6 CATALYSIS BY METALS By G. C. Bond (Johnson Matthey arid Co., Ltd., Exhibition Grounds, Wembley, Middlesex) THE literature on catalysis has been enriched by the publication during 1968 of books by Sir Eric Rideall and by Thomson and Webb,2 and of one with a strong practical bias edited by Ander~on.~ The regular publication of Aduances in Catalysis4 and of Catalysis Reuiews5 continues. The Fourth International Congress on Catalysis was held in Moscow and Novosibirsk; plans for publication of the papers are uncertain, but it seems likely that only the Rus- sian versions will appear. The three areas which have commanded most attention during the past year are (i) chemisorption of gases on clean metal surfaces, (ii) the structure of supported metal catalysts, and (iii) the mechanism of hydrogenation of unsaturated molecules and related processes.This year’s Report will therefore be devoted exclusively to a review of these three subjects. Chemisorption of Gases on Clean Metal Surfaces.-The study of the chemi- sorption of simple gas molecules on rigorously cleaned metal surfaces con- tinues to attract widespread attention more particularly from physicists than chemists : the increasing complexity and sophistication of the necessary apparatus perhaps makes this inevitable. The process is of great fundamental interest in its own right and knowledge acquired finds application or generates ideas in fields as separated as space research and catalysis, although only its relevance to the latter subject will concern this Report. Clean metal surfaces are highly reactive, and rapidly chemisorb at least a monolayer of oxygen in the presence of this gas: even gold now appears to have some attraction for atomic oxygen.6 To study the reaction of other less strongly adsorbing gases requires the metal surface to be cleaned and then maintained in ultra-high vacuum (u.h.v.) (less than 10-10 ton) for the duration of the experiment.Among the cleaning methods which continue to be applied are thermal desorption, ion bombardment (which severely disturbs the sur- face and has to be followed by annealing’), and thermal sublimation of ‘Concepts in Catalysis,’ E. K. Rideal, Academic Press, London and New York, 1968. ‘Heterogeneous Catalysis,’ S. J. Thomson and G. Webb, Oliver and Boyd, Edinburgh, 3 ‘Experimental Methods in Catalytic Research,’ ed.R. B. Anderson, Academic Press, ‘Advances in Catalysis and Related Subjects,’ eds. D. D. Eley, H. Pines, and P. B. 5 ‘Catalysis Reviews,’ eds. H. Heinemann, Marcel Dekker, New York, 1968, vol. 2(1). 6 R. R. Ford and J. Pritchard, Chem. Comm., 1968, 362; I. G. Murgulescu and M. 1. 7 R. L. Park and H. H. Madden, Surface Sci., 1968, 11, 188. 1968. London and New York. 1968. Weisz, Academic Press, London and New York, 1968, vol. 18. Vass, Rev. Roumaine Chim., 1968, 13, 373. 121122 G. C. Bond surface layers. It is of course important to have available a means of knowjng when a fully cleaned surface is obtained, and low energy electron diffraction (LEED) has proved very useful for this purpose. The problem of keeping the surface clean is not simple, especially since much of this work is now done with quite small areas (e.g.of single crystals), and high surface coverage by the background gas results in a matter of minutes rather than hours after cleaning is stopped even at 10-10 torr. Some cleaning procedures paradoxically have the effect of making the surface dirtier, as for example when they cause dissolved impurities such as carbon to migrate to the surface.* Metal films formed by condensing metal atoms, which have evaporated from an electrically heated wire, on to a cold surface continue to represent the easiest way of obtaining clean surfaces, especially since the films act as getters for residual gas during their formation. Such films when condensed onto glass are polycrystalline and expose no particular crystal plane, but nickel films condensed in u.h.v.on to sodium chloride preferentially expose the (100) plane: enthalpy and entropy of adsorption were obtained for xenon both on this film and on nickel-Pyrex as a function of coverage, and by use of a patch model of a heterogeneous surface it was concluded that no (100) planes