72. K. Koch, B. Bhushan, and W. Barthlott. Progr. Mater. Sci. 54:137 (2009).

73. N.L. Abbott, C.B. Gorman, and G.M. Whitesides. Langmuir 11:16 (1995).
74. S. Abbott et al. Langmuir 15:8923 (1999).
75. B.S. Gallardo et al. Science 283:57 (1999).
76. K. Ichimura, S.K. Oh, and M. Nakagawa. Science 288:1624 (2000).
77. J. Lahann et al. Science 299:371 (2003).

CATALYSIS BY SURFACES

9.1 Introduction
9.1.1 A Brief History of Surface Catalysisc
9.2 Catalytic Action
9.2.1 Kinetic Expressions
9.2.2 Selective Catalysis
9.2.3 Tabulated Kinetic Parameters for Catalytic Reactions on Metal
Surfaces
9.3 Catalyst Preparation, Deactivation, and Regeneration
9.3.1 Catalyst Preparation
9.3.1.1 Industrial Catalysts
9.3.1.2 Model Catalysts for Surface-Science Studies
9.3.2 Catalyst Deactivation
9.3.3 Catalyst Regeneration
9.4 Techniques to Characterize Catalyst Surface and Study the Reactivity of
Catalysts
9.4.1 Turnover Rate Measurements
9.4.2 High-Pressure Reactors
9.4.3 In Situ Characterization Techniques: High-Pressure Scanning
Tunneling Microscope, Sum Frequency Generation Spectroscopy, and
Ambient Pressure X-Ray Photoelectron Spectroscopy
9.5 Metal Catalysis
9.5.1 Trends Across the Periodic Table
9.5.2 Some Frequently Used Concepts of Metal Catalysis
9.5.3 Most Frequently Used Catalyst Materials
9.6 Case Histories of Surface Catalysis
9.6.1 Ethylene Hydrogenation Over Platinum Surfaces
9.6.1.1 Surface Species Involved in Ethylene Hydrogenation
9.6.1.2 The Role of Reaction Intermediates in Ethylene Hydrogenation
9.6.1.3 Structure Insensitivity of Ethylene Hydrogenation
9.6.1.4 Catalysis in the Presence of a Strongly Adsorbed Overlayer
9.6.1.5 Summary
9.6.2 Ammonia Syntheses
9.6.2.1 Thermodynamics and Kinetics
9.6.2.2 Catalyst Preparation
9.6.2.3 Activity for Ammonia Synthesis Using Transition Metals Across the
Periodic Table
9.6.2.4 Surface Science of Ammonia Synthesis
9.6.2.5 Mechanism and Kinetics of Ammonia Synthesis
9.6.2.6 Summary
9.6.3 Oxidation of Carbon Monoxide on Transition Metal Catalysts
9.6.3.1 Carbon Monoxide Oxidation Under UHV Conditions
9.6.3.2 Carbon Monoxide Oxidation Under High-Pressure Conditions
9.6.3.3 Carbon Monoxide Oxidation Under Oxygen-Rich Conditions
9.6.3.4 Carbon Monoxide Oxidation Over Nanoparticles
9.6.3.5 Carbon Monoxide Oxidation at High Temperatures
9.6.3.6 Summary
9.7 Selectivity in Multipath Heterogeneous Catalytic Reactions
9.7.1 Energetic View of a Heterogeneous Catalytic Reaction with Multiple
Products
9.7.2 Surface Structure and Selectivity
9.7.3 Alloy Catalysts and Selectivity
9.7.4 Adsorbate-Induced Surface Restructuring
9.7.5 Strong Metal Support Interaction
9.7.6 Oxidation States of Metal Catalyst and Selectivity
9.7.7 Reaction Intermediates
9.7.8 Reaction Conditions and Selectivity
9.7.9 Other Important Research Directions to Catalytic Reaction Selectivity
9.7.10 Summary
9.8 Summary and Concepts
9.9 Problems
9.10 References
List of Tables
Table 9.1 Ethane Hydrogenolysis
Table 9.2 Propane Hydrogenolysis
Table 9.3 Cyclopropane Ring Opening
Table 9.4 Cyclopropane Hydrogenolysis
Table 9.5 n-Butane Hydrogenolysis
Table 9.6 Isobutane Hydrogenolysis
Table 9.7 Methylcyclopropane Ring Opening
Table 9.8 n-Pentane Hydrogenolysis
Table 9.9 Isopentane Hydrogenolysis
Table 9.10 Neopentane Hydrogenolysis
Table 9.11 Cyclopentane Ring Opening and Hydrogenolysis
Table 9.12 n-Hexane Hydrogenolysis
Table 9.13 2-Methylpentane Hydrogenolysis
Table 9.14 3-Methylpentane Hydrogenolysis
Table 9.15 Cyclohexane Hydrogenolysis
Table 9.16 Methylcyclopentane Ring Opening
Table 9.17 Methylcyclopentane Hydrogenolysis
Table 9.18 Benzene Hydrogenolysis
Table 9.19 n-Heptane Hydrogenolysis
Table 9.20 Toluene Hydrodealkylation and Hydrogenolysis
Table 9.21 Other Hydrogenolysis Reactions
Table 9.22 Cracking Reactions Over Nickel Powder
Table 9.23 Ethylene Hydrogenation
Table 9.24 Hydrogenation Reactions of Terminal Olefins
Table 9.25 Benzene Hydrogenation
Table 9.26 Other Hydrogenation Reactions
Table 9.27 Cyclohexane Dehydrogenation to Benzene
Table 9.28 Other Dehydrogenation (D) Reactions
Table 9.29 n-Butane Isomerization (I)
Table 9.30 Isobutane Isomerization (I)
Table 9.31 n-Pentane Isomerization (I)
 Table 9.32 Neopentane Isomerization (I)
Table 9.33 n-Hexane Isomerization (I)
Table 9.34 2-MethyIpentane Isomerization (I)
Table 9.35 3-Methylpentane Isomerization (I)
Table 9.36 n-Heptane Isomerization (I)
Table 9.37 Other Isomerization (I) Reactions
Table 9.38 n-Pentane Dehydrocyclization
Table 9.39 n-Hexane Dehydrocyclization
Table 9.40 Other Dehydrocyclization Reactions
Table 9.41 Hydro- and Dehydroisomerization (DI) Reactions
Table 9.42 n-Hexane Conversion
Table 9.43 Cyclopropane Hydrogenation
Table 9.44 Propene Hydrogenation
Table 9.45 Neohexane Hydrogenolysis
Table 9.46 Neopentane Conversion
Table 9.47 Toolbox for Studying 2D or 3D Nanoparticles
Table 9.48 Several Structure-Sensitive and Structure-Insensitive Catalytic
Reactions
Table 9.49 Chemical Processes that are the Largest Users of Heterogeneous
Catalysts at Present and the Catalysts that are Utilized Most Frequently
Table 9.50 Chemical Processes that are the Largest Users of Homogeneous
Catalysts at Present and the Catalysts that are Utilized Most Frequently
Table 9.51 Kinetic Data for C2H4 Hydrogenation Over Various Pt Surfaces
Table 9.52 Bandgaps of Several Oxides

9.1 INTRODUCTION
In a surface catalytic process, the catalytic reaction occurs repeatedly by a
sequence of elementary steps, which include adsorption, surface diffusion,
chemical rearrangements (bond breaking, bond-forming, molecular
rearrangement) of the adsorbed reaction intermediates, and finally desorption of
the products.
Catalytic reactions play all important roles in our life. Most biological
reactions that build the human body, as well as the reactions that control the
functioning of the brain and other vital organs, are catalytic. Photosynthesis and
the majority of chemical processes that are utilized in chemical technology are
also catalytic reactions. These range from oil refining and the production of
chemicals by hydrogenation, dehydrogenation, partial oxidation, and organic
molecular rearrangements (isomerization, cyclization), to ammonia (NH3)
synthesis and fermentation.

9.1.1 A Brief History of Surface Catalysis

In 1814, Kirchhoff reported that acids aid the hydrolysis of starch to glucose.
The oxidation of hydrogen by air over Pt was observed by H. Davy (1817) and
E. Davy (1820), as well as by Dobereiner (1823), who constructed a “tinderbox”
to produce flame when a small dose of hydrogen generated by the reaction of Zn
and HCl reacts with air in the presence of Pt. His device sold handily in the early
part of the 19th century when matches were not yet available. Platinum was also
found to aid the oxidation of CO and ethanol (CH3OH) (Dobereiner).
Faraday was the first to carry out experiments to explore why Pt facilitates the
oxidation reactions of different molecules. He found that ethylene (C2H4)
adsorption deactivates the Pt surface temporarily while the adsorption of S
deactivates Pt permanently. He measured the rate of hydrogen oxidation,
suggested a mechanism, and observed its deactivation and regeneration. Thus,
Faraday was the first scientist to study catalytic reactions. In 1836, Berzelius
defined the phenomenon and called it catalysis. He suggested the existence of a
“catalytic force” associated with the action of catalysts [1, 2].
Catalyst-based technologies were introduced in the second half of the 19th
century. The Deacon process ( ) was discovered in
1860, and the catalyzed oxidation of SO2 to SO3 by Pt was discovered in 1875
by Messel. Mond introduced the nickel-catalyzed reaction of methane (CH4)
with steam ( )In the early 20th century, Ostwald
developed the process of ammonia oxidation ( )to
form nitric oxide (NO), the precursor to nitric acid manufacture (1902); and in
1902, Sebatier developed a process for the hydrogenation of ethylene (
). In 1905, Ipatieff used the catalytic action of clays to carry
out different organic reactions: dehydrogenation, isomerization, hydrogenation,
and polymerization.
A better understanding of thermodynamics established the limits of chemical
conversions in catalyzed reactions. A catalyst can bring a reaction closer to
equilibrium, but it cannot produce molecules in excess of equilibrium
concentrations. The NH3 synthesis from N2 and H2 became the testing ground
for both catalysis science and technology. The quality of the catalyst could be
tested based on how closely chemical equilibrium could be attained. High-
pressure reactors were designed to shift the chemical equilibrium during
catalyzed NH3 production.
Catalyzed reactions of CO and H2 were utilized to produce methanol (CH3OH)
( ) in 1923 and higher-molecular-weight liquid
hydrocarbons by 1930. The production of motor fuels became one of the chief
aims of catalysis during the period of 1930–1950. The cracking of long-chain
hydrocarbons to produce lower-molecular-weight products was achieved over
oxide catalysts composed mostly of alumina and silica. Acid-catalyzed
alkylation reactions provided high-octane fuel and important organic molecules.
Meanwhile, catalysis science was developed (1915–1940) through the efforts
of Langmuir (sticking probability, adsorption isotherm, dissociative adsorption,
role of mono-layers), Emmett (surface area measurements, kinetics of NH3
synthesis), Taylor (active sites, activated adsorption), Bonhoeffer, Rideal,
Roberts, Polanyi, Farkas (kinetics and molecular mechanisms of C2H4
hydrogenation, ortho–para hydrogen conversion, isotope exchange, intermediate
compound theories), and many others.
The discovery of abundant and inexpensive oil in Arabia in the early 1950s
focused the development of catalytic processes for converting petroleum crude
to fuels and chemicals. Oil and oil-derived intermediates (ethylene, propylene)
became the dominant feedstocks.
Platinum (metal)- and acid (oxide)-catalyzed processes were developed to
convert petroleum to high-octane fuels. Hydrodesulfurization catalysis removes
S from the crude to prevent catalyst deactivation. The discovery of microporous
crystalline alumina silicates (zeolites) provides more selective and active
catalysts for many reactions, including cracking, hydrocracking, alkylation,
isomerization, and oligomerization. Catalysts that polymerize ethylene,
propylene, and other molecules were discovered. A new generation of more
active bimetallic catalysts that are dispersed on high-surface-area (100–400 m2g
−2) oxides were synthesized. The new catalysts are also more resistant to

deactivation.
The energy crisis in the early 1970s renewed interest in chemicals and fuels,
producing technologies using feedstocks other than crude oil. Intensive research
was carried out utilizing coal, shale, and natural gas to develop new technologies
and to improve on the activity and selectivity of older catalyst-based processes.
Increasing concern about environmental quality led to the development of the
catalytic converter for automobiles and to other, NO-reducing catalysts.
Modern surface science developed during the same period and has been
applied intensively to explore the working of catalysts on the molecular level, to
characterize the active surface, and to aid the development of new catalysts for
new chemical reactions. Indeed, surface science provides the means to explore
the molecular structure and mechanisms of elementary reaction steps and to
provide for rational design for modification of catalyst activity and selectivity.
This is carried out usually by altering the structure of the surface and by using
coadsorbed additives as bonding modifiers for reaction intermediates on the
surface.
This chapter describes the important macroscopic and molecular concepts of
surface catalysis that emerged from studies of recent decades. We review what is
known about a few important catalytic reactions that provide case histories of the
state of modern surface science of catalysis and of catalytic science.
During the later part of the 20th century, surface-science research of metal
catalysis were mainly focused on increasing catalyst activity for one-product
reactions (e.g., NH3 synthesis, CO oxidation, and C2H4 hydrogenation). With
increasing concern for the need of clean air and water, minimizing chemical
waste by reaction selectivity becomes the frontier of surface-science research in
the 21st century. To form only one desired product out of several
thermodynamically possible waste byproducts is the motivation of “green
chemistry”, which attempts to achieve 100% molecular selectivity as fuels and
chemicals are produced from methane and carbon solids. The challenge of
catalysis science is how to build the catalyst architecture to steer the reaction
intermediates along the desired reaction path while inhibiting the formation of
other waste products.
The need to produce fuels and chemicals from renewable energy sources (e.g.,
biomass) and to harness sunlight to dissociate H2O and CO2 provides an
unprecedented challenge and opportunity for catalysis science. Exploring how
fuels and chemicals are produced in this way promises to lead to new science
and technologies through the intellectual ferment that always accompanies the
use of new feedstocks, such as when coal was introduced in the 19th century and
oil in the 20th century.
 9.2 CATALYTIC ACTION
One of the major functions of a catalyst is to aid in rapidly achieving chemical
equilibrium for certain chemical reactions. Two of the simpler, although
important, reactions that demonstrate this type of catalytic action are the
formation of water from oxygen and hydrogen ( ) and the
formation of ammonia from hydrogen and nitrogen (N2 + 3H2 → 2NH3). Water
has a standard free energy of formation ΔG0 = –58kcal mol−1 Yet O2 and H2 gas
mixtures may be stored indefinitely in a glass bulb without showing signs of any
chemical reaction. Just by dropping a high-surface-area Pt gauze into the
mixture, the reaction occurs instantaneously and explosively—as demonstrated
to the delight of freshman chemistry students in introductory chemistry courses.
The reason for this striking effect can be explained as follows: To form H2O,
one of the diatomic molecules, H2 or O2 must be dissociated first. In the gas
phase, the dissociation energies are 103 kcal mol−1 for H2 and 117 kcal mol−1 for
O2 [3], and are much larger than the thermal energy, RT ( ~0.6 kcal mol−1 at 300
K). Even after the dissociation of one of the diatomic molecules, the subsequent
atom–molecule reactions (H + O2 or H2 + O) still require an activation energy of
~ 10 kcal mol−1 [4]. Thus the gas-phase reaction is very improbable under any
circumstances. On a properly structured Pt surface, however, both molecules
dissociate to atoms with near-zero activation energies (H2 + 2Pt → 2H—Pt and
O2 + 2Pt → 2O—Pt), as shown by low-pressure surface studies [5, 6]. In
addition, the atom-atom or atom-molecule reactions that subsequently take place
on the surface have very low or no activation energies in contrast to the gas
phase [5]. Thus the surface catalytic action involves the ability to atomize the
diatomic molecules with large bonding energies by forming chemisorbed atomic
intermediates on the surface and to lower the activation energy for the
subsequent reactions on the surface.
Similarly, the synthesis of ammonia from dinitrogen and hydrogen (N2 + 3H2
→ 2NH3) requires the “activation” of the N≡N triple bond to dissociate the
molecule. The N atoms then must react with H atoms or molecules to produce
NH3. The very large dissociation energy of N2 (280 kcal mol−1) makes it
virtually impossible for this reaction to occur in the gas phase. On an Fe surface,
however, N2 dissociates on a properly structured surface [e.g., the (111) crystal
face] with a small activation energy (3 kcalmol−1), which is the key initiation
step for the catalytic reaction. Iron also readily atomizes the H molecules. The
chemisorbed N atoms then react with H atoms on the surface to produce NH,
NH2, and finally NH3 molecules that desorb into the gas phase.

9.2.1 Kinetic Expressions

Catalysis is a kinetic phenomenon; we would like to carry out the same reaction
with an optimum rate over and over again using the same catalyst surface. Let us
define the catalytic reaction turnover frequency, f, as the number of product
molecules formed per second. Its inverse, 1/f, yields the turnover time, the time
necessary to form a product molecule. By dividing the turnover frequency by the
number of active surface sites available on the catalyst surface, N, we obtain the
specific turnover rate, R (molecules site−1 s−1) =f/N (in the literature, R often is
called the turnover frequency). In experiments, this intrinsic turnover rate is hard
to obtain because the number of active sites for a given reaction is usually not
measureable [7, 8]. For single-crystal surfaces, the number of active surface sites
is replaced by the total number of surface atoms, N = nA, where n is the surface
atomic concentration of a given crystal surface, and A is the surface area of the
catalyst surface. This type of analysis assumes that every surface site is active.
Although the number of catalytically active sites could be much smaller (usually
uncertain) than the total number of available surface sites, the specific rate
defined this way gives a conservative lower limit of the catalytic turnover rate.
For industrial catalysts composed of small metal clusters embedded in porous
support materials, the number of active sites usually is estimated by the number
of available sites on the catalyst for the chemisorption of a simple molecule (e.g.,
H or CO).
If we multiply R by the total reaction time, δt, we obtain the turnover number,
the number of product molecules formed per surface site in a given time, δt. A
turnover number of one corresponds to one stoichiometric reaction. Because of
the experimental uncertainties, the turnover number must be on the order of 102
or larger for the reaction to be considered as catalytic.
While the turnover rate provides a figure of merit for the activity of the
catalyst sites, the reaction probability (RP) reveals the overall efficiency of the
catalytic process under reaction conditions. The reaction probability is defined
as:

(9.1)
Reaction probability can be readily obtained by dividing by the rate of molecular
incidence F, which is obtained from the kinetic theory expression, F =
P/(2πMRT)1/2 with P is the gas pressure of the reactant and M is the mass of the
reactant molecule.
The specific catalytic reaction rate R can often be expressed as the product of
the rate constant k and a reactant pressure (or concentration)-dependent term

(9.2)
where Pi is the partial pressure of the reactants. The rate constant for the overall
catalytic reaction may be a function of the rate constants of many of the
elementary reaction steps that precede the rate-determining step. Because the
slowest rate-reaction step may change as the reaction conditions vary
(temperature, pressure, relative surface concentrations of reactants, catalyst
structure), k also may change to reflect the changing reaction mechanism.
Nevertheless, an effective reaction rate k can be described by the Arrhenius
expression

(9.3)
where A is the temperature-independent pre-exponential factor and ΔE* is the
apparent activation energy measured under catalytic reaction conditions.
Ranges of turnover rates for hydrocarbon reactions are shown in Figure 9.1.
Turnover rates between 10−4 and 100 are used in the various technologies. Thus
the temperature employed is adjusted to obtain the desired rates. The more
complex isomerization, cyclization, dehydrocyclization, and hydrogenolysis
reactions have activation energies ΔE* in the range of 35–45 kcal mol−1; and
thus according to the Arrhenius expression, Eq. 9.3, high temperatures are
required to carry them out at the desired rates. Hydrogenation reactions have
activation energies of 6–12 kcal mol−1 and therefore may be performed at high
rates of 300 K or below. Thus, there are at least two classes of reactions
distinguishable by their very different activation energies that may be carried out
at high and at low temperature, respectively.
Figure 9.1. Block diagram of hydrocarbon conversion over a Pt catalysts
showing the approximate range of reaction rates and temperature ranges that are
most commonly studied.
 Combustion reactions are highly exothermic, and can be carried out at a higher
turnover rate (102–103) so that the mass transport of the reactants becomes the
rate-limiting step. Basically, the diffusion of the reactants in the gas phase to the
surface is much slower than the surface reaction process. Under these high-
temperature reaction conditions, the reactions between the gas-phase free
radicals may compete with the catalytic combustion on the surface. In order to
study the intrinsic properties of surface reactions, high turnover rates should be
avoided by controlling the catalytic reaction conditions, so that the effects of
mass transport in the gas phase can be neglected.
Figure 9.1 shows that the reaction probabilities can be very low for many
hydrocarbon conversion processes. The main reason for these low reaction
probabilities is the formation of various hydrocarbon intermediates on the
surface under high-pressure reaction conditions. For example, the dissociation of
CH4 on the Pt(1 11) surface may produce surface intermediates (e.g., —CH, —
CH3, —CCH, and —CCH3) at various temperatures (300–500 K) and high
pressures (1–10 Torr) as observed by sum frequency generation spectroscopy
(SFG) vibrational spectroscopy. These surface intermediates may block the
active sites for the dissociation of CH4 and cause the torturously low probability
(~10−8) of the dissociation process [9].
The rates of surface-catalyzed reactions are usually measured by monitoring
the concentrations of reactants and products as a function of time under steady-
state conditions. Such studies tell us relatively little about the elementary surface
reaction steps. Dynamic methods that alter the flow of reactants or introduce
pulses of isotopically labeled reacting species have been useful to distinguish
between reacting intermediates and adsorbed spectator species on surfaces.
These investigations are carried out by following changes of the concentrations
of adsorbates beginning when changes in flow rate commence, as a function of
time, and by monitoring the time-dependent changes in the concentrations of
isotopically labeled product molecules. Spectroscopic techniques that monitor
reaction intermediates on surfaces can further provide molecular details of
reactive species during catalytic reactions under operating conditions.

9.2.2 Selective Catalysis

A good catalyst is selective and permits the formation of only one type of
product when reactions may occur along several reaction paths. Both CO and H2
react to produce CH4 exclusively when Ni is used as a catalyst, whereas only
CH3OH is formed when the catalyst is Cu and ZnO. The catalysts Co and Fe
produce high-molecular-weight hydrocarbons from the same reactants. The
reaction of n-hexane in the presence of excess H over Pt catalysts can produce
benzene (C6H6), cyclic molecules, branched isomers, or shorter-chain species
(see Fig. 9.2). A selective catalyst will produce only one of these products.
In more general terms, catalyzed reactions involve either (a) successive kinetic
steps leading to the final product or (b) alternative, simultaneous reaction paths
yielding two or more products. The former reaction scheme may be represented
by

(9.4)
where Ri (i = 1, 2) are the turnover rates of two reaction steps, respectively. A
good example of this type of reaction is the stepwise dehydrogenation of
cyclohexane to cyclohexene and then to benzene.
Figure 9.2. Various organic molecules that can all be produced by the catalyzed
reactions of n-hexane [10].
 When two or more parallel reaction paths are operative, as is the case during n-
hexane conversion, the reaction scheme is

(9.5)
For a reaction with n pathways, we can define the fractional catalytic selectivity,
Si, as the fraction of reacting molecules that are converted along the ith pathway

(9.6)
An additional possibility is provided by competitive parallel reactions
 (2.7)
Here, the ratio of rates, R1/R2, defines the kinetic selectivity. The activity (rate)
and the selectivity are the key parameters of any catalytic reaction.

