 1930s by Brunauer, Emmett and Teller now
considers also the possibility of multilayer adsorption and condensation as shown below.

2. Thermodynamics of adsorption The BET adsorption isotherm

The forces responsible for the build-up of the second, third, and all following layers are van-der-Waalslike and are supported by low temperatures and high gas pressures. In contrast to the Langmuir
behaviour, no saturation occurs at high pressures, but a true condensation of a liquid phase.
Concerning the adsorption energies involved, one can clearly distinguish the adsorption enthalpy of the
very first layer (which may well behave Langmuir-like), i.e., Echem, and the heat of condensation (= heat
of evaporation) of all following layers, Econd . For the derivation of the BET isotherm it is essential to use
a strict terminology, and we distinguish the following quantities:

the rate of adsorption into the ith layer (clean surface = index o)

P = pressure, S(i) = free surface area ( to empty surface sites)

the rate of desorption from the ith layer

(i )
(i )
Rad
= k ad
P S(i )
(i )
(i )
Rdes
= k des
S(i ) e

E( i )
kT

2. Thermodynamics of adsorption The BET adsorption isotherm

We deal with a strict (dynamic) equilibrium between adsorbate and the gas phase containing the adsorptive
molecules, which is characterised by the identity of adsorption rate and desorption rate. Hence, for the first layer and
the bare surface sites we have:
(1)
(1)
k ad
P S ( 0) = k des
S (1) e

E( 1 )
kT

(eq. I )

If we now consider the principal situation we first have a look

how the surface S1 can change:
Altogether we have four processes that cause a change, namely:

adsorption of a particle on the bare surface (1.)

(1)
(1)
Rad
= k ad
P S0

desorption of a particle from the sample (2.)

(1)
(1)
Rdes
= k des
S (1) e

adsorption of a particle onto the first layer (3.)

( 2)
( 2)
Rad
= k ad
S (1) P

desorption of a particle from the second layer (4.)

( 2)
( 2)
Rdes
= k des
S( 2) e

E(1 )
kT

E( 2 )
kT

In equilibrium, however, the surface area S1 does NOT change, and all four processes taken together
must not change the overall concentration c. Hence, processes I and IV increase S1, while II and III
decrease it:
E( 2 )
E(1 )

k PS ( 0 ) + k
(1)
ad

( 2)
des

S( 2)e

kT

=k

(1)
des

S (1) e

kT

( 2)
+ k ad
S (1) P

(eq. (II))

2. Thermodynamics of adsorption The BET adsorption isotherm

Subtracting Eq. (1) from Eq. (2) yields immediately

( 2)
des

S( 2) e

E( 2 )
kT

( 2)
= k ad
S (1) P (eq. III )

Following this scheme, one realizes that we can continue these considerations for all following layers
and obtain a general expression of the kind:
( i +1)
k des
S (i +1) e

E( i +1)
kT

( i +1)
= k ad
S ( i ) P (eq. IV )

for all surface parts Si, with i = 1, 2, 3.... The total volume adsorbed is obtained when we multiply the
surface part Si of each layer with the number of layers below and additionally multiply this with the
specific volume per surface part S, V0

Vtotal = V0

i S
i =1

= V0 (S0 + 1S1 + 2 S2 + 3 S3 + ...)

2. Thermodynamics of adsorption The BET adsorption isotherm

Of interest is now the ratio between the total adsorbed volume Vtot and the volume of a perfect
complete monolayer, Vm (which is nothing but the coverage ) :

Vtotal
=
= i =1
Vmono

i Si

S
i =1

In order to carry out these summations we have to make some (plausible) assumptions: First, all
adsorption energies of the second and all higher layers will be set equal = EL, and second, we assume
that all the ratios of adsorption/desorption constants are equal to a constant g:
( 2)
( 3)
(n)
k des
k des
k des
= (3) = ... = ( n ) = g
( 2)
k ad k ad
k ad

as expected for a liquid-like behaviour. Note that the very first adsorbed layer is treated differently,
because its rate constants as well as its adsorption energy is largely different from those of the higher
layers. By defining
(1)

x=

k
P
E
E1
exp L with L 2 and y = ad
exp

(1)
k
RT
g
RT

des

and introducing the constant c = y/x we can express all surface parts Si by S0 according to:

