Faraday Discuss. Chern. SOC., 1990, 89, 107-117 Characterisation of Oxide-supported Alkene Conversion Catalysts using X-Ray Absorption Spectroscopy John Evans, 3. Trevor Gauntlett and 3. Frederick W. Mosselmans Department of Chemistry, The University, Southampton, SO9 5NH An X-ray absorption spectroscopy (XAS) cell has been developed to allow the loading of air-sensitive catalyst precursors in a glove box and their in situ activation and operation to be monitored. This has been used to investigate two alumina-supported propene metathesis catalysts derived from Mo,(OAc), and Li,[Mo,Me,]. The EXAFS data so derived has been ana- lysed by spherical wave methods using ab initio phase shifts and back- scattering factors. The detailed structural features obtained are sample- history dependent, but the following general observations can be drawn about the mean surface species present.In all cases the quadruple Mo-Mo bond is cleaved on the surface; some samples show evidence of one (or fewer) metal atoms at longer distances. For the latter sample activation in uucuo at 70°C yielded sites approximating to [MoOz(Me),(O-)]” . The addition of propene caused a change in the X g S spectra. There was evidence of a shell with Mo=C distance of ca 1.8 A that can be attributed to a terminal carbene ligand. For one of the Mo,( OAc),-derived catalysts there was also evidence for a molybdacyclobutane unit. The Li4[ Mo,Me,]-based materials also showed evidence of residual Mo=O bonds after exposure to propene. The Zr K-edge EXAFS of Zr(ally1)Joxide (oxide = silica and alumina) were very similar for a range of sample conditions, showing evidence for a residual ally1 group and Zr-0 bonds.X-ray absorption (XAS) has been successfully applied to provide structural detail on oxide-supported transition metal catalysts that had hitherto been unobtainable.’ It is one of the few techniques than can provide information about the local order of a metal site on the surface of an amorphous material, and has the added advantage that experiments may be performed in situ. Several different designs for a catalysis cell have been published, including that of Lytle’ using a BN or Be boat to contain a powdered sample, and the Koningsberger design employing pressed It is known that pressing discs modifies the surface structure of the support3 and also creates voids.‘ Since our test catalysis experiments were performed on free powders it seemed appropri- ate to carry out XAS measurements on samples in the same state.Hence we report a cell design which allows air-sensitive samples to be loaded in a glove box, activated and then exposed to a flow of reagent gas. The samples chosen for this study were two sets of alkene conversion catalysts that had been prepared from organometallic or coordination complex precursors.5 These were molybdenum-based propene metathesis and zirconium-centred ethene polymerisation catalysts. Olefin metathesis has been catalysed by supported-metal catalysts for many years. Using NM R, homogeneous catalysts have been characterised supporting the Chauvin mechanismh involving the transformation via a metallacycle of one alkylidene to another.’ The supported catalysts have generally been less well characterised, partly due to the harsher conditions under which they operate.Supported-molybdenum organometallics were first reported as being active in olefin metathesis by Whan and co-workers.8 However, no work on the structure of such catalysts was carried out at that time. XPS studies by Walton and co-workers were108 XAS Studies of Alkene Conversion Catalysts. inconclusive as to the nature of the catalyst derived from dimolybdenum tetra-acetate.' Iwasawa and co-workers have studied the interaction of ally1 molybdenum compounds with silica and alumina.lo The XAS studies, however, were undertaken on species after oxidation or reduction at high temperature, since their techniques did not enable them to study the highly unstable supported complexes." In the present work, using dimolyb- denum tetra-acetate and the more reactive tetralithium octamethyldimolybdenum as precursors, propene metathesis catalysts active at room temperature have been prepared.These have been activated in situ on an EXAFS beam line and then spectra of the active catalyst under propene have been