Catalysis Volume 8

A Review of Recent Literature

By G. C. Bond, G. Webb

The Royal Society of Chemistry

Copyright © 1989 The Royal Society of Chemistry
All rights reserved.
ISBN: 978-0-85186-594-2

Contents

Chapter 1 EXAFs in the Study of Catalysts By J Evans,
Chapter 2 Theoretical Approaches to the Study of Catalysed Reactions By E A Colbourn,
Chapter 3 Computational and Theoretical Studies on Zeolites By G H Grant and R J Abraham,
Chapter 4 Catalysis by Solid Acids and Bases By S Malinowski and M Marczewski,
Chapter 5 Complete Catalytic Oxidation of Volatile Organics By James J Spivey,

CHAPTER 1

EXAFS in the Study of Catalysts

BY J. EVANS

1. Introduction

EXAFS (Extended X-Ray Absorption Spectroscopy) offers a means of deriving interatomic distances and coordination numbers about a chosen element in either ordered or totally amorphous media. As such, it has very considerable applicability to the field of catalysis, particularly by metals. Many operating catalysts are either disordered solids or in solution and obtaining structural information on them by diffraction techniques is difficult. Without any clear structural model for the metal site then any account of mechanism or of the role of promoters or poisons can at best be hazy. Often research on model systems is performed on ordered arrays (either bulk crystalline solids or single-crystal surfaces) to reduce the complexity of the problem and to allow the use of single crystal diffraction (X-ray and neutron on bulk solids, Low Energy Electron Diffraction, LEED, on the surface of a gas-solid interface) . So because EXAFS can be applied to all of these types of materials, it can act as a bridge technique correlating information between model and “real” catalysts.

EXAFS may now be a vaunted acronym, but it represents only part of the information available in X-ray absorption spectroscopy, albeit the most understood part. Since this is the first article in this Specialist Periodical Review series on the technique, the first part of the chapter will be devoted to an introduction into the basis, methodology, information content of “mainstream” X-ray absorption spectroscopy. Then examples will be presented to illustrate its application to research in catalysis. Finally, some relatively new techniques will be described which seem to offer very substantial promise for catalysts investigations. Earlier articles by Joyner, Cox and Pettifer on the application of EXAFS to catalyst characterisation have been published in the monograph of Thomas and Lambert.

2. X-Ray Absorption Spectroscopy

2.1 X-Ray absorption spectra. – The absorption coefficient of a sample generally decreases as the frequency of X-radiation increases until there is a sharp rise in this value as the energy of an absorption edge of an element in the sample is reached. An example of this is shown in Figure 1 viz. the L (III)-edge spectrum of osmium powder.

This increase in absorption corresponds to the photoejection of a core electron. For EXAFS purposes generally the most convenient absorption edge is the K-edge, due to the ejection of a Is electron. A list of absorption edge energies and wavelengths of some elements of interest in catalysis are presented in Table 1. These should be related to the mass absorption coefficients for five representative matrix elements given in Table 2. Nitrogen, aluminium, nickel, silver and platinum are chosen to estimate the degree of absorption due to the atmosphere (or solvent.), an oxide support and metals in the 3 transition series respectively.

The first three elements listed in Table 1 (C, N and O) are best studied under UHV conditions. The absorption due to all other media is very high and contamination problems will also be severe. Light attenuation due to absorption by the atmosphere or window materials is also a problem in the soft X-ray region. This includes the K-absorpti on edges of the second row elements (Si to Cl) which are relevant to many catalyst, support, materials, or, in the case of chlorine, ate present in many preparations. Nevertheless, useful experiments can be performed on some systems in this wavelength band. Care must be taken in the design of the experiment though to avoid interference due to edges of heavier elements in the system (e.g, the near coincidence of the Cl K and Ru L (III) edges).

However, for the K-edges of virtually all the 3d and 4d metals and the L(III) edges of the 5d transition elements media absorption is a relatively minor problem. Measurements can be made under chosen atmospheric environments in self-contained reactors with suitable window materials (normally beryllium), and also in solution. For the harder edges (> 12keV), the absorption of a catalyst support also becomes a minor problem.

So X-ray absorption spectroscopy is applicable to a wide range of catalytic systems. Observing the edges due to common catalytic feedstocks (generally organics) is applicable only to surface science experiments. Some promoters and poisons (P, S and Cl) may be investigated in well chosen systems, but the entire transition element block is observable for single-crystal, metal film and supported catalyst samples.

The spectra themselves contain varied features viz. the energy of the absorption edge, some sharp peaks at or near the edge itself, and finally the broader, weaker oscillatory absorption changes which may extend for several hundred eV to the high energy side of the edge (EXAFS) . All contain different information which will be considered in turn.

