Lecture 18: Reactions at Solid Surfaces

For the next couple of lectures, we will consider reactions at surfaces. We will primarily look at
heterogeneous catalysis at gas-solid interfaces, where the solid surface serves to accelerate an
overall reaction with gas phase reactants and products. However, much of what we do will also
be applicable to situations where material is deposited or removed at a gas-solid interface
(chemical vapor deposition or etching, respectively) and in situations where the gas-solid
interface is replaced by a liquid-solid interface. This material is covered in chapter 2 of the
textbook by Froment and Bischoff.

A heterogeneous catalyst accelerates a reaction that has fluid phase reactants and products (so the
solid is nominally unchanged by the reaction). In general, it does this by stabilizing reactive
intermediates whose formation would be energetically unfavorable in the fluid phase. The great
advantage of heterogeneous catalysis over homogenous catalysis is that the since the catalyst is
in a separate phase from the reacting mixture, it can stay in the reactor while the reactants and
products flow through. We will not discuss the chemistry of particular catalytic systems, but will
try to develop some basic approaches that can be applied to a range of systems.

For a catalytic reaction occurring in a porous catalyst pellet (the traditional case in chemical
engineering applications), we can envision the overall reaction as a series of sequential steps:

(1) Transport of reactants to the exterior surface of the catalyst pellet.

(2) Transport of reactants to active sites inside the pores.
(3) Adsorption of reactants onto the surface at active sites.
(4) Reaction on the surface.
(5) Desorption of products from the surface.
(6) Transport of products out of the pores to the pellet surface.
(7) Transport of products from the pellet surface to the bulk fluid.

For the generic catalytic reaction A → B, this can be illustrated schematically as

We will first focus on steps 3, 4, and 5 – which we might call the reactive steps, as opposed to
steps 1, 2, 6, and 7, which involve only transport and not chemical transformations. In
subsequent lectures, we will consider the effects of transport limitations that can be both external
(steps 1 and 7) and internal (steps 2 and 6) to the catalyst pellet. For situations other than
heterogeneous catalysis with a porous pellet catalyst, some of these steps may be missing. In a
CVD reactor, for example, the surface is generally flat, smooth, and nonporous. Then, the
transport steps can be simplified or eliminated, but the reactive steps remain essentially the same.

Adsorption of gases on solid surfaces is often classified as chemisorption or physisorption,

depending on the strength of interaction between the surface and the adsorbed molecule.

Physisorption is characterized by interaction energies comparable to heats of vaporization

(condensation). The adsorbate is held to the surface by relatively weak van der Waals forces.
Multiple layers may be adsorbed with approximately the same heat of adsorption (∆H for the
reaction written as A + surface → adsorbed A). The heat of adsorption for physisorption is at
most a few kcal/mole.

Chemisorption is characterized by interaction energies between the surface and adsorbate

comparable to the strength of chemical bonds (tens of kcal/mole). The adsorbate forms a
chemical bond to a surface atom (or atoms). The adsorbate may dissociate or otherwise react
upon adsorption. Generally only a single molecular layer can be adsorbed. The heat of
adsorption is tens of kcal/mole.

Of course, this is a somewhat arbitrary distinction, but it has proven to be a useful one. In
general, physisorption behaves like condensation, while chemisorption behaves like chemical
reaction. For a surface-catalyzed reaction to occur, chemical bonds must be made and broken, so
we will focus primarily on chemisorption.

Adsorption Isotherms:
The simplest, most widely used, and most widely taught theory of chemisorption is the Langmuir
theory of adsorption, from which we derive the Langmuir isotherm. An isotherm describes the
dependence of the equilibrium surface concentration of an adsorbed molecule on its gas phase
concentration at constant temperature. The Langmuir isotherm is based on 3 primary
assumptions:

(1) The solid surface is made up of a uniform array of energetically identical adsorption sites.
(2) A maximum of one monolayer (one adsorbate molecule per site) can be adsorbed.
(3) There are no interactions between the adsorbed molecules.

A kinetic derivation of the Langmuir isotherm is as follows:

We write the adsorption as an elementary chemical reaction that obeys mass-action kinetics.

