Catalysis Volume 1

A Review of the Literature Published up to mid-1976

By C. Kemball

The Royal Society of Chemistry

Copyright © 1977 The Chemical Society
All rights reserved.
ISBN: 978-0-85186-534-8

Contents

Chapter 1 Catalysis on Well-defined Metal Surfaces and Non-metallic Substrates By S. J. Thomson, 1,
Chapter 2 Reactions of Hydrocarbons on Alloy and Bimetallic Catalysis By R. L. Moss, 37,
Chapter 3 Catalysis on Faujasitic Zeolites By R. Rudham and A. Stockwell, 87,
Chapter 4 Catalytic Properties of Aluminas for Reactions of Hydrocarbons and Alcohols By C. S. John and M. S. Scurrell, 136,
Chapter 5 Selective Oxidation of Hydrocarbons Over Mixed Oxide Catalysts By R. Higgins and P. Hayden, 168,
Chapter 6 Reactions on Sulphide Catalysts By P. C. H. Mitchell, 204,
Chapter 7 Ziegler Polymerization By A. D. Caunt, 234,
Chapter 8 Olefin Metathesis By J. J. Rooney and A. Stewart, 277,
Chapter 9 The Homogeneous Catalytic Activation of C-H Bonds By G. W. Parshall, 335,
Author Index, 412,

CHAPTER 1

Catalysis on Well-defined Metal Surfaces and Non-metallic Substrates

BY S. J. THOMSON

1 Introduction

The aim of this Report is the examination of recent papers published in surface physics and surface chemistry with a view to assessing their contribution to heterogeneous catalysis. What was sought in preparing the Report were advances in the understanding of catalytic processes through knowledge of adsorption, intermediates, and the role of the substrate. Also of major importance was the question of whether new principles would emerge and whether attitudes to solving problems in catalysis would be changed by these new methods of study.

The Report will deal mainly with applications to surfaces of: Electron spectroscopy for chemical analysis (ESCA), Photoelectron spectroscopy (PES), X-ray photoelectron spectroscopy (XPS), u.v. photoelectron spectroscopy (UPS), Auger electron spectroscopy (AES), low-energy electron diffraction (LEED), soft X-ray appearance spectroscopy (SXAS), ion-neutralization spectroscopy (INS), ion-scattering spectroscopy, (ISS), secondary-ion emission spectroscopy (SIMS), Auger emission angular profiles (AEAP), conversion electron Mössbauer spectroscopy (CEMS).

These techniques have been applied not only to well-defined single-crystal metal surfaces but also to clean polycrystalline metals and to selected non-metallic catalysts. Thus, when a catalyst chemist working with ill-defined supported catalysts, mixed oxides, or metal powders looks at the work of the surface physicist or chemist he might well be envious: clean surfaces, ultra-high vacuum, and precise spectroscopic observations all suggest a degree of exactitude and understanding which would lay down the foundations for the understanding of heterogeneous catalysis. A closer look, however, at LEED, AES, XPS, and UPS reveals that these methods all have their fundamental unresolved problems and shortcomings. LEED sees only that which is ordered on a clean surface and the other electron spectroscopies suffer from difficulties in interpretation of spectra and in the meaning of chemical shifts. As for the clean, well-defined surface, one has only to read the experimental section of almost any paper on AES or XPS to realize that even here there is controversy.

The transfer of ideas from surface physics to catalysis involves a controversial step. Fundamental studies usually involve gas pressures of 10-10 – 10-6 Torr whereas catalyst chemists, academic and industrial, must in this context seem to work at infinite pressure. Somorjai has given some attention to another problem, viz. whether or not a stepped single crystal can be regarded as a model surface for a catalyst. Cyclopropane hydrogenation over Pt was chosen as the reaction and it was found that rates at 1 atm for this surface compared favourably with those measured on dispersed supported Pt catalysts.

What follows must then be read in the light of these cautionary remarks: much of interest and significance does emerge from a subject which has not yet reached maturity.

A brief introduction to PES can be found in an article by Hercules: other conference reports and reviews are listed in reference 4. Specialist reports on particular aspects of the subject have also appeared. Reference to work on alloys is restricted, for this is the subject of a separate chapter in this volume.

2 The Potential of the Methods and Interpretation: the Current State of the Techniques

In this section a selection of recent remarks made by practitioners of the techniques is presented which illuminates the present positions for many of the methods.

LEED — A requirement for creating a model for adsorption or catalysis is a knowledge of the state of order or disorder of the adsorbate on a catalyst surface. Before presenting any results some cautionary statements are appropriate.

LEED suffers the disadvantage that it allows adsorbates to be examined only when they have long-range order: diffraction is sensitive to adsorbates having translational symmetry parallel to the surface. In addition, elastic scattering cross-sections for incident electrons may be significantly smaller for adsorbate atoms that for the substrate atoms, e.g. S and O on Ni. This reduces the sensitivity of scattered intensities to position changes in the adsorbed layer. It is for these reasons that Woodruff has given attention to the possible use of Auger emission angular profiles (AEAP). The angular features are highly sensitive to adsorption site and AEAP does not depend on long-range order. No doubt more will be heard of this promising technique.