2.1.1. Edge Positions. – There are many possible empirical prescriptions for the definition of the edge position within the experimental spectrum eg, the onset of the edge, the point of steepest slope, or the energy at half-height. The complex structure that may be associated with the edge makes an absolute statement impossible, but the point of steepest slope is clearly evident in the first derivative and is probably the most widely used definition. In principle the edge position might be related to the effective atomic number at the metal centre, and thus be correlated with oxidation state, as for the chemical shifts of x.p.s. spectra. Such correlations have been observed for early transition elements but are absent in, for example, some cobalt complexes and in the L(III) edge data of lead compounds. Changes in the coordination number, symmetry and the metal-ligand bonding can have a significant effect on the apparent edge position which can make it an unreliable indicator of oxidation state without corroborating evidence.

The relationship between the empirically determined edge position to the energy of the orbitals at the metal centre is also often unclear. While the absorption edge position is sometimes considered to represent the energy of the vacuum level (the position of Eo which is the onset of the continuum) , this seems by no means general. In cases where the absorption edge energy can be compared to the ionisation potential as measured by x.p.s., then the edge energy may be approximately 10 eV to lower energy; this has been observed from the carbon K-edge in ethene and ethyne. But in contrast to this, calculations on the [MO4] 2- (M = Cr and Mo) ions have indicated that the onset of the continuum is at the base of the metal K-absorption edges, and therefore at a lower energy than the normal definition of the edge position. Clearly care must be taken in considerations of absorption edge positions, but in some cases they may provide an indication of oxidation state.

2.1.2. The Edge Region. – The X-ray near edge structure, or XANES, may be considered as a combination of two contributions. To the low energy side of the continuum (below Eo) there are transitions to unoccupied valence states, which may be particularly intense when there is a high density of dipole allowed transitions (giving a “white-line” feature). Above the continuum there are other features, generally due to multiple scattering of the photoelectron. The multiple scattering paths will be sensitive to both the bond angles and bond lengths at the absorbing centre. In relatively simple structures, the latter region may be defined as being between Eo and a critical energy, Ec, such that the wavelength of the photoelectron of energy Ec-Eo, typically approximately 30 eV, is equal to the interatomic distance.

These featuies may be used as indicators of types of metal centre. For example, the tetrahedral [MO4]x- anions show a piominent pre-edge peak; this is due to a dipole allowed a1 (1s) to t2tiansition to a vacant orbital of metal d and ligand 2p character. In octahedral complexes, such as the [M(OH2)6]+, the d orbital derived t2g and eg orbitals of course do not provide dipole allowed transitions, and so there is no corresponding intense pre-edge feature.

There have also been reports that the intensity of a white-line at the absorption edge may be correlated with the d-state occupancy in some instances. For example, the L(II) and L(III) edges of the 5d transition element series generally include white-lines attributed to 2p to 5d transitions. A correlation was noted between the number’ of unfilled d-states and the area of the white line in the L(III) edge spectra of Ta, Ir, Pt and Au. The same argument has been extended to supported Os and Os-Cu catalysts. The white-line at the Os L(III) is more intense for an OS/SiO2 catalyst than for the bulk metal, suggesting that there are more unfilled 5d-states in the supported catalysts. However, in the Os-Cu/SiO2 alloy catalyst this increase is less pronounced, indicating that the osmium centres are less electron deficient with copper having the effect of decreasing the number of unfilled d-states associated with the osmium centres.

A note of caution is indicated though by some empirical observations on the L edge white-line intensities in some lanthanide compounds. In these compounds the white-lines at the L(II) and L(III) edges will be associated with 2p to 5d transitions. Even in lanthanide compounds, in which covalency and d orbital participation in bonding is much less pronounced than for transition elements, then the ligand type (and/or coordination number) also cause variation in these intensities.

Attempts to calculate the spectrum profiles in this region below Eo are relatively rare. Of particular note is an ab initio calculation for the metal K-edge of atomic Cu and CuCl2 which offers some guidelines to the analysis of other edges. Empirical assignments of some features as the dipole forbidden 1s to 4s excitation were shown to be ill-founded. Instead, the 1s to 4p allowed transition was accompanied by a series of simultaneous 11gand-to-meta1 shakedown processes. These two-electron transitions seem to contribute significantly to the edge structure.

The XANES features above EO may be used to fingerprint similar structures, because of the dependence of the multiple scattering paths upon the geometry at the absorption site. The resonances can be related to a R(M-E)-2 dependence. This factor has been used to assign features in the V K-edge XANES in a 0.1 wt % vanadium doped silica glass. By comparison with the features observed in compounds of known structure, the vanadium site was shown to be tetrahedral and the V-O distance was also estimated (1.77Å). Approaches to calculating XANES features have taken a variety of formalisms. Most commonly this has been a multiple scattering view, but band theory calculations have also been successful for metals; the connection between these two methods of calculating the same quantity has been drawn by Muller. Modifications of the Xα multiple scattering calculations which are well known to chemists have been employed fairly successfully to XANES analysis and have afforded useful results about the geometry at absorbing centres.