A + S ↔ A-S with rate = kaCACS – kdCA-S

where A is the gas phase reactant (adsorbate), S is an empty surface site, and A-S is a
chemisorbed A molecule. The gas-phase concentration of A, CA, has units of moles per volume
or molecules per volume, while the concentrations of the surface species, CS and CA-S, are
expressed in units of moles per surface area or molecules per surface area. The rate constants for
adsorption (ka) and desorption (kd) are assumed, in this theory, to be independent of the surface
concentrations. This follows from the assumptions that the surface sites are all identical and that
there are no interactions between adsorbed molecules. By our assumption that the surface
contains a fixed number of sites per area that can each accommodate one adsorbed molecule, CS
+ CA-S = Ct = constant total concentration of surface sites (occupied and unoccupied). At
equilibrium, the net adsorption rate is zero

kaCACS – kdCA-S = 0, or kaCACS = kdCA-S

Substituting CS = Ct - CA-S and solving for CA-S gives

This can be written in terms of the adsorption equilibrium constant Ka = ka/kd as

Instead of using the surface concentrations (CA-S, CS) as the concentration variables, it is often
preferable to use the fractional coverages, defined by θA = CA-S/Ct, and θS = CS/Ct. In terms of
the fractional coverage of A, the Langmuir isotherm is

The fractional coverage of empty sites is θS = 1 – θA. If Ka were expressed in terms of pressure
units rather than concentration units, then the gas phase concentration CA would be replaced by
the gas phase partial pressure pA. Isotherms are most often expressed in terms of partial pressure.

This adsorption isotherm (in terms of dimensionless variables) looks like

We will make extensive use of this isotherm, and variations on it, in writing approximate
expressions for the rates of reactions at surfaces.

Next, we will consider some simple extensions of the Langmuir isotherm. First, consider
competitive adsorption of two species onto the same type of site.

A + S ↔ A-S with rate = kaACACS – kdACA-S

B + S ↔ B-S with rate = kaBCBCS – kdBCB-S

At equilibrium, both rates are zero, so

kaACACS = kdACA-S
kaBCBCS = kdBCB-S

From which

CA-S = (kaA / kdA) CACS = KACACS

CB-S = (kaB / kdB) CBCS = KBCBCS

The total number of sites is still constant

CS + CA-S + CB-S = Ct

Writing the above 3 expressions in terms of site fractions gives,

θA = KACAθS
θB = KBCBθS
θS +θA +θB = 1

Substituting the first two expressions into the third one gives

θS + KACAθS + KBCBθS = θS (1 + KACA + KBCB ) = 1

From which

In general, it can be shown that the Langmuir isotherm for competitive adsorption of n species
on identical sites gives
Another frequently used modification is the Langmuir isotherm for dissociative adsorption. That
is

A2 + 2 S ↔ 2 A-S with rate = kaCA2 CS2  kd CA2 S

So, at equilibrium,

Or, in terms of site fractions

and we still have θA + θS = 1, so

kaCA2 (1   A )2  kd A2 , or K ACA2 (1   A )2   A2

And solving for θA gives

There are an almost unlimited variety of adsorption isotherms like this that can be derived.

Two other isotherms are the Freundlich isotherm and the Temkin isotherm.

The Freundlich isotherm is

where α and β are generally taken as fitting parameters. It has 2 parameters, which is one more
than the Langmuir isotherm, so it can often be made to fit adsorption data better over a limited
range.

The Temkin isotherm can be written as

where H ad 0
is the heat of adsorption at zero coverage, and α is a fitting parameter. This
isotherm is derived by assuming that the heat of adsorption varies linearly with surface coverage,
that is

Had  Had
0
(1   A )

There are, of course, many more sophisticated means of treating adsorption, but those listed
above are the most common. An important advantage of the Langmuir isotherm is that it can be
simply extended to competitive adsorption, dissociative adsorption, etc. while the other
isotherms mentioned here cannot be easily extended to these more complex situations.

Much more sophisticated models of adsorption can be obtained using statistical mechanical
treatments of the adsorbed molecules. One class of these methods is called lattice gas methods,
because they assume that the molecules are adsorbed on a regular array of lattice sites, and that
the interactions between molecules are like those in the gas phase. Detailed interactions between
the adsorbed molecules and the arrangement of the molecules on the surface can be considered.
Any serious consideration of these methods is more appropriate to a statistical mechanics course
than a course in reaction kinetics, so we will not pursue that topic further.

Rates of Reactions at Surfaces:

Having discussed adsorption, we will now proceed to write rate equations for surface catalyzed
reactions. If, in developing these rate equations, we assume that the adsorption and desorption
steps are in equilibrium and obey the Langmuir adsorption isotherm (or the appropriate
modification of it) then these rate expressions are called Langmuir-Hinshelwood kinetics. We
will also consider more general treatment of these surface reactions and simplifications obtained
by assuming different rate-limiting steps.

