J . Chem. Soc., Faraday Trans. 1, 1984, 80, 191-209 Mechanisms of 1,5-Dehydrocyclisation and Isomerisation of Alkanes on Iridium, Rhodium, Palladium and Platinum Films BY ODILLA E. FINLAYSON AND JOHN K. A. CLARKE Department of Chemistry, University College, Belfield, Dublin 4, Ireland AND JOHN J. ROONEY* Department of Chemistry, The Queen’s University of Belfast, Stranmillis Road, Belfast BT9 SAG, Northern Ireland Received 19th May, 1983 Deuterium tracer studies with a combined gas-liquid chromatography and mass spectrometry analysis of products and using 2,2,4,4-tetramethylpentane (TMP) as model reactant establish clearly that the selective cyclisation (SCM) of five-carbon chain alkanes on Ir, Rh, Pd and Pt is by the heterogeneous counterpart of reductive elimination of alkyls in organometallic chemistry.A comparison of cyclisation of TMP, 3,3-dimethylpentane and n-pentane on films of the same series of metals and an extended study of cyclisation of 3,3-dimethylpentane and of n-pentane on a series of Pt-Cu alloy films with variation of hydrogen partial pressure show that more than one mechanism of non-selective cyclisation (NSCM) takes place. The isotopic studies show clearly that a monoadsorbed intermediate is sufficient for bond-shift rearrangement of TMP to 2,2,5-trimethylhexane on Ir, Rh, Pd or Pt. An important major conclusion to be derived from the present work is that one surface metal atom comprises the active centre of the catalytic site for a variety of reactions in addition to simple hydrogenationdehydrogenation. The mechanisms of 1,5-dehydrocyclisation of alkanes and the reverse hydro- genolysis reaction on platinum surfaces have been debated actively in the recent literature. Two modes of reaction were recognised by Gault,’ namely a non-selective cyclic mechanism (NSCM) and a selective cyclic mechanism (SCM), and these were illustrated by the catalytic interconversion of methylcyclopentane, 2-methylpentane, 3-methylpentane and n-hexane.The NSCM allows interconversion between all four hydrocarbons whereas SCM does not allow methylcyclopentane and n-hexane to interconvert. It has been recognised that the nature of the catalyst and the experimental conditions employed determine the selectivity for NSCM or SCM. Further, while Pt has been found to be active for 1,5-dehydrocyclisation b either NSCM or SCM,l hydrogen only interconvert methylcyclopentane and 2- and 3-methylpentanes, i.e.exclusively SCM;2.3 Pd catalysts were suggested to cyclise by NSCM only.* Because the SCM involves two terminal C atoms Gault proposed a dimetallocarbyne intermediate, i.e. triple bonding of the two terminal carbons of the hydrocarbon chain to two contiguous metal site atoms. For the NSCM Gault proposed a metallo- 1 ,5-dicarbene specie^.^ Serious criticism of Gault’s arguments may be offered even on the basis of presently available experimental results. First, we note that carbenes are intermediates in the iridium catalysts of various particle sizes down to 15 K diameter and in excess 191192 CYCLISATION AND ISOMERISATfON OF ALKANES formation of carbynes.Further, the SCM is more important at relatively lower temperatures and higher hydrogen pressures whereas the NSCM is dominant at relatively higher temperatures and lower hydrogen pressures;lq therefore the NSCM must be more dehydrogenated than the SCM intermediate. Other intermediates suggested include the (highly dehydrogenated) pentadienyl radical for NSCM, alkyl/alkene insertion on Pd7b (implicitly NSCM)8 and an alkyl/carbene insertion reaction for Pt cyclisations8 for which Clarke and Rooney suggestedg reductive elimination of an ae-diadsorbed species by analogy with known organometallic reactionslO (see scheme 1). If an ae-type species were the intermediate for SCM, then cyclisation would be confined to the terminal C atoms for steric reasons only.Scheme 1. The present work is in two parts, first a mechanistic evaluation of the SCM on Ir, Rh, Pd and Pt films using deuterium exchange and combined gas-liquid chromato- graphy and mass spectrometry (g.l.c./m.s.), secondly a mechanistic investigation of NSCM with several reactant alkanes on the same series of metals and partly a comparative test with a series of Pt-Cu alloys. The sequence undertaken in the first part was as follows: (i) to prepare surfaces of Ir, Rh, Pd and Pt having a slow rate of alkane/deuterium exchange at the temperature required for C-C skeletal rearrangements and (ii) to analyse by g.l.c./m.s. the reaction product on these surfaces of a suitable model compound in excess deuterium gas : detailed mechanistic information was sought from such results mainly for the cyclisation reaction but also for isomerisation.The model compound chosen was 2,2,4,4-tetramethylpentane (TMP, 1) for the following reasons : (i) 1,5-dehydrocyclisation is confined to the terminal groups and (ii) the exchange pattern is expected to be simple owing to the presence of quaternary groups, and the exchange of the cyclic compound 1,1,3,3-tetramethylcyclopentane (TMCP, 2) is known to be essentially simple at low temperatures on Rh and Pt films." 2 is the cyclisation product from 1. The expected products of isomerisation are 2,2,4- trimethylhexane (2,2,4-TMH, 3) (by 1,2-methyl shift) and 2,2,5-trimethylhexane (2,2,5-TMH, 4) (by 1,2-neopentyl shift). The isomer 2,2,3,4-tetramethylpentane is much less probable because of the unfavourable 2,3-methyl shift required in its formation.Possible intermediates involved in the cyclisation are shown in table 1, which also gives the minimum number of deuterium atoms that would be incorporated into the product if cyclisation took place in excess deuterium rather than hydrogen. As will become apparent, suitable catalyst surfaces could be formed as described, and a decision on SCM proved to be possible for all four metals. As will be argued from the trends in reactivity of n-pentane, 3,3-dimethylpentane (DMP) and 1 on Ir, Rh, Pd and Pt in parallel with the isotopic work, there are indications that more than one NSCM is possible. Work by de Jongste and PeneP showed that ring opening of methylcyclopentane by SCM occurred on Pt-Cu alloys of 57 at.