2. Thermodynamics of adsorption The BET adsorption isotherm

S1 = y S0;
S2 = x S1 = x y S0;
S3 = x S2 = x2 S1 = x2 y S0 ;
S4 = x S3 = x3 S1 = x3 y S0; or generally:

Si = x Si - 1 = xi - 1 S1 = c xi S0

If we define the ratio of the total volume adsorbed / the volume of the monolayer as coverage

(as we did before) we obtain:

i
i

V
= total =
Vmono

c S0 i x
i =1

S 0 1 + c x i
i =1

c i x
i =1

1 + c xi
i =1

In the denominator we have the sum of an infinite geometric progression:

And similarly, we have:

xi =
i =1

x
1 x

d i
d x
x
i
x
x
x
x
x
x
x
x

=
+
+
+
=
+
+
+
=

x
=
x
2
3
...
1
2
3
...
=

i
dx i =1
dx 1 x (1 x )2
i =1
i =1
1

2. Thermodynamics of adsorption The BET adsorption isotherm

Vtotal
cx
1
cx
cx

=
=
=
=
2
2
cx
1

x
+
cx
Vmono (1 x ) (1 + 1 x ) (1 x ) ( 1 x ) (1 x ) (1 x + cx )
This expression can be rearranged to yield:

1
Vtot

c 1 1
1
x
+

=
x

c Vmono
1 x Vmono c

The unknown quantity x can be substituted by the fact that the volume Vtotal must become infinity for p =
p0. Since Vmono is always finite, x has to become 1 for this case. Otherwise, X = p/p0 . Since p as well as
p0 are measurable quantities, we have thus eliminated the unknown quantity x and finally arrive at the
famous BET isotherm in the common form:

1 p
1
c 1 1
p

=
+

Vtotal p0 p c Vmono
c Vmono p0

Two limiting cases can be distinguished:

a) the constant C >> 1, then the heat of adsorption of the very first monolayer is large compared to the condensation
enthalpy, and adsorption into the second layer only occurs once the first layer is completely filled. For C >> 1 and p <<
p0 the BET isotherm changes to the Langmuir isotherm.
b) Conversely, if C is small, then E1 is smaller or comparable to EL, and a multilayer adsorption already occurs while the
first layer is still incomplete.

2. Thermodynamics of adsorption The BET adsorption isotherm

For a successful determination of a BET surface the limiting case a) is required, i.e., C >> 1. In this
1
p
1
p
vs.
case, a plot of
yields a straight line with positive slope and intercept
from
Vtotal po p
p0
c Vmono
which the constant c and the monolayer sorption capacity can be obtained. Knowing the space
requirement of the adsorbed gas then enables one to calculate the specific active surface of the
respective catalyst material.

2. Thermodynamics: practical measurement of adsorption isotherms

Owing to the small single crystal surface areas it is difficult to directly determine the adsorbate
concentration (on a 0.5 cm2 surface, only ~ 1012...1014 particles are adsorbed, corresponding to
only 10-12...10-9 moles). Hence, indirect determinations are usually required. The adsorption of
particles on a surface is commonly accompanied by changes in the surface geometry, element
concentration, charge distribution, optical reflectivity, etc., and there exist various means to
indirectly monitor the number of adsorbed particles:

Intensities of adsorbate-specific electronic excitations (Auger, UV- and X-ray hotoelectron spectroscopy)
Intensities of adsorbate-specific vibrational modes (IR, HREELS)
Intensities of adsorbate-induced LEED superstructures (not always unambiguous)
Changes of the surface work function induced by the adsorbate [Tracy & Palmberg, JCP 51 (1969) 4852]
Changes of the surface conductivity (thin films!)
Changes of the paramagnetic resonance

In each case it is necessary to calibrate the respective quantity against the actually adsorbed
amount, e.g., by thermal desorption spectroscopy or (in case of solids) by a quartz microbalance
etc.). It is helpful, if the quantity measured is linearly correlated with the adsorbed amount.
In the following it will be exploited that the adsorbate-induced work function change
is often a quite convenient indirect monitor of the adsorbed amount, since it measures
the sum of the surface dipole moments of the adsorbed particles. The physical basis is
the validity of the Helmholtz equation

