J. CHEM. SOC. FARADAY TRANS., 1994, 90(11), 1541-1545 Catalytic Combustion of Methane :Copper Oxide supported on High-specific-area Spinels synthesized by a Sol-Gel Process Nolven Guilhaume" and Michel Primet Laboratoire d 'Application de la Chimie a I'Environnement, Unite Mixte CNRS -Universite Claude Bernard, 43 Boulevard du I I Novembre 1918,69622Villeurbanne Cedex, France Three spinels (MgAI,O, , ZnAl,O, and CaA120,) have been synthesized using a sol-gel process coupled with supercritical drying. After calcination at 1000 K, specific areas as high as 300 m2 g-' can be preserved. Deposi- tion of copper oxide on such carriers leads to the formation of catalysts active in methane combustion. The catalytic activity is strictly proportional, at low conversion levels, to the number of Cu2+ ions deduced from FTIR measurements of carbon monoxide adsorption.By comparison with spinels prepared by impregnation of alumina with nitrate salts, the use of a sol-gel process allows an increase in the dispersion state of the active phase. Spinel oxides such as ZnA120, exhibit interesting physical properties and are potentially useful as supports for hydro- carbon combustion catalysts based on transition-metal oxides : CuO deposited on ZnAl,O, is very active for methane combustion and is almost unaffected after ageing tests at 1340 K.' In comparison, CuO deposited on A1,0, has a higher activity in its fresh state, but shows severe deac- tivation after ageing under the same conditions. This has been explained by the fact that, because of the structure of ZnAl,O,, CuO particles cannot enter the spinel lattice, which is supposed to act only as a dispersing agent. In the case of CuO,/Al,O,, deactivation has been shown to be due to a solid-state reaction between the active phase and the support, leading to inactive copper aluminates.The dispersion of CuO on ZnA120, is mostly limited by the low specific surface area of the support (9 m' g-') which therefore disfavours the catalytic activity of the active phase in its fresh state compared with CuO/Al,O,. This is the reason for the investigation of new preparation methods in order to obtain supports with high specific surface areas. We report here the preparation of high-specific-area spinels MgA1204, ZnA120, and CaAl,O, by the sol-gel process.They were obtained as aerogel powders by supercritical extraction of the solvent2 which preserves the structural properties of gels such as high porosities and large surface areas. Their thermal behaviour under ageing at high tem- perature was also studied. These aerogels were used as sup- ports for the preparation of combustion catalysts based on copper oxide as the active phase. Their activity for methane combustion was compared with that of similar CuO/MgAl,O, and CuO/ZnAl,O, catalysts whose supports have been prepared by solid-solid reactions. Experimental Physicoc hernial Characterizations BET areas were measured by nitrogen adsorption at 77 K on samples previously evacuated under vacuum at 573 K.Powder XRD patterns were recorded with a D 500 Siemens diffractometer using monochromatized Cu-Ko! radi- ation. The patterns were recorded from 3 < 28/degrees < 70 with a scan rate of 1.2" min-'. The patterns were compared with JCPDS reference data for phase identification. Elemental analyses were obtained from the 'Service Central d'Analyses du CNRS'. IR spectroscopy studies of CO adsorption on the CuO/ spinel catalysts were performed with an FTIR spectrometer (Bruker IFS 110). The spectral range observed was 4000-lo00 cm-' and the resolution was set to 4 cm-'. The samples were pressed into thin discs of known weight (between 25 and 35 mg), and introduced into a cell allowing in situ treatment. The samples were pretreated at 773 K under flowing oxygen for one night, evacuated at 773 K under vacuum and cooled at room temperature in UQCUO.The background spectra were recorded (100 scans), then CO (40- 45 Torrj-) was introduced and spectra were recorded at differ- ent contact times (0, 1, 4 and 20 h).The samples were evacuated under vacuum and spectra recorded at room tem- perature after 10 s, 15 min, 1 and 3 h evacuation. Preparation of the Supports The synthesis and manipulation of alkoxide precursors were performed under an inert atmosphere, using standard Schlenk tubes and vacuum-line techniques. Solvents were dried and distilled before use. Drying of the gels was achieved in an autoclave under supercritical conditions.' The aerogels obtained were fired in a flowing stream of air at 973 K for 3 h (heating rate :2.5 K min -I).MgA1204 This oxide was prepared from the mixed alkoxide MgAl,(O-Bu), as described in ref. 3, with a slightly modi- fied procedure : magnesium turnings and aluminum