9.2.3 Tabulated Kinetic Parameters for Catalytic

Reactions on Metal Surfaces
A great deal of kinetic information has been obtained for different types of
catalyzed hydrocarbon reactions carried out over metal catalyst surfaces. These
reactions include dehydrogenation, hydrogenation, hydrogenolysis and cracking,
ring opening, dehydro-cyclization, and isomerization. The kinetic parameters for
these reactions are listed in Tables 9.1–9.46 (see page 655 to 750). In these
tables, the catalyst systems that were used are listed together with the
temperature range for the investigation. Because these reactions are always
carried out in the presence of hydrogen, both the hydrocarbon and hydrogen
partial pressures in torr are tabulated in these tables. The reaction orders in both
reactants are also displayed whenever they have been determined. From these
data we can determine the changes of the reaction rates with reactant
concentrations. The turnover rate of reaction (in molecules site−1 s−1) at a given
temperature, in the range used in the experimental study, is also calculated and
listed, together with the apparent activation energy for the reaction, ΔE* (in kcal
mol−1), and the logarithm of the pre-exponential factor, ln(A). From the rate and
reactant concentrations, an RP can be calculated and is also displayed in the
Tables 9.1–9.46 for the various catalytic reactions, as –ln(RP). Fractional
selectivities, S, are also supplied when reported. These are defined as the ratio of
the rate of the specific reaction to the total reaction rate.
There is a great deal of variation in the kinetic parameters obtained for a given
reaction on different catalyst systems. This is expected, since the structure and
bonding characteristics of the different metal catalysts vary widely. Nevertheless,
several conclusions may be reached from the inspection of the data. The reaction
probabilities are very low under the conditions where these reactions were
carried out. They range from 10−8 to 10−5 for hydrogenation to 10−12–10−8 for
most of the other reactions. The apparent activation energies are the lowest for
hydrogenation and cyclopropane ring opening, 9–15kcalmol−1. For
dehydrogenation of cyclohexane and for the hydrogenolysis of C4 to C6 alkanes,
ΔE* is in the range of 16–25 kcalmol−1. For most of the other reactions, which
include (1) hydrogenolysis (the most frequently studied reaction) of ethane,
propane, and other alkanes; (b) cracking of olefins and benzene; (c)
dehydrogenation of alkanes; and (d) isomerization of C5 to C6 hydrocarbons, the
apparent activation energies are in the range of 25–50 kcal mol−1.
The kinetic information displayed in Tables 9.1–9.46 (see page 655 to 750) can
be useful in establishing optimum reaction conditions and catalyst systems. It is
hoped that reliable kinetic parameters will become available for many other
important catalyzed hydrocarbon reactions in the near future.

9.3 CATALYST PREPARATION,

DEACTIVATION, AND REGENERATION
9.3.1 Catalyst Preparation
The higher the active surface area of the catalyst, the greater will be the number
of product molecules produced per unit time. Therefore, much of the art and
science of catalyst preparation deals with high-surface-area materials. Usually,
materials with 100 to 400-m2 g−1 surface area are prepared from alumina, silica,
or carbon; and more recently other oxides (Mg, Zr, Ti, and V oxides),
phosphates, sulfides, or carbonates also have been used. These are prepared in
such a way that they are often crystalline with well-defined microstructures and
behave as active components of the catalyst system in spite of their accepted
name “supports”. Transition metal ions or atoms are then deposited in the
micropores, which are then heated and reduced to produce small metal particles
10–102 Å in size with a large portion of the atoms located on the surface. The
surface structure of the metal particles can often be controlled by this method of
preparation. Usually, more than one metal component is used, with bimetallic
systems being the most popular in recent years [11, 14]. Frequently, another
oxide (e.g., TiO2) is dispersed on the high-surface-area oxide (alumina) to impart
unique catalytic properties as well. Additives that are usually electron donors
(alkali metals) or electron acceptors (halogens) are adsorbed on the metal or on
the oxide to act as bonding modifiers for these coadsorbed reactants [15]. These
complex and intricately fabricated catalyst systems can be used for hundreds or
thousands of hours and often millions of turnovers to produce the desired
molecules at high rates and selectivity before their deactivation.
9.3.1.1 Industrial Catalysts
The most common method of catalyst preparation in industry is the pore–volume
impregnation, often called “incipient wetness impregnation”. The impregnating
process is driven by capillary forces. A solution of the catalyst precursor (e.g.,
the platinum nitrate solution) is added to the porous support under continuous
stirring, until the incipient wetness point is reached. At this point, all the pores
are filled with the impregnating solution. Then, the catalyst is dried to remove
the solvent, and is calcined, reduced, or sulfurized, depending on the application.
The active component is the metal particles inside the pores of the support. The
size of the metal particles can be roughly controlled by the amount of metal
loaded into the catalyst. Figure 9.3 shows a typical TEM image and particle size
distribution of a 3 wt% Pt/Al2O3 catalyst prepared by incipient wetness.

Figure 9.3. The TEM image (a) and typical particle size distribution (b) for a 3
wt% Pt/ Al2O3 catalyst prepared by incipient wetness from tetrammine Pt(II)–
nitrate (calcination temperature 260°C).

Currently, the improvement of industrial catalysts is mostly through an

empirical trial-and-error process. In order to achieve the rational design of the
catalyst, surface-science techniques must be applied to obtain molecular-level
information on structure, composition, and oxidation state of the catalyst surface.
However, most industrial catalysts consist of nanoparticles hidden inside the
pores of a support, which inhibits the application of surface-sensitive techniques
such as scanning tunneling microscopy (STM), SFG, and AFM. Moreover,
oxidic supports are electrically insulating, which leads to serious loss of intensity
and resolution in many techniques such as X-ray photoelectron spectroscopy
(XPS), secondary ion mass spectrometry (SIMS), and Auger electron
spectroscopy (AES). In surface science research, much effort has been devoted
to develop model catalyst systems that are suitable for application of the various
surface-science techniques, and capture the essential properties of industrial
catalysts [16, 17].

9.3.1.2 Model Catalysts for Surface-Science Studies.

The evolution of model catalyst systems has experienced three phases (Fig. 9.4).
The first phase is the single-crystal surface study. The catalytic reactions are
studied on the surfaces of different transition metals and different crystal faces to
help us understand the catalytic activity across the periodic table, and the effect
of surface structure on the activity and selectivity for a given reaction. The
nanoparticles used in industrial catalysts can be viewed as a small crystallite
composed of well-defined atomic planes. By using a single-crystal surface, the
activities of different crystal faces for a given reaction can be examined in details
(Fig. 9.5).
A large amount of work has been done to make sure that the measured reaction
rates using the single-crystal model catalysts are compatible with reaction rates
obtained on large-surface-area catalysts. Table 9.43 (see page 747) shows the
turnover numbers and the activation energies obtained for the ring opening of
cyclopropane to form propane on small-area single-crystal Pt and on dispersed
Pt catalysts under identical experimental conditions. The agreement is indeed
excellent. This is a structure-insensitive reaction at high pressures that lends
itself well to such correlative studies. For structure-sensitive reactions, marked
differences are found on different crystal faces with the stepped and kinked
surfaces being much more active in general. Similarly, excellent agreements
among rates, activation energies, and the product distribution were obtained for
the hydrogenation of CO over polycrystalline Rh foils and dispersed, silica-
supported Rh catalyst particles [18, 19]. Figure 9.6a and b shows the agreement
reached between studies of the same reactions (cyclohexene hydrogenation over
Pt and CO hydrogenation over Ni) over low-surface-area model single-crystal
and high-surface-area dispersed catalysts.
For single-crystal surface studies, a variety of surface-sensitive techniques are
available to characterize the catalyst surfaces and even to monitor surface
morphology, the adsorbate coverage, and the oxidation state during the reaction
process by in situ techniques, for example, STM, SFG, and XPS. Additive
effects can also be studied by depositing a small amount of impurity on to the
crystal surface.
Figure 9.4. Schematics for three phases of model catalyst

Figure 9.5. Catalyst particle viewed as a crystallite, composed of well-defined

atomic planes, steps, and kink sites.

Figure 9.6. (a) Arrhenius plot of the rate of cyclohexene hydrogenation to

cyclohexane on Pt(111) crystal surfaces and on Pt particles dispersed on silica.
Both the rates and activation energies are similar [20]. (b) The Arrhenius plot of
the rate of CH4 production from H2 and CO over Ni(1 11) and Ni(100) compared
to the production over a supported Ni/Al2O3 dispersed catalyst. Both the rates
and the activation energies are the same [21].
 The second phase is a 2D catalyst system with nanoparticles deposited on a
flat support surface. This system mimics the nanoparticles properties in
industrial catalysts. With the advances in colloid chemistry, controlled
nanoparticle synthesis, transition metal nanoparticles with monodispersity, and
well-controlled shape can be routinely prepared in the solution [14, 22–24]. By
controlling the precursor concentration, the size of nanoparticles can be tuned
from 1 to 10 nm. These nanoparticles can be deposited as a 2D film onto a flat
support surface by using the Langmuir–Blodgett (LB) technique. Then the size
and shape effects on catalytic activity and selectivity can be studied in detail, and
molecular-level information can be obtained by in situ techniques, since the
nanoparticles are deposited on an external surface. By changing properties of the
support, the support effects can also be studied in this model system.
For example, using hexachloroplatinic acid or rhodium acetylacetonate as
precursor monomers, monodispersed metal nanoparticles can be produced, each
one coated with a polymer cap that prevents aggregation in solution [25–27].
One can show that under well-defined conditions, the particle size is
proportional to the monomer concentration and can be controlled by changing
the monomer concentration. By kinetically controlling nanoparticle synthesis
processes, it is also possible to control the shapes of Pt or Rh nanoparticles [28–
31]. Figure 9.7 shows Pt nanoparticles with controlled shape and size. High-
resolution transmission electron microscopy (HRTEM) along with electron
diffraction analysis reveals the shape of the nanoparticles [28]. These
monodispersed nanoparticles with uniform size and shape can be deposited as a
2D film by using the LB technique.
Figure 9.7. Platinum monodispersed nanoparticles with (a) size controlled from
1–7 nm and (b) well-controlled cubic or cuboctahedral shapes. The scale bars in
(a) are 10 nm, and in (b) are 50 nm.

Figure 9.8. Synthesis of dendrimer capped monodispersed Pt nanoparticles of

cluster sizes of 1 nm or less.
 Currently, small clusters of dendrimer encapsulated metal nanoparticles are
being explored for catalysis [32]. Structure and chemical properties of
dendrimers, a quasispherical hyperbranched polymer, can be controlled by
changing the core structure, the number and type of the duplicating units, and the
terminal functional groups. Poly(amidoamine) (PAMAM) is the most used for
the synthesis of metal nanoparticles (Fig. 9.8). Platinum nanoparticles ranging
from 1 nm can be synthesized within the cavities of high-generation PAMAM
dendrimers. The size distribution of metal nanoparticles synthesized within
dendrimers is very narrow due to the well-defined composition and structure of
the dendrimer template.
In the third phase, the nanoparticles can be incorporated into 3D supports, such
as mesoporous silica (SBA-15) and mesocellulous silica foam (MCF-17), by
sonication or by synthesizing the mesoporous channels around the nanoparticles
in the same solution phase (Fig. 9.9) [33]. By loading the nanoparticles into the
high-surface-area support, the conversion of the catalytic reactions with a low
turnover rate can be significantly increased. This method of synthesis may
produce a 3D catalyst with the shape-controlled nanoparticles, which is
impossible for traditional industrial methods. The shape control of nanoparticles
is one of the main gradients for achieving selectivity of catalytic reactions, as we
will discuss in Section 9.7.

9.3.2 Catalyst Deactivation

Catalysts can live long active lives, but they do not last forever. The type of
supported metal catalysts that are used in petroleum refining produces in the
range of 200–800 barrels of products per pound of catalyst (1 barrel = 42 gal).
Once the catalyst is deactivated, it is either regenerated or replaced. There can be
many reasons for the deactivation. At the operating temperatures, some of the
reactant hydrocarbons may completely decompose and deposit a thick layer of
inactive carbon on the catalyst surface (coke). For many catalysts, the
deactivation is slow enough that they are used in steady-state operation. The
liquid or gaseous reactants are passed through the catalyst with a well-defined
“space velocity” that is normally measured as the weight-hourly-space-velocity
(WHSV); that is, the number of pounds of liquids or gas passed over the unit
weight of catalyst per hour. For other active catalysts, deactivation is so rapid
that they are used in a cyclic fashion; the reactors “swing” between running the
catalytic reactions and regenerating. Thus understanding the causes of
deactivation and developing new catalysts that are more resistant to “poisoning”
are constant concerns of the catalytic chemist.
Figure 9.9. The TEM image of a 3D Pt nanoparticle encapsulated in mesoporous
silica (SBA-15).

Many of the catalyst poisons act by blocking active surface sites. In addition,
poisons may change the atomic surface structure in a way that reduces the
catalytic activity. Sulfur, for example, is known to change the surface structure
of Ni [34]. By forming chemical bonds of different strengths on the different
crystal planes, it provides a thermodynamic driving force for the restructuring of
the metal particles. Sometimes the rate of deactivation of metal catalysts from
small concentrations of S can be very dramatic. The automobile catalytic
converter necessitated the removal of tetraethyllead from gasoline, one of the
best anti-knocking agents, because it readily poisoned the Pt–Pd catalyst by
depositing lead sulfate on the noble metal surfaces. One of the major causes of
deactivation in crude oil cracking catalysts is the deposition on the catalyst
surface of metallic impurities that are present as compounds in the reactant
mixture. Vanadium- and titanium-containing organometallic compounds
decompose and not only deactivate the catalyst surface, but often plug the pores
of the high-surface-area supports, thereby impeding the reactant–catalyst contact
during petroleum refining.
A freshly prepared catalyst may not exhibit optimum catalytic activity upon its
first introduction into the reactant stream. There may be efficient, but
undesirable, side reactions that need to be eliminated. For this purpose, a small
amount of “poison” is often added to the reaction mixture or introduced in the
form of pretreatment. Thus deactivating impurities may also be used, in small
quantities, to improve the selectivity of the working catalyst.

9.3.3 Catalyst Regeneration

The regeneration treatment of the catalyst depends on the causes of deactivation.
Most frequently, carbon deposition is the primary source of deactivation during
hydrocarbon conversion reactions. In this circumstance, heating the deactivated
catalyst in air or in oxygen can burn off the carbon. The heat generated in this
exothermic combustion reaction can be used beneficially in the overall catalytic
process Sintering of catalyst particles due to exposure to high temperatures for
extended periods leads to loss of surface area. Oxygen can often oxidize the
metal component of the catalyst to alter the shape and size of the metal particles.
Metal oxides have lower surface energy than metals, and therefore oxidation
could lead to better “wetting” of the high-surface-area oxide support. Subsequent
reduction of the metal oxides in hydrogen may lead to redispersion of the metal
constituent as small particles with increased total surface area. Additives, such as
chlorine that may form volatile metal halides, can also help the redispersion of
some of the catalyst components.
At high enough temperatures, the micropores of the high-surface-area catalyst
may collapse by sintering or melting. It is therefore essential that the materials
chemistry be understood and that compounds with the proper surface and bulk
thermodynamic properties be chosen to maintain their thermal stability under
diverse (oxidizing or reducing) reaction conditions.
The removal of impurities that deposit from the reactant mixture poses
particular challenges. Sulfur, arsenic, phosphorous, and vanadium are often
deposited during oil refining. The reader is referred to publications that deal with
these special problems of catalyst deactivation and regeneration [35, 36].

9.4 TECHNIQUES TO CHARACTERIZE

CATALYST SURFACE AND STUDY THE
REACTIVITY OF CATALYSTS
9.4.1 Turnover Rate Measurements
In order to measure the turnover rate of gas-phase catalytic reactions, a reactant
mixture with given partial pressures is introduced to a catalyst surface at a given
temperature controlled by a heater (Fig. 9.10). The products are formed on the
catalyst surface and released in the gas mixture. The concentration of the
products in the reaction mixture then can be monitored by a gas chromatograph
(GC) to obtain the turnover rate of the reaction under controlled reaction
conditions. For 2D model catalysts (e.g., a single-crystal surface), and LB nano-
particle films, the amount of products produced per second is usually quite small
because of the relative small number of metal surface sites available on the 2D
model catalysts (e.g., the number of metal sites on a 1-cm2 Pt surface is on the
order of 1014–1015). In order to increase product concentration in the reaction
mixture so that a reliable product concentration can be measured by GC, the
reaction mixture has to be circulated in a closed gas loop by a recirculation
pump. This type of measurement is performed by an experimental setup called a
batch reactor (see Fig. 9.10a). The increase of production concentration is
measured at a sequence of time points. The slope of the concentration curve
gives information about the turnover frequency of the reaction.
Figure 9.10. (a) Experimental setup of catalytic batch reactor for measuring the
turnover rate of small area 2D catalysts, and (b) flow reactor for 3D high-
surface-area catalyst systems.
 For 3D catalysts, the number of available metal surface sites is usually on the
order of 1016–1018, which is two to four orders of magnitude greater than for 2D
catalysts, so the product concentration in the reaction mixture after a single pass
through the catalyst is large enough to be detected by GC. In this case, the flow
reactor (see Fig. 9.10b) can be used for catalytic turnover studies. In this type of
measurement, a reactant mixture with given partial pressures flows through the
catalyst at a constant flow rate. The product concentration after the reaction is
measured by GC. This concentration times the flow rate gives the turnover
frequency, f, of the catalyst.

9.4.2 High-Pressure Reactors

In the surface-science approach to catalytic reaction studies, it is imperative that
we determine the surface composition and structure in the same chamber where
the reactions are carried out, without exposing the crystal surface to the ambient
atmosphere. This necessitates the combined use of an ultrahigh-vacuum (UHV)
enclosure, where the surface characterization is to be carried out, and a high-
pressure isolation cell, where the catalytic studies are performed. Such an
apparatus is shown in Figure 9.11. The small-surface-area (~lcm) catalyst is
placed in the middle of the chamber (see the open state in Fig. 9.11b), which can
be evacuated to 10−9Torr. The surface is characterized by low-energy electron
diffraction (LEED), AES, and other desired surface diagnostic techniques. Then,
the lower part of the high-pressure isolation cell is lifted to enclose the sample in
a 0.5-L volume that is sealed by a Cu gasket (see the closed state in Fig. 9.11b).
The isolation chamber can be pressurized to 100 atm if desired and is connected
to a GC that detects the product distribution as a function of time and surface
temperature. The sample may be heated resistively, both at high pressure or in
UHV. After the reaction study, the isolation chamber is evacuated and opened,
and the catalytic surface is again analyzed by the various surface-diagnostic
techniques. Ion-bombardment cleaning of the surface or means to introduce
controlled amounts of surface additives by vaporization are also available. The
reaction at high pressures may be studied in the batch or the flow mode.
There are many different designs available for combined high-pressure
reaction studies and UHV surface-science investigations. Transfer rods that
move the sample from the environmental cells to the UHV chamber and reaction
cells that permit liquid- or gas-phase reaction studies have been described in the
literature [37]
Figure 9.11. (a) Schematic representation of one type of apparatus capable of
carrying out catalytic-reaction-rate studies on single-crystal surfaces of low
surface area at high pressures (atm) and also to perform surface characterization
in UHV. (b) Pictures of the open and closed high-pressure cell.
 9.4.3 In Situ Characterization Techniques: High-
Pressure Scanning Tunneling Microscope, Sum
Frequency Generation Spectroscopy, and Ambient
Pressure X-Ray Photoelectron Spectroscopy
A high-pressure STM (Fig. 9.12) can monitor the change of surface morphology
during catalytic reactions, since surface areas covered by different adsorbed
species exhibit different surface structures [38–41]. By monitoring the reaction
turnover rate at the same time, we may establish the correlation between the
turnover rate with the surface structure, and identify the active phase of catalyst
surfaces. Two in situ STM studies of CO oxidation on the Pt(110) surface and
C2H4 hydrogenation on the Pt(111) surface will be discussed later in Section 9.6.
The major limitations of this technique for in situ studies are the lack of time
resolution and chemical resolution. Various ways of circumventing these issues
have been attempted. Currently, radio frequency scanning tunneling microscopy
(RF–STM) is apromising technique to improve the time resolution. This
technique has a measurement time at each scan point of ~0.1 μs, which is 100
times faster than conventional STM [42].
Figure 9.12. The schematic of high-pressure STM.

Figure 9.13. (a) Schematic of a high-pressure SFG system, a vibrational

spectroscopic tool for probing the adsorbed species during the catalytic reaction.
(b) The frequency ωvis of the visible (vis) laser beam is kept fixed, and the
infrared (IR)-beam frequency ωIR is varied. When ωIR coincides with a
vibrational transition from |0 ; to |1 of an adsorbed molecule, the molecule is
excited to a virtual state |n and emits the sum frequency vSFG. (c) Because of
selection rules, the SFG signal is forbidden from a centrosymmetric medium.
(See color insert.)
 Sum frequency generation spectroscopy is a surface specific photon-in and
photon-out technique that is a premier tool for monitoring the reaction
intermediates on catalyst surfaces during reactions at high pressures and
temperatures (Fig. 9.13). Compared to the conventional linear optical techniques
(e.g., IR and Raman spectroscopy), SFG is highly surface specific due to the
selection rule for the second-order nonlinear optical process [43, 44]. It is
extremely valuable in the cases (e.g., the C2H4 hydrogenation), where the
reaction intermediates are too mobile on the surface for STM to produce the
surface images at atomic resolution. In the SFG spectrum, the adsorbed chemical
species are identified by their molecular vibrational frequencies. The SFG
intensity tells the coverage and ordering of reaction intermediates over a surface
area of ~1 cm2. Currently, the time resolution for conventional SFG is around
tens of seconds [45]. The picosecond time resolution may be achieved by the
pump-and-probe technique [46, 47].
Figure 9.14.Schematic of APXPS. An electron lens system with a differential
pumping stage is shown in the bottom panel.
 Ambient pressure X-ray photoelectron spectroscopy (APXPS) is an in situ
technique for monitoring the chemical composition and the oxidation state of
catalyst surfaces [48, 49]. In order to achieve surface sensitivity, the kinetic
energy of photoelectrons can be controlled by using the synchrotron light source
so that the mean free path of photoelectrons in the metal bulk is 1 nm. APXPS
utilizes an electron lens system and differential pumping to extract electrons
from a high-pressure cell and transfer them into an electron analyzer. Currently,
the high-pressure cell usually works at up to a pressure of 10 Torr (Fig. 9.14).
Surface science has a variety of techniques for ex situ and in situ studies of
surface reactions (Table 9.47 on page 751) that provide complimentary
information about reaction processes. The combination of several of these
techniques is usually necessary to achieve the overall understanding of a given
catalytic reaction process.
 9.5 METAL CATALYSIS
Transition metals and their compounds (oxides, sulfides, and carbides) are
uniquely active as catalysts, and are used in most surface catalytic processes.
The effective-medium theory of the surface chemical bond (Section 6.2)
emphasizes the dominant contribution of d-electrons to bonding of atoms and
molecules at surfaces. Other theories also point out that d-electron metals in
which the d-band is mixed with the s and p electronic states provide a large
density of low-energy electronic states and electron vacancy states [50]. This
condition is ideal for catalysis because of the multiplicity of degenerate
electronic states that can readily donate or accept electrons to and from adsorbed
species. Those surface sitestronic states have the highest densities of states are
most active in breaking and forming chemical bonds. These electronic states
have high charge fluctuation probability (configurational and spin fluctuations),
especially when the density of electron vacancy or hole states is high.