= 0

N ad
0 A

with 0 = initial dipole moment and 0 = dielectric constant; A = surface area

2. Thermodynamics: practical measurement of adsorption isotherms

The adsorption of the noble gas Xe decreases the W.F. of Ni(100) by ca. 340 meV. Stepwise increase of PXe at a
fixed temperature (93,6 K) shifts the adsorption - desorption equilibrium towards higher Xe coverages, which can
be followed by the corresponding W.F. changes. Stopping the xenon pressure (arrow) leads to complete removal of
the Xe atoms by thermal desorption. Next the corresponding pairs of (~ ) and P are plotted as a 93,2 K
isotherm [K. Chr. & J.E. Demuth, Surf. Sci. 120 (1982) 291].
Note: The proportionality between and has been checked independently by TDS and UPS (not shown here).

2. Thermodynamics: practical measurement of adsorption isotherms

One immediately realises that the adsorption - desorption equilibrium leading to a given Xe coverage (e.g., = 0,5)
can be achieved by different pressures and temperatures. This leads us to the important quantity hidden in the
adsorption isotherm, namely, the differential heat of adsorption: The strength of the adsorbate bond determines
how high the temperature must be to remove an adsorbed particle from the surface and to what extent one has to
increase the pressure in order to push this particle back to the surface.
At this point we recall the Clausius-Clapeyron equation describing the adsorption - desorption equilibrium:
hg had + OF d ln P
qst
1 dP
=

=
=
2
2
P dT
RT
dT RT

2. Thermodynamics: practical measurement of adsorption isotherms

The determination of the isosteric heat of adsorption requires an
integration of the Clausius-Clapeyron equation, which is simple
as long as qst is T-independent (tacitly assumed):

hg had +OF d ln P
q
1 dP
= st
=
=

2
1
P dT
RT
R
d ( T )

In the following, a few more

examples examples will be
shown of how one gets access
to an adsorption isotherm and
how one can deduce isosteric
heats of adsorption.

Plotting the ln P vs. the inverse absolute temperature should yield, for each coverage , straight lines with slope
- qst/R (left figure). In this way one obtains the heat of adsorption for all coverages of interest (right figure). This
important graph carries the information about the energetic homogeneity of the surface in question (see later).

2. Thermodynamics: practical measurement of adsorption isotherms

Example: CO + Pd(111)

G. Ertl & J. Koch, Z. Naturf. 25a (1970) 1906

2. Thermodynamics: practical measurement of adsorption isotherms

Example: Hydrogen on a Nickel(110) surface [K.Chr. et al., J.Chem. Phys. 60 (1974) 4528]

Adsorption isotherms

Adsorption isosteres

Coverage dependence of the isosteric heat of adsorption

2. Thermodynamics: calorimetric measurements of the heat of adsorption

In calorimetric
measurements certain
gas portions are
subsequently admitted
to the system while the
heat released is
monitored by a T
increase. Accordingly,
the heat portions sum
up to the integral heat of
adsorption. If the
surface is subjected to a
large gas portion
leading to saturation,
the whole integral heat
of adsorption is
liberated.

The famous Wedler glass cell for determining the heats of

adsorption on thin polycrystalline films by calorimetry. [F. Brcker
& G. Wedler, Disc. Faraday Soc. 41 (1966) 87

2. Thermodynamics: calorimetric measurements of the heat of adsorption

The technique of calorimetric measurements has been revived by modern instrumentation, c.f., M.C. Crowe & C.T.
Campbell, Annu. Rev. Anal. Chem. 4 (2011) 41 - 58 and references therein.
Pioneering first work performed in the group of D.A. King (Cambridge):
Stuckless JT, Al-Sarraf N, Wartnaby C, King DA. Calorimetric heats of adsorption for CO on
nickel single crystal surfaces. J. Chem. Phys. 99 (1993) 2202
Brown WA, Kose R, King DA., Femtomole adsorption calorimetry on single-crystal surfaces. Chem.Rev. 98
(1998) 797
(a) The Campbell groups single-crystal adsorption
calorimetry (SCAC) pyroelectric heat detector assembly,
which houses a polyvinylidene fluoride (PVDF) ribbon (4
mm wide, held in a 6-mm-diameter arc), shown
retracted from a single-crystal sample holder that exposes
an 8-mm-diameter area of the Pt(111) samples surface. A
linear translator is used to press the pyroelectric detector
into good thermal contact with the back
face of the sample during heat measurement. The ribbon is
fully coated on the front and back faces with a thin layer of
aluminum, as received from the manufacturer
(Goodfellow).
(b) Schematic of an SCAC experiment. The molecular
beam emits spatiotemporally resolved pulses of gas-phase
atoms or molecules that impinge on the sample surface;
the pyroelectric heat detector, shown pressed into contact
with the back face of a single-crystal sample, measures
temperature changes associated with adsorption events.
(c) Exploded view of the detector assembly, showing the
PVDF ribbon with patterned aluminum electrodes, ribbonmounting hardware, and a portion of the associated
electronic circuitry.