were exposed in nickel-Pyrex.9 The chemisorption of hydrogen on a metal surface is formally the most simple conceivable process in this area of study, but for a number of reasons the results exceed expectation in complexity. Almost every study reveals evidence for two or three different states of binding of hydrogen atoms: these have different energies and the relative population of each state can vary with surface coverage.There are also some severe practical difficulties in working with films, not the least of which is to secure uniform contact between the incoming gas and the film, especially with nonsintered, porous films. These points are exemplified in two recent studies of hydrogen on molybdenum10 and on nickel11 films. The sticking probability s of hydrogen on molybdenum films was determined at 78, 195, and 300"~: at the highest temperature the adsorbed &oms are mobile and s on a sintered film was 0-68, falling linearly with increasing coverage. However at 7 8 " ~ s fell rapidly from 0.73 to 0.40, providing evidence for a precursor state.1° The measurement of surface potentials by the static-capacitor method enables transient changes during adsorption to be followed.Hydrogen adsorbs on nickel at 9 0 " ~ forming atoms which are strongly bound and show a negative surface-potential: this is the so-called p-species which is preferentially adsorbed on one group of sites. On unsintered films, later doses are adsorbed partly on the remaining less-active sites and partly as molecules (cry-species) on top of atoms. Sin- tering forms a plane which will adsorb hydrogen in the cry-form but sur- prisingly not in the p-form.ll The adsorption of oxygen and of carbon * E. M. A. Willhoft, Trans. Faraday SOC., 1968, 64, 1925. 9 B. G. Baker and L. A. Bruce, Trans. Faraday SOC., 1968, 64, 2533. 10 D. 0. Hayward and N. Taylor, Trans. Faraday SOC., 1968, 64, 1904. l1 T. A. Delchar and F. C. Tompkins, Trans. Faraday SOC., 1968, 64, 1915.Cntalysis by Metals 123 monoxide has been examined at 7 7 " ~ ; ~ ~ the change of electrical resistance of a platinum film is linear with the number of molecules adsorbed, implying only one form which is thought to be the linear one.l3 There has been much concern in the last year or so concerning the surface structure of finely divided alloy powders, particularly those of nickel-copper.The occurrence of phase separation has been demonstrated, but the point of chief concern is whether the catalytic properties of the surface layer are those of an alloy, each atom having the same behaviour, or are the sum of the properties of the two different kinds of atoms acting independently of each other. A study of the low-temperature adsorption of hydrogen on granular nickel and copper and their alloys has been performed;14 the adsorption is activated on nickel but curiously not on copper, while the independence of the amount adsorbed on bulk composition between 10 and 80 % copper argues that phase separation occurred.One of the most disturbing uncertainties in the early days of the application of LEED and field emission or field-ion microscopy concerned the chemical nature of the surface species responsible for the visual effects. Farnsworth has pioneered the use of mass spectrometry with LEED, whereby the adsorbed species are characterised by their mass spectrum after flash desorption. This combination has been recently applied to confirm earlicr work on the adsorption of carbon monoxide on the (100) plane of nickel.15 Even more elegantly the combination of a time-of-flight mass spectrometry with field-ion microscopy permits the identification of individual surface atoms or species.16 The Structure of Supported Metal Catalysts.