% Pt and this changes to NSCM between 25 and 5 at.% Pt. Work with C13-labelled pentanes allowed the same authors with Gault to confirm that above 490.E. FINLAYSON, J. K. A. CLARKE AND J. J. ROONEY I93 Table 1. Possible mechanisms of 1,5-dehydrocyclisation by SCM and minimum deuterium incorporation by each mechanism intermediate 2Hmi, ref. a& elimination 0 9 0 aae insertion aam aaaE aaam 1 8 2 Gault NSCM5 2 - - n 3 I l l It M M 5 at. % Pt SCM takes place and gradually this changes to totally NSCM at 21 at. % Pt.13 Other workG shows clearly that NSCM is favoured over SCM on Pt at low hydrogen partial pressure (see above). In view of the finding that Pt-Au films caused cyclisation of n-pentane but not of 3,3-dimethylpentane14 it was thought promising to compare the cyclisation of n-pentane with that of 3,3-dimethylpentane on a series of Pt-Cu alloy films, first with a standard reaction mixture and then with a mixture lean in hydrogen.In this way a clarification might be found of the possible versions of the NSCM. EXPERIMENTAL MATERIALS A standard static reactor (ca. 800cm3) with provision for preparing single or binary evaporated films was connected to an in situ g.1.c. unit and via a glass capillary to an AEI MS 1Oc2 mass spectrometer. Iridium, palladium, platinum and copper wires used to prepare films were Johnson-Matthey ' spectrographically standardised '. Rhodium wire was Materials Research Corporation, Marz grade. 1 was an API standard sample supplied for an earlier study by the API Project at the Carnegie-Mellon University, Pittsburgh.3,3-Dimethylpentane was K and K and the purity was verified by g.1.c. n-Pentane was Fluka puriss. 2 was prepared by Wolff-Kishner reduction of 1,1,3,3-tetramethylcyclopentan-2-one. The sample contained ca. 10% impurity and was used mainly for calibration purposes. All hydrocarbons were subjected to repeated freeze-thaw cycles in vacuum before use. Commercial hydrogen or 99.99% deuterium as appropriate was palladium-diffused. ISOTOPIC STUDIES OF THE SELECTIVE CYCLISATION MECHANISM (SCM) A number of variations of film deposition and subsequent film treatment were used in efforts to slow the rate of the exchange reaction of the hydrocarbon while still allowing skeletal rearrangements to occur. On the one hand the degree of sintering was varied.Treatment of194 CYCLISATION AND ISOMERISATION OF ALKANES the deposited film with air or oxygen, with and without heating, and heating of the film in oxygen mixed with a large excess of H, (or D,) were other variations. The final procedure adopted involved deposition of the metal film at 773 K followed by heat treatment in 3-4 Torrt H, (or D,) at 773 K for a minimum of 2 h. The temperature was then reduced to the reaction temperature prior to admission of the hydrocarbon + H, (or D,) mixture. The reaction mixture was 10: 1 : : H, (or D,): hydrocarbon, with 0.8 Torr hydrocarbon corresponding to 1.30 x 1019 molecules of reactant in the reaction vessel. The cyclisation reaction of 1 was first run in H, to establish the reaction temperature and was then rerun in D,.Additional experiments were on (i) the exchange reaction of 2 on Pd films and (ii) the reaction of 3,3-dimethylpentane on Ir, Rh, Pd and Pt films prepared as before. A 15% squalane column operating at 378 K was used in the in situ g.1.c. to separate 1 and its products. These column conditions were inadequate to separate 2 and the isomer 4. These products could, however, be separated in the g.l.c./m.s. unit. In the mass spectrometer at 18eV 1 gave very small amounts of the parent ion but a large pseudo-parent ion at m/e 113 formed by loss of a methyl group, namely (CH,),C-CH,-C+(CH,),. Peaks at m/e 112 and 11 1 due to the loss of one or two hydrogen atoms were 2.5 and 1.1 %, respectively, of the m/e 113 peak. The reaction of 1 and deuterium was monitored by g.1.c. and by m.s.either until a peak due to the combined cyclic product and isomer was seen in the chromatogram or until there was an increase in the ratio of m/e 11 1 to 112 in the mass spectrum (in practice 5-20 min at the selected temperature). The presence of ,H,-,H, cycloalkanes should be manifest as an obvious outgrowth of the exchange contour, as the cycloalkane had a ten-fold greater sensitivity in the MSlOc2 than did the alkane. In order to obtain a detaild analysis the reaction mixture was then quenched and the hydrocarbon removed for g.l.c./m.s. analysis. For this, the sample was dissolved in ca. 1 cm3 diethylether and injected onto a capillary OVlOl fused silica column temperature-programmed from 293 K at 4 K min-l. The sample was then bled into a mass spectrometer operating at 20 eV.Partial separation of the exchanged hydrocarbons occurs on the g.1.c. column, with the heavier deuteroisomers being eluted first. Therefore several m.s. scans through the g.1.c. peaks were necessary to give the genuine starting point of the deuterium distribution of each compound. The mass spectrometer scanned the selected mass range every In studies of the reaction of 3,3-dimethylpentane, reactant and products were separated on a column of either 15% squalane on Chromosorb P (80-100 mesh) at 333 K or 10% E30 on silanised acid-washed diatomite C (100-120 mesh) at 336 K. 2-3 S. RESULTS As indicated, an extensive programme of tests was conducted in which films (of Pd and Pt, principally) were formed in different ways or treated following formation with oxygen, preadsorbed hydrocarbon etc.The general observation may be made that ‘unsintered’ films gave over-rapid deuterium exchange with 1 relative to cyclisation. Deuterium-number contours had a maximum at a relatively high deuterium number, and peaks in the ,H,-,H, range were small. Thus mechanistically meaningful information could not be obtained from the contours. Karpinski employed platinum15 and palladium16 films sintered for 3 h at 800 K in 4 Torr D, and thereby succeeded in reducing the deuterium exchange rate sufficiently to study the mechanism of isomerisation of neopentane. In the present work sintering of Pt and Pd films for 1 h at 773 K