recorded. Organozirconium-based polymerisation catalysts were reported by Yermakovi2 and Ballard." Again the metal centres were structurally ill-characterised, and similar experi- ments have been performed in the XAS cell, with tetra-allylzirconium as precursor. Experimental The catalysts are extremely oxygen-sensitive thus all experimental procedures, unless otherwise stated, were carried out under nitrogen using standard Schlenk techniques or in a nitrogen dry-box.The support employed for the metathesis reactions was aluminium oxide grade C (Degussa AG); this was preheated to 650 "C in air for 18 h prior to use. The surface area of aluminium oxide grade C is loo* 10 m' g-' and the surface density of hydroxyl groups was determined (by CH, evolution from MeMgCI) as 0.8 nm-I. The Zr experiments also used aluminium oxide C and Aerosil 200 silica; both were dried in vacuo at 200 "C for 12 h.14 Propene Metathesis The metathesis experiments were performed in a silica vessel (volume 128.4 cm3) attached to a conventional vacuum line which had an ultimate vacuum of 1 x lop5 mbar. The catalysis vessel was heated by a conventional oven.A typical catalytic run was initiated by putting approximately 0.1 g of pretreated support into the reaction vessel. This was then heated under vacuum at 200 "C for 1 h. The support was then cooled, under vacuum, to room temperature prior to the introduc- tion of 1 atmt nitrogen. The amount of molybdenum precursor introduced was normally sufficient to give a molybdenum: support mass ratio of 1 : 20. The tetralithium octamethyldimolybdenum was injected in a solution of Et,O (freshly distilled from sodium benzophenone) onto the support and left for 1 h. Then the solvent was removed by careful pumping, during which there were regular admissions of nitrogen to prevent the support being deposited on the sides of the reaction vessel.The catalyst was then pumped on at room temperature for 1 h before activation. However, for Mo,(AcO),, the substrate and the support were ground together in a nitrogen dry-box before being loaded into the catalysis vessel and transferred back onto the vacuum line. Activation of the catalyst was achieved by heating the catalysis vessel to the desired temperature in vacuo for 1 h. After cooling to room temperature, 40mbar of propene was admitted to the reaction chamber. Then after 1 h the pressure in the reactor was increased to 1 atm with nitrogen and samples were then injected into a Pye Unicam GCD chromatograph. This instrument was calibrated using a standard gas mixture of 1 .O% methane, 0.94% ethene, 1.05% propene and 1 .O3% 2-butene in nitrogen.The chromatograph contained a 2 m column filled with 'Porapak Q' and was operated at 150 "C using a flame ionisation detector.J. Evans, J. T. Gauntlett and J. F. W. Mosselmans Table 1. Propene metathesis catalysis results precursor (solvent) activation temperature turnover no. / "C /min-' 3 00 20 60 70 75 100 150 200 250 0.3 0.004 0.06 0.2 0.08 0.04 0.008 0.008 0.0006 109 The turnover numbers, expressed as molecules of propene disproportionating per molybdenum atom per min, for the metathesis of propene at room temperature (20 "C) for the precursors after activation at the most favourable temperature, are shown in table 1. The turnover rates are accurate to *5%. The turnover rates for the octamethyldimolybdenum( 1 1 ) salt and dimolybdenum tetra-acetate are of the same order as those found by Iwasawa et al.for tetra-allyldimolyb- denum on aluminaI5 (0.3 molecules min-'). X-Ray Absorption Spectra XAS spectra were recorded in fluorescence mode on Station 9.2 of the S.R.S., at the S.E.R.C. Daresbury Laboratory, using a double-crystal Si (220) monochromator. For the Mo K-edge data, a zirconium filter was used in front of a thallium-doped potassium iodide scintillation counter. The spectra were calibrated by the use of a molybdenum foil monitor, whose edge was taken to be at 20,003.9eV.t The Zr K-edge data were similarly recorded and calibrated, but without a filter. Background subtraction was achieved using the IBM PC resident program PAXAS,'~ and curve fitting was performed using the program EXCURVE" on the Daresbury Convex c220.Initially the spectra were aligned, averaged and background subtracted in PAXAS. They were then Fourier filtered in EXCURVE before refinement of trial structures. The type of atom of each shell, its abundance and the total number of shells are then progressively varied until a good model has been obtained. The relative importance of each additional shell can be validated by two methods. First, as a rough guide, its contribution to the total EXAFS can be calculated using the 'Fitstat' option in EXCURVE, in which the integral of the EXAFS due to each individual shell is obtained.'