-Many of the powerful physical methods for examining surface species mentioned above, and others not referred to, are applicable only to massive metal specimens and not to finely divided metal particles such as are present in supported metal catalysts. Thus, although the study of chemisorption on well-cleaned metal surfaces has been of value in suggesting possible or likely surface structures, its relevance to catalytic mechanisms has not been direct because of (i) the very different natures of the solids and of the purities of their environments and (ii) the fact that (with rare exceptions17) reacting systems have not been studied. A refreshing feature of the scene over the last two or three years has been the increased attention given to the study of practical supported cata- lysts, a study which in the Reporter's opinion has been too long neglected: undoubtedly the accurate description of a porous two-component supported catalyst is altogether more difficult than describing, for example a metal film, but the widespread industrial use of supported catalysts is gradually pricking the conscieiices of academic workers, with results which will now be re- viewed.l2 E. F. W. Thurston, Trans. Furuday SOC., 1968, 64, 2181. 13 T. Sugita, S. Ebisawa, and K. Kawasaki, Surface Sci., 1968, 11, 159. l4 D. A. Cadenhead and N. J. Wagner, J.Phys. Chem., 1968,72,2775. 15 M. Onchi and H. E. Farnsworth, Surface Sci., 1968,11, 203; see also D. Lichtman, F. M. Simon, and R. R. Kirst, ibid., 325. E. W. Miiller, J. A. Panitz, and S. B. McLane, Re!:. Sci. Itutr., 1968,39, 83. 17 R. F. Baddour, M. Modell, and U. K. Heusser, J . Pltys. Cltern., 1968, 72, 3621. E124 G . C. Borrd Experimental studies have had two main objectives; (i) an adequate description of the structure of catalysts, particularly of the mean size or size distribution of the metal particles by means of gas chemisorption, electron microscopy, X-ray diffraction and other methods ; and (ii) to establish whether the specific activity of metal crystallites is a function of their size, i.e. whether sites of different properties exist in concentrations which are dependent on the size of the particle.The above-mentioned methods of deriving information about the metal particle-size distribution are now almost routinely employed. The use of selective gas-chemisorption (usually either hydrogen or carbon monoxide) can, at best, however, give only a mean value of crystallite size, but the sim- plicity of the procedure makes it attractive as a routine means of quality control. The adsorption of hydrogen on Pt-AI203 has been measured as a function of temperature (250-450"~) and pressure (4 x 10-2-120 torr): saturation coverage is obtained at each temperature at 60-70 torr, but the ratio of number of hydrogen atoms adsorbed to total number of platinum atoms falls with increasing temperature as follows: 0.98 at 250°c, 0.71 at 350°c, and 0.57 at 45O"c.l8 A simple and rapid chromatographic method for determining the volume of hydrogen adsorbed by catalysts has been described,lg and differential thermal analysis has been applied to this prob- lem.20 Of course, the application of this method depends, for its success, on the gas not adsorbing on the support or migrating from the metal to the support during the measurement.Therz is a growing body of evidence to show that at sufficiently high temperatures such migration can occur. On heating in hydrogen Pd/A1203 or Pt/Alz03, to which has been added a small amount of ferric nitrate, an e.s.r. signal having g = 2.10 develops and this is attributed to metallic iron: it is suggested that the ferric oxide (formed by calcining the nitrate) is reduced by hydrogen atoms which have migrated from the metal particles.Au/A1203 is ineffective in producing an e.s.r. signal, which incidentally is not seen when iron is absent.21 Several other physical methods have been applied to the study of supported- metal catalysts in the past year, some for the first time. Nickel-chromia catalysts have been examined by using electrical conductivity and contact potential difference : the conductivity is metallic if the catalyst contains more than 96% nickel but it is semiconducting if the nickel content is lower. The contact-potential difference is a maximum at 92-93 nickel regardless of atmosphere.22 Strictly speaking, catalysts of these compositions should be termed 'promoted' rather than supported.Calcium ions in CaY zeolite have been ion-exchanged with Pt(NH3)42+ ions and the resulting material has l8 D. Cismaru and A. Fruma, Rev. Rournaine Chim., 1968, 13, 139, 679. l9 F. Figureas Rosa, L. de Mourgues, and Y. Trambonze, J . Gas Chromatog., 1968, 2O V. Zapletal, K. Kolomaznik, J. Soukup, and V. Ruzicka, Chem. listy, 1968,62, 210. 