only partly alleviated the exchange/cyclisation relative rate difficulty. For example, exchange of 1 with D, on Pd films sintered for 1 h at 773 K in 3-4 Torr D, gave a composite exchange contour$ centred at 2H2-2H, with maximum shifting to ,H, with time: ,H,, and ,H1 were very small.Significantly, f’ 1 Torr = 101 325/760 Pa. 1 Composite contour means the exchange distribution from 1, 2 and 4, and 3 if present, prior to g.1.c. separation.0. E. FINLAYSON, J. K. A. CLARKE AND J. J. ROONEY 195 sintering Pd films at 773 K for a minimum of 2 h gave a composite exchange contour from 2Ho-2H4 with maximum at 2H, shifting to 2H4 with time but [2Ho]TMP, [2Hl]TMP remain of significant size. Experience was similar with Rh and Ir, although these were much less extensively tested. The conditions described were therefore adopted for the detailed isotopic analysis work.The order of emergence of compounds on the OVlOl column was The first two peaks in the g.1.c. trace were thus due to [2H,l]- and [2H2]-2,2,5- trimethylhexane, respectively. The mass spectrometer scans the g.1.c. peak every 2-3 s and such multiple scanning through each peak is necessary to yield a genuine deuterium distribution pattern. Integration methods are available1' for the derivation of an average concentration of each deuteroisomer. It was not necessary to employ this for the present results, rather what was required was the starting point of the deuterium contour of the cyclic and isomeric products. No account was necessary of the dilution of the deuterium pool by hydrogen1*" because of the sharp cut-offs found at low deuterium number in the catalytic runs, as will be shown.Isotopic and fragmentation corrections were made on the results for TMP/D2 exchange only. With the other products it was ensured that the lowest deuterium number of the contour was genuine and not due to fragmentation of the next deuteroisomer of the contour. Positive identification of each reaction product was achieved by (i) preliminary identi- fication from g.1.c.-trace retention times coupled with the m/e value of the pseudo- parent ion (e.g. 11 1 for TMCP) and (ii) a comparison of the observed mass-spectral fragmentation pattern with that measured in preliminary calibration scans for each hydrocarbon. In particular, measurement of fragment peaks at m/e 70 and 71 is important for complete identification of 2 in the presence of considerable isomerised 1.(Thus at 20 eV the isomeric alkanes give a large peak at m/e 71, whereas 2 gives a large peak at m/e 70 and a smaller peak at m/e 71.) The procedure described is especially necessary because of g.1.c. overlap between compounds when deuterium substitution is extensive. IRIDIUM FILMS Both 3 and 4 as well as the cyclic product 2 are formed from the standard reaction mixture at 400 K (fig. 1). The amounts of 4 and 2 are approximately equal while the isomer 3 is present in much lesser amounts. The exchange contour of the isomer 4 clearly starts at m/e 114, i.e. the 2Hl isomer. This means that the bond shift (of a neopentyl group) requires only a monoadsorbed intermediate. A large mass-spectral peak at m/e 11 1 was present and had a maximum between the g.1.c.retention times of the two isomers 4 and 3. The assignment of this peak to the pseudo-parent of 2 was verified by the observation of a large fragmentation peak at m/e 70 associated with a small peak at m/e 71. Exchange of 2 was simple, and only small amounts of 2Hl were formed. This result proves that there is a mechanism of 1,5-dehydrocyclisation possible on Ir which involves direct ring closure from a di-a-adsorbed intermediate.196 100 0 t CYCLISATION AND ISOMERISATION OF ALKANES mass 120 100 120 I 1 i t I I I 15L I 162 L N I CI u - I u 1 I I I I I I I I I I I I - 170 I Fig. 1. Reaction of 1 on Ir at 400 K. Representative mass spectrometry scans (no. 454170) through a g.1.c. chromatogram. Relative concentration is given as a number for each high peak in this and later figures.The notation [2HI]-4 signifies monodeuteroisomer 4. 100 mass 120 100 120 1 I I 1 I i f - ' ' I L 2 c .- Y 5 s L t 1 6 2 155 L I v I' ", 160 / G I T= I Y L Fig. 2. Reaction of 1 on Rh at 513 K. Representative mass spectrometry scans (no. 155-171) through a g.1.c. chromatogram.0. E. FINLAYSON, J. K. A. CLARKE AND J. J. ROONEY 197 The isomer 3 (identity conclusively established from its fragmentation pattern) gave an exchange contour of the pseudo-parent ion insufficiently separated from the TMP 2Ho-2H, contour to give a mechanistically meaningful distribution. No detailed comment on the mechanism of this (methyl group) bond-shift reaction from the present deuterium contours is therefore possible. RHODIUM FILMS Reaction at 513 K (fig.2) gave 4 in much greater amounts than TMCP. Small amounts of the second isomer 3 were produced. The 4 was exchanged to 2H2, implying an aa-diadsorbed species but also small amounts of [2Hll]i~~mer were observed in accordance with the operation of the ap process. This feature of simultaneous maxima appearing in the [2H,]isomer and in the perdeuteroisomer (that possible by the ap process only) for exchange of alkane has previously been noted using unsintered Rh films at and below room temperature18b and can also be attributed to formation and migration of intermediate 2H, alkene to sites where it can only add back two 2H, atoms, but not undergo the ap process. Significantly, the contour for the 2,2,5-isomer clearly starts at m/e 114, i.e. the [2Hl]isomer.As for Ir, the bond-shift rearrangement reaction involves formation of a monoadsorbed intermediate. The cyclic product 2 was present as m/e 11 1 peak with a small amount of m/e 112. The identity of the cyclic product was confirmed by the fragmentation pattern as discussed for Ir . It may be concluded therefore, as for Ir, that the bond shift (of the neopentyl group) requires a monoadsorbed intermediate and that the 1,5-dehydrocyclisation involves the heterogeneous counterpart of reductive elimination of a 1,5-a-dialkyl intermediate. PALLADIUM FILMS It was found beneficial with both Pd and Pt to conduct runs using deuterium and g.l.c./m.s. analysis at ca. 50 K above the onset temperature of cyclisation rather than in the vicinity of that temperature because the relatively more rapid rate of isomerisation at the lower temperature tended to obscure the cyclisation product.3 and 4 were now found on Pd at 468 K but accompanied by a clear manifesta- tion of the deuterium contour of 2 (fig. 3). The contour of isomer 4 commenced clearly at m/e 114, i.e. the [2Hl]isomer. Therefore as on Ir and Rh a monoadsorbed intermediate permits the bond shift by neopentyl shift. 