* More rigorously the statistics test" of Joyner et al. may be applied as each shell is added. Thus, by a trial and error procedure the final model or models are obtained. There is a similarity in the backscattering properties of the carbon and oxygen atoms, but in most cases the differences were sufficient to allow them to be distinguished.When there are two shells at similar distances high (>0.8) correlations between the parameters of the separate shells might be expected. This was observed for two of the samples described below viz. Zr( allyl),/silica and Li4[ Mo,Me,]/alumina. Thus the statistical tests are essential to check the validity of each shell and this is used to justify some of the models despite some high correlations between parameters. However, in this situation the accuracy of some of the parameters may be less than normally achievable by EXAFS. Unless otherwise stated, the shells were found to have less than 1% probability of being + 1 e V = 1.602 I8 x 1 O - ' " J .110 XAS Studies of Alkene Conversion Catalysts.W///' "14 Fig. 1. Diagram of the XAS cell for the catalytic studies; ( a ) outer-housing with two small (25 mm outer diameter) Be windows for transmission measurements and a larger one at which a single scintillation counter may be placed. There is one vacuum connection flange, one for electrical connections and the third for the sample-bearing stem. ( b ) Central stem in which the sample is mounted between Be windows (4mm pathlength) in a stainless-steel block. The sample sits on a silica fritte and gas may be admitted through the central tube to flow down through the sample. The two smaller tubes are part of cooling coils. Heating elements and thermocouple connections are on the stainless-steel block.insignificant." The value of AFAC, the proportion of absorption resulting in EXAFS, was maintained at 0.85, the refined value for Mo foil. The values of the Debye-Waller factors are 2 u 2 , where u is the mean-square deviation in interatomic distances. Statisti- cally derived errors on the determined distances were all <0.01 A, but a more realistic estimate is 1.5%, or 0.03-0.04 A, for the distances quoted in this paper.,' The error bounds on coordination numbers are probably of the order of *30%. XAS Samples For the Mo EXAFS experiments a complex: support ratio of between 0.5 : 100 and 1 : 100 was used in order to try to achieve maximum uniformity in the state of the molybdenum atoms. The Li,[ Mo2Me8]-derived catalyst was prepared as above, however, after the solvent had been removed the catalyst was transferred to the XAS cell in a dry nitrogen glove box.The Mo,(AcO),-based catalyst was prepared in the same manner as for the catalytic reactions, with the mixed catalysts placed in the XAS cell. The sample is held between two beryllium windows 4mm apart in the controlled environment EXAFS cell (fig. 1) which allows the powder sample to be heated and cooled under vacuum (lo-' mbar) and for gas to be passed through the sample. Samples were evacuated for 1 h and then the activation was carried out by heating the samples to 300 "C [for Mo,(OAc),] and 70 "C (for Li4[Mo2Me8]) for 1 h. The cell was then allowed to cool to ambient temperature under vacuum, before spectra were collected on the activated catalyst prior to the introduction of propene.Finally, around 40 mbar of propene was admitted to the cell and spectra of the catalyst were recorded once more. Zr(allyl), was adsorbed from a pentane solution to saturate the available surface sites," ca. 8% by weight of complex. The materials were also dried in uucuo at room temperature. In situ reduction by H2 wasJ. Evans, J. T. Gauntlett and J. F. W. Mosselmans 1 1 1 Fig. 2. In situ monitoring of the activation of the 0.6% Moz(AcO),/alumina sample at 300°C. These Mo K-edge XANES spectra have been shifted along the energy scale and down the