21 K. M. Sancier and S. H. Inami, J . Catalysis, 1968, 11, 135. 22 D. Tarina, E. Weissman, and D. Barb, J . Catalysis, 1968, 11, 348. 6, 161.Catalysis by Metals 125 been examined by the X-ray absorption-edge method.23 After reduction all the platinum is zero-valent, but 60% of the metal is soluble in cold concen- trated HCl (particle size about lOA) while X-ray diffraction shows the re- mainder to be ca. 60 8, in size. Exposure of the catalyst to hydrogen at 300"c produces a larger change in the absorption edge than does exposure at 100°c, due, it is suggested, to the formation of stronger bonds.Hydrogen adsorbed on Pt-SiO2 or Pt-Al203 gives two i.r. bands, at 2040 and 2110-2120 cm.-l; the intensity of the latter band is much enhanced if the catalyst is pretreated with oxygen, and it is proposed that the band should be assigned to weakly adsorbed hydrogen on areas of platinum oxide.24 One of the most promising of the newer techniques to be applied to the study of adsorption and cata- lysts is Mossbauer spectroscopy;25 its application to examining the prepara- tion of a supported gold catalyst has been described.26 It has sometimes been reported in the past that the ease of reduction of a supported oxide depends on the nature of its support; a detailed study of the reduction of nickel oxide on various supports has now a~peared.~7 Rates were measured manometrically in a closed circulating system; rates of reduc- tion at 400"c decreased in the sequence: SiO2 > Si02-Al203 > A1z03.With nickel oxide on silica, increasing temperature of treatment (300-7OO"c) before reduction decreased the rate of reduction at 400"c, and co-precipitated catalysts were shown to be much more difficult to reduce than impregnated catalysts. A Pt-SiOz catalyst prepared by ion exchange of silica with [Pt(NH3)4](0H)2 and reduced in hydrogen at 500"c has the metal in the form of about 15 A particles as determined by gas adsorption (5-30 A by electron microscopy): this is considerably smaller than the particles in a catalyst prepared by impregnation with H ~ P ~ C ~ C .~ ~ It has again been reported that 'extractable' platinum in Pt-Al203 catalysts shows activity specifically for dehydrocyclisation of straight chain alkanes but not for their dehydro- genat ion .29 Little has appeared during 1968 on the vexed question of whether specific activity (i.e. activity per unit surface area) is a function of crystallite size, although several papers presented at the Fourth International Congress on Catalysis bore on this point.30 The general consensus appears to be that specific activity is remarkably constant over a large range of average particle sizes for a number of different processes, with perhaps the exception of the hydrogznolysis of ethane.31 In an important papeF Boudart and his associates have shown that selectivity for isomerisation of neopentane to isopentane 23 P.H. Lewis, J. Catalysis, 1968, 11, 162. 24 D. D. Eley, D. M. Moran, and C. H. Rochester, Trans. Faraday Soc., 1968,64,2168. 25 W. N. Delgass and M. Boudart, Catalysis Rev., 1968, 2, 129. 26 W. N. Delgass, M. Boudart, and G. Parravano, J . P h p . Chern., 1968, 72, 3563. 27 V. C. F. Holm and A. Clark, J . Catalysis, 1968, 11, 305. 28 H. A. Benesi, R. M. Curtis, and H. P. Studer, J . Catalysis, 1968,10, 328. 29 N. R. Bursian, S. B. Kogan, and Z. A. Davydova, Kinetika i Kataliz, 1968, 9, 661. 30 Anon., Platinum Metals Review, 1968, 12, 136. 31 D. J. C. Yates and J. H. Sinfelt, J . Catalysis, 1967, 8, 348. 32 M. Boudart, A.W. Aldag, L. D. Ptak, and J. E. Benson, J . Catalysis, 1968, 11, 35.126 G. C. Bond (the other reaction being hydrogenolysis to isobutane and methane) on a number of platinum catalysts at 307 O c decreases significantly with increasing dispersion of the platinum, but this is chiefly because activity for hydro- genolysis varies by 300 while activity for isomerisation varies only by fifteen. Another vexed problem relating to supported-metal catalysts has been at least partly resolved in 1968. It is of course well known that the activity of a metal varies greatly depending on the chemical nature of its support and it has often been suggested that some electron transfer from metal to support or vice versa, modifying the electronic structure of the metal, is at least somewhat responsible.33 By comparing the electrical conductivities and activities for methanol oxidation of silver, zinc oxide and Ag-ZnO, it has been conclusively shown that valency induction indeed occurs.34 Mechanism of Hydrogenation of Unsaturated Molecules and Related Processes.