2 appeared at m/e 11 1, i.e. 2H,-cyclic, with some exchange to m/e 1 12, i.e. 2Hle-cyclic. The identity of the 1 1 1 peak was confirmed by the fragmentation pattern in the region m/e 70. Therefore cyclisation can occur on Pd by reductive elimination of a-diadsorbed intermediate. It is interesting that the isomer 3 was formed at approximately the same rate as 4.The former had exchanged to 2H12 with a maximum at 2H,-2H,. The minimum deu- terium content of this contour was not clear because of inadequate separation of the deuteroisomers of 1 and 3. The pattern of exchange of the parent compound 1 is also unclear owing to this lack of separation. However, in contrast to Ir, Rh and (see below) Pt, the exchange of 1 seems to tend towards multiple exchange with small [2Hl]deuteroisomer, maximum at 2H,-2H8 and exchange to 2Hlo. This suggests that an ay-type multiple exchange may be possible on Pd at these temperatures (where Pd exists as the a-hydride phase under conditions of the catalytic run). The formation of 3 appears to imply then a tendencyg for 1,2-methyl shift on Pd through a 1,3-diadsorbed species and a metallacarbonium-type intermediate (scheme 2).This mechanism permits methyl shift of 1 without having a strong repulsion of the bulky neopentyl group by the surface.198 CYCLISATION AND ISOMERISATION OF ALKANES mass 120 100 100 120 I i. c Fig. 3. Reaction of 1 on Pd at 468 K. Representative mass spectrometry scans (no. 100-116) through a g.1.c. chromatogram. mass 100 120 100 120 I 1 I 1 ' 1 I I 1 4 I n r" \ Y I l l I I I I I I I 1 1 1 1 1 I I I I I I - - c I n 'I 11 0 I . . I" I u I I 1 I l l l l l 0 t Fig. 4. Reaction of 1 on Pt at 523 K. Representative mass spectrometry scans (no. 102-110) through a g.1.c. chromatogram. m/e 1 15-1 25 represent a distribution of 2H,-2H,, deuteroisomers of product 4 (scans 102 and 105) or of products 3 and 4 (scans 107 and 110).0.E. FINLAYSON, J. K. A. CLARKE AND J. J. ROONEY 199 C c c \ / I,c /c\ ;\c - c c b + \w/ b- The reaction temperature was cyclisation (453 K). However, at Scheme 2. 2,2,4 - T M H PLATINUM FILMS raised by ca. 70 K above the onset temperature of this temperature isomerisation of 1 to both 3 and 4was very rapid. To poison the isomerisation reaction the hydrocarbon mixture was first admitted at 493 K and the temperature was raised to 543 K. A second hydrocarbon mixture was then admitted at 523 K and the reaction followed as previously described. Both 3 and 4 occurred as products along with 2 at 523 K (fig. 4). The isomer 4 gave a deuterium contour starting at m/e 114, i.e. the [2Hl]deuteroisomer with a maximum at the [2Hll]i~~mer (cut-off) moving to a maximum at the [2H,]isomer with successive mass-spectral scans.The peak at m/e 11 1 in scan 107 is due to the [2Ho]isomer of 2 as confirmed by the fragmentation pattern as for Ir, Rh and Pd. Some 2 with m/e 112 and 113 was also produced arising from exchange to the cyclic [2H2]isomer. The peak at m/e 111 in earlier scans (e.g. 102) was not due to 2 but to the loss of a methyl group from the exchanged 4 (i.e. m/e 126). The isomer 3 showed as a contour with a local maximum at 2Hl, (e.g. scan 107). Again inadequate separation conditions prevent the minimum deuterium content of this isomer from being determined. We note that the formation of 4 was favoured over 3. It may be concluded that one of the bond-shift reactions on Pt (neopentyl shift) requires a monoadsorbed intermediate.Also, 1,5-cyclisation can occur by elimination of a diadsorbed intermediate. PLATINUM-COPPER ALLOY STUDIES Alloy films were prepared as previously reported for Pt-Au.14 Some difficulty was experienced in preparing films of high Pt content owing to the relatively high volatility of copper. The Pt-Cu alloy system forms a continuous series of solid solutions at equilibrium at all temperature^.'^ The films prepared in the present work, which had undergone a homogenisation stage at 773 K, showed two regions of composition in the nominal composition range 50-80 at. % Pt. This is in contrast also to the results of de Jongste,20 who prepared Pt-Cu alloy catalysts which were single-phased at all compositions and which obeyed Vegard’s law. The two-composition region characteristic of the present films is probably due to an insufficient homogenisation temperature.Higher temper- atures are not feasible because of the Pyrex construction of the vessel. At compositions of < 50 at.% Pt lattice parameters agreed with Vegard’s law, indicating bulk homo- geneity on an atomic scale. No indication of superlattice formation was detectable in X-ray diffraction. From the foregoing, X-ray diffraction could be used to test lateral homogeneity of films. Thus for one film having a nominal composition of 23 at.% Pt the mean deviation was f 1.7 at.% Pt with a maximum deviation of k6.5 at.% Pt in samples of metal sampled from around the vessel.200 CYCLISATION AND ISOMERISATION OF ALKANES Table 2. Reaction of 3,3-dimethylpentane + hydrogen on Pt-Cu films initial product distribution (wt%) (10: 1 hydrogen to hydrocarbon ratio) at.% Pt T/K C,-C6 alkanes MCPa 1 , 1 -DMCPb Xc Tol/MCHd k/% min-l 77 518 550 567 59 5 35 497 535 573 600 638 28 576 603 636 22 18 635 0 695 98.7 1.2 86.1 3.9 54.6 2.0 76.2 - 100.0 - 100.0 - 27.4 - 9.3 95.8 89.8 - 88.7 - 57.7 - inactive to 609 K 100.0 - 100.0 - - - - - 70.9 0.3 90.7 - 3.5 10.2 4.2 4.6 36.9 - - - 10.0 14.9 22.0 - 1.5 0.8 - 2.5 5.3 0.22 0.28 0.33 0.002 0.001 0.0 10 0.0 19 0.007 0.004 0.006 0.007 0.56 0.002 0.002 a MCP, methylcyclopentane; * 1,l-DMCP, 1,l-dimethylcyclopentane; X, other C,-ring isomers including unsaturateds; Tol/MCH, toluene + methylcyclohexane. It is probable that film surfaces are more copper-rich than indicated by the nominal alloy composition even in the range of bulk homogeneity.Two ion-scattering spectroscopy studies2'? 