9.5.1 Trends Across the Periodic Table

One prediction of these theoretical models is that the heat of chemisorption of
atoms should increase from right to left in the periodic table. This trend is well
documented in Section 6.2, and there is good agreement between experiment and
theory. Thus, one important function of transition metals in catalytic reactions is
to atomize diatomic molecules and then to supply the atoms to other reactants
and reaction intermediates. The diatomic molecules of importance, in order of
increasing dissociation energy, are H2, O2, N2, and CO. The high strength of
bonding of H, C, N, and O atoms on transition metal surfaces provides the
driving force for the atomization and for the release of atoms for reactions with
other molecules. If the surface bonds are too strong, the reaction intermediates
block the adsorption of new reactant molecules because of their long surface
residence times, and the reaction stops. For adsor-bate-surface bonds that are too
weak, the necessary bond-scission processes may be absent. Hence, the catalytic
reaction will not occur. A good catalyst is thought to be able to form chemical
bonds of intermediate strength. These bonds should be strong enough to induce
bond dissociation in the reactant molecules. However, the bond should not be too
strong, thereby ensuring only short residence times for the surface intermediates
and rapid desorption of the product molecules, so that the reaction can proceed
with a large turnover number.
These considerations are strikingly demonstrated by the volcano-shaped
pattern of a variation of catalytic activity, as shown schematically in Figure 9.15.
While the heat of adsorption is steadily decreasing from left to right, the catalytic
reaction rates peak at the Group VIII metals in the periodic table. Figure 9.15
shows the pattern of variation of catalytic reaction rates across the series of
transition metals Re, Os, Ir, Pt, and Au for the hydrogenolysis of the C—C bond
in C2H6, the C—N bond in CH5N, and the C—Cl bond in CH3Cl.

Figure 9.15. Catalytic activities of transition metals across the periodic table for
the hydrogenolysis of the C—C bond in ethane (C2H6), the C—N bond in
methylamine (CH5N), and the C—Cl bond in methyl chloride (CH3Cl) [10].

The influence of the electronic structure of surface atoms shows up not only in
producing the volcano-shaped trends of transition metal catalytic activity across
the periodic table, but also in producing the structure sensitivity of certain
catalytic reactions on a given transition metal. A catalytic reaction is defined as
structure sensitive if the turnover rate changes markedly as the surface structure
of the catalyst is changed. Reaction studies on single crystals revealed the
importance of steps of atomic height and of kinks in the steps in increasing
reaction rates for H2/D2 exchange, for dehydrogenation and hydrogenolysis.
Theoretical studies indicate large changes in the local density of electronic states
at the surface defect sites that correlate with changes in catalytic activity.

9.5.2 Some Frequently Used Concepts of Metal

Catalysis
During the operation of complex catalyst systems, several macroscopic
experimental parameters have been uncovered that provide useful practical
information about the nature of the catalyst or the catalyzed surface reaction. A
catalytic reaction is defined to be structure sensitive if the rate changes markedly
as the particle size of the catalyst changes [51]. Conversely, the reaction is
structure insensitive on a given catalyst if its rate is not influenced appreciably
by changing the dispersion of the particles under the usual experimental
conditions. In Table 9.48 (see page 751), we list several reactions that belong to
these two classes. Clearly, variations of particle size give rise to changes of
atomic surface structure, in which the relative concentrations of atoms in steps,
kinks, and terraces are altered.
One way to investigate the effect of surface structure on a given reaction is to
perform a reaction kinetic study on different crystal faces under the same
reaction condition. The CO oxidation by O2 was proved to be structure
insensitive by the observation of the virtually identical reaction rate on Rh(111)
and Rh(100) surfaces under the partial pressure ratio pco/po2 close to 2:1.
However, the CO oxidation by NO on Rh(1 11) and Rh(100) exhibits marked
different reaction rates, and is a structure-sensitive reaction [52].
Recent experimental evidence suggests that the structure sensitivity of a given
reaction may also depend on reaction conditions, such as the partial pressures of
reactants and the reaction temperature. For example, the CO oxidation by O2 on
Pt(110) may switch to a high reaction rate under the O2 rich condition [53, 54].
Moreover, the ignition temperatures for the CO oxidation on Pt(100) and Pt(1
11) are 500 and 620 K, respectively, which indicates the structure sensitivity of
this reaction at high temperatures [55].
Two types of reaction processes have found general acceptance during
development of mechanistic interpretations of catalytic reactions using the
macroscopic rate equations determined by experiment. In one, the rate-
determining surface reaction step involves interaction between two atoms or
molecules, both in the adsorbed state. This reaction process is called the
Langmuir–Hinshelwood mechanism [56, 57]. In the other (the Eley–Rideal
mechanism), the rate-determining reaction step involves a chemical reaction
between a molecule from the gas phase and one in the adsorbed state [58]. Most
reactions have rate equations that fit one of these two mechanisms. Some
reactions may change their reaction mechanisms under different reaction
conditions. The oxidation of CO has been identified by molecular-scale studies
as obeying the Langmuir–Hinshelwood reaction mechanism [59, 60]. However,
correlation of these reaction mechanisms (suggested by inspection of the
macroscopic rate equations) with molecular-level studies of the elementary
surface reactions remains one of the great challenges of catalysis [61].
An interesting phenomenon called the compensation effect, was found during
studies of a given catalyzed reaction over catalysts prepared in different ways
[62]. If we use the Arrhenius expression for the rate constant, both the pre-
exponential factor and the activation energy for the reaction have been found to
vary greatly from catalyst to catalyst. However, they vary in such a way as to
compensate each other, so that the rate constant (or the reaction rate under the
same conditions of pressure and temperature) remains almost constant. For
example, for the methanation reaction (i.e., the hydrogenation of CO), the
following empirical relationship has been found to hold between A and ΔE*:

(9.8)
where α is a constant and Θ is called the isokinetic temperature, at which the
rates on all the catalysts are equal. For the methanation reaction [63], α ~ 0 and
Θ = 436 K. Thus ln ~ l.lΔE*kcalmol−1. Figure 9.16 shows the compensation
effect for the methanation reaction for eight different metal catalysts. The ln
versus ΔE* plotsyield a straight-line relationship. Figure 9.17 shows the
compensation effect for the hydrogenolysis reactions whose rates are displayed
in Figure 9.15.
Figure 9.16. Compensation effect for the methanation reaction. The logarithm of
the pre-exponential factor is plotted againt the apparent activation energy, ΔE*,
for this reaction over several transition metal catalysts [63].
 Possible explanations for the compensation effect have been explored
extensively and are reviewed by Bond et al. [64]. One of the interpretations of
the compensation effect cited by Bond et al. is a model proposed by Larsson.
This model assumes there is a transfer of energy into the vibrational mode of the
reactant that “most effectively distorts the molecule toward the structure it has in
the “activated complex” of the reaction” [65, 66]. In another interpretation,
Bligaard et al. suggest the compensation effect arises from “a switching of
kinetic regimes”, meaning there is a monotonic relationship between “the
activation energy of the rate-limiting step and the stability of the reaction
intermediates on the surface” [67].
Figure 9.17. Compensation effect for the hydrogenolysis reactions. The
logarithm of the pre-exponential factor is plotted against the apparent activation
energy, ΔE*, for this reaction over several transition metal catalysts. The
squares, triangles, and circles represent values for C5H6, CH5N, and CH3Cl
hydrogenolysis, respectively [10].
 During most reactions, the surface of the active metal catalyst is covered with
a strongly chemisorbed overlayer that remained tenaciously bound to the surface
for 102–106 turnovers. During hydrocarbon reactions, this is a carbonaceous
overlayer with a composition of (H/C) ~ 1, during NH3 syntheses it is
chemisorbed N, and during hydrodesulfurization it is a mixture of S and C. It is
believed that this overlayer may play a role in restructuring the surface to create
new active sites and in altering the bonding of reactants, intermediates, and
products.
It is observed that, during hydrocarbon conversion reactions over Pt,>80% of
the metal surface is covered with the carbonaceous deposit [68]. Increasing
evidence shows that the uncovered metal sites as well as the carbonaceous layer
are active parts of the working catalyst. Only when this carbonaceous layer is
totally dehydrogenated will it deactivate the catalyst by forming a cross-linked
graphite coating. This notion has been used to explain the structure insensitivity
of some hydrocarbon conversion reactions (e.g., the hydrogenation of C2H4)
[69]. Basically, if the reactions take place on top of the strongly chemisorbed
overlayer, the structure of the underlining metal surface will not affect the
reaction rates directly. The metal surface only participated indirectly by aiding
the dissociation of molecular hydrogen.
Structure and bonding modifiers are often introduced as important additives
when formulating complex catalyst systems. Structural promoters can change the
surface structure that is often the key to catalyst selectivity. Aluminum oxide
facilitates the restructuring of Fe in the presence of nitrogen to produce surfaces
that are most active during NH3 synthesis. Alloy components may not participate
in the reaction chemistry, but modify structure and site distribution on the
catalyst surface. Site blocking can improve selectivity, which has been proven
for many working catalyst systems. Sulfur and silicon or other strongly adsorbed
atoms that seek out certain active sites can block undesirable side reactions.
Bonding modifiers are employed to weaken or strengthen the chemisorption
bonds of reactants and products. Strong electron donors (e.g., K) or electron
acceptors (e.g., Cl) that are coadsorbed on the catalyst surface are often used for
this purpose. Alloying may create new active sites (mixed-metal sites) that can
greatly modify activity and selectivity. New catalytically active sites can also be
created at the interface between the metal and the high-surface-area oxide
support. In this circumstance, the catalyst exhibits the so-called strong metal-
support interaction (SMSI). Titanium oxide frequently shows this effect when
used as a support for catalysis by transition metals. Often the sites created at the
oxide–metal interface are much more active than the sites on the transition
metal.

9.5.3 Most Frequently Used Catalyst Materials

Some 80% of chemical processes use catalysts whose sales are
~$10,000,000,000 per year, but which is less, 1 % of the revenue from the
products they create [70]! It may be instructive to review how widely catalysts
are applied in the various technologies and to identify some of the most
frequently used materials. There are three major areas of catalyst application at
present: automotive, fossil-fuel refining, and production of chemicals [71–73].
Table 9.49 (see page 752) lists the chemical processes that are the largest users
of heterogeneous catalysts and the catalyst systems that are employed most
frequently at present.
The automotive industry uses mostly noble metals (Pt, Rh, and Pd) for
catalytic control of car emissions: unburned hydrocarbons, CO, and NO. These
highly dispersed metals are supported on oxide surfaces, and the catalyst system
is specially prepared to be active at the high space velocities of the exhaust gases
and over a wide temperature range. In petroleum refining, zeolites are most
widely used for cracking of hydrocarbon in the presence of hydrogen. The
important hydrodesulfurization process uses mostly sul-fides of Mo and Co on
an alumina support. The “reforming” reactions to produce cyclic and aromatic
molecules and isomers from alkanes to improve the octane number are carried
out mostly over Pt or Pt-containing bimetallic catalysts (e.g., Pt—Re and Pt—
Sn). Sulfuric and hydrofluoric acids are the catalysts for alkylation. In chemical
technologies, steam reforming of natural gas (mostly CH4) to produce H2 and
CO is an important large-volume catalytic process. The purified natural gas is
reacted with steam to form CO and H2, mostly over a supported Ni catalyst. The
water–gas shift reaction (CO + H2O → CO2 + H2) is then employed to produce
more hydrogen. The most frequently used catalyst for this purpose is iron based.
Methanol (CH3OH) is produced from CO and H2, and NH3 is produced from H2
and N2. Copper and zinc oxide are also used for the shift reaction, as well as for
the production of CH3OH from CO and H2. Nickel is the catalyst for
methanation from CO and H2, and Fe is the major catalyst for the NH3 synthesis.
Catalytic hydrogenation primarily uses Ni and Pd as catalysts. Hydrogenation
of nitrile groups to amines and various edible and inedible oils for the
preparation of margarine, salad oils, and stearine are some of the major
applications. Selective hydrogenation of olefins is also an important catalytic
process. Among the larger-volume oxidation reactions, the oxidation of NH3 to
NO to produce nitric acid (HNO3) uses noble metals: Pt, Pt—Rh, and Pt—Pd—
Rh. The oxidation of SO2 to SO3 to produce sulfuric acid (H2SO4) uses mostly
vanadium oxide as catalyst. Ammoxidation, which makes acrylonitrile from
propylene, oxygen, and ammonia, uses bismuth and molybdenum oxides as
catalysts. The oxychlorination process to make vinyl chloride from acetylene
and HCl uses copper chloride as a catalyst. Polymerization reactions of ethylene
and propylene are catalyzed by titanium trichloride, aluminum alkyls, chromium
oxide on silica, and peresters. While these are the catalysts that are used in the
largest quantity, many other highly selective catalysts serve as the basis of entire
chemical technologies. In fact, the value of a very selective catalyst that aids a
complex chemical transformation such as the production of precious lifesaving
pharmaceuticals is without comparison.
Most of the catalysts employed in these chemical technologies are
heterogeneous. The chemical reaction takes place on surfaces, and the reactants
are introduced as gases or liquids. Homogeneous catalysts, which are frequently
organometallic molecules or clusters of molecules, also find wide and important
applications in chemical technologies [74, 75]. Some of the important
homogeneously catalyzed processes are listed in Table 9.50 (see page 752).
Carbonylation, which involves the addition of CO and H2 to a Cn olefin to
produce a Cn+1, acid, aldehyde, or alcohol, uses Rh and Co complexes. Cobalt,
copper, and palladium ions are used for the oxidation of C2H4 to acetaldehyde
and to acetic acid. Cobalt(II) acetate is used mostly for alkane oxidation to acids,
especially butane. The air oxidation of cyclohexane to cyclohexanone and
cyclohexanol is also carried out mostly with Co salts. Further oxidation to adipic
acid uses Cu(II) and V(V) salts as catalysts. The hydrocyanation of butadiene to
adiponitrile uses zero-valent Ni complexes. Polymerization technologies also
frequently use homogeneous catalysts. The manufacture of polyethylene
terephthalate uses Sb salts, and the copolymerization of ethylene and propylene
toproduce rubber usesalkylvanadiumcompounds.

9.6 CASE HISTORIES OF SURFACE

CATALYSIS
Surface-science studies have succeeded in identifying many of the molecular
ingredients of surface-catalyzed reactions. Each catalyst system that is
responsible for carrying out important chemical reactions with a high turnover
rate and selectivity has unique structural features and composition. In order to
demonstrate how these systems operate, we will review what is known about (a)
ethylene hydrogenation on Pt, (b) NH3 synthesis catalyzed by Fe, and (c)
oxidation of CO on transition metal surfaces.

9.6.1 Ethylene Hydrogenation Over Platinum Surfaces

The catalytic hydrogenation of ethylene over platinum, , was
first observed by Sabatier and Senderens as reported in an article by Horiuti and
Miyahara at the end of the 19th century [76]. Since then, a continued effort has
been made to understand this simplest example for hydrocarbon conversion
reactions. The hope is that the molecular level understanding of its chemistry
will provide insights about the hydrogenation, exchange, dehydrogenation, and
isomerization of more complicated alkenes [77].
Ethylene hydrogenation is thermodynamically favorable with a heat of
formation about –32.5kcalmol−1. It is one of the fastest heterogeneous catalytic
reactions known. Over Pt surfaces, turnover rates are typically on the order of 10
molecules site−1 s−1 at room temperature and atmospheric pressures of H2 and
C2H4 (Table 9.51 on page 753). Activation energies are between 8.6 and
10.8kcalmol−1 [80]. Reaction orders in H2 and C2H4 are about –0.5 and 1.0,
respectively, under the relatively high H2/C2H4 partial pressure ratios and
temperatures around or higher than room temperature. At low temperatures or
low H2/C2H2 partial pressure ratios, the reaction orders may be sensitive to
temperature changes. It was observed over a Pt/Cab-O-Sil catalyst that, at 223 K,
the reaction orders in C2H4 and H2 are –0.17 and 0.48, respectively [80]. The
approximately zero order of C2H4 indicates that the surface is saturated by
certain inactive surface species (the surface species do not turnover to form
C2H6) from the C2H4 adsorption, which makes the reaction turnover rate
insensitive to the C2H4 partial pressure. These surface species also partially
block the surface sites for hydrogen adsorption, which lowers the reaction order
in hydrogen from ~1.0–0.48.
In contrast to the structure sensitivity of C2H4 decomposition over various
crystal faces of Pt, the macroscopic turnover rate measurements demonstrate the
surface-structure insensitivity of C2H4 hydrogenation (Table 9.51 on page 753).
For example, the measured turnover rates at room temperature are ~11 and 12
molecules site−1 s−1 for Pt(111) and Pt(100), respectively, under partial pressures
of 100 Torr of H2, 35 Torr of C2H4, and 625 Torr of He [81].
The macroscopic reaction studies mentioned above put forth molecular level
questions, for example, what surface species are present on Pt surfaces during
the hydrogenation reaction; which of them is the reaction intermediate; and why
C2H4 decomposition is structure sensitive while C2H4 hydrogenation is not.
The first molecular level reaction mechanism was proposed by Horiuti and
Polanyi in 1934 [82]. The Horiuti–Polanyi mechanism (Fig. 9.18) envisions
C2H4 being adsorbed on the surface of a transition metal by breaking one of the
carbon–carbon double bonds and forming two σ-bonds with the metal surface.
This intermediate is known as di-σ-bonded C2H4. This di-σ-bonded species
would be stepwise hydrogenated by atomic hydrogen (formed from
dissociatively adsorbed molecular hydrogen) through an ethyl intermediate to
form C2H6, which is desorbed into the gas phase.
Figure 9.18. The Horiuti–Polanyi mechanism of C2H4 hydrogenation on the
transition metal surface.
9.6.1.1 Surface Species Involved in Ethylene
Hydrogenation.
Surface techniques, such as ultraviolet photoemission spectroscopy (UPS),
LEED, high-resolution electron energy loss spectroscopy (HREEL), infrared
spectroscopy (IR), STM, and SFG, have since been employed to study the
surface intermediates involved in the process of C2H4 hydrogenation [81, 83–
89]. Three stable surface species found on the Pt surface after C2H4 adsorption
are π-bonded C2H4, di-σ-bonded C2H4, and ethylidyne (C2H3).
The π-bonded C2H4 is a physisorbed species with a carbon–carbon bond
parallel to the surface. On Pt(111), this species exists at temperature 52K under
UHV conditions, as observed by UPS experiment. The molecule interacts
weakly with the surface through the π orbital of C2H4. The structure of adsorbed
C2H4 remains almost unchanged with respect to the gas-phase C2H4.
Upon annealing the Pt(111) surface >52K, one of the carbon–carbon bonds of
the π-bonded C2H4 is broken and the C atoms attain nearly sp3 hybridization to
form two σ-bonds with the metal surface. As suggested by a LEED study, this
di-σ-bonded species resides in a face-centered cubic (fcc) threefold hollow site
with the C—C bond axis tilted up slightly from the plane of the surface. The
saturation coverage of this species is 0.25 ML. The SFG spectrum of a Pt(1 11)
covered by di-σ-bonded C2H4 shows a peak at 2910 cm−1, which can be
assigned to CH2 symmetric stretch of the molecule (Fig. 9.19a).
The di-σ-bonded C2H4 species is stable on Pt(1 11) up to 240 K, at which
temperature it begins to dehydrogenate to form C2H3 (ethylidyne) by losing one
H and transferring a second H to the other C atom to form a CH3 group. The
C2H3 molecule on Pt(111) resides in an fcc threefold hollow site with a C—C
bond normal to the surface, as determined by LEED. Three Pt atoms forming a
bond with one of the C atoms are pulled slightly out of the surface (Fig. 6.20).
The saturation coverage of C2H3 on Pt(111) is 0.25 ML. The SFG spectrum of
C2H3/Pt(111) is shown in Figure 9.19b. The peaks in the spectra are a good
match to the vibrational spectrum of C2H3 on an organometallic Os cluster and
his was originally used to identify the surface species. The 2886-cm−1 feature in
theSFG spectrum is the CH3 symmetric stretch of the terminal CH3 group.
Ethylidyne is observed to be stable on Pt(111) up to 450 K, where it
dehydrogenated to C2H and CH species, and then to graphite .>800 K.

Figure 9.19. The SFG spectra of saturation coverages of (a) the di-σ-bonded
C2H4 at 202 K on Pt(111) and (b) C2H3 at 300K on Pt(111) [90].

Both experimental and theoretical evidence exist showing that C2H4 is very
mobile on Pt(111) at 300 K. Scanning tunneling microscopy experiments on
ethylidyne-covered Pt(111) at room temperature have shown that C2H3 is not
visible under these conditions [88] most likely due to the mobility of the
ethylidynes at room temperature. In fact, STM images of C2H3/Pt(1 11) at 300 K
look very similar to images of the clean Pt(111) surface. It was only upon
cooling the Pt crystal that the C2H3 became visible. Extended-Huckel
calculations have been performed for an C2H3 moiety on Pt(111) to determine
the activation barrier for surface diffusion. The results show a low-activation
barrier of 0.11 eV, thus explaining the mobility of C2H3 on the surface at 300 K
[91].
Ethyl is the reaction intermediate proposed in the Horiuti–Polanyi mechanism.
The spectrum of C2H5/Pt(111) at 193 K is shown in Figure 9.20. Several CH
stretch features are observable on the surface. The two lower frequency features
(2860 and 2920 cm−1) have been assigned to a Fermi resonance and the CH3
symmetric stretch of the adsorbed C2H5 group.