-The simplest catalytic process which a molecule can undergo is the substitution of its hydrogen atoms by those of an isotope (e.g.D from deuterium gas or heavy water, T from tritiated hydrogen). Even if the mole- cule is unsaturated, it is possible to find conditions under which substitution can be effected without addition. The exchange of propane with deuterium on a platinum-containing fuel-cell electrode has been studied with a view to establishing how propane is and exchange of neopentane with deuterium has been studied with especial regard to diffusion effects in porous catalyst^.^^ Results concerning the exchange of cycloheptane and cyclo- octane have also been rep0rted.3~ Garnett and his associates continue to examine the exchange of unsaturated molecules with heavy-water catalysed by the Group VIII metals.38 The unresolved problems concerning the mechanism of hydrogenation of alkenes continue to excite much interest, and a number of groups of workers are now employing isotopic-tracer techniques. However, isotopic methods are not essential, and much interesting information comes from looking at double-bond isomerisation during hydrogenation and from measuring comparative rates of reaction.Rates of hydrogenation of a large number of pairs of alkenes over Pt-SiOz have been measured:39 if the ratio of rates for two molecules A and B is RAB elc., it was shown that RAB x RBC = RAC. Ru-C is reported to show very different rates of hydrogenation for alkenes of different structure.40 In the liquid-phase hydrogenation of a number of 33 F. Solymosi, Catalysis Rev., 1967, 1, 233.34 G. M. Schwab and K. Koller, J . Amer. Chem. SOC., 1968,90, 3078. 35 H. J. Barger and A. J. Coleman, J . Phys. Chem., 1968, 72, 2285. 36 F. G. Dwyer, L. C. Eagleton, J. Wei, and J. C. Zahner, Proc. Roy. SOC., 1968, A , 301, 253. 37 B. S. Gudkov, E. P. Savin, and A. A. Balandin, Zzoest. Akad. Nauk S.S.S.R., Ser. khim., 1968, 509. 38 G . E. Calf and J. L. Garnett, Austral. J . Chem., 1968, 21, 1221; G. E. Calf, B. D.Fisher, and J. L. Garnett, ibid., 947; G. E. Calf, J. L. Garnett, and V. A. Pickles, ibid., 961. 39 R. Maurel, J.-M. Elene, J.-F. Mariotti, and J. Tellier, Compt. rend., 1968, 266, C, 599; R. Maurel and J. Tellier, Bull. SOC. chim. France, 1968,4191. 40 L. Kh. Freidlin, E. F. Litvin, and S. K. Tulayev, Neftekhimiya, 1968, 8, 155.Catalysis by Metals I27 cycloalkenes, Pt-Ah03 causes very little isomerisation : dialkylcycloalkenes yield 43-58 % of the corresponding cis-cycloalkane,41 while Rh-C gives slightly more isomerisation and 52-61 % of cis-cy~loalkanes.~~ The be- haviour of all the noble Group VIII metals for isomerisation of branched alkenes has been described,43 the metals falling in the now familiar sequence: ruthenium and osmium are confirmed to be quite active for double-bond migration.44 A revised theory for calculating product distributions from the reaction of ethylene with deuterium has been described.45 The reaction of cyclohexene with deuterium has been discussed,46 and the Horiuti-Polanyi mechanism has been successfully extended to describe the formation of products from the reaction of methyl oleate with de~terium.4~ The reaction of ethylene with tritiated hydrogen has been studied but, unfortunately, only under condi- tions where mass-transport limitation is either partial or complete.4s Anoma- lous effects are observed when ap-unsaturated carbonyl compounds are reduced with tritiated hydrogen.49 Heats of hydrogenation of three- and four-membered rings are reported,50 and the mechanism of reduction of substituted cyclopropanes has been discussed.51 When the spiro-ketone (1) is hydrogenated, no cleavage of the cyclopropane ring occurs with Raney nickel, the product being the saturated spiro-ketone (2), but with Pd-C in a number of solvents cleavage occurs with the intermediate formation of the alkene (3).52 Mann and his co-workers have reported the kinetics of the hydrogenation of propyne, but-1-yne and allene over a number of catalysts.53 Selectivity in the reduction of the three double-bonds in cyclohepta-1,3,5-triene falls in the sequence Pd> Rh> Pt,54 this being the same as with linear diolefins.4L A. S. Hussey, G. P. Nowack, G. W. Keulks, and R. H. Baker, J . Org. Chent., 1968, 33, 610. 