absorbance axis to facilitate comparison. The time intervals between runs was ca. 7 min. Results Mo,(AcO),-derived Catalysts Two experiments were performed using concentrations of molybdenum of 0.52 and 0.60% by weight.Since the molybdenum(i1) acetate catalyst is prepared by a dry mix method, only two states of the catalyst were studied in detail: after activation and under propene. There is a noticeable change after the addition of propene, which suggests that the coordination sphere of the molybdenum is being affected by the propene in some manner. XANES spectra were recorded continuously during thermal activation to monitor the progress of the reaction; these spectra, which had a duration of ca. 7 min, are shown in fig. 2 (for the 0.6% sample). The elevated sample temperature produces a broader spectrum but changes in the XANES are apparent as the activation progresses. The most notable of these changes is the enhancement of the pre-edge peak.This is generally associated with a ls-5p/4d electronic transition. This implies that during heating the coordination environment of the molybdenum centres is changing, with a resultant loss of symmetry to allow a low-lying metal-centred orbital of mixed p/d character. This is most accentuated for molybdenum in high-oxidation-state tetrahedral centres. The EXAFS results for the two ‘post-propene’ experiments are shown in table 2. A solution of the ‘pre-propene’ EXAFS has proved impossible to solve for both experi- ments; an R factor of 40% could not be achieved for either set of data. This probably indicates a large disorder in the coordination environment of the molybdenum atoms after activation. Another possibility is that one or other of these collections of spectra may be contaminated by oscillations due to beam movement.** Both of these spectra, however, have a component at high k.It is possible that after activation there are molybdenum neighbours at between 2.7 and 3 A. There is no sign of any molybdenum atoms within bonding distance in the EXAFS data of the ‘post-propene’ samples, suggesting that on activation the multiple molyb- denum-molybdenum bonds are broken as the metal atoms are dispersed around the surface of the alumina. The data from the more dilute sample could be satisfactorily fitted by a three-shell model (C, 0 and C, in increasing radius). As discussed below, this first shell is at a distance expected for a terminal alkylidene ligand. The second sample, though, provided evidence for six shells (fig. 3 ) , including a molybdenum atom112 XAS Studies of Alkene Conversion Catalysts. Table 2.Results of EXAFS analysis for the Mo2( AcO), system after exposure to 40mbar of propene Debye- Waller atom type shell radius/A coordination number factor/A2 C 0 C C 0 C C C Mo 1.88 2.06 2.58 1.81 2.04 2.26 2.64 2.98 3.01 sample I, 0.52% Mo" 2.0 2.0 3.2 sample 11, 0.60% Mot' 0.8 1.6 1.9 2.5 2.3 0.5 0.013 0.003 0.012 0.004 0.008 0.007 0.017 0.008 0.012 ~~ " Fourier window 1.3-3.0 A, fit index 0.37, R = 24.8%, E, = 27.9 eV, VPI = -6.0 eVh Fourier window 0.6-3.7 A, fit index 0.05, R = 10.7%, E, = 14.6 eV, VPI = -5.0 eV. - 4 1 4 6 8 10 '12 v14 ' 161 shell radius/ A Fig. 3. The best fit for the Mo K-edge EXAFS of sample I1 of the Mo2(AcO),/alumina sample after exposure to propene (40 mbar).(a) k3-weighted EXAFS and (b) Fourier transforms, phase corrected for the first shell. (-) Observed data; (- - -) curved wave theory (a). at around 3 A. This is too long for a metal-metal bond but could be indicative of an oxygen-bridge dimer. The inner three shells could be fitted as either all 0 sites, or the chemically more plausible C , 0, C sequence, as for the first sample. The remaining two carbon shells only passed a statistical test (on the raw data) at the level of a less than 5% probability of being insignificant. Li,Mo2Me, .4Et20 as a Catalyst Precursor Here also, two separate sets of data were obtained. In each case spectra were recorded after activation at 70°C in vacuo and under propene.The two experiments involved 0.87 and 0.70% by weight of molybdenum, respectively. The results of the EXAFS analyses are shown in table 3, and the fits for the sample I are presented in fig. 4. InJ. Evans, J. T. Gauntlett and J. F. W. Mosselmans 113 Table 3. Results of EXAFS on Li,Mo2Mex .4Et20-derived catalysts Debye- Waller