Many LEED papers are devoted to the fundamentals of electron scattering at surfaces: only one illuminating and thoughtful comment will be reproduced here on this topic which arises in a study of the structure of overlayers, as determined by LEED, for N2 on Mo{001}. This system has been critically examined by Ignatiev et al. They examined a c(2 x 2) structure (i.e. the periods along cubic directions in the plane of the overlayer are doubled with respect to the substrate and the unit mesh contains a central atom) in four-fold, bridging, and top atom positions on Mo. Whilst the best agreement between experimental and calculated LEED spectra was obtained for the four-fold structure with N in pyramidal hollows formed by four adjacent Mo atoms, the fit between theory and experiment was good only for certain beam energies, and much less good for others. The paradox that they see in LEED interpretation is worth quoting: ‘The fit between theory and experiment seems to be too good, for some beams, for the proposed structure to be wrong, yet too poor, for some other beams, for the proposed structure to be entirely right’. Their views are elegantly stated.

Alterations in adsorption and desorption by the effect of an electron beam striking a surface, as in AES and LEED, are discussed by Margoninski et al. Caution is therefore required in interpretations based on these methods where authors do not specifically deal with this matter.

ESCA (PES) Methods — In surface analysis by AES and XPS it is necessary to know the depths sampled by the two techniques: Coad et al. report the mean sampling depths as ~0.7 and ~1.4 nm respectively. They point out that for a monolayer of impurity on a surface there will therefore be a contribution of 20% to the AES analysis, 10% to the XPS analysis: UPS, LEED, and SIMS can be claimed to be techniques which sample surface layers.

Auger electrons have finite escape depths and thus signals from multilayers are not linearly related to depth. In spite of this, few papers have appeared in the past year on intercalibrations: a notable exception is that by Peralta et al. who used a radioactive method for AES intercalibrations for S on Mo(110). Morabito has enumerated the difficulties he sees as inherent in quantitative AES: the detection system does not measure the absolute Auger current directly; Auger escape depths have not been measured for many elements; backsputtering correction factors are not easily available; ionization cross-sections are not available; there has been insufficient study of the effect of surface roughness.

Fuggle et al. have published on the quantitative aspects of XPS. In an XPS study of oxide growth on a metal surface the amount of oxide can be taken as proportional to the intensity of an oxygen photoelectron peak as long as the atoms are on the surface. Oxygen atoms in deeper layers may yield only 1/e of the contribution from surface atoms. Thus care has to be exercised in making quantitative evaluation of surface concentrations in the XPS method.

Bonding and Chemical Shifts — The determination of energy-level shifts which accompany chemisorption is of key importance in understanding the bonding of chemisorbed species. Electron spectroscopy provides kinetic energy information on electrons ejected from the surfaces of solids on which chemisorption has occurred, and peaks in the energy distribution curves correspond to ionization of electrons initially in the orbitals of the surface complexes. The orbital energies depend on hybridization, nature of bonding, image, and relaxation effects. Thus energy-level changes between free and adsorbed molecules represent basic information on bonding.

The fundamental problems associated with relating PES to the bonding of adsorbed species are succinctly stated by Yates et al. There are two major factors which determine chemical shifts in the binding energy of core electrons of atoms bound at surfaces.

First, there is the initial charge state of the adsorbed atom and its surroundings. If it is assumed that a bound atom is partially ionic in nature, with the excess or deficiency of charge associated with it, qe, located in a shell of radius r, then the change in binding energy associated with this charge is ΔEB = qe2/r. If there are neighbouring charges then summation of these over the neighbours must also be made, i.e.

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

for all neighbours j of charge qje at distance rj.

Secondly, the chemical shifts are also determined by the final state of the core hole produced by photoemission. In photoemission from A,

hv + A [right arrow] (A+)* + e-

where (A+)* represents the core hole ion which has a total energy E(A+)*.

The photoelectron has a measured energy EK and since EB = hv — EK it follows that

EB = E(A)+* — EA = (final-state total energy) — (initial- state total energy)

It should not be assumed that variations in EB for different adsorbed states arise only from changes in the initial-state energy. Variations also occur in E(A+)* because of relaxation processes which are ascribed to polarization effects arising from different surroundings or modes of bonding for A. These extra atomic relaxations (XR) alter the true binding energy thus: EB = E’b – EXR. Since charge distribution for an adsorbed atom or molecule cannot be determined exactly E‘B cannot be calculated exactly: neither can EXR be determined exactly and so what is determined experimentally is the total chemical shift [EB – EB (free atom)] = ΔEB. This is then related qualitatively to bonding character for adsorbed species.