2. Thermodynamics: calorimetric measurements of the heat of adsorption

Measured heat of Ag atom adsorption on oxide surfaces, versus the diameter of the Ag particle to which it adsorbs, for four different
surfaces: two 4-nm-thick CeO2(111) films with different extents of surface reduction (CeO2-x , x = 0.1 and 0.2), a 1-nm-thick CeO 1.9(111)
film, and a 4-nm-thick MgO(100) film. The first three films were grown on Pt(111), and the fourth was grown on Mo(100). (Right)
Structural models for perfect CeO2(111) and MgO(100); black lines indicate their unit cells. After Farmer JA, Campbell CT. 2010. Ceria
maintains smaller metal catalyst particles by strong metal-support binding. Science 329:93336

2. Thermodynamics of adsorption: initial heat of adsorption and a-priori heterogeneity

Initial heats of adsorption for some
selected adsorption systems
Adsorbate
hydrogen H2
oxygen O2
carbon dioxide CO2
xenon Xe
xenon Xe
hydrogen H (from H2)
hydrogen H (from H2)
hydrogen H (from H2)
carbon monoxide CO
oxygen O (from O2)
oxygen O (from O2)

Surface
Ru(0001)(1x1)-H
Ru
Cu
Ni
Pd
Pt
Co, Ni, Ru, Rh, Pd
Mo, W, Re
Ni, Pd
Ru, Pd
Rh, Re

E0 from...to [kJ/Mol]
2...5
10...20
10...30
15...25
20...40
60...80
30...100
125...175
100...150
200...250
200...300

copper Cu

Ru, W, Re

250...330

The so-called initial adsorption energy E0 is released by the very first

particle arriving at the clean surface. It is a characteristic quantity for any
given adsorption system and reflects the principal strength of the
interaction between adsorbate and substrate (cf. Table on the left-hand
side).

On a given surface there can inherently exist adsorption sites with different
local geometry providing different chemical coordination; good examples
being surfaces with regular steps. Usually, the strength of adsorbate
substrate interaction increases with the coordination number. Example:
H/Pd.

Real-space model
of a real surface
with various defects
(steps, kinks, holes
etc.)

Coverage dependence
of Ead for Pd(111)/H: full
circles: smooth (111)
surface; open circles:
periodically stepped
Pd(111) surface (after
Conrad et al.)

2. The adsorption energy: a-posteriori heterogeneity;

induced lateral interactions
Even the very first adsorbing particle may alter both
shape and depth of the adsorption potentials of
neigh-boring sites, due to lateral interactions. The
respective forces can be attractive or repulsive.
The distance dependence of the total potential (c) is
a superposition of the periodic potential of the
uncovered crystal surface (a) and the particle pair
potential (b).
The interactions can be mediated either by direct
orbital orbital repulsions (direct interactions) or
through the substrate (indirect interactions). The
reason is that the charge associated with a given
adsorption site is shared between adjacent adparticles, as illustrated below.

Schematic representation of the potential modulation

Often, lateral interactions become feasible only

beyond a certain critical adsorbate surface
concentration. Since they are mostly repulsive,
one observes a sudden drop in the heat of
adsorption. Example: CO adsorption on
Ru(0001) [Pfnr et al.]: Up to a 3 x 3 -R30
structure the CO molecules can occupy
equivalent sites; thereafter the adsorbed COs feel
the repulsive forces to the neighbors.
Consequently, the heat of adsorption falls
abruptly by ca. 50 kJ/Mol.