9.6.1.2 The Role of Reaction Intermediates in

Ethylene Hydrogenation.
As discussed previously, the reaction order in C2H4 is close to zero at relative
high partial pressures of C2H4 or low-temperature conditions, indicating that the
surface is saturated by certain inactive surface species with relatively high-
adsorption energies. Both in situ IR and SFG measurements have observed that
C2H3 and di-σ-bonded C2H4 are the abundant species on the surface during the
reaction. Figure 9.21a shows the in situ SFG spectrum of Pt(111) under
conditions of 110 Torr H2 and 35 Torr C2H4 at 295 K. The dominant peak at
2878 cm−1, which is observed on the C2H3 covered Pt(1 11) (Fig. 9.19b), is the
CH3 symmetric stretch of C2H3. The small peak at 2910 cm−1 also was observed
on the di-σ-bonded C2H4 covered Pt(111) (Fig. 9.19a) and is clearly due to the
CH2 symmetric stretch of di-σ-bonded C2H4. By calibrating SFG peak
intensities of surface species with knowncoverages, the coverage of C2H3 can be
obtained as 0.15 ML or 60% of the saturation coverage, and the coverage of di-
σ-bonded C2H4 as 0.08 ML.

Figure 9.20. The SFG spectrum of saturation coverage of ethyl (C2H5) groups
on Pt(111) at 193 K under the UHV condition [90].

Figure 9.21. (a) The SFG spectrum of the Pt(111) surface during C2H4
hydrogenation with 100 Torr of H2 and 35 Torr of C2H4 at 295 K. (b) The SFG
spectrum under the same conditions as (a), but on a Pt(111) surface that was
precovered in UHV with 0.25 ML of C2H3 [90].
 Ethylene hydrogenation has been performed on the C2H3 precovered Pt(1 11)
under the same partial pressures and temperature to investigate the role of C2H3
in the reaction process. The SFG spectrum in Figure 9.21b shows that the peak
intensity of the di-σ-bonded C2H4 becomes smaller. The coverage of di-s-bonded
C2H4 is 0.02 ML, which is a factor of four drop from the surface coverage
during the reaction on a clean surface. This result can be understood since C2H3
and di-σ-bonded C2H4 are competing for the fcc threefold hollow sites on Pt(1
11). The preadsorbed C2H4 blocks the adsorption sites for the di-σ-bonded C2H4,
so the relative coverage of these two species depends on the prehistory of the
metal surface.
However, the measured reaction turnover rates on the clean and C2H3
precovered Pt(111) are almost the same (11 and 12 ethane molecules per Pt site−1
s−1, respectively), indicating that once the surface is saturated by the two
strongly adsorbed species, their relative coverage does not affect the rate of
C2H6 formation. The inactive role of C2H3 in the process of C2H4 hydrogenation
has been confirmed by several techniques. Under room temperature and 1 atom
of hydrogen, the measured direct hydrogenation rate of chemisorbed C2H3 is
orders of magnitude slower than the rate of C2H4 hydrogenation (Fig. 9.22).
The di-σ-bonded C2H4 is readily hydrogenated under these reaction
conditions. The fact that the coverage of the di-σ-bonded C2H4 is not directly
correlated with the reaction ratesuggests that it is not the reaction intermediate
leading to the activation barrier of the hydro-genation reaction.
Figure 9.22. Turnover rates for C2H4 hydrogenation, the rehydrogenation of
C2H3, and the deutera-tion of the CH3 group of C2H3 on Pt and Rh crystal
surfaces [20]. Note that C2H4 hydrogenation rates are orders of magnitude faster
than the rate of removal of chemisorbed C2H3.
 At present, which reaction intermediate is directly responsible for the
formation of ethane is still under debate. The general properties of the
intermediate include (1) it must be a weakly adsorbed species, since, if it were a
strongly absorbed species like C2H3, it should be easily identified in the IR or
SFG spectra; (2) It must be bonded to a single atom on the metal surface
because, if it were bonded to multiple atoms, the intermediate would have to
compete with C2H3 and the di-σ-bonded C2H4 for adsorption sites, meaning that
the coverage of C2H3 and the di-σ-bonded C2H4 should have a huge effect on the
hydrogenation rate.
The two most-likely active intermediates are the p-bonded C2H4 and C2H5.
The π-bonded C2H4 shows up in the SFG spectrum (Fig. 9.21) as a weak and
broad hump ~3000 cm−1. Its surface coverage is only ~4% ML. The peak
intensities of this species are similar during the reaction over the clean and the
C2H3 precovered Pt(111), which agrees with the unchanged reaction rate on
these two surface. Therefore this species is likely to be the key intermediate in
C2H4 hydrogenation. However, recent efforts by surface techniques other than
SFG could not confirm the existence of this species under similar reaction
conditions because of the very small adsorption energy of this species [92].
The C2H5 group shows up in the SFG spectrum in Figure 9.21 as a shoulder
~2850 cm−1, corresponding to the Fermi resonance peak (see Fig. 9.20). The
existence of a C2H5 group is more evident when the reaction is performed under
a higher partial pressure of hydrogen and at room temperature. Under these
reaction conditions, the surface coverages of C2H3 and the di-σ-bonded C2H4 are
very low and the two peaks at 2850 and 2925 cm−1 become evident in the SFG
spectrum (Fig. 9.23). Recently, time-resolved Fourier transform infrared (FTIR)
spectra of transient-state C2H4 hydrogenation over an alumina-supported Pt
catalyst have been recorded at 25-ms time resolution [93, 94]. Based on the
observation that, at 323 K, the decay time of the surface C2H5 concentration
(~122 ms) coincides with the rise time of C2H6 concentration (~144 ms), it has
been suggested that hydrogenation of surface C2H5 is the rate-limiting step in the
C2H4 hydrogenation process.

9.6.1.3 Structure Insensitivity of Ethylene

Hydrogenation.
Ethylene hydrogenation, like most olefin hydrogenation reactions, exhibits
surface-structure insensitivity. The turnover rate of C2H4 hydrogenation on
Pt(100) under 35 Torr of C2H4 and 100Torr of H2, at room temperature is ~12
C2H6 molecules Pt site−1 s−1, which is almost identical to that on Pt(111). The in
situ SFG spectrum on Pt(100) (Fig. 9.24) shows that the ratio between the
coverage of the di-σ-bonded C2H4 and C2H3 derived from the relative peak
intensity of these two species is markedly different from that on Pt(111) (Fig.
9.21). On Pt(100), there are more di-σ-bonded C2H4 and less C2H3 species. The
results again demonstrate the inactive role of these two strongly adsorbed
species in C2H4 hydrogenation. Basically, the reaction rate is determined by the
sites available for weakly bonded intermediates on Pt(111) and Pt(100), which
gives a sound explanation for the structure insensitivity of this reaction.
There are other explanations for the observed structure insensitivity of
hydrogenation of olefins over a variety of metal surface [95, 96]. Ethylene may
be adsorbed weakly on the top of the hydrocarbon overlay formed by strongly
bonded reaction intermediates (e.g., C2H3 and the di-σ -bonded C2H4). The H
atom is transferred through the hydrocarbon overlayer to directly hydrogenate
the weakly bonded C2H4, so that the turnover rate is not sensitive to the
structural details of the underlying metal surface [97, 98]. Alternatively, the
reaction may still involve the hydrogenation of the weakly adsorbed
intermediates on the metal surface. However, the rate-limiting step is
transportation to hydrogen through the hydrocarbon overlayer, which is not
sensitive to the structure of the metal surface. These two explanations are
possible only when there is a hydrocarbon overlayer formed during the reaction,
so they cannot account for the structure insensitivity observed in the cases where
the metal surface is mainly covered by hydrogen under the conditions of high
partial pressures of hydrogen or relatively high temperature [77].
Figure 9.23. The SFG spectrum of the Pt( 111) during C2H4 hydrogenation
under 727 Torr of H2 and 60 Torr of C2H4 at 295 K. Under this hydrogen-rich
condition, the surface coverage of C2H3 and the di-σ-bonded C2H4 is low. The
two peaks marked with arrows are features characteristic of an ethyl (C2H5)
species [90].

9.6.1.4 Catalysis in the Presence of a Strongly Adsorbed

Overlayer.
Ethylene hydrogenation may provide an example of an important class of
catalytic reactions that could occur at low temperatures or high pressures with
the metal surface covered by a near-monolayer amount of carbonaceous deposit.
In order to determine the surface residence time of the carbonaceous deposit,
the Pt surface was dosed by the l4C-labeled organic molecules under the reaction
conditions. Carbon-14 is a β-particle emitter. A β-particle detector was used to
monitor its surface concentration as a function of time during the catalytic
reaction. The H content of the adsorbed organic layer is determined by detecting
the amount of desorbing H with a mass spectrometer. These investigations reveal
that the residence time of the adsorbed carbonaceous layer depends on its
hydrogen content, which in turn depends on reaction temperature (Fig. 9.25).
Figure 9.24. The SFG spectrum of adsorbed species during C2H4 hydrogenation
on Pt(100) under 35 Torr of C2H4 and 100 Torr of H2, at room temperature [81].

Although the amount of deposit does not change much with temperature, the
composition does; it becomes much poorer in hydrogen as the reaction
temperature is increased. The adsorption reversibility decreases markedly with
increasing temperature as the carbonaceous deposit becomes more hydrogen
deficient. As long as the composition is ~CnH1.5n and the temperature is above
450 K, the organic deposit can be removed readily with hydrogen. With
increasing reaction temperatures above 450 K, it converts to an irreversible
adsorbed deposit with a composition of C2nHn that can no longer be readily
removed (hydrogenated) in the presence of excess hydrogen [100].
Nevertheless, the catalytic reaction proceeds readily in the presence of this
active carbonaceous deposit [101, 102]. Above 750 K, this active C layer is
converted to a graphitic layer that deactivates the metal surface, and all chemical
activity for any hydrocarbon conversion reaction ceases. Hydrogen-exchange
studies indicate rapid exchange between the hydrogen atoms in the adsorbing
reactant molecules and hydrogen in the active, but irreversibly adsorbed, deposit.
Only the C atoms in this layer do not exchange. Thus, one important property of
the carbonaceous deposit is its ability to store and exchange hydrogen [101–
103].
The structure of the adsorbed hydrocarbon monolayers was submitted to
detailed studies by LEED and HREELS [104]. In the temperature range of 300–
400 K, the adsorbed alkenes form C2H3 molecules (shown in Chapter 6). The C
—C bond closest to the metal is perpendicular to the surface plane, and its 1.5-Å
length corresponds to a single bond. The C atom that bonds the molecule to the
metal is located in a threefold site 2.0 Å equidistant from itsnearest metallic
neighbors [105]. This bond is appreciably shorter than the covalent metal–
carbon bond (2.2 Å) and is indicative of multiple metal-carbon bonds of the
carbene or carbyne type. Although this layer is ordered, on being heated to ~
100°C it disorders and hydrogen evolution is detectable by a mass spectrometer
attached to the system. As the molecules dehydrogenate, the disordered layer is
composed of CH2, C2H, and CH type fragments that can be identified by
HREELS [104]. Only after being heated to ~400°C do the fragments lose all
their hydrogen and a graphite overlayer forms. These sequential bond-breaking
processes, which occur as a function of temperature, are perhaps the most
important and unique characteristics of the surface chemical bond (Section 6.3).
Although the surface remains active in the presence of organic fragments of C2H
stoichiometry, it loses all activity when the graphite monolayer forms.
Figure 9.25. Carbon- 14-labeled C2H4 (or other alkenes) was chemisorbed as a
function of temperature on a flat Pt( 111) crystal face. The H/C ratio of the
adsorbed species was determined from hydrogen thermal desorption. The
amount of preadsorbed alkene that could not be removed by subsequent
treatment in 1 atm of hydrogen represents the irreversibly adsorbed fraction. The
adsorption reversibility decreases markedly with increasing adsorption
temperature as the surface species become more hydrogen deficient. The
irreversibly adsorbed species have long residence times, on the order of days
[99].

How is it possible that the hydrocarbon conversion reaction exhibits great

sensitivity to the surface structure of Pt, while under the reaction conditions the
metal surface is covered with a near-monolayer of carbonaceous deposit? In fact,
often more than a monolayer amount of carbon-containing deposit is present, as
indicated by surface-science measurements. Recent STM studies that were
carried out at high hydrocarbon and hydrogen pressures and typical reaction
temperatures indicate that CH2, C2H, and CH fragments are mobile on the
surface. These fragments move around by surface diffusion in the presence of
coadsorbed molecular reactants. While they do not desorb, their mobility makes
the active metal sites on the surface available to the molecular reactants. When
the carbonaceous species polymerize at higher temperatures to form a graphite
deposit, they lose their mobility and deactivate the metal surface by permanently
blocking the active sites.
In order to determine how much of the Pt surface is exposed and remains
uncovered, the adsorption and subsequent thermal desorption of CO was
utilized. This molecule, although readily adsorbed on the metal surface at 300 K
at low pressures, does not adsorb on the carbonaceous deposit. The results
indicate that up to 10–15% of the surface remains and uncovered metal sites
decrease slowly with increasing reaction temperature. Under the reaction
conditions, the structure of these uncovered metal islands is not very different
from the structure of the initially clean metal surface.
As a result of catalyzed hydrocarbon conversion reaction studies on Pt crystal
surfaces, a model for the working Pt reforming catalyst can be proposed (see
Fig. 9.26). Between 80 and 95% of the catalyst surface is covered with an
irreversibly adsorbed carbonaceous deposit that stays on the surface much longer
than the reaction turnover time [106, 107]. The structure of this carbonaceous
deposit varies continuously from 2D to 3D with increasing reaction temperature.
There are Pt patches that are not covered by this deposit. These metal sites can
accept the reactant molecules that then compress the carbonaceous deposit by
surface diffusion to free up the active sites where the reactions occur. Upon
desorption of the products, the carbonaceous species may diffuse back to cover
the metal sites. The adsorption of new reactant molecules repeats the process;
compression of the carbonaceous deposit by surface diffusion, reaction at the
metal sites, and product desorption. There is evidence that the carbonaceous
deposit participates in some of the reactions by hydrogen transfer, in which it
provides a site for rearrangement and desorption while remaining inactive in
other reactions; its chemical role requires further exploration.
Figure 9.26. Model for the working structure and composition of a Pt
dehydrocyclization catalyst.Most of the surface is continuously covered by a
strongly bound carbonaceous deposit whose structure varies from 2D to 3D with
increasing reaction temperatures. Uncovered patches or ensembles of Pt surface
sites always exist in the presence of this carbonaceous deposit. Bond breaking
and chemical rearrangement in reacting hydrocarbon molecules take place
readily at these uncovered sites [102].
9.6.1.5 Summary
Ethylene hydrogenation is a hydrocarbon conversion reaction in which weakly
adsorbed species are directly responsible for a high turnover rate at room
temperature. Strongly adsorbed species, such as ethylidyne and s-bonded
ethylene, are extremely mobile on platinum surfaces under reactions, but do not
directly participate in the reaction. The reactive weakly adsorbed species have
very low surface coverages. On Pt(111), for example, SFG under reaction
conditions has determined that the surface coverage of π-bonded ethylene is only
~4%. The high turnover rate of this reaction (~10 C2H4 molecules site−1 s−1) at
such low coverage of active reaction intermediates is due to the high mobility of
the strongly bound surface species under reaction conditions. This high mobility
ensures that surface active sites are frequently made available for the adsorption
of the catalytically-active weakly bound species. Surface defect sites are released
much less frequently than terrace sites due to relatively high adsorption energies
on defect sites. Thus, the majority of turnover events for weakly bonded species
occur on terrace sites. As a result, this catalytic reaction is insensitive to catalyst
surface structure.

9.6.2 Ammonia Syntheses

9.6.2.1 Thermodynamics and Kinetics.
The reaction of nitrogen and hydrogen to pro duce ammonia, ( ),
is somewhat exothermic. The free energy of NH3 formation as a function of
temperature is shown in Figure 9.27. The reaction is carried out over an Fe
catalyst that is frequently “promoted” by adding alumina and K. The reaction
temperature is ~400°C, and total pressures utilized are in the range of 150–300
atm.
Based on the observed dependence of the reaction rate on N2 and H2 pressure,
several rate laws have been proposed. The best known dependency is perhaps
the one by Temkin and Pyzher [109, 110]. An extension of this rate law by
Nielsen yields [111, 112]

(9.9)
Figure 9.27. The free energy of NH3 formation as a function of temperature
[108].

where w = 1.5 and α = 0.75. The parameters k, Ka, and K3 are constants. The rate
of NH3 formation depends in a rather complex manner on the partial pressures of
N2, H2, and NH3, mostly because of the possibility of a back-reaction. When the
partial pressures are far from equilibrium, this may be neglected and under this
circumstance the rate depends only on the N2 pressure. This indicates that the
rate-limiting step is the dissociative adsorption of N2 on the catalyst surface, a
conclusion shared by most of the practitioners.
Other important rate equations that are applicable in a variety of experimental
conditions have been proposed by Ozaki et al. [113]. The net activation energy
for the reaction is 76kJmol−1, which is in excellent agreement with the 81-kJmol
−1 value determined using single-crystal Fe surfaces [108, 114–117].

9.6.2.2 Catalyst Preparation.

The industrial catalyst is prepared by the reduction of Fe oxide, Fe3O4 (94 wt%).
It consists of small porous particles with a surface area in the range of 10–15
m2g−1. Additives that improve its performance include Al2O3 (2.3 wt%), K2O
(0.8 wt%), and often CaO (1.7 wt%), MgO (0.5 wt%), and SiO2 (0.4 wt%). Even
though the addition of Al, Mg, Ca, and Si oxides may somewhat decrease the
iron surface area the oxide additives stabilize the pore and surface structure of
the iron catalyst, and greatly increase the ammonia yield at 613 K from 0.2 to
0.34 mol%.

9.6.2.3 Activity for Ammonia Synthesis Using Transition

Metals Across the Periodic Table.
There are two factors that are all important in determining the NH3 synthesis
rate. One is the N2 dissociative sticking probability. Dinitrogen dissociation turns
out to be rate limiting, and at low conversions the total rate of the reaction equals
the dissociation rate of N2. The other factor is the N atom chemisorption energy.
Chemisorbed atomic nitrogen is by far the most stable reaction intermediate.
Therefore, the surface is mainly covered by N atoms up to 90% of a monolayer;
and the number of free sites on the surface where the nitrogen molecule can
adsorb is proportional to (1 – θN), where θN is the atomic nitrogen coverage.
If we use the kinetic model that was reported by Nielsen, the NH3 formation
rate can be calculated as a function of the number of d-electrons in the transition
metals [112]. The results are shown in Figure 9.28a. It produces a volcano curve
similar to that observed experimentally by Ozaki and Aika, who plotted the
variation of the activity of various transition metals for the NH3 synthesis
reaction as a function of the degree of filling of the d-band (Fig. 9.28b) [118].
The calculated results agree very well with experiments. On the right side of the
maximum in the volcano curve, the NH3 production decreases because the rate
of N2 dissociation drops as a consequence of the increase in the activation
energy for dissociation. To the left of the peak of the volcano, the dissociation
rate increases; but since the nitrogen chemisorption bond also increases in
strength, the number of surface sites on which the nitrogen molecule can
dissociate decreases so fast that the overall rate decreases.

9.6.2.4 Surface Science of Ammonia Synthesis

9.6.2.4.1 Structure Sensitivity of Ammonia Synthesis.
A UHV chamber equipped with a high-pressure cell has been developed to study
the NH3 synthesis reaction on an Fe single-crystal surface. A single crystal is
enclosed in a high-pressure cell that constitutes part of a microbatch reactor.
High pressures of gases (e.g., 15 atm of hydrogen and 5 atm of nitrogen) are
introduced and the sample is heated to reaction temperatures (600–700 K). The
NH3 production is monitored using a selective photoionization detector with a
photon energy that ionizes NH3 and not N2 or H2. After the reaction is
completed, the reaction loop is evacuated and the cell is opened, returning the
sample to the UHV environment, where surface characterization is performed by
AES, LEED, and temperature-programmed desorption.
Figure 9.28. (a) The calculated NH3 concentration for a fixed set of reaction
conditions as a function of the number of d-electrons [119]. (b) The activity of
various transition metals forNH3 synthesis as a function of the degree of filling
of the d-band [118].
 In Figure 9.29, the rates of NH3 synthesis are shown over five Fe crystal
orientations. The Fe(111) and Fe(211) surfaces are by far the most active in NH3
synthesis and are followed in reactivity by Fe(100), Fe(210), and Fe(1 10) [120].
Schematic representations of the idealized unit cells for these surfaces are shown
in Figure 9.30. There are two possible reasons for the high activity of the (111)
and (211) faces compared to the other (210), (100), and (110) orientations: their
exceptionally high surface roughness or the presence of unique active sites the
other crystal faces may not possess.
The (111) surface can be considered a rough surface, since it exposes second-
and third-layer atoms to reactant gases in contrast to the (110) surface that only
exposes the first-layer atoms. Work functions are related to the roughness of a
surface, and it is useful to quantify the corrugation of a plane in this way [121].
Open faces, such as the (111) surface, have lower work functions than close-
packed faces, such as the (110) surface. The work functions of all the Fe faces
are not currently available, but they are for W, another body-centered cubic (bcc)
metal that also shows structure sensitivity for NH3 decomposition [122, 123].
The order of a decreasing work function (φ) is as follows: φ110 >φ211 > φ100 >
φ111 >φ210. However, the order of a decreasing work function from crystal face
to crystal face does not correlate with variations of catalytic activity.
 The second possible explanation for the structure sensitivity of the NH3
synthesis rate of Fe involves the nature of the active sites. The (111) and (211)
faces of Fe are the only surfaces that expose C7 sites (Fe atoms with seven
nearest neighbors) to the reactant gases. Theoretical work by Falicov and
Somorjai has suggested that highly coordinated surface atoms would show
increased catalytic activity due to low-energy charge fluctuations in the d-bands
of the highly coordinated surface atoms [50], which is the key to the structure
sensitivity of NH3 synthesis over Fe.
Figure 9.29. Rates of NH3 synthesis over five Fe single-crystal surfaces with
different orientations: (111), (211), (100), (210), and (110) [120].