atom type shell radius/ 8, coordination number factor/A2 0 C C Mo 0 c 0 Mo 0 C Mo pre-propene exposure sample I, 0.87% Mo" 1.71 (1.72) 2.6 1.86 (1.88) 1.0 (2.0) 2.1 1 (2.08) 2.0 ( 1.4) 1.75 2.8 2.03 3.3 2.20 2.2 2.60 0.8 sample 11, 0.70% Mob post-propene exposure sample I, 0.87% Mo' 1.66 1 .o 1.80 1.8 2.44 2.1 3.08 0.9 sample 11, 0.70% Mo" 1.75 3.9 2.09 2.4 2.59 0.9 0.017 0.009 (0.0 17) 0.013 (0.010) 0.006 0.006 0.006 0.016 0.008 0.005 0.016 0.023 0.015 0.0 14 0.015 " Fourier window 0.5-2.5 A, fit index 0.92 (0.60), R = 23.5 (24.2)%, E,, = 33.8 (31.5) eV, VPI = -3.0 eV.h Fourier window 0.6-3.2 A, fit index 0.96, R = 16.0%, E,, = 23.6 eV, VPI = -5.0 eV.' Fourier window 1.3-3.5 4, fit index 0.74, R = 20.6%, E, = 28.1 eV, VPI = -5.0 eV." Fourier window 0.7-2.8 A, fit index 0.84, R = 19.0%, 15,) = 32.4 eV, VPI = -5.0 eV.this table more than one fit is quoted for the spectrum of the first sample, as it proved impossible to distinguish whether one of the shells consisted of carbon or oxygen atoms ( o r a mixture of the twoj. For this catalytic system the pre-propene data could be solved suggesting a more ordered structure after activation. After activation the catalyst appears to consist of molybdenum $toms tethered by relatively short molybdenum-oxygen bonds at between 1.7 and 1.9 A.There may also be some carbon ligands a t distances that vary from 2.03 through 2.1 1 to 2.20 A. While the last is indicative of a molybdenum-carbon single bond, the first two may represent either a strong single bond or a weak double bond. Coordination numbers are less than precise for these systems and, in particular, the model for the second experiment has rather more atoms than are feasible in the coordination sphere. This experiment also shows indications of a molybdenum atom at a distance similar to a molybdenum- molybdenum single bond length. In the first experiment when propene is added there are signs of a carbene group (1.80 A ) similar to that in the acetate catalyst, though the molybdenum-oxygen bonds are still shorter than in that system (1.70 * 0.0: A).In the second experiment, however, there are carbon atoms at a distance of 2.09 A. There are signs in the first experiment of a molybdenum atom outside the immediate coordination sphere but possibly linked by an oxygen bridge. In the second there is a nearby molybdenum atom, which could either be directly bonded or linked by an oxygen bridge; in this model there are a high number of oxygens in the coordination sphere.114 XAS Studies of Alkene Conversion Catalysts- 0 1 2 3 4 5 shell radius/ A B , I shell radius/a Fig. 4. The best fit for the Mo K-edge EXAFS of sample I of the Li,[ Mo2Me,]/alumina sample (A) after activation at 75 "C; k'-weighted EXAFS and ( b ) Fourier transforms, phase corrected for the first shell.(-) Observed data; (- - -) curved wave theory. ( B ) Similarly presented data on sample I after exposure t o propene (40mbar). Zr(allyl),-derived Catalysts Several series of spectra were recorded on alumina and silica at differing zirconium concentrations. All spectra were very similar, both with and without hydrogen reduction, showing low EXAFS amplitude. This indicates low coordination numbers and/or high disorder in the coordination sphere. A fit ofoone of these data sets is presented in fig. 5. This shows ca. three Zr-0 bonds at 2.0 A, and 3 C atoms at ca. 2.25 A, suggesting that in these samples, on average, one of the ally1 groups is retained on the metal centre. Discussion There were two aims to this project viz.to identify the species present after catalyst activation and during catalysis. The first aim was only achieved to a limited extent for one example, namely the Li,[ Mo,Me,]-derived catalyst. The likely presence of some carbon ligands on the activated state strongly suggests that on impregnation some methyl groups remain on the molybdenum centres. Methane gas was found to be evolvedJ. Evans, J. T. Gauntlett and J. F. W. Mosselmans 115 shell radius1 8, Fig. 5. The best fit for the Zr K-edge EXAFS of a saturated sample of Zr(allyl),/silica in uacuo. ( a ) k3-weighted EXAFS and ( b ) Fourier transforms, phase corrected for the first shell. (-) Observed data; (- - -) curved wave theory ( a ) . Fit index 1.53, R = 33%, E,= 26.6 eV, VPI = -1.0 eV, Debye-Waller factors 0.017 and 0.018 A* for 