No general agreement exists on how to calculate ionization energies for adsorbed states from experimental observations. Indeed Hagstrum says that there is confusion and error in the literature. Thus he sets forth the basic concepts and his prescription which may be stated as follows. For simple adsorption systems the measurable macroscopic parameter which comes closest to giving the ionization energy of a surface orbital is the binding energy of the orbital with respect to the Fermi level, plus the work function of the saturated, uniformly covered surface. The debate on this topic will undoubtedly continue.

The information available from PES can be summarized briefly thus: XPS yields information on core level and valence levels whereas UPS can probe the valence region with great sensitivity and resolution. In UPS interpretation is not easy for the total emission comes from substrate and adsorbate. XPS has the advantage over AES that electron-beam-induced changes in the surface are absent. Ejection of electrons in PES leaves behind holes. Relaxation processes then involve the outer orbital electrons which adjust towards the hole created in, say, XPS in order to screen the excess positive charge. The ejected photoelectron thus acquires additional energy. In solids and condensed states both extra-atomic relaxation and relaxation within the atom have to be considered. ‘Shake off’, i.e. when more than one electron is ejected, and ‘shake up’, where one or more electrons are excited to bound states, all require consideration.

Changes in the Surface Density of States. — Attention in this section is given mainly to problems associated with electron energy distributions in substrates. Although the surface density of states cannot in general be extracted from a measured photoemission spectrum, attempts have been made to estimate these changes. Figure 1, taken from Plummer, shows that there are indeed large changes in distribution curves for electron energies in photoemission when, for example, oxygen is adsorbed on W. The problem is that the emission is partly from surface states, partly from bulk states.

Plummer plots ‘optical surface density of states’ in which he assumes that the shape of the clean energy distribution does reflect the density of states of the surface, after removal of the inelastic electron contribution. He assumes that there are no changes in the matrix elements which describe the transitions from initial to final states and that 40% of the signal comes from surface atoms, and that only the surface atoms change their local density of states upon adsorption. The results are shown in Figure 2.

The measured difference curves for adsorption of O and C were used in calculation of the new ‘surface density of states’ after adsorption. The author says clearly that these curves should not be construed as surface density of states but they do show that dramatic changes are occurring for electron states of surface atoms. One point of significance which emerges is that the difference in the curves for O and C may indicate adsorption on different sites on the metal.

Fingerprint Technique. — In spite of all the problems which have so far been discussed it is now possible to demonstrate how PES can lead to an increased knowledge of adsorbed states. An example is shown in Figure 3 in which the effect of heating CO adsorbed on W(100) in a molecular state becomes apparent. When the adsorbed layer is heated to 1100 K part of the CO is desorbed and the spectrum of the remaining species can then be compared with the spectra of adsorbed O, and of adsorbed C produced by C2H4 dissociation. Difference spectra are plotted in Figure 3 and it can be seen that the upper spectrum from CO heated to 1100 K is well reproduced by a combination of spectra from adsorbed C and adsorbed O. Thus the fingerprint technique is obviously of value in finding the state for an adsorbed species, which in this case is clearly not molecular.

Examples of Interpretation Problems in Applications of PES. — Problems of interpretation in PES have now been stated in general terms in the previous sections. Some specific examples will now be presented and, although they are not from catalytic studies, they illustrate the difficulties which will arise in assessing the validity of papers dealing with reactions on surfaces. Brundle has reviewed the techniques of UPS and XPS from the point of view of the information obtainable on detection of different states of adsorption and interpretation of structure and bonding.

A basic study was made by Yu et al. They measured spectra of condensed gases by UPS where there was little likelihood of chemical bonding. They chose C6H6, C5H5N, CH3OH, C2H5OH, HCHO, H2O, and NH3 condensed on MoS2 at liquid nitrogen temperature: they concluded that for all the condensed gases the spectra were similar to those of the corresponding gas-phase spectra, except that all energy levels were shifted by 1 — 1.65 eV.

The object of another paper by Yu et al. is to relate UPS spectra with molecular orbital models of bonding and with thermodynamic data on heats of adsorption. Together with information already published, they have now assembled information on adsorption of H2, O2, CO, C2H4, and C2H2 on Fe, Ni, and Cu. Fingerprint spectra are shown for the adsorbed species. For the hydrocarbons, shifts in the π-levels do not correlate well with measured heats of adsorption and the authors then formulate the notion that one should examine changes in the substrate electronic levels in order to interpret trends in heats of adsorption. Unfortunately this is not yet theoretically possible.

The importance of d-electrons in the substrate has often been referred to in attempts to understand catalysis on metals. As Melius says the correlation of d-bonding character or d-band vacancies with chemisorption and catalytic activity is poor. He proposes a new model in which d-electrons do not participate in forming dsp hybrid bonds with the adsorbate but in which they remain localized. The model has been examined by sophisticated theoretical techniques.

(Continues…)Excerpted from Catalysis Volume 1 by C. Kemball. Copyright © 1977 The Chemical Society. Excerpted by permission of The Royal Society of Chemistry.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.