3. Adsorption and desorption as kinetic phenomena

As in any chemical reaction, surface

processes involve breaking and making of
bonds. This can involve substantial
amounts of energy.
However, as one can see from the stability
of a mixture of hydrogen and oxygen gas,
there are many metastable reaction
systems which cannot (at least not without
external support) reach equilibrium.
Reason: Large activation barriers slow
down the reaction rate.
This is what catalysis is all about: A
catalyst provides an easier reaction path
offering a greatly reduced activation
energy barrier. In a sense, an active
surface which readily adsorbs and
dissociates molecules from the gas phase
is nothing but an efficient catalyst.

3. Adsorption and desorption kinetics:

The rate of adsorption terminology

Consider the rate of an ordinary chemical reaction of type A B. The general definition of the
reaction rate is change per time, i.e. number of molecules per time, or concentration change of a
certain species per unit time interval:

Rate R =

moles
dN A
dN P
particles
d [ A]
d [ P]
]
=+
in [ 3 ] or R =
=+
in [
dt
dt
dm s
dt
dt
Volume s

For surface reactions, two-dimensional concentrations must be considered instead (quantities

related to unit area)

dN S , A
dN S , P
particles
d
moles
d [ A]
d [ P]
=+
=+
]=
in [
in [ 2 ] or R =
Rate R =
unit area s
dt
dm s
dt
dt
dt
dt
Often, rates are expressed in terms of change of coverage =

N ad
N ad , max

, with 0 < < 1.

3. Adsorption and desorption kinetics:

The rate of adsorption terminology

In an adsorption experiment, an initially bare surface is exposed to a certain gas pressure P, and the
rate of collision of the gas particles with the unit surface is given by kinetic theory:
Collision frequency

dN
=
A dt

P
particles
2
2 mkT m s

The decisive process now is the so-called trapping or sticking: Will the impinging particle stay on the
surface or will it be reflected? The respective probability is called (initial) sticking probability s0 and
varies between zero and one. As the surface is gradually covered, s decreases simply because the
number of empty adsorption sites becomes smaller.
If an adsorbed particle statistically occupies
a single site, s() = 1 - ;

1st order adsorption;

if it dissociates, two sites are blocked by one collision event, and accordingly
s() = (1- )(1- ) = (1 - )2 2nd order adsorption.
*

Sometimes, the adsorption reaction requires a certain activation energy Ead . Only particles having at
least this energy will be able to stick.

3. Adsorption and desorption kinetics: The rate of adsorption terminology

The rate of adsorption is then:

*
Ead
s0
d
P
kT
Rad =
=
f ( )
e
dt
N max
2 mkT

(for a first-order process, f () = 1 - , for a second-order process, f () = (1 - )2). This rate

expression allows to calculate the actual coverage after a certain time of gas exposure, simply by
integrating the rate equation. For a non-activated 1st order process, one obtains

(t ) = 1 e

k ad t

, with

k ad =

s0
N max

P
2 mkT

which is a typical saturation function. The larger the initial sticking coefficient, the more rapidly the
surface becomes covered. Metal single crystal surfaces (Ni) contain ca. 1019 adsorption sites/m2.
A rough estimate neglecting the coverage dependence of the sticking function (constant unity
sticking probability) yields that a surface would be completely covered in one second, if one
maintains a pressure of ~10-6 mbar.

3. Adsorption and desorption kinetics: The rate of adsorption terminology

The initial sticking probability s0 is an interesting quantity; it contains all dynamical and steric effects
and is governed by the ability or effectiveness of a given particle to dissipate its kinetic energy to the
heat bath of the surface (phonon excitation; electron hole pair excitation). The sticking of diatomic
molecules, hydrogen in particular, is of special interest, since the sticking at finite temperatures includes
the dissociation reaction (activated and non-activated dissociation paths).
An extensive consideration of the quantum-dynamical background of sticking can be found in the book
by A. Gro (Theoretical Surface Science, Springer Berlin 2003)
Complications can arise when the colliding particle is trapped for some time (typically microseconds) in
a weak potential in which it can freely move across the surface and search for an empty adsorption site.
This weakly bound state is called a precursor. Major consequence of a precursor state: The sticking
coverage function f() is no longer linear, but has a convex shape: At not too large coverages the
sticking remains high, but as the diffusion length in the precursor state becomes shorter than the mean
diameter of the already covered area (island), the particles can no longer be accommodated and are
finally reflected back into the gas phase before they can find an empty site.