22 vindicate theoretical expectation^^^ in showing moderate surface enrichment with copper. Thus bulk concentrations of 13 and 18 at.% Cu showed surface compositions of 48 and 5 1 at. % Cu. A preliminary Auger electron spectroscopic study on 25 at.% Pt-Cu gave a surface concentration of 15 at.% Pt.24 Surface copper enrichment of Pt-Cu alloys has been found to be dependent on crystallite size. 25 The reactions of 3,3-dimethylpentane and n-pentane were studied on the Pt-Cu films at hydrogen to hydrocarbon partial-pressure ratios of 10: 1 (tables 2 and 3) and 4: 1 (tables 4 and 5). The overall activity of the surfaces (total reaction) for both 3,3-dimethylpentane and n-pentane decreased (by 20-fold or more) as Cu was added in both series of experiments. Cyclisation was generally the main process on all films at both partial-pressure ratios : however, in detail a distinction becomes apparent between the two reactants.By comparison of the 3,3-DMP and n-pentane cyclisation reactions at 10: 1 ratio it is seen that 3,3-DMP failed to cyclise to 1,l-DMCP at 22 at.% Pt whereas n-pentane continued to form cyclopentane on alloys of 18 at.% Pt. [We estimate that transition compositions of alloys for catalysis were significant to ca. 2% Pt even though the composition spread is greater (see above).] 100% Cu films failed to cyclise 3,3-DMP to 695 K. By reducing the hydrogen to hydrocarbon ratio from 10: 1 to 4: 1 the overall activity of the films decreased by a factor of ca.2-4. 3,3,-DMP failed to cyclise at a higher Pt content, i.e. below 39 at. % Pt (cyclic product not produced at 27 at. % Pt) whereas n-pentane continued to undergo cyclisation at lower Pt compositions ( < 23 at. % Pt). There is thus a widening of the range in which 3,3-DMP fails to cyclise while n-pentane continues to do so as the hydrogen pressure is decreased. It may thus be deduced that0. E. FINLAYSON, J. K. A. CLARKE AND J. J. ROONEY 20 1 Table 3. Reaction of n-pentane+ hydrogen on Pt-Cu films initial product distribution (wt %) (1 0 : 1 hydrogen to hydrocarbon ratio) at.% Pt T/K C,-C, i-C, c-dieneu c-eneb c-CSc > CSd SHe SIe SCe k/% min-' 100 546 575 607 30 559 589 603 23 540 57 1 609 18 5 56 580 608 644 673 22.3 26.9 - - 50.8 - 0.22 0.27 0.51 69.9 14.4 0.7 4.0 10.8 - 0.70 0.14 0.16 67.1 7.4 3.7 5.7 16.1 - 0.67 0.07 0.26 8.3 8.9 - 2.6 80.1 - 0.08 0.09 0.83 5.4 5.2 2.9 7.5 79.0 - 0.06 0.05 0.89 9.0 1.1 27.1 46.6 16.2 - 0.09 0.01 0.90 15.1 6.4 - 7.2 71.3 - 0.15 0.07 0.78 8.7 1.0 20.3 26.6 43.2 0.4 0.09 0.01 0.90 13.8 1.4 51.6 19.8 13.4 - 0.14 0.01 0.85 75.9 9.3 - - 11.3 3.5 0.64 0.08 0.10 15.5 3.1 - 4.6 75.5 1.3 0.16 0.03 0.80 8.8 1.0 25.9 15.1 49.2 - 0.09 0.01 0.90 28.7 1.9 56.6 11.5 1.1 0.2 0.29 0.02 0.69 26.8 - 30.4 18.6 24.2 - 0.27 - 0.73 0.29 0.13 0.29 0.01 1 0.05, 0.036 0.003 0.0 16 0.017 0.001 0.005 0.010 0.006 0.008 c-diene, cyclopentadiene; c-ene, cyclopentane; c-C,, cyclopentane; > C, hexane and S,, S,, S,, selectivity for hydrogenolysis, isomerisation and cyclisation, respectively. Table 4.Reaction of 3,3-dimethylpentane + hydrogen on Pt-Cu films benzene; initial product distribution (wt%) (4: 1 hydrogen to hydrocarbon ratio) at.% Pt T/K C,-C, alkanes MCP" 1,l-DMCP' Xu Tol/MCH' k/% min-' 100 483 516 551 589 39 552 576 618 27 581 612 645 28.9 43.5 53.8 1.3 100.0 35.9 51.7 67.5 100.0 77.3 - 7.2 63.9 2.2 51.6 - - 30.7 0.7 51.6 0.8 58.4 6.7 - 39.8 6.7 26.0 6.0 - - - - - - - - - - - - - 2.7 14.8 46.3 - 1.8 6.5 16.7 - 0.032 0.042 0.024 0.034 0.001 0.004 0.003 0.001 0.003 0.008 See table 2. the n-pentane molecule has available to it an additional mechanism of cyclisation not possible for 3,3-DMP, and that this mechanism has an intermediate which is more dehydrogenated than that for 3,3-DMP. It is consistent with this deduction that n-pentane cyclised more rapidly on Pt, Pd, Ir and Rh films (table 6) than did 3,3-DMP (table 7). In the reaction of n-pentane, cyclisation is accompanied by considerable monoene and diene formation both on the present Pt-Cu films and on the Pt-Au films of previous work,14 where two Pt-Au films ( 5 and 33% Pt, respectively) were active in n-pentane cyclisation but inactive for 3,3-DMP cyclisation because development202 CYCLISATION AND ISOMERISATION OF ALKANES Table 5.Reaction of n-pentane + hydrogen on Pt-Cu films initial product distribution (wt%) (4: 1 hydrogen to hydrocarbon ratio) at.% Pt T/K C,-C, i-C, c-dienea c-enea c-C, > CSa k/% min-l 37 533 2.9 5.9 0.6 90.6 - 0.018 566 2.3 3.6 1.4 4.8 87.5 0.4 0.042 599 9.6 - 15.6 53.2 20.1 1.5 0.010 23 538 95.0 - 0.003 5.0 - 97.0 3.0 0.0 10 57 1 603 2.5 0.4 42.7 31.4 12.5 10.5 0.014 619 2.9 0.4 31.9 29.2 32.6 3.0 0.0 14 - - - - - - - a See table 3.Table 6. Reaction of n-pentane + hydrogen on sintered metal films initial product distribution (wt%) metal T/K C,-C, alkanes i-C, c-C,-enea c-C, > C,b k/% min-l ref. Ir 52 1 73.0 2.2 - 24.8 - 0.8, 7 556 49.0 7.2 43.8 - 1.3, - 1.3, 589 100.0 658 91.0 0.3 trace 0.4 8.3 0.75 605 37.8 1.5 13.6 47.1 - 0.06, - - - - Rh 574 99.0 0.5 trace 0.5 - 0.37 26 Pd 573 28.1 3.4 68.5 - 0.18 7 Pt see table 3 - a c-C,-ene cyclopentene and cyclopentadiene: > C , n-hexane and benzene. of the additional unsaturation required is prevented by the gem-dimethyl group. We note, lastly, that de Jongste et a1.12 found for a series of Pt-Cu alloys that ring opening of MCP (hydrogen:MCP: : 17: 1) was by NSCM only between 25 and 5 at.% Pt but because surface compositions are not known for either set of alloy catalysts, so that 'matching' of bulk compositions is not possible, we restrict ourselves to the comment that a NSCM must be operating (at the least) at < 22 at.% Pt on the alloy surfaces of the present work at 10: I hydrogen: hydrocarbon pressure ratio.