Figure 9.30. Schematic representations of the idealized surface structures of the

(111), (211), (100), (210), and (110) orientation of Fe single crystals. The
coordination of each surface atom is indicated [120]. (See color insert.)
 The reaction rates (see Fig. 9.29) show that the (211) face is almost as active as
the (111) plane of Fe, while Fe(210) is less active than Fe(100). The Fe(210) and
Fe(1 11) faces are open faces that expose second- and third-layer atoms. The
Fe(211) face is more close-packed, but exposes the C7 sites. If either surface
roughness or a low work function were the important consideration for an active
NH3 synthesis catalyst, then the Fe(210) would be expected to be the most active
face. However, in marked contrast, Fe(1 11) and Fe(211) faces are much more
active, indicating that the presence of C7 sites is more important than surface
roughness in an NH3 synthesis catalyst.
The idea of C7 sites being the most active site in NH3 synthesis on Fe has been
suggested in the past. Dumesic et al. found that the turnover number for NH3
synthesis was lower on small Fe particles than on larger ones [124]. Pretreatment
of an Fe/MgO catalyst with NH3 enhanced the turnover number over small Fe
particles, but did not affect the larger particles. This result was explained by
noting that the concentration of C7 sites would be expected to be higher on the
smaller Fe particles and that restructuring induced by NH3 enhanced the number
of these sites on the catalyst.
9.6.2.4.2 Kinetics of Dissociative Nitrogen.
Because the adsorption step is rate determining for NH3 synthesis, considerable
effort has been expended on its detailed investigation. It has turned out to be of
great complexity so that, even now, complete understanding of the underlying
microscopic dynamics is still lacking, although there exists general agreement
about the experimental findings.
In Figure 9.31, the variation in the relative surface concentration of
chemisorbed N atoms (as monitored by AES) with N2 exposure at elevated
temperatures for the Fe(1 10), Fe(100), and Fe(111) surfaces is shown [125,
126]. The slopes of these curves yield the sticking coefficients for dissociative
chemisorption that are obviously very small and depend markedly on surface
orientation. More specifically, the initial sticking coefficient (at 683 K) changes
from 7 × 10−8 to 2 × 10−7 to 4 × 10−6 in the sequence Fe(110) −4Torr) is in
agreement with that found for the rate of NH3 production at high pressure
(20atm) described in Section 9.6.2.4.1. Moreover, the sticking coefficients are
approximately of the same orders of magnitude as the reaction probabilities
derived from high-pressure work. This remarkable result demonstrates that
kinetic parameters derived from well-defined single-crystal surfaces are
transferable over the “pressure gap” and it confirms that the dissociative nitrogen
adsorption is indeed the rate-limiting step, since the rate of NH3 formation
equals that of dissociative nitrogen adsorption. Similar conclusions had already
been reached many years ago by Emmett and Brunauer, who measured the
uptake of nitrogen by commercial catalysts and concluded likewise that the
sticking coefficient is only on the order of 10−6 [127].
Figure 9.31. Variation of the relative surface concentration of atomic nitrogen as
a function of N2 exposure [108] [1 L (Langmuir) = 1026 Torrs].
 The sticking coefficient(s) can be formulated in terms of the usual Arrhenius
equation for a rate constant, s = A exp( – ΔE*/RT), with the pre-exponential A
and activation energy ΔE* as parameters. Measurements at different
temperatures revealed that the differences between the three crystal planes can
essentially be traced back to differences in the net activation energy E* for the
overall process N2 → 2Nads, which in the limit of zero coverage was found to be
~27kJmol−1 for Fe(110), ~21 kJmol−1 for Fe(100), and ~0kJmol−1 for Fe(111).
These activation energies increase continuously with increasing coverage, in
qualitative agreement with previous measurements using a supported Fe
catalysts [128].

9.6.2.4.3 Effects of Aluminum Oxide in Restructuring Iron Single-

Crystal Surfaces for Ammonia Synthesis.
The initial rate of NH3 synthesis has been determined over the clean Fe(111),
Fe(100), and Fe(110) surfaces with and without aluminum oxide. The addition of
Al2O3 to the (110), (100), and (111) faces of Fe decreases the rate of NH3
synthesis in direct proportion to the amount of surface covered [129]. This
suggests that the promoter effect of Al2O3 involves reaction with Fe that cannot
be achieved by simply depositing Al2O3 on an Fe catalyst.
Remembering that industrial catalysts are prepared by fusion of 2–3 wt% of
Al2O3 and K with iron oxide (Fe3O4), experiments have been performed with the
AlxOy/Fe single-crystal surfaces which are pretreated in an oxidizing
environment prior to NH3 synthesis. These experiments are carried out by
depositing ~2mL of AlxOy on Fe(111), Fe(100), and Fe(110) surfaces and then
treating them in varying amounts of water vapor at 723 K in order to oxidize the
Fe and to induce an interaction between iron and aluminum oxide. After removal
of water vapor, high pressures of nitrogen and hydrogen are added to determine
the rates of NH3 synthesis. The rate of NH3 synthesis over AlxOy/Fe surfaces
pretreated with water vapor prior to NH3 synthesis is shown in Figure 9.32. The
initially inactive AlxOy/Fe(110) surface restructures and becomes as active as the
Fe(100) surface after a 0.05-Torr water-vapor treatment, and as active as the
Fe(111) surface after a 20-Torr water-vapor pretreatment. This is about a 400-
fold increase in the rate of NH3 synthesis compared with clean Fe(110) [120].
The activity of the AlxOy/Fe(100) surface can also be enhanced to that of the
highly active Fe(1 11) surface by utilizing a 20-Torr water-vapor pretreatment,
and this high activity is maintained indefinitely as in the case for the restructured
AlxOy/Fe(110). Little change in the activity of the Fe(111) surface is seen
experimentally when it is treated in water vapor in the presence of AlxOy.

Figure 9.32. Rates of NH3 synthesis over clean Fe single-crystals and water-
induced restructured AlxOy/Fe surfaces. Restructuring conditions are given in
the figure [120].
 The activity of the Fe(110) and Fe(100) surfaces for NH3 synthesis can also be
enhanced to the level of Fe(111) by water-vapor pretreatments in the absence of
aluminum oxide, but under these circumstances the enhancement in activity is
only transient. Figure 9.33 shows the rate of NH3 synthesis as a function of
reaction time for restructured Fe(110) and AlxOy/Fe(110) surfaces. Both surfaces
have an initial activity similar to that of the clean Fe(111) surface. The
restructured AlxOy/Fe(110) surface maintains this activity for >4h while the
restructured Fe(110) surface loses its activity for NH3 synthesis within 1 h of
reaction.

9.6.2.4.4 Characterization of the Restructured Surfaces.

The observation that the AlxOy/Fe(110) and AlxOy/Fe(100) become as active as
the Fe(111) surface for NH3 synthesis suggests that new crystal orientations are
being created upon restructuring the AlxOy/Fe(1 10) and AlxOy/Fe(100) surfaces
in water vapor. A suggested increase in surface area cannot account for the
enhancement in rate, since it has been shown that ~40% less CO adsorbs on
restructured AlxOy/Fe(110) and AlxOy/Fe(100) relative to the clean respective
surfaces (i.e., the Fe surface area actually decreases) [130].
Electron spectroscopies, LEED, temperature-programmed desorption (TPD),
and scanning electron microscopy (SEM) have been used to characterize the
restructured surfaces. The SEM micrographs for restructured Fe(110) and
AlxOy/Fe(110) surfaces following a 20-Torr water-vapor pretreatment show that
the surfaces seem to be completely recrystal-lized. Auger electron spectroscopy
finds that only ~5% of the Fe surface is covered by AlxOy, and sputtering the
surface with Ar ions reveals AlxOy beneath the Fe surface.
Figure 9.33. Deactivation of the restructured Fe(1 10) surface occurs within 1 h
while the restructured AlxOy/Fe(110) surface maintains its activity under NH3
synthesis conditions [120].
 Temperature-programmed desorption of NH3 from Fe single-crystal surfaces
following high-pressure NH3 synthesis proves to be a sensitive probe of the new
surface binding sites formed upon restructuring. Ammonia TPD spectra for the
four clean surfaces are shown in Figure 9.34. Each surface shows distinct
desorption sites. The Fe(1 10) surface displays one desorption peak (β3) with a
maximum at 658 K. Two desorption peaks are seen for the Fe(100) surface (β2
and β3) at 556 and 661 K. The Fe(111) surface exhibits three desorption peaks
(β1, β2, and β3) with peak maxima at 495, 568, and 676 K, and the Fe(211) plane
has two desorption peaks (β2 and β3) at 570 and 676 K. The TPD spectra for the
AlxOy/Fe(110), AlxOy/Fe(100), and AlxOy/Fe(111) surfaces restructured in
20Torr of water vapor are shown in Figure 9.35. A new desorption peak, β2,
develops on the restructured AlxOy/Fe(110) surface, and an increase in the β2
peak occurs on the restructured AlxOy/Fe(l00) surface. The β2 peaks from the
restructured AlxOy/Fe(110) and AlxOy/ Fe(100) surfaces grow in the same
temperature range as the β2 on the Fe(111) and Fe(211) surfaces.
 The ammonia TPD results point toward the formation of surface orientations
that contain C7 sites during water-vapor-induced restructuring. The growth of the
β2 peaks upon restructuring of the Fe( 110) and Fe( 100) surfaces suggests that
the surfaces change orientation upon water-vapor treatment. The β2 peaks also
reside in the same temperature range as the Fe(1 11) β2 peak. It seems likely that
the TPD peaks in this temperature range act as a signature for the C7 sites
because the Fe(211) surface that contains C7 sites is highly active in the NH3
synthesis reaction and also exhibits a β2 peak after NH3 synthesis, with a peak
maximum at 570 K. These results suggest that surface orientations that contain
C7 sites, such as the Fe(1 11) and Fe(211) planes, are formed during the
reconstruction of clean and AlxOy treated Fe surfaces, but only in the presence of
AlxOy does the active restructured surface remain stable under the NH3 synthesis
conditions.
Figure 9.34. Ammonia TPD after high-pressure NH3 synthesis. The low-
temperature peaks exhibited by Fe(111) and Fe(211) (β1 and β2) are attributed to
the presence of C7 sites [120].
 With the addition of AlxOy, the mobility of Fe is increased and restructuring
can occur at a lower water vapor pressure. The SEM micrographs suggest that Fe
forms crystallites on top of the restructured AlxOy/Fe(110) surface [as opposed to
the uniform appearance of the restructured clean Fe(110) surface]. Auger
electron spectroscopy finds little AlxOy on the surface, suggesting that the Fe has
diffused through the AlxOy islands, covering them. These findings can be
explained by considering wetting properties and the minimization of the free
energy for the iron oxide-aluminum oxide system. The formation of iron
aluminate (i.e., FeAl2O4 the presence of an oxygen source was also postulated
on the basis of microelectron diffraction data [131].
The formation of an iron aluminate during reconstruction of the Fe surface
may be responsible for the stability of the restructured AlxOy/Fe surfaces. The
presence of iron aluminate has been postulated from XPS studies on Fe— Al2O3
and Fe3O4—Al2O3 systems, as well as in numerous studies on the industrial NH3
synthesis catalyst [132–136]. The low coverages of AlxOy on the restructured
surfaces suggest that FeAl2O4 plays the role of support on which Fe surfaces
grow in (111) orientation, which is most active in NH3 synthesis. This result is
supported by the fact that ion sputtering the restructured surfaces reveal
subsurface AlxOy This model for the role of alumina as a structure modifier of Fe
for NH3 synthesis is shown in Figure 9.36.
Figure 9.35. Ammonia TPD following NH3 synthesis from restructured
AlxOy/Fe(100) surfaces exhibit low-temperature peaks similar to those of Fe(1
11) and Fe(211). Thus, restructuring by water vapor creates active C7 sites [120].

9.6.2.4.5 Effect of K on the Dissociative Chemisorption of Nitrogen

on Fe Single-Crystal Surfaces in UHV.
The rate-determining NH3 synthesis reaction is widely accepted to be the
dissociation of N2, [113,137–139]. Consequently, the direct interaction between
N2 and Fe has been studied together with the addition of submonolayer amounts
of K [ 125, 126, 138, 140]. All the work that will be referred to in this section
has been carried out in a UHV chamber, which limits the pressure range between
10−4 and 10−10 Torr.
With the use of both Fe single crystals and polycrystalline foils, the sticking
probability of molecular nitrogen on Fe is found to be on the order of 10−7. This
result reveals why, in addition to thermodynamic considerations, NH3 synthesis
from the elements is favored at high reactant gas pressures. Because the sticking
probability of dissociating nitrogen is so low on Fe, higher pressures of nitrogen
enhance the kinetics of the rate-limiting step in NH3 synthesis. The structure
sensitivity of the reaction is also revealed in nitrogen chemisorption studies. It
was found that the Fe(1 11) surface dissociatively chemisorbed nitrogen 20 times
faster than the Fe(100) surface and 60 times faster than the Fe(1 10) surface.
This agrees well with the structure sensitivity of NH3 synthesis and adds more
credence to dissociative chemisorption being the rate-limiting step. The addition
of submonolayer amounts of elemental K has dramatic effects on the nitrogen
chemisorption properties of the (110), (100), and (111) faces of Fe.
Figure 9.36. The scheme of the restructuring process of Fe induced by water
vapor and the presence of aluminum oxide. The oxidation of Fe permits the
migration of the metal on top of the aluminum oxide. The formation of FeAl2O4
may facilitate this process. Upon reduction in N2 and H2, Fe is left in active and
stable (111) orientation for NH3 synthesis on top of FeAl2O4.

The effect of K on the initial sticking coefficient (S0) of nitrogen on an Fe(

100) surface is shown in Figure 9.37. For clean Fe(100), S0 is 2 × 10−7, but with
the addition of potassium S0 increases almost linearly, until at a K concentration
of 2.5 × 1014 K atoms cm−2, where S0 maximizes at a value of 3.9 × 10−5, a
factor of 195 enhancement is seen. Higher coverages of K start to decrease S0,
presumably due to the blocking of Fe sites by K, which would otherwise
dissociatively chemisorb nitrogen. The maximum increase in S0, due to K
adsorption on Fe(111) is about a factor of 10 (S0 = 4 × 10−5) at a K concentration
of 2 × 1014 atoms cm−2. The potassium-induced enhancement of S0 on the
Fe(110) surface is greater than that observed on either Fe(1 11) or Fe(100), so
that the differences in activities for nitrogen dissociation seen on the clean
surfaces is much smaller in the presence of K.
Figure 9.37. Variation of the initial sticking coefficient S0 of N2, with the
addition of K to Fe(100) at 430 K. The N2 sticking coefficient can be enhanced
by a factor of 280 relative to clean Fe(100) [138].

The mechanism by which K promotes nitrogen chemisorption is usually

attributed to the lowering of the surface work function in the vicinity of a K+ ion.
This effect is greatest at sufficiently low coverages (0.15 ML), where the K—Fe
bond has strong ionic character, so that the local ionization potential of the
surface Fe atoms is lowest. This allows for more electron density to be
transferred to the nitrogen 2π* antibonding orbitals from the surface. This
phenomenon increases the adsorption energy of molecular nitrogen and
simultaneously lowers the activation energy for dissociation. For example, on
the Fe(100) surface, the addition of 1.5 × 1014 K atoms cm−2 decreases the work
function by ~1.8 eV and increases the rate of nitrogen dissociation by more than
a factor of 200. This enhancement in rate is accompanied by an increase in the
adsorption energy of nitrogen on Fe(100) by 11.5kcalmol−1, which decreases the
activation barrier for dissociation, in the presence of K, from 2.5kcalmol−1 to
~0kcal mol−1

9.6.2.4.6 : TPD Studies ofNH3from Fe Surfaces in the Presence of

K.
The TPD of NH3 from clean Fe(111) and K/Fe(111) is shown in Figure 9.38
[141]. Ammonia desorbs through a wide temperature range, resulting in a broad
peak with a maximum rate of desoption occurring at ~300 K. With the addition
of 0.1 ML of K, the temperature of the peak maximum is reduced by ~40 K.
Assuming first-order desorption for NH3, the 40 K decrease corresponds to a
2.4-kcal mol−1 drop in the adsorption energy of NH3 on Fe in the presence of
0.1-ML potassium. The peak maximum continuously shifts to a lower
temperature with increasing amounts of coadsorbed K. At a coverage of 0.25
ML, a new desorption peak appears at ~189 K. Increasing coverages of K now
increase the intensity of the new peak (it also shifts to lower temperatures) and
decreases the intensity of the original NH3 desorption peak. At a K coverage of
~1.0ML, only a weakly bound NH3 species is present, with a maximum rate of
desorption occurring at 164 K. This observation of decreasing adsorption energy
for NH3 with the coadsorption of K on Fe is similar to what is found for NH3
desorption from Ni and Rh with coadsorbed Na [142, 143].
Figure 9.38. Ammonia TPD from clean Fe(1 11) and K/Fe(1 11) surfaces. The
desorption temperature of NH3 from Fe(1 11) is lowered in the presence of K.
Thus K lowers the adsorption energy of NH3 on the Fe surface [141].
9.6.2.4.7 Effects of K on NH3 Synthesis Kinetics.
Extensive research has been completed in which the effects of K on NH3
synthesis over Fe single-crystal surfaces of (111), (100), and (110) orientations
have been determined [141]. The apparent order of NH3 and H2 for NH3
synthesis over Fe and K/Fe surfaces has been determined in addition to the effect
of K on the apparent activation energy (Ea) for the reaction. In all the
experiments, K was coadsorbed with O2 because only ~0.15 ML of K
coadsorbed with O2 is stable under NH3 synthesis conditions (20-atm total
pressure: 3:1 H2 to N2: T = 673 K) [129, 141, 144]. It has been shown that the
addition of 0.15 ML of K to Fe(111) and Fe(100) increases the slope of NH3
partial-pressure dependence curve from –0.60 for the clean Fe surfaces to –0.35
for the 0.15-ML K/Fe(111) and 0.15 ML K/Fe(100) surfaces under high-pressure
NH3 synthesis conditions (Fig. 9.39a). The apparent order in hydrogen partial
pressure has been found to decrease from 0.76 for clean Fe(1 11) to 0.44 for the
0.15-ML K/Fe(111) surface (Fig. 9.39b). The Fe(110) is inactive for NH3
synthesis under these conditions with or without K. These changes in both the
apparent order of H2and NH3 pressure dependence occur with no change in the
activation energy, which suggests that K does not change the elementary steps of
NH3 synthesis (Fig. 9.40). The data show that the promotional effect of K is
enhanced as the reaction conversion increases with increasing NH3partial
pressure.
Figure 9.39. (a) The apparent order in NH3 for NH3 synthesis over Fe(100) and
K/Fe(100) surfaces. The order in NH3 becomes less negative when K is present.
The same values were found for Fe(1 11) and K/Fe(111) surfaces. (b) The
apparent reaction order in H2 for NH3 synthesis over Fe(111) and K/Fe(111)
surfaces. The order in H2 decreases in the presence of K [141].

Figure 9.40. The activation energy for NH3 synthesis on Fe(111) and
K/Fe(111).Within experimental error there is no change, suggesting that K does
not change the reaction mechanism of NH3 synthesis [141].
 These results are consistent with the earlier literature in which the effects of K
on doubly promoted (AlxOy and K) catalysts were studied [145, 146]. It was
shown that the turnover number for NH3 synthesis is roughly the same over
singly (AlxOy) and double promoted Fe when 1 -atm reactant pressure of N2 and
H2 is used [ 146]. This implies that at low-pressure conditions, the gas-phase
NH3 concentration is not high enough for K to exert a promoter effect. As higher
reactant pressures are achieved (95–200 atm), the promoter effect of K becomes
significant. It was found that doubly promoted catalysts became increasingly
more active than catalysts without K when the concentration of NH3 increased in
the gas phase [ 145]. This implies that K makes the apparent reaction-order
dependence in NH3 partial pressure less negative over commercial catalysts, in
agreement with the single-crystal work.

9.6.2.4.8 Effects of K on the Adsorption of NH3 on Fe Under NH3

Synthesis Conditions.
As shown in Figure 9.39a, the NH3 partial-pressure dependence of the
reactionrate over K/Fe(111) is different from that over Fe(111). The changes in
the apparent reaction-order dependence in the NH3 partial pressure suggest that
in order to elucidate the effects of K on both Fe single crystals and industrial
catalysts it is necessary to understand the readsorption of the gas-phase NH3 on
the catalyst surface during NH3 synthesis. Once adsorbed, NH3 has a certain
residence time (τ) on the catalyst, which is determined by its heat of adsorbtion
(ΔHads) on Fe, τ= τ0 exp(ΔHads/RT) [147]. During this residence on the catalyst,
NH3 can either diffuse on the surface or decompose to atomic nitrogen and
hydrogen [109, 110, 113]. In both cases, the species produced by NH3 might
reside on surface sites that would otherwise dissociatively chemisorb gas-phase
N2 and thereby decrease the rate of NH3 synthesis [109, 110, 113, 118]. The
promoter effect of K then involves lowering the adsorption energy of the
adsorbed NH3 so that the concentration of adsorbed NH3 is decreased. This
finding is supported by the TPD results, which show that NH3 desorption from
Fe(111) shifts to lower temperatures when K is adsorbed on the surfaces. Even at
a 0.1-ML coverage of K (coverage roughly equivalent to that stable under NH3
synthesis conditions), the adsorption energy of NH3 is decreased by 2.4kcalmol
−1. Thus, the residence time for the adsorbed NH is reduced and more of the
3
active sites are available for the dissociation of N2. At higher coverages of K, the
adsorption energy of NH3 decreases to an even greater extent, but these
coverages could not be maintained under NH3 synthesis conditions. There also
seems to be an additional adsorption site for NH3 when adsorbed on Fe at high
coverages of K, as indicated by TPD results. The development of a new
desorption peak with coverages of K > 0.25 ML might result from NH3
molecules interacting directly with K atoms, with the negative end of the NH3
dipole interacting with the K+ ion on the Fe surface [142]. This interaction
appears to be weak, since at the K coverage of 1 ML, NH3 desorbs from the
surface at 164 K.
Additional experimental evidence supporting the notion that NH3 blocks active
sites comes from the postreaction Auger data. Within experimental error, there is
no change in the intensity of the nitrogen Auger peak between an Fe surface and
a K/Fe surface after a high-pressure NH3 synthesis reaction. This finding
suggests that K does not change the coverage of atomic N2, but instead the
presence of K helps to inhibit the readsorption or promote the desorption of
molecular NH3 on the catalyst. High-pressure reaction conditions are probably
needed to stabilize this NH3 product on the Fe surface at 673 K, so it will not be
present in the UHV environment. Thus, only the more strongly bound atomic
nitrogen will be detected by AES in UHV.

9.6.2.5 Mechanism and Kinetics of Ammonia Synthesis.

If all the experimental evidence presented in the preceding sections is put
together, the reaction scheme for the catalytic synthesis of NH3 on iron-based
catalysts can unequivocally be formulated in terms of the following steps:

(9.10)
where * schematically denotes a vacant site on the catalyst surface.
Figure 9.41. Schematic energy profile of the progress of NH3 synthesis on Fe (in
KJmol−1) [108].
 The progress of the reaction may be rationalized in terms of its energy profile
as reproduced in Figure 9.41.
Attempts at theoretical modeling of the kinetics along these lines were recently
performed independently by two groups: Bowker et al. [148, 149] and Stolze and
Norskov [150–155]. The latter group starts with the experimentally well-
established fact that dissociation of adsorbed nitrogen is rate-limiting. The
overall rate can then be calculated from the rate of this step and the equilibrium
constants of all the other steps. This reduces the number of input parameters
significantly. The adsorption-desorption equilibria are treated with the
approximation of competitive Langmuir-type adsorption and by evaluation of the
partition functions for the gaseous and adsorbed species. The data for the
potassium-promoted Fe(111) surface are used for the rate of dissociative
nitrogen adsorption and are also representative of the other crystal planes of the
promoted catalyst, as outlined above. The active area of the commercial catalyst
is assumed to equal that derived from selective carbon monoxide chemisorption
as a well-established standard procedure. A particular strength of this model is
the fact that experimental data from single-crystal studies (such as TPD traces)
are reproduced well with the same set of parameters and the same model as used
for the determination of the rate under “real” conditions. Comparison of the
resulting yields against those determined experimentally with a commercial
catalyst yielded general agreement to within a factor better than 2. In Figure
9.42, a compilation of data over a wide range of conditions is presented that
demonstrate this excellent agreement.
A general conclusion from these models based on single-crystal data is that the
most abundant surface species under practical synthesis conditions will be
adsorbed atomic nitrogen (>90%), despite the fact that its formation is the rate-
limiting step of the overall reaction.
Figure 9.42. Comparison of calculated and measured NH3 production over
commercial iron-based catalysts for a broad range of temperatures, pressures,
N/H ratios, and gas flows [155].