0 and C shells, respectively.during the absorption, thus it seems probable a partial hydrolysis by hydroxyl groups occurs Mo-Me+S-0-H + Mo-0-S+MeH. This tethering is probably by two or perhaps three hydroxyl groups, though on activation some of these may become terminal ligands. The range of Mo-O,,,~ bond distances given by Iwasawa" is large, ranging from 1.72 to 2.10 A, hence definite assignment of the bond types would seem difficult, especially as not all single molybdenum-oxygen bonds are necessarily tethering, they may be part of an oxygen bridge. Something of the range of Mo-0 distances may be identified from the complex K,[Mo,05(oxa- late)z(OH2)2].24 Each Mo centre cpntains two terminal M=O near 1.70A, with an Mo-0-Mo bond length of 1.88 A, and terminal Mo-0 distances of ca.2.15 8, to oxalate and 2.33 8, to the water ligands. A very tentative description of the mean site might be taken as [MoO,(M~)~(O-)]" . It is worth noting that neither in this nor in the acetate system could any surface aluminium atoms be seen in the EXAFS, though had the spectra been recorded at lower temperatures this situation might have been different. The second question then concerns the catalytic species during metathesis. Modifying the Chauvin mechanism to allow for the Schrock observation of the co-existence of two types of metallacyclobutane structures during alkene metathe~is,~' the possible observ- able species may be described as 1-3 in scheme 1. Although there are virtually no structural data on model Mo compounds, there is now a representative set of structure determinations on W complexes; the close similarity of the atomic radii of these elements allows the latter to be acceptable guides of interatomic distances, as presented in the The two sets of spectra for the Mo,( AcO),-derived catalyst in the presence of propene are not identical, but the results are similar; there are carbon atoms at between 1.8 and 1.9 A in both experiments, the distance expected for a carbene carbon.The molybdpum atoms are probably tethered to the surface by two oxygen atoms at around 2.05 A. In sample I the third shell at 2.58 A is difficult to assign. This distance is too long for a direct bond between the two atoms but too short for there to be a single atom between these two groups; it would imply an angle of around 100" at the carbene carbon.A mean structure of 4 may represent these results. The differences between sample I116 XAS Studies of Alkene Conversion Catalysts- R OH" ' 0 (4) Scheme 1 and sample I1 might be rationalised if one of the carbene units is transformed into a metallacyclobutane ( 5 ) ; presumably a slightly different coordination site alters the position of the surface equilibria shown in scheme 1. The carbons at 2.98 A may be attached to the carbene carbon, suggesting a Mo-C-C angle of around 126", which is in thc expected range. Although there is evidence of another molybdenum atom at 3.01 A in one of the spectra, there are not enough oxygens around the centres for a bridge between these atoms to be likely.The Li4[ Mo,Me,]-based catalysts differed in that there was strong evidence of Mo=O bonds (ca. 1.7 A) after exposure to propene. Again carbene formation is likely (the first shell for sample I1 may well have both 0 and C components). There is reasonable evidence of paired species present in the catalyst, though whether both molybdenum atoms play a part in the reaction is undetermined. The systems that have been studied here are perhaps not ideal subjects for XAS studies in that they are not clean and well ordered. However, without oxidation or reduction at elevated temperatures many adsorbed systems are complicated and dis- ordered. No assignment can be made about the oxidation state of the catalyst. Although the results obtained are not as clear-cut as might be desirable, a few common threads run through them which provide some insight into the metal coordination spheres.The actual structure of the catalyst is possibly dependent on the precursor, but the active catalyst is probably tethered by two oxygen atoms to the alumina surface and in some cases is part of a paired species. The pre-edge peak in the active catalyst XANES indicates a significant distortion from a centro-symmetric species. The use of promoting agents such as Sn( CH3)4 to activate