powder (1 :2 atomic ratio) were refluxed in butanol until all the metals were consumed. After cooling, the reacting mixture was filtered and the concentration of the solution was adjust- ed to 0.15 mol I-' by addition of butanol. Hydrolysis of the OBu groups was achieved at room temperature by slow addi- tion of the stoichiometric amount of water diluted in butanol (5% volume). The gel obtained was dried under supercritical conditions (593 K, 60 atm). Analysis: found: Mg, 15.31; Al, 35.73; C, 0.92; H, 1.54.Mg/Al = 0.476. MgA120, requires: Mg, 17.08: Al, 37.93; Mg/Al, 0.50. ZnA120, Commercial Al(OPr'), was dissolved in refluxing isopro- panol. The solubility of the product was rather poor. Zinc acetylacetonate (1 : 2 molar ratio) dissolved in acetone was added and the mixture was refluxed for 2 h to yield a clear solution. After cooling, the stoichiometric amount of diluted water (5% in isopropanol) was added slowly and the gel obtained was dried under supercritical conditions (523 K, 50 t 1Torr = 101325/760 Pa. atm). Analysis: found: Zn, 34.45; Al, 27.85; C, 0.58; H, 0.69; Zn/Al = 0.51. ZnA1204 requires: Zn, 35.66; Al, 29.43; Zn/Al, 0.50. CaA1204 Metallic calcium (40.14 mmol) was suspended in 100 ml 2- methoxyethanol with two crystals of HgCl, .The mixture was refluxed overnight to yield a pale yellow solution. After cooling, Al(OPr'), (81 mmol) was added and the mixture was refluxed again for 30 min, giving a clear, pale yellow solution. This solution was hydrolysed at room temperature with the stoichiometric amount of water diluted in 2-methoxyethanol. The gel was dried at 603 K and 58 atm (supercritical conditions). Analysis: found: Ca, 26.12; Al, 35.59; C, 1.54; H, 1.70. Ca/Al = 0.49. CaAl,04 requires: Ca, 25.36; Al, 35.59; Ca/Al, 0.50. Ageing of the Supports The samples were aged by firing at 1273 K under flowing air for 24 h. Preparation of the CuOlspinel Catalysts Catalysts consisting of 5 wt.% CuO deposited on the spinel were prepared by impregnation of the supports by an aqueous solution of the required amount of copper nitrate, evaporation of the solvent under reduced pressure, followed by calcination in flowing air at 773 K for 2 h.Analysis: 3.78, 3.85 and 3.62 wt.% Cu (or 4.73, 4.82 and 4.53 wt.% CuO) for CuO/MgA1204,ZnA1204 and CaA1204 ,respectively. Catalytic Activity Measurements The catalytic activity was measured under isothermal condi- tions, in the range 623-1023 K. The temperature was increased by steps of 50 K and the catalysts kept at each temperature for 3 h. In order to avoid overheating during the reaction, the reactant mixture consisted of diluted methane and oxygen in nitrogen (1 vol.% CH,, 4 vol.% 0,, N, as balance), with a total flow rate of 6.4 dm3 h-'.Carbon mon- oxide, carbon dioxide and unreacted methane were separated on a Carbosieve S (60-80 mesh) chromatography column. CO and CO, were methanized, on an Ni/MgO catalyst maintained at 753 K, and analysed by flame-ionization detec- tion. The catalyst (500 mg) was loaded into a U-shaped quartz reactor and calcined again at 673 K under flowing oxygen for 1 h prior to the test. Blank experiments with an empty reactor showed the absence of methane conversion up to 873 K. Beyond this temperature, the methane conversion was 0.4% at 923 K and 2% at 973 K. Comparison of the catalysts was performed using the Ts0,defined as the tem- perature corresponding to 50% methane conversion. Results and Discussion Preparation and Characterization of the Supports Preparation of the solids by the sol-gel process starting from metal alkoxides allowed us to obtain oxides with large surface areas compared with more conventional preparation methods: ZnAl,O, was only 9 m2 g-' when prepared by impregnation of high surface area alumina with zinc nitrate.' MgA1204 could be obtained with a surface area of 60 m2 g-' when prepared with an improved method4 of exfoliation of the solid also prepared by impregnation of alumina with magnesium nitrate. Besides the sol-gel process, a high-temperature aerosol decomposition process has been described' which allows the preparation of MgA1204 with a J.CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 specific area of 250 m2 g- '. Finally, MgAl,O, was prepared with an area of 290 m2 g-' by a sol-gel method starting from a modified Mg/Al double alkoxide, M~A~,(OBU),(PEG),~ with supercritical drying of the gel.This method seemed to be the most efficient to obtain solids with high specific areas. X-Ray Diflraction The aerogels obtained after supercritical drying were com- pletely amorphous to X-rays. After calcination at 973 K, the XRD patterns of the magnesium and zinc spinels (Fig. 1) showed the main lines of MgAl,O, and ZnAl,04. They are very broad (FWHM = 