9.6.2.6 Summary.
Surface science research has provided molecular-level detail of ammonia
synthesis which has lead to the improvement of industrial catalysts for this
reaction. The rate-limiting step of this reaction is the dissociative adsorption of
N2. Because N2 is an inert reactant, ammonia synthesis proceeds at higher
temperatures and pressures over metal surfaces capable of forming stronger
chemisorption bonds, as compared to reactions with weakly adsorbed reactants,
like ethylene hydrogenation. In order to assist the dissociative adsorption of N2,
highly active sites with specific atomic arrangements, such as C7 sites on
Fe(111), are required. As a result of this specificity, this reaction is very sensitive
to the catalyst surface structure. For Fe-based catalysts in harsh reaction
conditions, the active sites can be stabilized by the presence of aluminum oxide.
Adding K to Fe catalysts further optimizes the ammonia synthesis process by
both enhancing N2 adsorption and assisting NH3 desorption so as to prevent
product poisoning.

9.6.3 Oxidation of Carbon Monoxide on Transition

Metal Catalysts
The oxidation of CO by O2 over transition metal catalysts is of great practical
importance in pollution control. In the three-way catalytic converter, CO is
removed as CO2 from the automobile exhaust by reacting with O2 or NO on Pt,
Pd, and Rh catalysts [72, 156]. Meanwhile, the relative simplicity of this reaction
makes it an ideal model system for studying heterogeneous catalysis.
This reaction is thermodynamically favorable with a free energy of formation
of CO2 of about –122kcalmol−1. In the gas phase, this reaction is prohibited by
the high activation energy of O2 dissociation (~117 kcal mol−1). On transition
metal surfaces, O2 can be dissociated with zero activation energy and the
adsorbed O atoms are ready to react with CO to produce CO2.
In past decades, a large body of surface science and catalysis work has been
devoted to this reaction in order to understand elementary reaction steps at the
molecular level. The adsorption and desorption of reactants and products, and
the transient-state, and steady-state reaction kinetics has been investigated in
UHV and a high-pressure environment. It is found that, in general, this reaction
proceeds through the Langmuir–Hinshelwood mechanism involving adsorbed
CO and O atoms, and that, at the O2/CO partial-pressure ratio close to the
stoichiometric reaction conditions, the reaction mechanism at high pressures is
similar to that found at UHV conditions [59, 157, 158]. Most recent research is
focused on conditions with large O2/CO partial-pressure ratios [159]. Under
these oxygen-rich conditions, several interesting phenomena have been observed
and ascribed to the formation of a surface oxide on the catalyst surfaces. In the
following sections, first we discuss the reaction mechanism under UHV
conditions and the high-pressure conditions with the O2/CO ratio close to one-
half. Then, we discuss recent studies under oxygen-rich conditions.
9.6.3.1 Carbon Monoxide Oxidation Under UHV
Conditions.
The first step of surface reactions involves the adsorption of reactants. On the
close-packed late transition metal surfaces, CO adsorbs nondissociatively with
initial sticking probabilities close to 1. The adsorption energies on the Pt, Pd, and
Rh close-packed surface range from 33 to 36kcalmol−1 [160]. The relatively low
adsorption energies indicate that, at temperatures ~500 K, the desorption of CO
should be taken into account under steady-state conditions. The adsorption
process is described by Eq. 9.11

(9.11)
where k1 = γpCOSCO is the adsorption rate constant that is proportional to the CO
partial pressure pCO and the sticking probability SCO;γ is a constant relating the
CO pressure to the impingement rate; k2 = v2exp(–E2/RT) is the desorption rate
with v2 the pre-exponential constant; and E2 the desorption activation energy that
is equal to the adsorption energy. The sticking probability is coverage dependent.
Due to the repulsive interaction between neighboring CO molecules, the sticking
probability decreases as the coverage increases, and becomes zero when the
saturation coverage is reached.
Dioxygen adsorbs dissociatively on the late transition metal surfaces [161]. On
the Pt, Pd, and Rh close-packed surfaces, the adsorption energies are 58, 56, and
47kcalper mole of oxygen atoms, respectively. These energies are higher than
CO adsorption energies, so O2 desorption takes place at a much higher
temperature, and under most reaction conditions, the desorption of O2 can
usually be neglected. The higher adsorption energies also indicate that the
adsorbed O atoms are less mobile than the CO molecules. The adsorption
process,

(9.12)
takes two vacant sites on the surface. The initial sticking probabilities are ~0.5
on these metal surfaces. Due to the strong repulsive interaction between
negatively charged O atoms, the sticking probability decreases dramatically as
the O surface coverage increases. Compared to the adsorbed CO molecules, the
distance between neighboring O atoms at saturation coverage is much larger than
that between CO molecules [162–167]. For example, the saturation coverage of
CO on the Pd(111) surface is 0.66 and the distance between neighboring CO
molecules is ~3 Å, while the saturation coverage of O atoms is 0.25 and the
nearest-neighbor distance is 5.5 Å.
The coadsorption of CO and O2 on the surface is rather complicated due to
strong interactions between COads and Oads [168–170]. The rate constants k1, k2,
and k3 are functions of the surface coverage θCO and θO. On the CO precovereed
surfaces, the adsorption of O2 strongly decreases with the increase of CO
coverage. On the (111) surfaces of Pt, Pd, and Rh, the O2 adsorption is
completely prohibited when the CO coverage, θCO ≥ 0.33. If θCO , 0.33 initially,
the adsorption of O2 leads to the separate islands of Oads and COads on the
surface. On O precovered surfaces, the adsorption of CO is not prohibited even
if the surface is initially saturated with Oads. On (111) surfaces, a small amount
of CO adsorption forces the O to compress into small islands, and leads to the
separate islands of Oads and COads. At high CO exposures, CO molecules start to
occupy the vacant sites inside the O islands, and to form a mixed phase where
the two species are in intimate contact. The coadsorption behavior depends on
the adsorption sequence, which can be understood as follows. On a surface
precovered with a high coverage of CO, it is hard to find an open space for the
dissociative adsorption of O2, which requires two vacant neighboring sites. On a
surface precovered with high coverage of Oads, CO can still find a vacant site in
between O adatoms because the O—O distance is quite large even at saturation
coverage.
Figure 9.43 shows in situ STM images of an O precovered Pt(1 11) surface
during its reaction with CO at 247 K [ 171 ]. The Pt( 111) surface is initially
covered with a submonolayer of oxygen that consists of small O islands with the
(2 × 2) O structure (Fig. 9.43a). At the initial stage of CO exposure, the O
islands are compressed to form larger islands, and domains of CO emerges with
no resolvable structure (Fig. 9.43b). Upon further exposure to CO, large CO
islands with the c(4 × 2) CO structure form (Fig. 9.43c). The length of the
boundary between the islands of the two species continuously changes during
the reaction.
There are two possible reaction mechanisms for CO oxidation. First is the
Langmuir–Hinshelwood process, where the two adsorbed species react on the
surface to produce CO2,
 (9.13)
Figure 9.43. (a) The STM image of aPt(111) surface covered with a
submonolayer of oxygen. (b) The STM image after 140-s exposure to 5 ×
10−8mbar CO at 247 K. (c) The STM image after 600-s exposure. Image sizes
are 180 × 170 Å [171].

Carbon dioxide has a negligible adsorption energy, so it desorbs easily once it is

formed on the surface. Second is the Eley–Rideal process, where the CO gas
molecule directly reacts with the adsorbed O atom to form CO2,

(9.14)
Under UHV conditions, modulated molecular beam experiment have been used
to identify the Langmuir–Hinshelwood process as the reaction mechanism [172,
173], a result also confirmed by isotope-labeling experiments [174].
According to the Langmuir–Hinshelwood process, the reaction rate of CO2
production

(9.15)
if the coadsorption phase of CO and O is a random mixture with adsorbed CO
and O in intimate contact. However, under most reaction conditions, the random
mixture phase is hardly achieved. The coadsorption may lead to separate islands
of different species (see Fig. 9.43), and the reaction rate is found to be
proportional to the boundary length of islands on the surface. This result
suggests that the reaction occurs on the boundary of islands, where COads and
Oads are in intimate contact. The traditional way to address this problem is to
assume that the rate constant also depends on the coverages; that is, k = k(θO,
θCO).
A simple reaction rate model can be derived under the assumption that the
dominant surface species is CO [158]. The approximate reaction rate is given by

(9.16)
where A is the temperature independent constant, and ECO,des is the desorption
energy of CO. The rate is first order with respect to the partial pressure of O2,
and negative first order to the partial pressure of CO. The rate equation can be
rationalized as follows. On a CO dominant surface, the rate-limiting step is the
O2 dissociative adsorption on the vacant sites that are left after the desorption of
CO molecules. Thus the rate increases with the O2 partial pressure, and the
apparent activity energy of the reaction is close to the desorption energy of CO.
Furthermore, the increase of CO partial pressure increases the CO coverage, and
decreases the number of vacant sites, causing a decrease of reaction rate.

9.6.3.2 Carbon Monoxide Oxidation Under High-

Pressure Conditions.
The steady-state kinetic studies of CO oxidation have been performed on various
transition metal surfaces under high-pressure conditions, with the total pressure
up to tens of torr. Most of the earlier studies have been performed under the
partial pressure ratio po2 =pCO close to or less than the stoichiometric ratio of 2,
so the dominant species on the catalyst surface is CO [157]. For example, the
CO oxidation on the Rh(111) and Rh(100) surfaces shows first order with
respect to the O2 partial pressure and negative first order to the CO partial
pressure (Fig. 9.44). In the Arrhenius plot (Fig. 9.45), there is excellent
agreement between the two catalyst systems in both the turnover rate and the
apparent activation energy of ~25.4 kcal mol−1, which indicates this reaction
surface-structure insensitive. This surface-structure insensitivity, which was
observed in studies under similar reaction conditions on the Pt, Pd, and Ir
catalysts, can be understood by theby the rate equation given in Section 9.6.3.1
[157].Basically, on the CO dominant surfaces, the reaction is limited by the O
adsorption rate, and the reaction activation energy is similar to the desorption
energy of CO. Since, at high CO coverages, the CO desorption energy does not
vary much with the change of crystal face, the reaction rate is insensitive to
surface structures.
Figure 9.44. (a) Rates of CO oxidation on Rh(111) and Rh(100) as a function of
O2 partial pressure, (b) Rates as a function of CO partial pressure [52]. (TOF =
turnover frequency).

Figure 9.45. Arrhenius plot of the CO + O2 reaction rates on Rh(111) and

Rh(100) surfaces.
9.6.3.3 Carbon Monoxide Oxidation Under Oxygen-
Rich Conditions.
On later transition metals, the increase of the CO oxidation rate has been
observed on increasing the partialpressure of O. Under UHV conditions, this
increase of reaction rate may be attributed to theformation of a mixed phase with
COads and Oads in a close contact [59]. Kinetic studies show the surface reaction
COads + Oads to CO2 can be activated at low temperatures. At high pressures, it
is not clear what exact mechanism is responsible for the high reaction rate. An
exampleof the high reaction rate in the oxygen-rich environment is demonstrated
in Figure 9.46 [ 160]. The reaction is carried out over a Pt(110) surface at 525 K
with an initial po2pCO ratio of 5and a total pressure of 80 Torr in a batch reactor.
Figure 9.46a shows that, at the beginning, theCO and O2 are consumed at a
stoichiometic ratio of 2 to produce CO2. The total pressure is decreasing and
po2/pCO is increasing slowly. At a critical value of po2/pCO ~ 10, the increase of
the ratio suddenly speeds up, and indicates that the reaction proceeds much
faster than at the early stage (Fig. 9.46b). One possible explanation for this
phenomenon is the formation of certain surface oxides as the highly active phase
[ 175–177]. Figure 9.47a shows the critical partial-pressure ratios for Pt, Pd, Rh,
and Ru crystal surfaces. The metal that has the greater adsorption energy
difference between O2 and CO is more easily transformed from the CO covered
surface to the O covered surface, and can achieve a highly active phase at lower
criti cal po2/pCO ratios (Fig. 9.47b).
The molecular-level nature of this highly active surface phase has been studied
by in situ STM and surface X-ray diffraction (SXRD) [53, 54]. In the STM
study, CO oxidation is carried out over a Pt(110) surface under a varying mixture
of CO and O2 at a total pressure of 0.5 bar and a temperature of 425 K. Figure
9.48 shows STM images taken at different partial pressures of a gas species
during the reaction. At the beginning, the CO covered surface consists of smooth
terraces with the (l×l) bulk-terminated structure under a condition where the CO
partial pressure is much higher than the O2 partial pressure. After the gas flow
change to an O2 rich mixture, the CO2 pressure increases, and the reaction
proceeds over the metallic surface at the rate Rmetal, which is evident from the
smooth surface shown in Figure 9.48b. As the reaction proceeds, the O2/CO ratio
increases. At a certain point, the CO2 partial pressure has a sudden jump,
indicating that a more active surface phase has been formed. The STM image
(Fig. 9.48c) at this point shows that a roughening transform occurs, which is
attributed to the formation of surface oxides. The surface roughness increases as
the reaction proceeds at the reaction rate Roxide (Fig. 9.48d). Finally, as the gas
flow changes to the CO rich mixture, the reaction rate decreases and the surface
turns back to the CO covered smooth surface. The in situ SXRD study suggests
that the highly active surface phase is a surface oxide with commensurate
structure (see Fig. 9.49) instead of the incommensurate PtO2 surface oxide.

9.6.3.4 Carbon Monoxide Oxidation Over

Nanoparticles.
Based on many early studies of CO oxidation on different single-crystal surfaces
and supported catalysts, it has been suggested that this reaction is structure
insensitive [52, 178–180]. However, note that many observations of structure
insensitivity were under the conditions where CO is the dominant species on the
surfaces, so that the CO desorption is the rate-limiting step. Recent studies on
Pt(110) under oxygen-rich conditions suggested that the turnover rate can be
significantly higher on a certain surface oxide. Since it is known that the
oxidation state of nanoparticles depends on the size of the nanoparticles (the
smaller the size, the easier the surface of nanoparticles is oxidized). The question
is whether, under the oxygen-rich reaction conditions, the reaction is still
structure insensitive. With advances in synthetic techniques, the size-controlled
nanoparticles with a narrow size distribution have become available. Several
studies on the nanoparticle size dependence of this reaction have been performed
under oxygen-rich conditions. Over the Rh and Pd supported catalysts, it was
found that the turnover rate increases as nanoparticle size decreases [29, 160]. In
one recent study, CO oxidation has been performed over the Rh nanoparticles
with sizes between 2 and 11 nm under an O2/CO ratio of 2.5. A sevenfold
increase of the turnover rate is observed as the nanoparticle size decreases from
11 to 2 nm (Fig. 9.50). The correlation between the turnover rate and the
thickness of the surface oxide was established by ambient pressure XPS (AP-
XPS). As the nanoparticle size decreases, the thickness oxidized layer on the
nanoparticles increases, which correlates well with the increase of turnover rate.
Figure 9.46. A typical experiment to measure the CO oxidation rate by
monitoring the total pressure change. (a) Total pressure decrease (solid line) and
change in the O2/CO ratio (dashed line) as a function of reaction time. (b)
Carbon dioxide formation rate (dotted line) and the change in the partial
pressures of O2 and CO (solid line and dashed line, respectively) as a function of
reaction time. The left vertical line indicates the time point at which CO2
formation rate has a sudden jump as the O2/CO partial pressure ratio approaches
~ 10. The reaction was carried out over a Pd(110) surface at 525 K with an initial
O2/CO ratio of 5 and a total pressure of 80 Torr [160].
Figure 9.47. (a) Carbon dioxide formation rates as a function of the O2/CO ratio
over Pt, Pd, Rh, and Ru crystal surfaces. For Pt, Pd, and Rh, the initial total
pressure is 80 Torr, and the initial O2/CO ratio is 5. The reaction temperatures
are 550, 525, and 525 K, respectively. (b) Plot of adsorption energy differences,
EO – ECO versus ln(po2/pCO) required to achieve the highly active phase [160].
Figure 9.48. The STM images and the partial pressures of the gas-phase species
measured simultaneously during CO oxidation on Pt(110) at a temperature of
425 K in a flow reactor. The STM images are 210 × 210 nm [54].

Figure 9.49. Two possible oxide structures on a Pt(110) surface during the CO
oxidation reaction. The left side is the incommensurate PtO2 surface structure.
The right side is the commensurate surface structure with the O—Pt—CO
complex formed [53].
 Currently, it is believed that a certain oxidation state is responsible for the
turnover rate increase. However, earlier kinetic studies over single-crystal
surfaces (e.g., the result shown in Fig. 9.44) showed that the metal surface may
be deactivated under a high O2 partial pressure due to the formation of the stable
metal oxide. Also note that most of the studies over single-crystal surfaces that
show the highly active surface phase are under transient state conditions rather
than the steady state. On single-crystal surfaces, it is still in question whether the
highly active surface phase exists under steady-state conditions [159].
One interesting aspect of the nanoparticle size effect, which needs further
investigation, is the coadsorption phase of COads and Oads on the nanoparticle
surface. On single-crystal surfaces, COads usually forms islands with sizes
typically >5 nm (Fig. 9.43b). The formation of islands implies that there is a
long-range interaction between adsorbed CO molecules that stabilize this
aggregation. There could be a critical size of the islands below which the
formation of islands becomes energetically unfavorable and formation of a CO
—O mixture phase becomes favorable. If the size of a nanoparticle is small
enough, we would expect that the formation of a CO—O mixed phase is more
favorable than the formation of these CO islands due to limited surface area of
the nanoparticle. The lower activation energy for the CO2 formation from this
mixed phase may be responsible for the increased turnover rate. Particle size
effect on the formation of the nitrogen island formation during CO + NO
reaction over Rh(111) has been studied by Zaera and co-workers [181]. The
studies using in situ STM, molecular beam, and field ion microscopy (FIM) on
the nanoparticles are expected to address this problem in the future [38, 182,
183].
Figure 9.50. (a) Turnover frequency relative to Rh foil at 50 Torr O2, 20 Torr
CO at 200°C, and activation energy for CO oxidation. (b) The thickness of the
oxide shell scales with particle size; 2 and 7-nm nanoparticles are illustrated here
with the oxide layers shown to scale, as determined by AP-XPS [29].

9.6.3.5 Carbon Monoxide Oxidation at High

Temperatures.
Above a critical temperature, CO oxidation may become self-sustaining and
proceeds at a constant high temperature without the need of external heating
because of the high exothermicity of this reaction. This phenomenon is called
“ignition”, and the critical temperature is called the ignition temperature. Above
this ignition temperature, the reaction activation energy become much lower than
that at lower temperatures and implies a reaction mechanism change (Fig.
9.51a). The ignition temperature usually decreases with the increase of the
O2/CO ratio (Fig. 9.51b). Another indication of the reaction mechanism change
is that the reaction orders of O2 and CO at temperatures above the ignition
temperature are different from those at low temperatures. Above the ignition
temperature, half-order dependence on the partial pressure of both reactants has
been observed over Pt(111) with the total pressure kept at 140 Torr (Fig. 9.52),
while below the ignition temperature, the negative-order dependence on CO and
the positive-order dependence on O2 were observed.
In Figure 9.51b, there is an unstable regime for the reaction conditions. When
the reaction is performed in this regime, the turnover rate may oscillate between
a high and a low value [185–187]. Several models have been proposed to
understand this rate oscillation behavior. The oxidation model introduced by
Sales et al. [188], and the carbon model by Chabal and co-workers [189] can be
applied to the reaction under high-pressure conditions [190]. The reconstruction
model has been shown to be the valid mechanism for Pt singlecrystal surfaces at
low pressure [191, 192].
Figure 9.51. (a) Arrhenius plot for the CO oxidation on Pt( 111) at 100 Torr of
O2/40 Torr of CO. Two different activation energies can be observed for the
reaction below and above the ignition temperature. Below the ignition
temperature, the activation energy is ~42 kcal mol−1. Above the ignition
temperature, it becomes 14 kcal mol−1(b) Partial-pressure dependence of the
ignition temperature on Pt(111). The total pressure of CO and O2 is 140 Torr
[184].

Figure 9.52. Reaction order of CO oxidation above the ignition temperature on

Pt(111), (a) for CO, (b) for O2. The total pressure of CO and O2 is 140 Torr
[184].