heterogeneous metathesis catalysts" has provided strong evidence for the Chauvin metathesis mechanism in the heterogeneous reaction but there appear to be no previous reports of the presence of carbenes in active heterogeneous metathesis catalysts. The presence of carbon atoms at distances appropri- ate for carbenes from 1.80 to 1.96 A is thus the most important finding of this work.Although on this limited evidence it is impossible to comment on the mechanism of carbene formation or on the step from alkene and carbene to metallacycle and vice versa, the presence of such species on a heterogeneous catalyst strongly suggests that this is the propagating step in such catalysis. We thank the S.E.R.C. for support (to JTG and JFWM). Both the S.E.R.C. and the staff of the Daresbury Laboratory are thanked for providing the facilities at the S.R.S. We are grateful to Degussa AG for providing the oxide supports.J. Evans, J. T. Gauntlett and J. E W. Mosselmans 117 References 1 ( a ) Characterisation of Catalysts, ed. J.M. Thomas and R. M. Lambert (Wiley, Chichester, 1980); ( b ) Bimetallic Catalvsts: Discoveries, Concepts and Applications, J. H. Sinfelt (Wiley, New York, 1983); ( c ) X-Ray Absorption: Principles, Applications and Techniques of EXAFS, SEXAFS and X A NES, ed. D. C. Koningsberger and R. Prins (Wiley, New York, 1988); ( d ) J. Evans, in Catalysis, ed. G. C. Bond and G . Webb (Royal Society of Chemistry, Cambridge, 1989), vol. 8, p.1. 2 F. W. Lytle, G. H . Via and J. H. Sinfelt, in Synchrotron Radiation Research, ed. H. Winich and S. Doniach (Plenum Press, New York, 1980), p. 49. 3 J . L. van d e Venne, J. P. M. Rindt and G. J. M. M. Coenen, J. Colloid Interface Sci., 1980, 74, 287. 4 W. C. Conner, E. L. Weist, T. Ito and J. Fraissard, J. Phys. Chem., 1989, 93, 4138.5 Catalysis by Supported Metal Complexes, ed. Yu. I . Yermakov, B. N. Kuznetsov and V. A. Zhakarov, (Elsevier, Amsterdam, 1981). 6 J. L. Herrison and Y. Chauvin, Makromol. Chem., 1970, 141, 161. 7 R. R. Schrock, J. Organomet. Chem., 1986, 300, 249. 8 J. Smith, W. Mowat, D. A. Whan and E. A. V. Ebsworth, J. Chem. Soc., Dalton Trans., 1974, 1742. 9 S. A. Best, R. G . Squires and R. A. Walton, J. Catal., 1979, 60, 171. 10 Y. Iwasawa, S. Ogasawa and M. Soma, Chem. Lett., 1987, 1039; Y. Iwasawa, H. Ichinose, S. Ogasawa and M. Soma, J. Chem. Soc., Faraday Trans. I , 1981, 77, 1763. I I Y. Iwasawa, N. Ito, H. lshii and H . Kuroda, J. Chem. Soc., Chem. Commun., 1985, 827. 12 Yu. I . Yermakov and V. A. Zhakarov, Adv. Catal., 1975, 24, 173. 13 D. G . Ballard, Adv. Catal., 1973, 23, 263. 14 J. Schwartz and M. D. Ward, J. Mol. C'atal., 1980, 8, 465. 15 Y. lwasawa and S. Ogasawa, J. Chem. SOC., Farada-v Trans. 1 , 1979, 75, 1465. 16 N. Binsted, PAXAS, Programme for the Analysis of X-Ray Abssorption Spectra (University of 17 N. Binsted, J. Campbell, S. J. Gurman and I . Ross, EXCURVE, (S.E.R.C. Darebury Laboratory, 1988). 18 N Binsted, S. L. Cook, J . Evans, G. N. Greaves and R. J. Price, J. Am. Chem. Soc., 1987, 109, 3669. 19 R. W. Joyner, K. J. Martin and P. Meehan, J. Phys. C, 1987, 20, 4005. 20 J . M. Corker, J. Evans, H . Leach and W. Levason, J. Chem. Soc., Chem. Commun., 1989, 181. 21 V. A. Zhakarov, V. K. Dudchenko, E. A. Paukshtis, L. G. Karakchiev and Yu. I . Yermakov, J. Mol. Catal., 1977, 2, 421. 22 Phantom Oscillations, K. I . Pandya, J. van Grondelle, A. MuSoz Paez and D. C. Koningsberger (Eindhoven University of Technology, 1989). 23 Y. Iwasawa, Adv. Catal., 1987, 35, 187. 24 F. A. Cotton, S. M. Woodhouse and J. S. Wood, Inorg. Chem., 1964, 3, 1603. 25 J. Feldman, W. M. Davis and R. R. Schrock, Organometallics, 1989, 8, 2266. 26 R. R. Schrock, R. T. DePue, J . Feldman, C. J . Schaverian, J. C. Dewan and A. H . Liu, J. Am. Chem. Sor.., 1988, 110, 1423; M. T. Youinou, J. Kress, J. Fischer, A. Aguero and J. A. Osborn, J. Am. Chem. Soc., 1988, 110, 1488; M. R. Churchill, A. C. Rheingold, W. J. Youngs, R. R. Schrock and J. H. Wengrovius, J. Organornet. Chem, 1981, 204, C17; J. Feldman, J. S. Murzdek, W. M. Davis and R. R. Schrock, Organometallics, 1989, 8, 2260. 27 K-I. Tanaka and K. Tanaka, in Homogeneous and Heterogeneous Catalysis, ed. Yu. 1. Yermakov and V. Likholobov (VNU Science Press, Utrecht, 19861, p. 245. Southampton, 1989). Paper 9/05425A; Received 18 th December, 1989