(In drawing this conclusion the assumption is made that cyclisation and ring scission by NSCM both have the same mechanism). The cyclisation activity with n-pentane, 3,3-DMP and TMP are compared for metal films of Ir, Rh, Pd and Pt (tables 6-8). At each of several temperatures the rates increase in the order0. E. FINLAYSON, J. K. A. CLARKE AND J. J. ROONEY 203 Table 7.Reaction of 3,3-dimethylpentane + hydrogen on sintered metal films initial product distribution (wt%) metal T/K C,-C, alkanes MCP 1,l-DMCP Tola k/% min-I Ir 405 433 49 1 Rh 507 533 557 Pd 439 460 484 514 Pt 473 509 542 580 52.2 66.5 95.1 91.8 75.5 89.3 52.9 20.7 7.1 22.2 34.0 54.1 73.3 - 47.8 10.5 0.9 2.0 1.7 4.6 6.3 15.6 4.6 4.1 1.7 - 23.0 4.0 6.1 22.8 10.7 trace 47.1 74.7 86.6 62.2 57.1 32.6 11.5 - 0.007 - 0.016 - 0.12 - 0.09 - 0.05, - 0.08 - 0.002 - 0.006 - 0.014 - 0.05, 4.3 0.06, 9.2 0.10 13.5 0.09 - - a Tol, toluene, methylcyclohexane and other dimethylcyclopentanes (C, isoalkanes are not resolved from reactant peak in g.1.c.). Table 8. Reaction of 2,2,4,4-tetramethylpentane on sintered metal films initial product distribution (wt%) from g.l.c./m.s.: see footnotes partial: see footnotes 2,2,5-TMH k/% 2,2,5-TMH + k/% 1,1,3,3-TMCP: min-lb metal T/K C,-C, alkanes 1,1,3,3-TMCP min-l 2,2,4-TMH (corr.) Pt 459 487 Pd 442 479 Ir 400 430 462 504 Rh (i) 494 513 Rh (ii) 517 1.9 9.3 100.0 19.3 89.5 90.4 89.2 98.6 96.5 92.2 96.8 98.1 90.7 80.7 10.5 9.6 10.8 1.4 3.5 7.8 3.2 - 0.07, 0.06, 0.003 0.05, 0.003 0.008 0.028 0.26 0.027 0.05, 0.30 78: 1:9 (523 K) 84: 1:94 (468 K) 25:l:l (400 K) - 8: 1:0 (413 K) 0.09 0.08 0.003 0.1 1 0.003 0.01 0.03 0.27 0.03 0.06 0.30 a Estimate for each metal of the ratio of the three rearrangement products (from g.l.c./m.s.): 2,2,5-TMH and 1,1,3,3-TMCP are not resolved in the in situ g.1.c. and 2,2,4-TMH is not resolved from the 2,2,4,4-TMP reactant. Total conversion rate corrected using product ratios in previous column and assuming these ratios to apply at the several reaction temperatures.204 CYCLISATION AND ISOMERISATION OF ALKANES Also the onset temperature for cyclisation decreases in the same order.The a& mechanism of cyclisation known to occur with TMP on these metals is also possible for 3,3-DMP and n-pentane. Pt gives comparable rates of cyclisation for the substituted pentanes while Pd cyclises 3,3-DMP at a temperature 20-30 K lower than that required to cyclise TMP. Muller and Gault similarly showed a lower temperature for cyclisation of 2,2,4-TMP and 2,2,3-TMP than for tetramethylpentane on Pd films.* These results support the existence of a cyclisation mechanism in addition to the a& mechanism, one of which is available to (b) above but not to (a).This further mechanism cannot involve a completely dehydrogenated intermediate as previously proposed7 because the presence of the gem-dimethyl group blocks its formation. Finally, the Pt-Cu alloy results show a clear distinction between cyclisation of 3,3-DMP and n-pentane, the latter having available a route requiring a substantially dehydrogenated intermediate. DISCUSSION SELECTIVE l75-CYCLISATI0N MECHANISM For Pt catalysts the NSCM mechanism is generally favoured over SCM at higher metal dispersion,l higher temperature' and lower hydrogen partial pressure.6 Under various conditions of dispersion, pressure and temperature, however, iridium has been shown to cyclise only by the SCM.2v TMP will necessarily cyclise by the SCM because only the terminal carbon atoms are available for cyclisation. The appearance of sub- stantial cyclic [2Ho]isomer at the onset of the deuterium contour of 2 on Ir (fig.1) shows beyond doubt that cyclisation of 1 occurs by formation of an ae-di-o-adsorbed intermediate as in scheme 3. Organometallic analogues are known for this mechanism Yr ' Ir Scheme 3. of reductive elimination. C,, C, and C, rings, respectively, are reductively eliminated from the corresponding polymethylene di-a-complexes of platinum in which electron withdrawing ligands are also coordinated.l0* 2 7 9 28 Although carbynes are reasonably well documented, molecular dicarbynes have still to be reported requiring necessarily a binuclear or larger metal complex. Rhodium, palladium and platinum have also been found to cyclise 1 to 2 by the same a& reductive elimination mechanism which involves removal of the minimum number of hydrogen atoms that is required before ring closure is possible.Other experimental facts apart from the definitive cyclic 2Ho compound produced on each metal support this mechanism, as follows. Assuming an aam intermediate for the formation of 2 then the hydrogenolysis of the cyclic compound (thermodynamically the more favoured direction of reaction) which involves carbene formation should occur at the same temperature as cyclisa- tion of the alkane. On Pd films 2 failed to hydrogenolyse below 493 K whereas 1 cyclised at 436 K. Nickel films also failed to hydrogenolyse 2 at 419 K.ll The temperature for the hydrogenolysis reaction is experimentally higher then than the temperature for cyclisation. Therefore cracking of 2 and cyclisation of 1 must involve different mechanisms. The law of microscopic reversibility is not violated here.The cycloalkane is chemisorbed by the relatively easy C-H rupture and the resulting monoadsorbed species becomes a/? diadsorbed etc. rather than sustaining the more difficult metal-atom insertion reaction. Cracking may then proceed via the dicarbene0. E. FINLAYSON, J. K. A. CLARKE AND J. J. ROOMY 205 intermediate visualised by Gault. [Significantly, exchange of 2 with deuterium on Pd films at 440 K gave a large percentage of 2Hl (and also some 2H2-2H5) suggesting that carbene formation was not occurring at this temperature.] Gault noted29 that ring closure across the transannular positions of Cg-C,, ring compounds (scheme 4) to give bicycl~alkanes~~ is not possible by dicarbene or dicarbyne formation owing to conformational strain.ThePt-catalysed formation at Scheme 4. 