The ignition temperature is surface-structure sensitive. At 100 Torr of O2 and

40 Torr of CO, the ignition temperatures on Pt(111), Pt(557), and Pt(100) are
~673, 640, and 500 K, respectively [55]. A nanoparticle usually has a lower
ignition temperature and a lower melting point than the single-crystal surface.
Improving the thermal stability of a nanoparticle under the ignition reaction
condition is of practical importance in high-temperature catalysis applications.
For example, the temperature in a three-way catalytic converter may reach 700
K [156]. A promising method is to coat the metal nanoparticles with thermally
stable material (e.g., silica) [193].
Under ignition conditions, oxygen becomes the dominant species on the
surfaces because of the higher heat of adsorption of atomic oxygen. On the
surfaces with high oxygen coverages, the reaction activation energy is quite low.
For example, the activation energy for the reaction on Pt(111) with high oxygen
coverages is ~ 11.7kcalmol−1, which is close to that under ignition conditions
(14kcalmol−1). Due to the high turnover rate under ignition conditions, the
transport of reactants to the surface may become the rate-limited step.
 Recent SFG studies suggested another possible reaction mechanism under
ignition conditions [55, 184]. The SFG spectra shown in Figure 9.53 indicates
that, below the ignition temperature (642 K), the Pt(111) surface is mainly
covered by chemisorbed CO, evidenced by the peak ~2100 cm−1 for the CO
adsorbed on top of the Pt atom. Above the ignition temperature, the peak from
CO on top of the Pt atom disappears, and three broad bands ~2045, 2130, and
2240 cm−1 show up as the turnover rate increases dramatically. The band at 2130
cm−1 can be attributed to the stretch mode of CO adsorbed at oxidized Pt sites,
which has previous been shown not to be important in CO oxidation. The broad
band at 2045 cm−1 can be assigned to the incommensurate CO overlayer together
with the terminally bonded CO at defective sites. The latter could be a multiply
bonded carbonyl cluster-like species [Pt(CO)n, n >1]. The 2240-cm−1 feature has
been assigned to the stretching mode of CO multiply bonded to Pt on an
oxidized surface. By monitoring the coverage change of the incommensurate CO
overlayer at different CO/O2 relative partial pressures, it has been shown that the
reaction rate is proportional to the coverage of this incommensurate CO
overlayer. Therefore, this species must be directly responsible for CO oxidation.
Figure 9.53. Temperature dependence of the SFG spectra of CO oxidation on
Pt(1 11) under 40 Torr of CO/100 Torr of O2/600Torr of He. The significant
increase of turnover rate (TOR) at temperatures >590 K indicates the onset of
ignition [184].
 The nature of the carbon–oxide overlayer has been further investigated on
Pt(111), Pt(100), and the stepped Pt(557) under ignition conditions [55]. The
development of the carbon-oxide overlayer can be characterized by the increase
of SFG background ~2075 cm−1. Figure 9.54 shows that the onset of the ignition
on Pt(557) is accompanied by a sudden increase of the carbon–oxide overlayer.
It has been proposed that the C deposition needed to form this overlayer came
from the Boudouard reaction,
(9.17)
Figure 9.54. Temperature profile and SFG background signal as a function of
time for Pt(557) at a pressure of 40 Torr of CO and 100 Torr of O2. The
temperature and SFG signal were monitored as the crystal temperature was
raised to the ignition temperature. As the ignition took place, a sudden increase
of surface temperature was accompanied by a sudden increase of the SFG
background at 2075 cm−1 [55].

which is facile once the Pt carbonyl is formed on the defects on the Pt surface.
The rapid and highly exothermic gasification of surface C,

(9.18)
may provides another reaction channel for the ignition process. Since the
ignition temperature on the initially C covered Pt(557) surface is lower than that
on the clean Pt surface under the same reaction condition provides addition
evidence that C deposition is an important step for the onset of ignition.

9.6.3.6 Summary.
Research on CO oxidation suggests that this reaction proceeds via the
Langmuir–Hinshelwood mechanism for most transition metal surfaces at
temperatures below ignition and with partial pressure ratio, pCO/po2, close to the
stoichiometric ratio of 2. Under these reaction conditions, this reaction is surface
structure insensitive over most transition metal catalysts. Recent studies under
oxygen-rich conditions on Pt(110) by in situ STM, however, have shown that the
formation of a specific surface oxide phase may result in an increase in the CO
oxidation turnover rate. In addition, the turnover rate over a series of size-
controlled Rh nanoparticle catalysts has been found to increase with decreasing
particle size. This trend correlates with the increase of an Rh surface oxide, as
observed with in situ XPS. Under ignition conditions, recent studies by in situ
SFG have shown that the ignition temperature of this reaction is surface structure
sensitive over Pt single crystal surfaces, which correlates with the surface
structure sensitivity of CO dissociation temperatures over Pt(100), Pt(557), and
Pt(111) (500, 548, and 673 K, respectively). These new findings exemplify the
complexity of the reaction mechanism under relevant reaction conditions and
demonstrate the existence of regimes of both surface structure sensitivity and
surface structure insensitivity for this reaction.

9.7 SELECTIVITY IN MULTIPATH

HETEROGENEOUS CATALYTIC
REACTIONS
As pointed out in the introduction (Section 9.1), the development of new
heterogeneous catalysts for carrying out multipath reactions with high selectivity
in order to gain high-energy efficiency, and meanwhile eliminate undesirable
byproducts, is the goal of catalysis research in the 21 st century [194–197]. The
huge economic impact of developing a catalyst with high selectivity has been
demonstrated by the discovery of zeolites as catalysts for fluid catalytic cracking
of heavy petroleum distillates in 1962 [198]. The new zeolitic catalysts are not
only orders of magnitude more active than the previously used amorphous
silica–alumina catalysts, but they also selectively increase in the yield of
gasoline, the most valuable product of the petroleum refining industry. It has
been estimated that, as a whole, the cost of petroleum refining worldwide would
be higher by at least $10 billion/year, if zeolite catalysts were not available
today.
In the following sections, we will focus on the selectivity of transition metal
catalysts. First, we present the energetic processes involved in the multipath
reactions and discuss how the fundamental energetic parameters may control the
final product distribution. Then, we discuss how the fundamental energetic
parameters can be tuned by atomic level surface properties of metal catalysts
(the surface chemical composition, the surface structure, and the adsorbate-
induced surface restructuring); the reaction conditions (reaction pressure,
reactant pressure ratio, and reaction temperature); and the properties of catalyst
support materials.

9.7.1 Energetic View of a Heterogeneous Catalytic

 Reaction with Multiple Products
Catalysts accelerate the chemical reactions by lowering the height of the
activation free energy barrier along the reaction path. For a multipath reaction,
there is an activation energy barrier associated with each path. The inset of
Figure 9.55 depicts a simplified energy diagram for a reaction with two products.
The absolute turnover rates, R1 = A1 exp(–ΔG1/RT) and R2 = A2exp(–ΔG2/RT),
for two products are controlled by the barrier heights ΔG1 and ΔG2, respectively.
Here, A1 and A2 are the pre-exponential factors, R is the gas constant, and T is
reaction temperature. However, the selectivity to product 1,

(9.19)
is determined by the difference of the two barrier heights, ΔG1 – ΔG2. Here, c =
A2/A1. At 300 K, RT ~0.6 kcal mol−1, so a small variation of the difference of the
two barrier heights may change the selectivity significantly (Fig. 9.55).
The relative activation barrier heights among reaction paths determine the final
product distribution; that is, the selectivity for a reaction over a given catalyst.
These fundamental parameters depend on the electronic structures of adsorbed
reactants on the metal surface, which can be tuned by changing the properties of
the catalysts, such as the chemical composition of the metal surface, the surface
structure of the metal catalysts, and the oxide support of metal catalysts; and the
reaction conditions, such as temperature and the partial pressures of reactants.
We may call these tunable molecular level properties of catalysts and reaction
conditions the controllable factors for catalytic selectivity. The purpose of
selectivity research is to find the optimal combination of these controllable
factors for each reaction of economic interests.
Figure 9.55. Illustration of the key role that relative heights among activation
free energy barriers for different reactions play in determining the selectivity of
catalytic processes. The diagram on the on of a reactant to two possible products.
The rates of those reactions are determined by the absolute activation barriers,
indicated here as ΔG1 and ΔG2, but selectivity between the two is controlled by
the difference between those two values. The calculated selectivity of this two-
reaction system, plotted as a function of ΔG1 – ΔG2 on the left side of the figure,
shows how a variation of the barrier height difference by 2 kcal mol−1 leads to a
switch in selectivity from the exclusive formation of one product to the other
[181].
 It is fair to say that most current research in this field is focused on
establishing the correlation between these controllable factors and reaction
selectivity, and that detailed understanding of how these factors affect the
electronic structures of the adsorbed reactant and further affect the conversion of
reaction intermediates to produce different products, are still limited. Here, we
first discuss some important correlations between controllable factors and
reaction selectivity established on the model catalytic systems. Current advances
in the nanoparticle synthesis that allows us to study the selectivity of size- and
shape-controlled catalysts will be emphasized.

9.7.2 Surface Structure and Selectivity

Platinum is the most important catalyst for hydrocarbon reforming in industry.
Hydrocarbon reforming on a Pt surface yields several products (Fig. 9.2). How
does the reaction selectivity depend on the atomic structure of the Pt catalyst
surface? To answer this question, reaction rate studies using flat, stepped, and
kinked single-crystal surfaces with a variable surface structure are very useful.
For the important aromatization reactions of n-hexane to benzene and n-heptane
to toluene, it was discovered that the hexagonal Pt(111) surface, where each
surface atom is surrounded by six nearest neighbors is three to seven times more
active than the Pt(100) surface with a square unit cell [100, 199]. Aromatization
reaction rates increase further on stepped and kinked Pt surfaces. Maximum
aromatization activity is achieved on stepped surfaces with terraces about five
atoms wide with a hexagonal orientation, as indicated by reaction rate studies
over >10 different crystal surfaces with varied terrace orientation and step and
kink concentrations (Fig. 9.56).
Figure 9.56. The structure sensitivity of dehydrocyclization of alkanes to
aromatic hydrocarbons. The bar graphs compare reaction rates for n-hexane and
n-heptane catalyzed at 573 K and atmospheric pressure over the two flat Pt
single-crystal faces with different atomic structure. The Pt surface with a
hexagonal atomic arrangement is several times more active than the surface with
a square unit cell over a wide range of reaction conditions [199]

The reactivity pattern displayed by Pt crystal surfaces for alkane isomerization

reactions is different from that for aromatization. Studies revealed that maximum
rates and selectivity (rate of desired reaction/total rate) for butane isomerization
reactions are obtained on a flat crystal face with a square unit cell. Isomerization
rates for this surface are four to seven times higher than those for the hexagonal
surface. Isomerization rates are increased to only a small extent by surface
irregularities (steps and kinks) on the Pt surfaces (Fig. 9.57).
For the undesirable hydrogenolysis reactions that require C—C bond scission,
the two flat surfaces with highest atomic density exhibit very similar reaction
rates. However, the distribution of hydrogenolysis products varies sharply over
these two surfaces. The hexagonal surface displays high selectivity for scission
of the terminal C—C bonds, whereas the surface with a square unit cell always
prefers cleavage of C—C bonds located in the center of the reactant molecule.
The hydrogenolysis rates increase markedly (three- to fivefold) when kinks are
present in high concentrations on the Pt surfaces.
Because different reactions are sensitive to different structural features of the
catalyst surface, we must prepare the catalyst with the appropriate structure to
obtain maximum activity and selectivity. The terrace structure, the step or kink
concentrations, or a combination of these structural features, is needed to
achieve optimum reaction rates for a given reaction. Studies indicate that H—H
and C—H bond-breaking processes are more facile on stepped surfaces than on
flat crystal faces, while C—C bond scission is aided by kink sites that appear to
be the most active for breaking any of the chemical bonds that are available
during the hydrocarbon conversion reactions. Because molecular rearrangement
must also occur, in addition to bond breaking, it is not surprising that the terrace
structure exerts such an important influence on the reaction path that the
adsorbed molecules are likely to take. The difference in chemical behavior of
terrace, step, and ledge atoms arises not only from their different structural
environment, but also from their different electronic charge densities that result
from variation of the local atomic structure. Electron spectroscopy studies reveal
altered density of electronic states at the surface irregularities; there are higher
probabilities of electron emission into vacuum at these sites (lower work
function), indicating the redistribution of electrons [200]. The d-band theory has
been proposed to understand the ability of surface irregularities to break strong
chemical bonds.
Figure 9.57. The structure sensitivity of light alkane isomerization and
hydrogenolysis. Shown here are the reaction rates of isobutane catalyzed at 570
K and atmospheric pressure over four Pt surfaces. Isomerization is favored over
Pt surfaces that have a square atomic arrangement. Hydrogenolysis rates are
maximized when kink sites are present in high concentrations on the Pt surface
[199].
 Industrial catalysts are nanoparticles. Their size and shape are difficult to
control by conventional synthesis methods. With the advances in nanoparticle
synthesis, metal catalysts with narrow size distribution and different shapes can
now be produced routinely. These advances give us opportunities to control the
selectivity by changing the shape and size (or the density of surface
irregularities) of metal nanoparticles. The study of the benzene hydrogenation
reaction on the shape-controlled Pt nanoparticles gives a good example in this
regard [201]. The turnover rates given in Figure 9.58 shows that cyclohexane is
the only product on the (100)-facet-rich cubic Pt nanoparticles, while, both
cyclohexene and cyclohexane can be produced on the (111)-facet-rich
cuboctahedra Pt nanoparticles (Fig. 9.58b). This finding correlates well with
selectivity on the single-crystal (111) and (100) surfaces (Fig. 9.58a) [202].
It has been observed in the selectivity study of cyclohexene hydrogenation and
dehydro-genation (Fig. 9.59a) on single-crystal Pt (111) and (100) surfaces under
10-Torr cyclohexene and 100-Torr hydrogen that the open (100) surface
produces more benzene, the dehydrogenation product, than the close-packed
(111) surface [81]. This trend was also observed on Pt nanoparticles, as the
particle size becomes smaller (Fig. 9.59b) [203]. As the particle size becomes
smaller, the density of the sites with low coordination increases on the
nanoparticle surfaces. These two examples demonstrate that the surface
structural dependence of selectivity on the nanoparticles should be, in general,
similar to that of single crystal surfaces
Figure 9.58. Turnover rates of cyclohexene and cyclohexane formation during
the hydrogenation of benzene on (a) Pt(1 11) and Pt(100) surfaces, as well as (b)
the cubic and cuboctahedra Pt nanoparticles [201].
Figure 9.59. (a) Cyclohexene hydrogenation and dehydrogenation. (b) The
dependence of the selectivity of cyclohexene hydrogenation and
dehydrogenation on the size of Pt nanoparticles. The reaction conditions are 10-
Torr cyclohexene and 200-Torr hydrogen, at 423 K [203].

9.7.3 Alloy Catalysts and Selectivity

Many transition metal catalysts can promote hydrocarbon hydrogenation, but
they usually exhibit different catalytic selectivity. Nickel, for example, favors
undesirable cracking products in hydrocarbon reforming, while Pt based
catalysts lead to the selective production of useful reforming products. The
selectivity of Pt based catalysts can be finely tuned by alloying them with a
second metal (e.g., Rh, Au, or Ir) [204–206].
Compared to pure Pt, bimetallic alloys (e.g., Pt–Rh and Pt–Au) frequently
exhibit superior activity, selectivity, and deactivation resistance while catalyzing
reforming reactions. The influence of Au on hydrocarbon conversion catalysis
by Pt was studied by evaporating Au onto Pt single-crystal surfaces [207]. At
low temperatures, Au forms epitaxial overlayers on Pt, but upon heating it
dissolves to form an alloy in the near surface region. This Pt–Au alloy displays
markedly different activity and selectivity for the conversion of n-hexane (see
Fig. 9.60). Isomerization activities increase substantially as compared to those
for clean Pt, whereas the aromatization and hydrogenolysis rates decrease
exponentially with increasing Au surface concentration. This remarkable change
in catalytic behavior can be explained by a change in the geometric distribution
of Pt sites that are present in the (111) alloy surface. Substitution of Au atoms
dilutes the surface Pt atoms such that the high-coordination threefold Pt sites are
eliminated much faster than the twofold bridge and single-atom top sites. This
change in the distribution of the available reaction sites is frequently called the
ensemble effect [207]. As a result of this effect, catalyzed reactions that involve
adsorption and rearrangement at threefold sites are eliminated, whereas reactions
that require one or two atom sites are attenuated to a much lesser extent.
Although minor changes in electronic structure may also occur at the alloy
surface sites, most of the reaction results can be explained by this high-
coordination-site elimination model. Similar results revealing pronounced
changes in catalytic behavior with alloy composition were reviewed by Ponec
[204] and Sinfelt [205]. In most cases, the geometrical ensemble effect is
decisive in controlling the reaction selectivity for a variety of hydrocarbon
reactions that are catalyzed over metal films and high-area-supported catalysts.
Figure 9.60. Rates of formation of various products from n-hexane conversion
as a function of fractional Au coverage for Au–Pt alloys that were prepared by
vaporizing Au onto Pt(111) and Pt(100) crystal surfaces, respectively [20, 207].
 The effect of alloying is also surface-structure sensitive, as shown by studies
where Au was the alloying constituent in the Pt(100) crystal face instead of the
Pt(111) surface [208]. The (100) surface has a square unit cell that contains
fourfold bridge and top sites, and unlike the (111) surface it does not have
threefold sites. When this surface is alloyed with Au, all reaction rates decline in
proportion to the concentration of inactive Au on the Pt(100) surface, when n-
hexane was used as a reactant. This is shown in Figure 9.60. Thus the
enhancement of the isomerization activity requires a presence of threefold sites.
When Au is used as an alloying agent, there are three types of threefold sites
available. One contains only Pt atoms, whereas the other two mixed Pt—Au sites
contain one and two Au atoms, respectively. Thus alloying produces new mixed-
metal sites with catalytic behavior that can modify the selectivity. Figure 9.60
clearly indicates that the high isomerization rate of n-hexane is sustained until
the surface was covered by up to two-thirds of the Au monolayer [208]. Thus all
three threefold sites are active for isomerization. The mixed Pt—Au sites are
then responsible for the enhanced isomerization activity of the Pt—Au alloy.
 The large possible combinations of different metals for alloy catalysts provide
opportunities to find new low-cost materials to replace precious metal catalysts.
In this regard, the theoretical model developed by Norskov and co-workers is
playing an increasingly important role in searching for alloy combinations that
exhibit desirable selectivity for given reactions [209–212]. This model
characterizes the electronic structure of alloy surfaces by a small number of
descriptors (bonding energies of key reaction intermediates, activation energy
barriers, and pre-exponential factors), which can be calculated by density
functional theory (DFT) and scaling relations [213]. By using these descriptors,
the heights of activation barriers for various reaction paths then can be predicted
by the scaling relations, such as the Brønsted–Evans–Polanyi (BEP) relation.
Finally, the energy potential surface for a multipath reaction can be constructed,
and the selectivity of the alloy surface can be predicted. The major advantage of
this model is its high computational efficiency, since the time-consuming DFT
calculation is only needed for a small number of descriptors. These descriptors
can be employed to screen large amount of bimetallic combinations with
affordable computational effort. Recently, this model has successfully helped in
identifying a nonprecious alloy catalyst, NiZn, for selective hydrogenation of
trace C2H2 in an C2H4 stream [214]. Compared to the current industrial PdAg
catalyst, the new alloy material provides a huge cost-efficiency improvement.

9.7.4 Adsorbate-Induced Surface Restructuring

Introduction of a small amount of surface impurity may help the selectivity of
catalysts. For example, the strongly adsorbed adsorbates can be used to block the
steps and kinks on the catalyst surface, and to reduce the undesirable cracking
products from hydrocarbon reforming. In general, the restructuring of the surface
induced by strong adsorbates makes it difficult to establish the correlations
between reactivity or selectivity and surface structure, as well as metal alloy
composition.
In the chapters discussing surface structures and surface chemical bonding, we
have shown a number of examples of how the strongly adsorbed intermediates
can displace the surface metal atoms. These adsorbate effects are expected to be
more prominent on nano-particle surfaces, since the density of low-coordination
sites on the nanoparticle surfaces are higher than that on the single-crystal
surfaces.
For the bimetallic surfaces, surface metal composition is sensitive to the
chemical properties of the adsorbates. Under high-vacuum conditions, in situ
XPS studies on the Pt—Rh crystal surface have demonstrated that the adsorption
of hydrogen induces the increase of Pt content on the surface, and that the
adsorption of O increases the Rh content (Fig. 6.19).
Under high-pressure conditions, the adsorbate-induced change of surface
composition on nanoparticles is more dramatic [215]. Figure 9.61 shows that the
evolution of surface composition of Pt—Pd nanoparticles under oxidizing
conditions (100mTorr NO or O2) and catalytic conditions (100mTorr NO and
100mTorr CO) at 300°C. In the top part of Figure 9.61, red diamonds (big blue
dots) indicate the rhodium (palladium) atomic fractions in the near-surface
region under different chemical environments. Under the oxidizing conditions,
the near-surface region is rich in Rh. Under the catalytic conditions, the surface
layer has a Rh–Pd composition close to 0.5/0.5. A variation as much as 30% of
the metal surface composition can be seen in Figure 9.61. This prominent
variation is driven by the change in surface energy. Rh oxide has a lower surface
energy than Pd oxide. Under oxidizing conditions, the formation of Rh oxide in
the near-surface region is thermodynamically favorable as shown by red
diamonds in the bottom part of Figure 9.63. Introduction of CO reduces the Rh
oxides to Rh metal. In the meanwhile, the atomic fraction of oxidized Pd (big
blue dots in the bottom part) does not change significantly. The reduced Rh
atoms diffuse into the bulk in order to lower the surface energy of the alloy
surface since Rh metal has a much higher surface energy than the Pd oxide.
Figure 9.61. (Top) Evolution of Rh and Pd atomic fractions in the near-surface
region of the Rh0.5Pd0.5 NPs with the size ~15nm at 300°C under oxidizing
conditions (100mTorr NO or O2) and catalytic conditions (100mTorr NO and
100mTorr CO) denoted in the x-axis. (Bottom) Evolution of the fraction of the
oxidized Rh (left y-axis) and Pd atoms (right y-axis) in the examined region
under the same reaction conditions as the top part of the figure. All atomic
fractions in this figure were obtained with an X-ray energy of 645 eV for Rh3d
and Pd3d, which generates photoelectrons with a MFP of ~0.7 nm. Schematic
diagrams above the top of the figure show the reversible segregation of Rh and
Pd under alternating oxidizing and catalytic conditions [215]. (See color insert.)
 All above examples indicate that it is extremely difficult to retain certain
surface structures or surface composition of catalysts optimal reactivity and
selectivity under practical reaction conditions. In order to stabilize the preferred
surface structure under harsh reaction conditions, the common methods are using
a specific oxide as the catalyst support and a small amount of additive. A good
example in this regard is the Fe based catalysts for NH3 synthesis. One way of
retaining a certain surface composition of bimetallic catalysts is to use the alloys
with a large miscible gap so that one metal always prefers to segregate to the
surface, even under high temperatures. For example, AuNi catalysts exhibit a
stable surface composition with Au segregating to the surface during the steam
reforming process (mainly, CH4 + H2O → CO + 3H2) at high temperatures
[216]. The presence of Au atoms in the surface layer reduces the adsorption
energy of the C atom significantly, and, therefore suppresses the graphite
formation on the surface that can poison the stream reforming on the pure Ni
surface. Another way to stabilize the surface alloy composition is to prepare
nano-particles with their size small enough so that the majority of atoms are
surface atoms, and then surface segregation has a minimal effect on the surface
composition. This method seems more versatile and can be applied to the alloys
with miscible components.