Consider the overall reaction:

A↔B

with the assumed reaction mechanism

A + S ↔ A-S with rate = r1 = kaACACS – kdACA-S

A-S ↔ B-S with rate = r2 = kfCA-S – krCB-S
B-S ↔ B + S with rate = r3 = kdBCB-S – kaBCBCS

The rate equations for the five species (2 gas phase, 3 surface) can be written as
Since this is a surface process, the reaction rates all have units of moles per area per time.
Therefore, the rate equations for the fluid phase species must include a surface to volume ratio
that relates the rates of the surface processes to the rates of change of concentrations in the fluid
phase. Here, we consider the gas phase concentrations to be uniform throughout the volume of
the system (perfect mixing) and the surface phase concentrations to be uniform throughout the
system, so we still need not consider any spatial dependence of the concentrations. In that case,
the appropriate factor is just the surface to volume ratio of the system (considering only surfaces
where reaction occurs).

The equations for the surface concentrations could be re-written in terms of the site fractions as

If we would like, we can replace one of these three differential equations with the surface site
balance (assuming the total number of surface sites is conserved). Instead of using one of the
rate equations, we use the algebraic relationship

θA + θB +θS = 1

The conceptually simplest but mathematically most difficult method of treating these rate
equations is to simply solve them. This generally cannot be done analytically, but we know how
to integrate a set of rate equations like this numerically.

The next most justifiable and most straightforward treatment is to assume that the surface
concentrations are in steady-state (do not change with time). If the reaction is taking place in a
reactor operating at steady state, then this will be rigorously true. To see this, recall that the
species balance equation for a component in the fluid phase in a reactor will have a form like
Where the three terms on the right-hand-side represent convection (with velocity v), diffusion or
dispersion (with effective diffusion coefficient D), and production by chemical reaction. The
production rate by chemical reaction is the rate equation that we would write for a batch reactor,
i.e. for this problem the production rate of A would be

At steady-state, the time derivative is zero, and we have

The same equation could be written for the surface species, but the solid surface does not move
(v = 0), and the diffusion on the surface is usually negligibly slow (D = 0), so for a surface
species, this just becomes

Even for time-dependent problems, it is often the case that the time scale for changes in the
surface coverages is much faster than the time scale for the overall reactor dynamics, so that the
assumption that the surface concentrations are in steady state is still a good one. In that case, it is
equivalent to the pseudo-steady-state approximation that we made for gas phase species that
participate in fast reactions.

If we assume that the surface coverages are in steady state, we obtain four equations for the
surface site fractions, of which 3 are independent (any of the 1st three equations below can be
written as a combination of the other two).

Solving these equations for the surface coverages involves substantial algebra, but presents no
fundamental difficulties. The solution was found to be
This could be re-written using the adsorption equilibrium constants, but we will not do so here.
The overall reaction rate is equal to r1, r2, or r3 (they are all equal), and may be written as

This is a reasonably compact analytical expression for the overall reaction rate that is rigorously
valid in a steady-state reactor, and approximately valid for a range of other conditions.

An even simpler approximation is to assume that the adsorption steps are near equilibrium, so
that the surface coverages are given by the Langmuir isotherm. Note that they cannot be exactly
at equilibrium, since then the reaction rate would be zero. This assumption is equivalent to
assuming that the surface reaction step (A-S ↔ B-S) is the rate-limiting step.

In this case, the surface coverages are given by

So, the reaction rate is given by

This same expression could be obtained from the general expression derived above, by assuming
that kf and kr are much smaller than all of the adsorption and desorption rate constants (so that
surface reaction is the rate-limiting step).
Rate expressions derived using the assumption of adsorption equilibrium are often called
Langmuir-Hinshelwood kinetics, after the originators of this technique.

We could likewise assume that the adsorption step is rate-limiting, so that the other two steps are
(approximately) in equilibrium

r2 = Ct (kfθA – krθB) = 0
r3 = Ct (kdBθB – kaBCBθS) = Ct (kdBθB – kaBCB(1-θA -θB)) = 0

In the second equation, we have used the overall site balance to substitute (1-θA -θB) for θS.
Solving for θA and θB gives

So, the overall reaction rate is

where Kr is the equilibrium constant for the surface reaction (A-S ↔ B-S)
In Froment and Bischoff, it is stated that for surface catalyzed reactions with a mechanism based
on chemisorption as we have discussed here and a single rate determining step, the reaction rate
can always be written as a combination of three groups. These are a kinetic group, a driving-
force group, and an adsorption group. The rate is expressed as

For the above example with the surface reaction taken to be the rate-limiting step, these groups
were

Kinetic factor = kfKACt

Driving force = CA – (KBCB)/(KAKr) = CA – CB/K
Adsorption group = 1 + KACA + KBCB
Where K (= KB/(KAKr)) is the equilibrium constant for the overall reaction (A ↔ B).