703 K of triamantane by 1,6-cyclisation of a polycyclic alkene structurally related to it reported by Burns et aL31 is also not possible by carbene or carbyne intermediates owing to steric hindrance, Gault suggested a 1,5-di-a-adsorbed intermediate for transannular ring closure. He dismissed the possibility of such a mechanism of ring closure for small molecules such as n-hexane because he considered it not to give an explanation of the changes observed in selectivity on large-sized particles of Pt, on alloying with a Group IB metal, or the changes in selectivity of the different metals Ir, Pd and Pt.29 Hydrogen-pressure dependences were not considered and these indeed provide the key to these problems, as developed in the following paragraphs.The mechanism of SCM on Ir, Rh, Pd and Pt involving reductive elimination of an a&-diadsorbed species requires only one metal atom. Conclusions arrived at previously as to the surface-structure sensitivity of SCM on Pt are affected by the confusion that the ring-opening reaction is in practice by a different preferred mechanism to the ring-closure direction of reaction as already discussed : experimental results were in the main for the former in Gault’s programme. For Ir, however, the cyclisation reaction (SCM) was, qualitatively at least, independent of Neither the SCM nor the NSCM reaction took place on a PtAu/Aerosil alloy having good activity for the bond-shift reaction and ring enla~gernent,~~ suggesting a surface-structure sensitivity which was not satisfied on this severely heat-treated surface. The question of the surface-structure sensitivity of selective 1,5-cyclisation cannot be regarded as settled, but on balance we incline to the view that it is surface-structure sensitive.However, this may be resolved because as the intermediate lies edge-on to the surface, steric hindrance by methyl or larger alkyl groups in the 1 or 5 position of the alkane chain will block the formation of the a&-species, as may readily be appreciated from a molecular model. The need for diterminal carbon atoms for SCM is not therefore due to dissociation of the hydrogen atoms but is a steric factor only.Finally, in disagreement with Gault and coworkers,* who suggested that Pd catalysts cyclise by NSCM only, Pd films have now been shown to cyclise 1 at 436 K upward. Muller and Gault failed to find cyclisation of 1 at 573 K on Pd films.8 NON-SELECTIVE 1 ,SCYCLISATION The NSCM is important at lower hydrogen pressures and generally occurs at higher temperatures than the SCM. The SCM involves an ae-intermediate, and therefore a206 CYCLISATION AND ISOMERISATION OF ALKANES more dehydrogenated intermediate than this is required for NSCM. In the paragraphs which follow the possible subdivision within the general class of non-selective I ,5-cyclisations will be considered. As discussed in the Results section the cyclisation of 3,3-DMP occupies an intermediate position between n-pentane and 1 in the rate of reaction at a given temperature and in the onset temperature for cyclisation.The 3,3-DMP cyclisation may be intermediate also in the degree of dehydrogenation of the reaction intermediate as follows. The results with Pt-Cu films show that 3,3-DMP fails to cyclise on alloys at low % Pt while n-pentane continues to do so. Decrease of hydrogen partial pressure brings about a widening of the region of alloy composition in which this distinction applies. This result implies that the additional route available for n-pentane cyclisation involves a more dehydrogenated intermediate than for 3,3-DMP cyclisation. Ponec et ~ 2 Z . l ~ have concluded that NSCM-type cyclisation takes place at < 21 at.% Pt in PtCu alloys in their work.We suggest that both n-pentane and 3,3-DMP cyclise by NSCM at low percentages of platinum in our Pt-Cu films, the additional mechanism for n-pentane being the pentadienyl -, cyclopentenyl route7 precluded for 3,3-DMP by the gern-dimethyl group. Only under extreme conditions of low hydrogen concentration, e.g. alloying with Cu or Au, or at high reaction temperatures,14 is the completely dehydrogenated (pentadienyl) intermediate of importance. This is the end species in a progressive series of partly dehydrogenated intermediates : a&, a@, a&, a@$&. The ap& mechanism available for 3,3-DMP may be an alkyl/alkene or a& reductive elimination from an initially n-monoadsorbed chain having lengthened surface sojourn time to permit successful chelation to the site atom.The statistical ratios of the hydrogenolysis products of methylcyclopentane by NSCM may be explained by a completely dehydrogenated intermediate (n-bonded to a Pt atom) shown in scheme 5 or by the partly dehydrogenated intermediates alkyl/alkene ( a p ~ ) or alkyl/allyl (aby~) insertion. [We note here that the latter two modes (abc) or (abyc) may effect direct 1,6-closure as well (NSCM).] The conversion from n-ally1 to 7 ‘ I J Scheme 5.0. E. FINLAYSON, J. K. A. CLARKE AND J. J. ROONEY 207 pentadienyl species or from a-adsorbed alkyl to n-adsorbed alkene/a-adsorbed alkyl gives an equal chance of breaking all bonds in the MCP ring. Decreasing the surface hydrogen pressure should favour the NSCM by the above mechanisms over SCM.We may summarise the possible intermediates for various pentane chains as follows : cyclisation intermediate SCM only a& apc apy& SCM, NSCM + a& SCM, NSCM a& as& apy& SCM, NSCM n Some deductions may be noted. For n-pentane cyclisation the ease of formation of the possible intermediates is in the order apyde > ap& > a&, the first of these resulting from progressive dehydrogenation, i.e. a -+ ap -+ apy + apyd -+ apydc, whereas a/% requires the monoadsorbed species to become diadsorbed, i.e. it is not progressive. Therefore n-pentane may cyclise under conditions where 3,3-DMP fails to. Further, factors which alter the surface hydrogen concentration, e.g.carbiding, al- loying, temperature and possibly aromatics, will change the selectivity of the catalyst for the cyclisation reaction. The dispersion of the catalyst may also be important for selectivity for cyclisation; it certainly is for ring opening. Gault and his group have found29 that NSCM (ring opening) occurs at Pt diameters of < 25 A, and not at all for > 25 A. As noted earlier, a PtAu surface subjected to high-temperature reduction retained activity for bond-shift rearrangement and for dehydrogenation but was inactive for n-pentane cy~lisation,~~ suggesting that sites necessary for both SCM and NSCM had been removed. Even highly dispersed Ir catalysts, however, are inactive in NSCM (both ring opening and ring scission).273 From the foregoing remarks the existence of the required sites may be a necessary but not sufficient condition for non-selective cyclisation and the ability of the surface to effect the required degree of dehydrogenation of the reactant may be equally necessary.1,2-BOND-SHIFT REARRANGEMENTS The present work clearly confirms that an adsorbed alkyl is already at a sufficient degree of dehydrogenationl59l6 for bond shift as first claimed from studies of isomerisation reactions of caged hydrocarbon^^^ (scheme 6). This is potentially very - 1 'ct - \ C"3 M CH,- C .