3. Adsorption and desorption kinetics: Precursor effects

Kisliuk precursor kinetics:

(P.J. Kisliuk 1955)

pd"
K = '
pch + pd'

s ( ) = s0

1
1 + 1 K

pch = probability that the particle adsorbs from an intrinsic precursor

(no adsorbed atom underneath)
pd = probability that particle desorbs from intrinsic precursor to gas phase
pd = probability that particle desorbs from extrinsic precursor (adsorbed
Particle underneath)

3. Adsorption and desorption kinetics: The rate of desorption terminology

The removal of particles from the adsorbed state into the gas phase is called desorption. It can be achieved by
thermal energy (thermal desorption, temperature-programmed desorption), electron impact (EID, DIET), ion impact,
resonant photon irradiation etc. Here, only thermal desorption is considered. Again, desorption is understood as a
normal chemical reaction and described by the respective kinetic formalism: Aad Agas with rate constant kdes.

Upon introducing again the coverage

particles
dN ad
= k des N adx

2
dt
m s

= Nad/Nmax

we obtain:

d
x 1
= k des x N max
dt

As in any chemical reaction with an activation barrier, kdes can be expressed as a product of a pre-exponential factor,
des , and an exponential term containing the activation energy :
Edes

k des = exp

kT

Inserting this in the first equation yields the well-known Polanyi-Wigner equation which is the basis for a determination
of both energetic and kinetic quantities from a thermal desorption spectrum:

Rate of desorption: R = R (, T)

Edes
d
kT
x
x 1

= x N max e
dt

3. Adsorption and desorption kinetics: The rate of desorption

application of the Polanyi-Wigner equation

R = rate of desorption: R = R (, T) =

Edes
d
kT
x
x 1

= x N max e
dt

Of interest: To determine the kinetic quantities such as the

order of the desorption, x, which contains valuable information on the mechanism of the desorption
process
Pre-exponential factor which allows conclusions on the configuration and mobility of the adsorbed
phase (mobile or immobile adsorbed layer)
Activation energy (= energy to remove a particle from the bottom of the potential energy well =
adsorption energy, which we have already discussed)
Furthermore: Note that the peak integrals reflect the adsorbed amount prior to application of the
temperature program. Using a coverage calibration (LEED pattern etc.), absolute coverages are
accessible.
In addition, by plotting the TPD peak integrals versus the exposure, one can determine the
sticking probability as a function of coverage.

3. The rate of desorption: Visualisation of the Polanyi-Wigner equation

The desorption rate of Cu from Re(0001) as a function of two variables T and [R. Wagner + K.C. Surf. Sci. 469 (2000) 55]

3. The experimental procedure: Taking TPD spectra

Thermal Desorption spectroscopy

is probably the most frequently used method
to gain information on the energetics and
kinetics of adsorbed particles.
The adsorbate-covered surface is heated
(mostly in a time-linear fashion), while the
desorbing particles are collected and
monitored with a mass spectrometer. In a
pumped recipient one obtains pressure
maxima right at those temperatures, where
most of the substrate adsorbate bonds are
thermally cleaved. The obtained desorption
peaks contain all relevant information about
activation energies for desorption, reaction
orders and surface concentrations as pointed
out before.

Typical experimental set-up for a TD experiment

3. The experimental procedure: Evaluating TPD spectra

There exist different procedures to evaluate TD spectra. Relatively simple is the Redhead analysis [P.A. Redhead,
Vacuum 12 (1962) 201], which neglects coverage dependences. More accurate, but more complex, are line-shape
analyses such as the ones described by E. Bauer [Surf.Sci. 53 (1975) 87] or by D.A. King [Surf.Sci. 47 (1975) 384],
which explicitly consider dependences. Here briefly commented: line shape analysis by King.
a) Family of thre TD curves. After various
desorption times t different amounts of
adsorbates t are left on the surface. These
times vary linearly with the surface
temperatures T. At times t1, t2, t3 different
desorption rates d/dt are obtained for each
individual TD curve.
b) For each TD curve 1, 2 and 3, t is
plotted against time t or temperature T; one
obtains s-like descending curves. Horizontal
cuts at various t values establish the
condition that a fixed coverage is obtained
at different temperatures T1, T2, and T3 for
each TD curve.
c) A plot of ln(d/dt) vs T-1 yields isosteric
straight lines whose slopes contain
thedesorption energy Edes at the respective
coverage t. If either the pre-exponential v
or the desorption order x is known, the
respective other quantity can be evaluated
from the intercept. [D.A. King, Surf.Sci. 47
(1975) 384]