42 A. S. Hussey, T. A. Schenach, and R. H. Baker, J . Org.Chem., 1968, 33, 3258. 43 M. Abubaker, I. V. Grostunskaya, and B. A. Kazanskii, Vestnik Moskov. Univ., 1968, 106; M. Abubaker, Z. S. Khrustova, I. V. Gostunskaya, and B. A. Kazanskii, ibid., 148. 44 G. C. Bond, G. Webb, and P. B. Wells, Trans. Faraday SOC., 1968,64,3077. 45 C. Kemball and P. B. Wells, J . Chem. SOC. (A), 1968, 444. 46 D. Moger, G. Mink, and F. Nagy, Magyar Ke'm. Folyoirat, 1968, 74, 315, 318. 47 H. J. Dutton, C. R. Scholfield, E. Selke, and W. K. Rohwedder, J . Catalysis, 1968, 4* L. Guczi and P. TCtknyi, Zeit. phys. Chem. (Leipzig), 1968, 237, 356. 49 H. Simon and 0. Bernguber, Tetrahedron Letters, 1968, 471 1. 50 R. B. Turner, P. Goebel, B. J. Mallon, W. Von E. Doering, J. F. Coburn, and 5 l W. J. Irwin and F. J. McQuillin, Tetrahedron Letters, 1968, 2195.52 M. T. Wuesthoff and B. Rickborn, J . Org. Chem., 1968, 33, 131 1. 53 R. S. Mann and K. C. Khulbe, Canad. J. Chem., 1968, 46, 623 idem, J. Catalysis, 1968,10,401; R. S. Mann and D. E. To, Canad. J . Chem., 1968,46, 161. 54 B. D. Polkovnikov, 0. M. Nefedov, E. P. Mikos, and N. N . Novitskaya, Zzoest. Akad. Nauk S.S.S.R., Ser. khim., 1968, 1240. 10, 316. M. Pomerantz, J. Amer. Chem. SOC., 1968, 90, 4315.128 G. C. Bond Selective poisoning of palladiuni catalysts by cadmium salts alters the position of attack in 6-methylhepta-3,5-dien-2-one from the 5,6 double-bond to the 3,4 d~uble-bond.~~ A number of interesting papers have appeared on the hydrogenation of aromatic rings. The exchange of hydrogen between 13C-labelled benzene and unlabelled cyclohexane occurs in the vapour-phase over Pt-AI203 at 157-182"c and over Au-MgO at 205-235"C.56 Self-exchange of mono- deuteriotoluene takes place at 60-100"c over platinum and nickel : transfer of D to and from the ovtho-position is markedly slower than to and from the nzeta- and para- positions.57 This observation may mean that some earlier work on the exchange of alkylaromatic compounds will have to be recon- sidered.Weitkamp has described a most comprehensive study of the stereo- chemistry of the hydrogenation of substituted naphthalenes;58 1,2-di-t-butyl- benzene is reduced over Rh-C to the corresponding cyclohexane which is almost all the cis-isomer although up to 45% of an intermediate olefin (2,3-di-t-butylcyclohexene) is also formed.59 This is a remarkable demonstra- tion of the effect of substituents on the course of a catalytic reaction. Selectivity aspects of the reduction of p-terphenylGO and of diphenyl ether6I have been described. Several careful studies of the hydrogenolysis of ethane have been re- ported.62 It now seems well established that activity for this reaction at 150-3OO"c is greater for the noble Group VIII metals than for the base metals, and that activity falls on passing from left to right through each row. This is in contrast to most hydrogenations where activity is maximum in the third column of Group V111. 55 L. Kh. Frcidlin, N. V. Boriinova, L. I . Gvintcr, and S. S. Danielova, Zhirr. Jiz. 56 G. Parravano, J . Catalysis, 1968, 11, 269. 5' K. Hirota, T. Veda, T. Kitayama, and M. Itoh, J . Phys. Chern., 1968, 72, 1976. 5* A. W. Weitkamp, Adv. Catalysis, 1968, 18, 1 . 59 B. van der Graff, H. van Bekkum, and B. M. Wepster, Rec. Trav. chim., 1968, 87, 60 Y . Bahurel, G. Descotes, and J. Sabadie, Bull. SOC. chim. France, 1968, 4259. 61 P. N. Rylander and M. Kilroy, Engelhard Industries Technical Bulletin, 1968, 9, 1. 62 J. H. Sinfelt and W. F. Taylor, Trans. Faraday SOC., 1968,64, 3086; J. H. Sinfelt and D. J. C. Yates, J . Catalysis, 1968,10, 362; G. K. Starostenko, T. A. Stovochotova, A. A. Balandin, and K. A. el Chattib, Vestnik Moskoc. Univ., 1968, 52. Khitn., 1968, 42, 98. 777.