- CHYA-C-' RCH2-C -CH3 I M 1 '3 M Scheme 6.208 CYCLISATION AND ISOMERISATION OF ALKANES important when viewed in the light of the fact that various alkyl-cobalt complexes which are intermediates in Vitamin B,, catalysed vicinal interchange reactions, e.g.Q-methyl aspartate e glutamate (cJ neopentane + i~opentane),~~ also undergo 1,2- bond-shift is~merisations.~~ Another very important aspect is that while it was necessary to use heavily sintered films in this work each metal still exhibits its characteristic behaviour (most particularly for the ap process), as noted over twenty years ago for unsintered films at or near ambient ternperat~res.~~ We can therefore be confident that the nature of the sites is the same, only the numbers per m2 have changed. Pd still gives very extensive multiple exchange in contrast to Rh. CONCLUSIONS Firm evidence has been provided here that in the mechanisms of ‘non-destructive’ reactions such as cyclisation and 1’2-bond-shift rearrangements of hydrocarbons one metal atom only is required as the catalytic centre, just as previously argued for hydrogenation and dehydr~genation.~~ Therefore the claim that catalysis is often due to interconversion of various intermediates as ligands of one surface metal atom or ion also extends to the higher-temperature reactions as well.This is a very important philosophy because of the current importance attached to such terms as ‘facile’, ‘demanding’ and ‘structure sensitive’ and an over-ready acceptance of such jargon without critically examining what it means as far as surface sites are concerned. In the hydrocarbon field multiple bonding at the edges or faces of surface ensembles or clusters may only be important when carbene, carbyne and carbide intermediates are involved, i.e.in ‘destructive’ reactions such as hydrocracking. Even in the reverse re- action of hom~logation~~ the key homologation step of addition of methylene to an alkene may still be mediated by one metal atom. We thank Mr D. Clements and Gallaher Ltd for the g.l.c./m.s. analyses. G. Maire, G. Plouidy, J. C. Prudhomme and F. G. Gault, J. Catal., 1965, 4, 556. F. Weisang and F. G. Gault, J. Chem. Soc., Chem. Commun., 1979, 519. F. Weisang, Thbe Docteur-IngLnieur (University of Strasbourg, 1979). M. Hajek, G. Maire, A. 0. Cinneide, C. Corolleur and F. G. Gault, J. Chim. Phys., 1974, 71, 1329. F. G. Gault, Gazz. Chim. Ital., 1979, 109, 255. 0. V. Bragin, Z. Karpinski, K. Matusek, Z. Paal and P. Tetenyi, J. Catal., 1979, 56, 219. (b) F. E. Shephard and J. J. Rooney, J.Catal., 1964, 3, 129. J. M. Muller and F. G. Gault, J. Catal., 1972, 24, 361. J. K. A. Clarke and J. J. Rooney, Ado. Catal., 1976, 25, 125. ’I (a) C. O’Donohoe, J. K. A. Clarke, and J. J. Rooney, J. Chem. SOC., Faraday Trans. I , 1980,76,345; lo G. B. Young and G. M. Whitesides, J. Am. Chem. SOC., 1978, 100, 5808. l1 F. G. Gault, J. J. Rooney and C. Kemball, J. Catal., 1962, 1, 255. H. C. de Jongste and V. Ponec, Proc. 7th Znt. Congr. Catal., Tokyo, 1980, ed. T. Seiyama and K. Tanabe (Elsevier, Amsterdam, 1981), p. 186. l3 H. C. de Jongste, V. Ponec and F. G. Gault, J. Catal., 1980, 63, 395. l4 A. F. Kane and J. K. A. Clarke, J. Chem. SOC., Faraday Trans. I , 1980,76, 1640. l5 Z. Karpinski and L. Guczi, J. Chem. Soc., Chem. Commun., 1977, 563. l6 Z. Karpinski, Nouv. J. Chim., 1980, 4, 561. l7 R. S. Dowie, C. Kemball, J. C. Kempling and D. A. Whan, Proc. R. SOC. London, Ser. A , 1972,327, l8 (a) C. Kemball and J. C. Kempling, Proc. R. SOC. London, Ser. A , 1972, 329, 391 ; (b) F. G. Gault Is A. Schneider and U. Esch, Z. Elektrochem., 1944, 50, 290. *O H. C. de Jongste, F. J. Kuijers and V. Ponec, Proc. 6th Int. Congr. Catal., London, 1976, ed. 491. and C. Kemball, Trans. Furaday SOC., 1961, 57, 1781. G. C. Bond, P. B. Wells and F. C. Tompkins (The Chemical Society, London, 1977), p. 915.0. E. FINLAYSON, J. K. A. CLARKE AND J. J. ROONEY 209 21 M. J. Kelley, D. G. Swartzfager and V. S. Sundaram, J. Vac. Sci. Technol., 1979, 16, 664. 22 H. H. Brongersma, M. J. Sparnaay and T. M. Buck, Surf: Sci., 1978, 71, 657. 23 V. S. Sundaram and P. Wynblatt, Su$. Sci., 1975, 52, 569. 24 A. D. Van Langeveld, cited by F. Stoop, F. Toolenaar and V. Ponec, J. Chem. SOC., Chem. Commun., 25 J. H. Anderson, P. J. Conn and S. G. Brandenberger, J. Catal., 1970, 16, 326. 26 J. F. Taylor, Ph.D. Thesis (National University of Ireland, 1976). 27 P. Foley and G. M. Whitesides, J. Am. Chem. Soc., 1979, 101, 2732. 28 P. W. Hall, R. J. Puddephatt, K. R. Seddon and C. F. H. Tipper, J. Organomet. Chem., 1974,81,423. 2B F. G. Gault, Ad3. Catal., 1981, 30, 1 . 30 B. A. Kazanskii, E. A. Shokova, S. I. Khromov, V. T. Aleksanyan and Kh. E. Sterin, Dokl Akad. 31 W. Burns, M. A. McKervey and J. J. Rooney, J. Chem. Soc., Chem. Commun., 1975, 965. 32 J. K. A. Clarke, A. F. Kane and T. Baird, J. Catal., 1980, 64, 200. 33 A. Heumann, M. Reglier and B. Waegell, Angew. Chem., Znt. Ed. Engl., 1979, 18, 866. 34 P. Parayre, V. Amir-Ebrahimi and F. G. Gault, J. Chem. Soc., Faraday Trans. I , 1980, 76, 1723. 35 M. A. McKervey, J. J. Rooney and N. G. Samman, J . Catal., 1973, 30, 330. 36 J. M. Pratt, in B Twelve, ed. D. Dolphin (Wiley, New York, 1982), vol. 1 , p. 325. 37 R. Hamilton and J. J. Rooney, J. Mol. Catal., 1982, 17, 29. 38 C. Kemball, Bull. Soc. Chim. Belg., 1958, 67, 373. 39 J. J. Rooney and G. Webb, J . Catal., 1964, 3, 488. 1981, 1024. Nauk SSSR, 1960, 133, 1090. (PAPER 3/808)