When the adsorption step is rate controlling, then the groups are

Kinetic factor = kaACt

Driving force = CA – (KBCB)/(KAKr) = CA – CB/K
Adsorption group = 1  (1  K r1 ) K B CB =1 + KACB/K + KBCB

On pages 79-80 of Froment and Bischoff, there are tables of these groups for many situations.
The tables are in terms of partial pressures, rather than concentrations, and assume that the rate
constants are defined such that the total site concentration (Ct) does not show up in the rate
expression.

The same approach can be applied in reactions that follow an Eley-Rideal mechanism. This is a
mechanism in which a gas phase species is assumed to react with a surface species directly,
without adsorbing, as in

A + B-S ↔ AB + S

We will conclude the topic of reactions at surfaces with a brief discussion of determination of
rate parameters for these reactions. There is an extensive discussion of this in Froment and
Bischoff, from the classical chemical engineering point of view where kinetics of the overall
process of interest are measured. Experiments are performed to measure the rate of the overall
process (including adsorption, desorption, and surface reaction). Great care must be taken to
ensure that mass transfer to and from the catalyst surface is sufficiently fast that there are no
mass-transfer effects on the rate. Concentrations are measured vs. time and compared to the
predictions of a reaction mechanism. Parameters in the mechanism are adjusted (fit) to minimize
the sum of the squares of the difference between model and experiment. Goodness-of-fit is taken
as a criterion for acceptance or rejection of the model. This is all the same as what we did before
for parameter fitting in general. The discussion in Froment and Bischoff goes well beyond what
we covered and includes important discussions of experimental design and statistical analysis of
data.

An alternative approach is to try to measure at least some of the elementary rate parameters in
separate experiments that isolate a single reaction. For example, an adsorption rate constant can
be characterized by a sticking coefficient or sticking probability. The adsorption rate is

rad = ZASA(θS)

where ZA is the flux of A hitting the surface (moles per area per time) and SA(θS) is the
probability that a given molecule will ‘stick’ to the surface (which depends on empty site
fraction, θS). For an ideal gas, the flux to the surface is
The adsorption rate constant is then given by

As written here, this rate constant is not really constant, but depends on the fractional coverage
of open surface sites. We often assume that the sticking coefficient S(θS) is just proportional to
the fractional coverage of open sites – that is S(θS) = γθS. Then

Unfortunately, γ is also often called the sticking coefficient. Sometimes, it is called (more
correctly) the zero-coverage sticking coefficient.

In a modern surface science experiment, the sticking coefficient at a particular temperature could
be measured by directing a molecular beam of the reactant at a clean, single-crystal surface in
ultra-high vacuum and measuring how much sticks. In these experiments, the sticking
coefficient could even be measured as a function of the angle of incidence, velocity, and internal
energy of the molecules.

Similar sophisticated surface science techniques can be applied to isolate other elementary steps
and measure their kinetics. However, this is relatively expensive and time consuming, and has
only really been done for a few systems. In principle, rates of these elementary reactions can
also be calculated using quantum mechanics and transition-state theory. Everything is
conceptually the same as in the gas phase, except that one of the molecules is replaced by an
infinitely large ‘molecule’ – the solid surface. That is not the case if the catalyst is amorphous or
if the active sites are some sort of ‘special’ locations on the surface like molecular steps or some
sort of defect. Also, many surfaces ‘reconstruct’ so that they have a different structure from
what we would get if we simply took a slice through a bulk crystal along a given crystal plane.

After completing your study of these lecture notes you should be able to:
(1) Understand and explain the difference between chemisorption and physisorption.
(2) Derive and use the Langmuir adsorption isotherm for a single species and for more general
cases of competitive adsorption, dissociative adsorption, etc.
(3) Write rate equations for a set of elementary reactions involving both gas phase and surface
phase species
(4) Solve rate equations for steady state surface species concentrations
(5) Derive Langmuir-Hinshelwood rate expressions for surface catalyzed reactions
(6) Derive an overall rate expression for a surface catalyzed reaction with a single rate-limiting
step
(7) Know what a sticking coefficient is and how it is related to the adsorption rate