9.7.5 Strong Metal Support Interaction

Metal catalysts dispersed on reducible oxide supports usually exhibit
enhancement of catalytic activity and selectivity. This effect was observed first
by Schwab and is commonly referred to as strong metal-support interaction
(SMSI) [217–221]. Tauster et al. reported a large enhancement of activity of Ni
catalyst supported on TiO2 for CO hydrogenation [219, 220]. Figure 9.62 shows
the turnover rates to CH4 (the filled bars), the turnover of hydrocarbon with two
or more carbon atoms (the unfilled bars), and the total turnover rate for CO
conversion (the total bar height) over Ni catalysts supported by various oxides.
The fact that the Ni/Al2O3 and Ni/TiO2 catalysts have 3–30 times greater
turnover rates than that of unsupported Ni and Ni supported on inert SiO2
provides unambiguous evidence that SMSI increases the activity of Ni for CO
hydrogenation. The SMSI also has a strong effects on product selectivity. The
turnover rates of hydrocarbons with two or more carbon atoms are significantly
higher on the catalysts with SMSI.
Figure 9.62. Effects of support on the CH4 turnover rate and selectivity at 525 K
on Ni catalysts: the filled bar is for CH4 turnover, the unfilled bar is for the
turnover of hydrocarbons with two or more carbon atoms and the total bar height
for CO conversion [222].
 The SMSI effect has also been observed on the catalysts with inverted
structure (Fig. 9.63a); that is, with the metal surface partially covered by
reducible oxides. Because the same catalytic behavior can be obtained by
depositing the metal on the oxide support or by deposition of oxide islands on
the transition metal, the oxide–metal periphery area is implicated as the active
site responsible for the increased reaction rates. A typical reaction rate behavior
exhibits a maximum with increasing oxide coverage over a transition metal
catalyst (see Fig. 9.63b) for CO2 hydrogenation over TiO2 on Rh. The oxide
alone is inactive while the metal is active for CH4 formation. At ~50% of a
monolayer of oxide coverage, which corresponds to the optimum oxide–metal
interface area, the reaction rate exhibits a maximum [223, 224].
A study has been carried out to compare the promotion of CO2 hydrogenation
over a Rh foil decorated with submonolayer quantities of AlOx, TiOi, VOx, FeOx,
ZrOx, NbOx, TaOx, and WOx. Based on the extent of turnover enhancement at
the half-monolayer coverage of oxides, these oxides can be divided into three
groups: TiOx, TaOx, and NbOx showing the highest promotion effect (~12-fold
enhancement compared to the pure Rh foil); ZrOx, VOx, and WOx showing the
modest promotion effect; and FeOx with no enhancement at all. These results
indicate a correlation between the Lewis acidity of oxides and the enhancement
effect: the stronger the Lewis acidity of the oxide, the greater the activity of the
oxide–metal interface catalyst. Another interesting correlation can be established
when looking at the band gaps of these oxides (see Table 9.52 on page 753).
Compared to the three oxides showing the highest enhancement, the oxides with
modest or no enhancement have the band gaps either too high or too low. The
band gap of oxide determines the height of the Schottky barrier at the metal–
oxide interface, which further determines the electron-transfer behavior at the
interface. As we will discuss later in this section, the electron transfer at the
metal–oxide interface has been proposed as the origin of the SMSI effect.
Figure 9.63. (a) Inverted metal–oxide catalyst. (b) Carbon-dioxide
hydrogenation turnover rate over the Rh/TiO2 catalysts as a function of oxide
coverage on the metal [223].

For some multipath reactions, the active sites on the oxide-metal interface area
may produce a reaction product different from the product produced on the pure
metal sites. The atomic fraction of atoms in the oxide–metal interface area
increases as the size of the metal catalyst decreases. Thus, it is expected that the
selectivity of these multipath reactions can be tuned by changing the size of the
metal catalysts [227]. For example, the selectivity of ring opening of
methylcyclopentane over Pt/MgO catalysts shows a very strong dependence on
the size of Pt particles, which is not observed over Pt on the inert SiO2 support
(Fig. 9.64). The increase of n-hexane as the particle size decreases can be
attributed to the preferential formation of n-hexane at the oxide–metal interface,
which has been confirmed by a series of experiments in which the interfacial
area was deliberately poisoned by NH3, CO, and so on [228, 229].
In the oxide–metal interface area, the metal may easily grab O atoms from the
reducible oxide to form certain partially oxidized states, and to leave oxygen
vacancies on the oxide surface. The electronically modified metal atoms in the
interfacial area or the oxygen vacancies on the oxide surface can serve as active
sites for the formation of specific products [230, 231]. For some reactions, it is
believed that the presence of both of them is necessary [232].
Compared with the pure metal sites, the adsorption energies of reactants are
usually changed significantly on these interfacial sites due to their unique
electronic structures. It was proposed by Schwab and Koller that the SMSI effect
can be traced back to an electron exchange between the support and the catalyst
[233]. This idea was developed further by Akubuiro and Verykios, who
suspected that the Schottky-like barrier at the metal–oxide interface may be
responsible for the formation of the unique electronic structure at the interface
[225]. Recently, the successful fabrication of a catalytic nanodiode, provided for
the first time a way to measure the electron transport through the metal–oxide
interface during exothermic catalytic reactions [234–236]. By establishing the
correlation between the electron transport and the reaction activity and
selectivity for different reactions, the studies using catalytic nanodiodes are
expected to provide a more mechanistic understanding of the SMSI effect [237,
238].
Figure 9.64. Ring opening of methylcyclopentane in excess hydrogen at 513 K:
turnover ratio of branched hexanes [2-methylpentane (2-MP) and 3-
methylpentane (3-MP)] to n-hexane versus mean-particle size of Pt deposited on
silica, alumina, magnesia, and acidic alumina supports. The strong size
dependence of the selectivity was observed on the catalysts with strong metal
support interactions [225, 226].
 9.7.6 Oxidation States of Metal Catalyst and
Selectivity
A partially oxidized metal surface may serve as an active phase for catalytic
reactions. One example is the highly active surface oxide phase during a CO
oxidation reaction that we discussed earlier in Section 9.6.3.3. From the
thermodynamics point of view, metal nanoparticles are more easily oxidized as
their size becomes smaller. Figure 9.65 shows XPS spectra of Pt nanoparticles
with a size of 1.0 and 1.5 nm. In the smaller particles, ~60% of the Pt atoms are
oxidized, while only ~10% of Pt atoms are oxidized in the bigger nanoparticles.
The oxidation state of these Pt nanoparticles is believed to play a key role in the
size dependence of the selectivity in pyrrole hydrogenation (Fig. 9.66) [239].
The pyrrole hydro-genation to pyrrolidine is followed by ring opening to n-
butylamine, and the scission of the N—C bond to form butane and ammonia. On
single crystal Pt surfaces, this reaction is known to be poisoned by strongly
adsorbed n-butylamine. The N of n-butylamine is more electron-rich than other
intermediates such as pyrrolidine or pyrroline and thus can form stronger bonds
with the surface and consequently inhibit turnover. Figure 9.66 shows the n-
butylamine formation is more facile over larger Pt nanoparticles. The pyrrolidine
formation occurs more easily on the smaller Pt nanoparticles. The change of the
selectivity becomes significant as the size of nanoparticles is reduced below
about 2 nm, which correlates with the significant increase in the oxidation state
of these smaller nanoparticles. It has been suggested that the higher oxidation
state on the small nanoparticles reduces the adsorption energy of n-butylamine
and consequently increases the surface coverages of weakly adsorbed
intermediates for pyrrolidine formation.

9.7.7 Reaction Intermediates

During catalytic reactions, various reaction intermediates may coexist on the
surface. Some of them are not detectable in the gas phase. For example, Figure
9.67a shows all the reaction intermediates possibly converted from the
chemisorbed iso-butyl by elementary reaction steps. For each reaction product,
there usually exists a reaction intermediate that is directly responsible for its
formation [240]. This reaction intermediate is called the reactive intermediate for
the product. In order to control the selectivity to a given product, we must know
what the reactive intermediate is for this product, and how to enhance the surface
conversion step for this intermediate by tuning the properties of the catalysts.
Figure 9.65. X-ray photoelectron spectra of (a) the 1.0-nm, and (b) the 1.5-nm
Pt nanoparticles. The smaller Pt nanoparticles are in higher oxidation states than
the bigger particles. The dentrimer Pt sample is synthesized by a method
demonstrated in Figure 9.8. The capping agent of the 1.5-nm Pt nanoparticles is
poly(N-vinyl-2-pyrrolidine) or PVP.

Figure 9.66. (Top) The reaction scheme for pyrrole hydrogenation. (Bottom)
The nanoparticle size dependence of selectivity to the reaction products of
pyrrole hydrogenation [239]. (See color insert.)
 Under the UHV conditions, TPD combined with techniques such as HREELS
and reflection–absorption infrared spectroscopy (RAIRS) can be used to identify
possible reaction intermediates and to correlate the intermediates with the
desorbed product at a different temperature [181]. For multipath reactions, the
resulting vibrational spectra of the surface species are usually very complex,
andinterpretation of the spectra becomes difficult [195]. In this regard, the
vibrational frequencies of surface intermediates predicted by DFT calculations
become more and more useful for spectral interpretation [241,242]. Once the
spectral signatures of different intermediates have been determined, they serve as
references for the reaction intermediates under high-pressure reaction conditions.
Figure 9.67. (a) Potential elementary steps available to alkyl moieties (iso-butyl
in this example) when chemisorbed on metal surfaces. All these reactions, with
the exception of b-methyl elimination, have been seen in surface science studies.
(b) Elementary hydride elimination steps lead to different reforming products
[240].
 Studies on the intermediates during hydrocarbon reforming suggested that the
β-hydride elimination typically dominates the chemistry of adsorbed alkyl
groups on most transition metal surfaces. The reaction step leads to the
formation of alkene products (Fig. 9.67b). Compared to other metal catalysts, Pt
based catalysts have the unique ability to carry out α- and γ-hydride elimination,
which leads to hydrogenolysis, cyclization, and isomerization products. This is
the reason Pt is the most versatile catalyst for selective reforming.
In situ monitoring of the intermediates during catalytic reactions under high-
pressure conditions can be performed by SFG and polarization-modulated
RAIRS (PM–RAIRS) [243–245]. Figure 9.68a shows the SFG spectra of
cyclohexene hydrogenation and dehydro-genation over Pt(111) under 10Torr of
hydrogen and 1.5Torr of cyclohexene [107]. There are three intermediates: 1,4-
cyclohexadiene, π-allyl c-C6H9, and 1,3-cyclohexadiene (C6H8), which are
detectable at different temperature ranges. At 303 K, 1,4-cyclohexadiene is the
major surface species and small features from 1,3-cyclohexadiene are also
visible. In the temperature range from 303 to 400 K, the turnover rates in Figure
9.68b show cyclo-hexane as the major product.
The corresponding major surface intermediate is π-allyl c-C6H9. At higher
temperatures, the onset of 1,3-cyclohexadiene formation on the surface is
correlated with the production of benzene in the gas phase, which suggests that
1,3-cyclohexadiene is the reactive intermediate.
The selective hydrogenation of the C=O carbonyl group in the presence of a
C=C bond can produce the unsaturated alcohol, an important intermediate in the
pharmaceutical and fragrance industries. Figure 9.69 compares the selectivity of
the hydrogenation of cro-tonaldehyde and prenal over Pt(111) [259]. It is shown
that the formation of the unsaturated alcohol products are more favored in the
hydrogenation of prenal than of crotonaldehyde. The bulky heads of two CH3
groups in prenal may weaken the di-σ(CC) surface bonds of intermediates and
cause the difference in selectivity. Indeed, the SFG spectra of the Pt(111)
surfaces during the two reactions show marked differences (Fig. 9.70). During
the hydrogenation of crotonaldehyde, the major intermediate is the η2-cis-
crontoaldehyde species that is characterized HETEROGENEOUS by peaks at
2860, 2915, 2965, and 2995 cm−1 (Fig. 9.70a). No O–H stretching mode is
observed in this case. During the hydrogenation of prenal (Fig. 9.70b), the O–H
stretch mode is visible as a broad peak centered at 3380 cm−1, which indicates
the existence of chemisorbed unsaturated alcohol on the surface. The existence
of the η3-di-σ(CC)-σ(O)-cis species or η4-di-σ(CC)-di-σ(CO)-cis species, that
form three or four bonds with the surface are indicated by the peak at 2945 cm−1
at higher temperatures. The fact that the O atom in the highly coordinated η3 and
η4 species is close to the surface may cause the preferential hydrogenation of the
C=O bond to selectively produce the unsaturated alcohol product in the
hydrogenation of prenal.
Figure 9.68. (a) The reaction intermediates detected by SFG during cyclohexene
hydrogenation and dehydrogenation over Pt(111) at different temperatures. The
reaction conditions are 1.5 Torr for cyclohexene and 15 Torr for hydrogen. (b)
The turnover rates for benzene and cyclohexane at different temperatures.
Correlation between surface intermediates and formed reaction products may
help identifying the reactive intermediates for the products [107]. (See color
insert.)

Figure 9.69. Selectivity of hydrogenation of (a) crotonaldehyde and (b) prenal

on Pt( 111) at different temperatures. Open squares for unsaturated aldehyde
products, open circles for unsaturated alcohols, filled diamonds for cracking
products, and filled triangles for completely hydrogenated products. Compared
to the hydrogenation of crotonaldehyde, the hydrogenation of prenal is more
selective to the desired unsaturated alcohol product [246].

Figure 9.70. The SFG spectra of Pt(111) during hydrogenation of (a)

crontonaldehyde and (b) prenal. The major intermediate of crontonaldehyde
hydrogenation is the η2-di-σ(CC)-cis species shown by the ball and stick model
in (a). During the hydrogenation of prenal, the chemisorbed unsaturated alcohol
is detectable on the surface [see the ball and stick model in (b)][246].
 9.7.8 Reaction Conditions and Selectivity
Reaction conditions, such as temperature and reactant pressures, are important
parameters to control reaction selectivity. As shown in Figure 9.68b, the product
distribution of cyclohexene hydrogenation and dehydrogenation changes
significantly with the reaction temperature. At lower temperatures, cyclohexane
is the main product. At higher temperatures, the dehydrogenation product
(benzene) becomes dominant. The selectivity of this reaction also has very
strong reaction pressure dependence even when the partial-pressure ratio of
hydrocarbon to hydrogen is fixed. On the Pt(223) stepped surface and with the
partial-pressure ratio of cyclohexene to hydrogen fixed at 1:10, it has been
observed that cyclohexane is the predominant product under a total pressure .>1
Torr. When the pressure is 10−4 Torr, C6H6 becomes an almost exclusive product
[247].
In the discussion of CO oxidation of metal surfaces in Section 9.6.3, we have
shown how reaction conditions affect the reaction activity through changing the
coverages and distributions of adsorbates. During catalytic hydrocarbon
conversion processes under atmospheric pressures, the clean metal surfaces
become covered by carbonaceous species at the initial stage of the reactions.
These strongly bonded surface species may block the surface sites with low
coordination, and induce surface restructuring, and change the chemisorption
characteristics of reactants. Currently, the exact role of this overlayer in changing
the reforming selectivity is not fully understood. In the hydrogenation of olefins,
alkylidynes are the strongly adsorbed species that form the carbonaceous
overlayer. Alkylidynes are not directly involved in hydrogenation and
dehydrogenation processes at low temperatures [248]. At high temperatures,
these species may be directly involved in hydrogenolysis and reforming
processes [249, 250].
During the conversion of cyclohexene over Pt(1 11), the reaction probability of
hydrogenation to cyclohexane is approximately independent on the partial
pressure of cyclohexene. Meanwhile, the reaction probability of
dehydrogenation to C6H6 decreases significantly with the partial pressure
increase. Thus, it can be argued that, compared to hydrogenation process, the
dehydrogenation process requires that surface ensembles consist of a relatively
larger number of empty sites. The development of a carbonaceous overlayer at
high hydrocarbon pressures blocks the active surface sites for dehydrogenation,
and leads to the marked change in selectivity.
For a long time, it has been recognized that one of the key functions of metals
as catalysts in hydrogenation, dehydrogenation, and skeletal rearrangement
reactions, is to activate molecular hydrogen to produce chemisorbed atomic
hydrogen [251, 252]. The dissociation adsorption of hydrogen on the metal
surface can be impeded by the formation of a carbonaceous overlayer. Currently,
it is not clear how atomic hydrogen participates in thehydrocarbon conversion
processes, but there is no doubt about hydrogen pressure as a key parameter for
tuning the selectivity of hydrocarbon conversions.

9.7.9 Other Important Research Directions to

Catalytic Reaction Selectivity
Heterogeneous catalysis is a very rich research area. Currently, various
promising approaches are under intensive research in order to achieve high
selectivity. Here, we briefly mention several of these approaches. Zeolites are
silica-lumina-based microporous materials containing channels and cages with
dimensions of 0.2 to 1 nm. The pores of zeolites and molecules interacting with
the surface of zeolites have dimensions of the same order of magnitude. This
geometric confinement leads to unique effects of zeolites in catalysis. A generic
term shape selectivity is in use today to refer these effects [198]. The reactant
molecules with sizes greater than the pore size of a zeolite cannot access the
internal surface of the zeolite to take part in reactions; the reaction paths
involving large-size reaction intermediates, or producing large-size product
molecules, are also prohibited.
In catalytic reactions, a specific chemical bond in a reactant molecule can be
preferentially attacked to form the desired product. This type of selectivity is
called regioselectivity. An example is acrolein hydrogenation (see Fig. 9.71).
The C=C and C=O bonds in acrolein can be hydrogenated to form
propionaldehyde and allyl alcohol, respectively [253, 254]. The unsaturated
alcohol product is an important intermediate in both the pharmaceutical and
fragrance industries. On Pt group metals, the selectivity to the desired
unsaturated alcohol product, allyl alcohol, is usually relatively low. The
regioselectivity of C=O bond hydrogenation can be improved upon alloying it
with another metal, as well as upon the addition of various promoters (e.g., K)
[254, 255].
The chiral selection of reaction product (usually referred as enantioselectivity)
is an important issue in the research to understand the chiral nature of
biomolecules and in industries ranging from synthetic drugs to the
biodegradation of packaging materials [258, 259]. Compared to extensive
studies in homogeneous catalysis, enantioselectivity in heterogeneous catalysis,
a topic emerging in the past decade, is a relatively new field. To achieve
enantioselectivity in heterogeneous catalysis a catalyst surface can be prepared
with specific chirality so that only reactants with certain chirality can be
preferentially adsorbed on the surface and the product with specific chirality can
be formed. The chiral surfaces can be prepared by irreversibly adsorbing chiral
molecules over an achiral surface, where the chiral adsorbates impart chirality on
the surface (Fig. 9.72a); [258] or by cleaving a crystalline material to expose
crystal faces with a specific chirality (Fig. 9.72b) [257, 260].
Figure 9.71. Hydrogenation pathways for acrolein.

Figure 9.72. (a) The 108 ×108-Å STM images showing mirror chiral surfaces
created by (R,R)- and (S,S)-tartaric acid adsorbed on Cu(110) [202]. (b) Three-
dimensional ball models of the Ag(643) surface and its mirror image, the
Ag(643) surface [203].
 Currently, metal oxide catalysts are being explored for carrying out
hydrocarbon conversion selectively [261]. Metal oxide catalysts are used in
oxidative dehydrogenation of alkanes to alkenes and selective partial oxidation
of hydrocarbons to oxygenated products. The challenge in these applications is
to avoid the generation of CO2 from the complete oxidation of desired products.
The activity and selectivity of metal oxide catalysts can be correlated with
structure, the oxidation states of the metal, and basicity of the oxide support
[262-265.

9.7.10 Summary
Achieving high catalytic selectivity is the goal of catalysis science in 21st
century. For transition metal catalysts, we have shown that the advances in
nanotechnology enable us to control the concentration of the active sites for
different reaction products by tuning the size and shape of nanoparticles. The
active site for a given product could be a certain atomic arrangement of metal
atoms, the surface atom in a certain oxidation state, bimetallic surface site with
specific composition, or the interface between the metal nanoparticle and the
oxide support. Under reaction conditions, active sites may reconstruct due to the
elevated temperature, or be blocked by strongly adsorbed intermediates. All
these complexities make it essential to develop in situ techniques, and to identify
the active site under reaction conditions.
The development of surface science techniques allows us to study the
molecular factors which control the selectivity of catalytic reactions under
practical reaction conditions. These molecular factors include: 1) Surface
structure (the size and shape of nanoparticles), 2)Surface composition, 3)
Adsorbate mobility, 4) Adsorption-induced surface restructuring, 5) Reaction
intermediates, 6) Surface oxidation state, and 7) Charge transport at metal/
support interfaces.

9.8 SUMMARY AND CONCEPTS

Surface catalysis aims to carry out the same reaction repeatedly at high
rates (activity) with desired selectivity.
The catalytic process can be characterized by its kinetic parameters
(turnover rate, rate constant, pre-exponential factor, activation energy,
reactant pressure dependencies, and reaction probability).
The preparation, activation, deactivation, and regeneration of high-surface-
area catalyst materials are dominant concerns of surface catalysis. There are
three phases in the development of a model system for catalysis research:
single crystal, 2D nanoparticle system, and 3D nanoparticle system. Various
surface science techniques can be applied to these model systems.
Turnover rate measurement is essential to characterize the activity of a
catalyst. In order to characterize the surface properties of a catalyst before
and after catalytic reaction, a high-pressure reactor has to be carefully
designed to avoid surface contamination. Developing in situ
characterization techniques, which enable us to study surface properties
under reaction conditions, is a major challenge in modern catalysis
research.
Catalysis by transition metal surfaces exhibits trends across the periodic
table whereby metals that form chemical bonds of intermediate strength
have the highest activities.
Important catalytic reaction concepts include structure sensitivity and
insensitivity of reactions, mechanistic classifications (Langmuir–
Hinshelwood, Eley–Rideal), the compensation effect, the presence of a
strongly chemisorbed overlayer, and the roles of structure and bonding
modifier additives (promoters).
Three catalytic reactions: C2H4 hydrogenation, NH3 synthesis, and CO
oxidation, are discussed in details. These examples demonstrate how to
obtain molecular-levelknowledge of catalytic reactions by applying the
surface-science approach.
During C2H4 hydrogenation under high pressures, multiple surface species
are coad-sorbed on the surface. It is not necessary for the most strongly
adsorbed species to be the key intermediate responsible for the turnover to
ethane. These strong adsorbed species form a carbonaceous overlayer that
affects the reaction in an indirect way. The structure insensitivity of this
reaction is believed to be due to the formation of the car bonaceous
overlayer.
Ammonia synthesis over an Fe surface is a structure-sensitive reaction. In
order to stabilize the active surface phase under high pressures and
temperatures, additives (e.g., aluminum oxide) must be added. Addition of
K may change the adsorption of reactants and the desorption of product,
and improve the activity of the catalyst.
Most knowledge of CO oxidation was obtained under UH V conditions and
can be used to understand the reaction under high-pressure conditions. A
highly active catalytic surface may be formed under oxygen-rich
conditions, which may be related to an O covered metal surface formed
under these conditions. Several recent studies suggested that the turnover
rate of this reaction can depend on the particle size of the catalyst. These
observations contradict the structure insensitivity of this reaction indicated
bythe earlier studies. At high temperatures, this reaction may become self-
sustained with a high turnover rate (so-called ignition). The ignition
temperature depends on the partial-pressure ratio and the surface structure
of the catalyst.
Understanding and controlling the selectivity of multiple-path catalytic
reactions are the major challenge for surface science in 21st century. The
selectivity may be tuned by changing the size and/or shape of nanoparticle
catalysts, by alloying, by using reducible oxides as support materials, and
by changing reaction conditions (e.g., reactant pressures and temperature).
In order to understand the selectivity of a catalyst, it is essential to study the
reaction intermediates on the catalyst surface under the reaction condition.