3. The experimental procedure: Evaluating

TPD spectra

D. Schlatterbeck et al,
Surf.Sci. 418 (1998) 240

TD spectra of Ag from Re(0001)

-dependence of and Edes Ag/Re

Physisorption and chemisorption as monitored

by thermal desorption spectroscopy
Depending on the strength of the interaction energy one
distinguishes between physisorption and chemisorption. This
distinction is, however, not rigid. Below, say, 20 kJ/Mol there
is genuine physisorption, the interaction is dominated by vander-Waals-forces. Example: Rare gas adsorption on
graphite, hydrogen adsorption on Ru(0001)/(1x1)-H. Beyond
50 kJ/Mol there dominates chemisorption with chemical
binding forces. Example: oxygen adsorption on a
rhodium(110) face.
Notice the extremely low
desorption temperatures !

TD spectra of H2 from a Ru(0001) surface (after Frie et al.)

The heats of adsorption and desorption range only between

1,6 and 2,9 kJoule/Mol !
Notice the extremely high
Desorption temperatures!

The heats of adsorption and desorption

range between 200 and 300 kJoule/Mol !

Oxygen desorption from a Rh(110)-surface

Summary and outlook

One additional remark about the importance of the adsorption in heterogeneous catalysis
In heterogeneous catalysis surfaces help to concentrate the reactants
from the gas phase at phase boundaries and, more importantly, to
stretch or even break inner-molecular bonds and to facilitate bondmaking with coadsorbed species. However, too loosely bound
molecules may stay only very shortly on the surface and are
immediately lost by desorption, while too strongly bound species are
immobile and cannot take part in Langmuir-Hinshelwood surface
reactions. Consequently, it is advantageous, if adsorbates are bound
moderately strong. This is illustrated in the right-hand frame showing
the activity for methane formation from synthesis gas (CO+H2). These
typical vulcano curves are known for various adsorbates.
A second effect is crucial: In surface reactions, often homonuclear diatomic
molecules with strong covalent bonds must be cleaved. Their potential energy
minima lie far away from the surface, and there exist strong activation barriers for
dissociation, an example being nitrogen N2 interacting with Fe surfaces. Adding
electropositive or negative species changes the local surface charge density and
often increases the adsorption energy of the molecules. Consequently, activation
barriers for dissociation are reduced.
An effect of this kind was found, for example, by Ertl and coworkers for nitrogen
molecules interacting with clean and potassium-covered Fe single crystal surfaces.
It can also be made responsible for the acceleration of CO hydration reactions by
coadsorbed alkali metals.

Some literature for further reading

1) G. Ertl & J. Kppers, Low-energy Electrons and Surface Chemistry, 2. Auflage, Verlag Chemie,
Weinheim 1985
2) W. Gpel & M. Henzler, Oberflchenphysik des Festkrpers, Teubner-Verlag, Stuttgart 1993
3) K. Christmann, Introduction to Surface Physical Chemistry, Steinkopff-Verlag, Darmstadt 1991
4) M.W. Roberts & C.S. McKee, Chemistry of the Metal-Gas Interface, Clarendon, Oxford 1978
5) A. Clark, The Theory of Adsorption and Catalysis, Academic Press, New York 1970
6) A. Clark, The Chemisorptive Bond - Basic Concepts , Academic Press, New York 1974
7) E. Shustorovich, Hrsg., Metal - Surface Reaction Energetics, VCH-Verlagsges. Weinheim 1991
8) G. Ertl, in: The Nature of the Surface Chemical Bond (G. Ertl & T.N. Rhodin, eds.), North Holland
Publishing Company, 1979, Ch. V, pp. 315 - 380.
9) D. Menzel in: Interactions on Metal Surfaces (R. Gomer, ed.), Series: Topics in Applied Physics,
Vol. 4, Springer-Verlag Berlin, Heidelberg, New York, 1975, Ch. 4, pp. 101 - 142
10) R.I. Masel, Principles of Adsorption and Reaction on Solid Surfaces, Wiley, New York 1996
As well as various original publications and textbooks on adsorption/desorption phenomena.