2.1.3 EXAFS. – The region above the critical energy, Ec, is that ascribed to EXAFS. When the photoelectron has a wavelength of less than the dominant interatomic distance, the principal scattering mode is considered to be at only one centre. Constructive interference between the outgoing photoelectron and back-scattered waves lead to an increase in the X-ray absorption probability; correspondingly destructive interference causes a reduction in that probability. (Figure 2).

The periodicity of this modulation of the absorption will be dependent upon the interatomic distance between the absorbing and back-scattering atoms, Rj, and the phase shifts, δij, encountered when the photoelectron experiences the potentials at these centres. Its intensity will be governed by the number of back-scatterers, Nj and their’ back-scattering amplitudes, Fj(k). Finally the amplitude is dampened by disorder (thermal and static), σ2, in the interatomic distance and any inelastic processes (related to the mean free path of the electron, λj. The recognition of the structural information intrinsic to this phenomenon and the derivation of a tractable formula for the estimation of interatomic distances was the result of the work of Sayers, Lytle and Stern. This has resulted in the majority of the attention in the analysis of X-ray absorption spectroscopy being devoted to this region of the spectrum. The principal attributes of the technique for determining local structures are:

i) element specificity,

ii) applicability to ordered, amorphous and fluid samples,

iii) it yields radial interatomic from the absorption centres (to a precision of ca. 0.02Å and over a range of ca. 3.5 – 6Å),

iv) differentiation between back-scattering elements of different rows in the period table (eg O versus S) and,

v) estimation of coordination numbers (to a maximum precision generally of ca. 10%). In terms of a typical metal catalyst supported on an amorphous oxide surface this is all highly desirable. The surface selectivity required is then derived from the element specificity.

2.1.4 EXAFS Theory. – The observed EXAFS, (k), is defined empirically as:

(k) = [μ(E) – μo(E)] / μo (E) (1)

where μ and μo are the observed and estimated “background” absorption respectively at the photon energy E. Rather than be represented as a function of the photoelectron energy, the EXAFS, (k), is conveniently presented as a function of the photoelectron vector, k, in units of Å-1 such that:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (2)

The most commonly adopted formalism for calculating (k) from structural parameters has been according to a single-scattering event representing the photoelectron as having a plane wave front. This affords Equation 3:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (3)

In addition to the terms defined above, there is an amplitude reduction factor Si which allows for the multi-electron processes at the central atom such as shake up and shake off. Difficulties arise from the estimation of the back-scattering factors and phase shifts. These are sometimes considered to be empirical parameters which require evaluation from close model compounds. Alternatively, theoretical values can be used, either per se or as a basis for empirical adjustment. There has been speculation about the transferability of these factors between sites of different chemical environment, with the general conclusion being that the deficiencies in the theoretical phase shifts are largely attributable to the approximation of the photoelectron as forming a plane wave. The more thorough description in terms of a series of spherical wavelets of differing angular momentum values was presented by Lee and Pendry. However, this exact solution is computationally more complex and analyses using it are rare. But a rapid procedure applicable to samples other than single crystals has been developed.

A second approximation was the consideration of single-scattering events only. In many cases, especially when there are linear arrays of atoms, multiple scattering can be not merely significant, but dominant. Recently, the rapid spherical wave algorithm was extended to include multiple scattering, and found to replicate the experimental EXAFS in metal carbonyl complexes possessing linear M-C-O units to within 14 eV of the absorption edge. This programme is available in a FORTRAN 77 version from the Daresbury Laboratory. Within it are ab initio phaseshifts and back-scattering factors calculated using electron distributions from a relativistic Hartree-Fock procedure which appear to have an acceptable degree of transferability between chemical systems.

So within the chemical literature there are now two main alternative analysis procedures. The first is based upon the plane wave equation using empirical adjustments to theoretical phaseshifts to account for inaccuracies due to spherical wave and multiple scattering factors. If the model compound used for the adjustment is a very close analogue to the site in question, then this can give accurate results. However, there are many situations m catalysis for which there may be no such model available; even a change in bond angle of a few degrees can modify multiple scattering components very markedly. In those situations, some of these empirical procedures may be prone to error, and the alternative more exact spherical wave based approach seems clearly preferable.

The spherical wave method seems to offer the best procedure for EXAFS analysis. All approaches though must always be used with care so that the information distilled from an analysis is both valid and plausible. This will involve internal checks of the quality of the background subtraction and the statistics of the analysis. However, EXAFS studies are best carried out in conjunction with other analytical techniques so that information may be cross referenced.

(Continues…)Excerpted from Catalysis Volume 8 by G. C. Bond, G. Webb. Copyright © 1989 The Royal Society of Chemistry. 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.