1.85" in 26 and 1.55-1.66' in 28, respectively), indicating very small crystalline particles. The CaAl,04 spinel was still amorphous at this temperature. The same supports prepared by impregnation of alumina with magnesium6 or zinc nitrates' had to be calcined at 1273 K for 24 h and 4 days, respectively, in order to form the spinel phases, and still some zinc oxide was present together with ZnA1204 after this treatment.' The sol-gel process allows mixing of the precursors at the molecular scale in solution, by formation of double alkoxides [the alkoxides MgAl,(OEt), and CaAl,(OEt), are known and characterized]., The hydrolysis/gelification reactions, which start building the oxide network by hydrolysis of the alkoxo groups, preserve the homogeneity of the precursors.The desired spinel phases can thus be obtained at much lower temperatures than in the case of solid-solid reactions. Specijic Areas The three spinels obtained by sol-gel methods had high spe- cific areas compared with those prepared by impregnation of alumina with metal nitrates (Table 1).Drying the gels under 200 1 5 100 .AB 10 20 30 40 50 60 2Oldegrees Fig. 1 XRD patterns of the aerogel samples after calcination at 973 K. (a)MgAl,O,; (b) ZnAl,O,; (c) CaAl,O,. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 Table 1 Comparison of the BET areas (m2 g- ') of the supports pre- pared by sol-gel methods (before and after ageing) and by impregna- tion of commercial alumina (102 mz ggl)' support impregnation sample aerogel aerogels after ageing MgA1Z04 ZnAl,O, 49 9 296 147 27 59 CaAl,O, - 230 1 supercritical conditions is one way of preserving the high porosity of the gels and providing solids with high surface area. It affords very homogeneous powders' which can react at low temperatures to form the oxide network, and to keep higher specific areas.Effect of Ageing tbe Supports After ageing the supports by firing them at 1273 K for 24 h, the three spinels were well crystallised: the XRD patterns of the samples (Fig. 2) present all the characteristic diffraction lines of the spinel phases. No other phase than ZnAl,O, was detected. In the pattern of MgA1204, two very small peaks can be attributed to the two main lines of MgO. All the dif- fraction lines of CaA1,0, were present, but also some small peaks could be attributed to the phase Ca,Al,,O,, . No CaO was detected. This ageing procedure led to a severe decrease in the spe- cific surface area (Table 1).The loss is not proportional to the initial area : the zinc aluminate appears more thermostable than the other spinels. The decrease in specific area is particu- v)c 5 5008 0 10 20 30 40 50 60 70 29/degrees Fig. 2 XRD patterns of the aerogel samples after ageing at 1273 K. (4MgAl20, (+,MgO); (b)ZnAlz04;(4 CaAl,O, (V,Ca,AI,,O,,). larly important in the case of CaAl,O,, this solid having a poor resistance to thermal sintering. CuOFpineI Catalysts 5 wt.% CuO was deposited on each spinel by an impregna- tion procedure. The XRD patterns of the three catalysts remained unchanged after the calcination step at 773 K per-formed to form the copper oxide: no diffraction lines of CuO can be detected, suggesting a good dispersion of the active phase on the supports.The specific areas of the catalysts were smaller than those of the supports alone (cf: Tables 1 and 2). It is known' that when aerogels are immersed in a liquid, they collapse immediately because the large-scale structure of the network is weak and easily collapsed by capillary pres- sure. After drying, the resulting xerogel has lost a portion of its larger pores, but the micropores and mesopores that produce the high surface area remain, in a more rigid network. This could explain why, during the impregnation step, a decrease in specific area was observed. Here again the calcium aluminate is the least resistant to pore collapse. Despite the above-mentioned collapse, the aerogel-based catalysts had large surface areas compared with those pre- pared by solid-solid reactions (Table 2).Catalytic Activity Fig. 3 shows the methane conversion as a function of the catalyst temperature for various CuO/spinel samples. Experi- ments performed with MgAl,O, alone showed that the con- version does not exceed a few per cent of ca. loo0 K. Results from previous work on CuO/ZnAl,O, and CuO/MgAl,O, have been included for comparison. Table 2 Comparison of the BET areas (mZg-') of the CuO-based catalysts supported on carriers prepared by sol-gel methods and by impregnation of a commercial y-Alz03 CuO/spinel prepared sample CuO/aerogel by solid-solid reaction CuO/MgAl,O, CuO/ZnAl,O, CuO/CaAl,O, 227 108 103 50 10 - 00 80 60 40 20 0 600 700 800 900 1000 1100 T/K .,Fig.3 Catalytic activity for methane combustion. CuO/ MgAl,O, aerogel; A, CuO/ZnAl,O, aerogel; 0, CuO/CaAl,O, aerogel; U, CuO/MgAl,O, 16; A, CuO/ZnAl,O, .' Carbon dioxide was the single oxidation product. The activity of the two CuO/MgAl,O, (high and moderate spe- cific areas) were identical. In the case of CuO/ZnAl,O,, the activity for CH, combustion was strongly improved for the catalyst supported on ZnAl,O, aerogel, as can be seen with the decrease in Ts0of ca. 55 K. CuO/CaAl,O,, despite its high specific area, had the poorest activity. This could be caused by a reaction between this support and CuO, which cannot be detected by XRD since the samples are amor- phous. These results suggest that the dispersion of CuO was already optimum when deposited on MgA120, having a spe- cific area of 49 m2 g-', and that an increase in the specific area of the support does not improve the dispersion state of the active phase.In the case of CuO/ZnAl,O,, the specific area of the support prepared by solid-solid reaction was very low (9 m2 g-I), and CuO could be detected in the XRD pattern of the catalyst, indicating that rather big crystallites of copper oxide were formed. When the same loading (5 wt.% CuO) was deposited in ZnA120, aerogel, no diffraction lines of copper oxide could be detected. The higher dispersion of CuO on this support is responsible for the better activity of this catalyst. IR Spectroscopy of CO Adsorption After oxygen pretreatment at 773 K, desorption at the same temperature and cooling at 298 K, carbon monoxide was admitted onto the three samples under a pressure close to 40 Torr.Whatever the support, a v(C0) band between 2105 and 2120 cm-'was observed, in addition to the bands due to gaseous carbon monoxide. The intensity of the v(C0) bands increases slightly with contact time, but most of the sites able to chemisorb CO are covered as soon as the CO is intro- duced. Table 3 shows the position and intensity of the v(C0) band as a function of time. The intensity of the previous bands are expressed in absorbance units referred to 1 g cata- lyst, with a CuO content normalized to 1 wt.%. Note that the wavenumber of the v(C0) bands does not vary when the CO coverage changes, suggesting that adsorption occurs on ionic species.Upon desorption at room temperature, the absorbance of the v(C0) bands of the three samples decreases quickly during the first 15 min, but only slowly for longer evacuation times (Table 4). The absorption of CO, studied by IR spectroscopy, is a very useful tool for analysis of the surface properties of sup- ported active phases, either in a metallic or an ionic The position of the v(C0) band depends on the oxidation state as well as on the surroundings of the adsorption centre; its intensity is proportional to the number of accessible adsorption sites.' ' According to literature data, and as far as ionic oxidized copper species are concerned, it seems that Cu2+ and Cu' ions lead to well separated v(C0) bands only when these ions are located in a zeolitic structure.In a recent paper, Sarkany Table 3 v(C0) band position and intensity (absorbance units referred to 1 g CuO/spinels catalysts and normalized to 1 wt.% CuO) after CO adsorption at 25 "C, at several contact times absorbance sample v(CO)/cm-' t=O t= 1 h t=4h t=20h CuO/MgAl,O, CuO/ZnAl,O, CuO/CaAl,O, 2110(m) 2120 (s) 2105 (w) 1.25 3.22 0.59 1.64 3.685 0.77 1.895 4.13 0.92 2.47 4.74 1.29 s = strong, m = medium, w = weak. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 Table 4 Absorbance of the v(C0) bands (absorbance units referred to 1 g catalysts and normalized to 1% CuO) after different times of evacuation under vacuum time of evacuation sample 10 s 15 min lh 3h ~ ~~~~ CuO/MgAl ,04 CuO/ZnAl,O, 2.26 4.29 1.555 3.28 1.35 2.91 1.18 2.575 CuO/CaAl,O, 0.80 0.22 0.12 0.07 et al.' investigated CO adsorption onto excessively ion- exchanged Cu/Na-ZSM 5.They observed v(C0) bands at 2180 and 2157 cm-' attributed to CO adsorption onto Cu2+ and Cu+, respectively. Upon hydrogen reduction at ca. 773 K, the formation of metallic copper was evidenced by a v(C0) band close to 2108 cm-'. In the case of copper-containing species deposited on more conventional supports such as silica or alumina, CO adsorp- tion led to v(C0) bands which are not so different according to copper oxidation state. For metallic copper, CO adsorp- tion leads to the formation of v(C0) bands in the spectral range 2110-2063 cm-', according to the nature of the support and to the CO ~overage.'*'~-'~ In the case of sup-ported CuO, a main v(C0) band close to 2125 cm-' was assigned to Cu2+ ions belonging to the cupric oxide lattice.l4-I8 Cu+ ions have not been definitively identified by an FTIR study of CO adsorption.Several authors proposed to assign a v(C0) band at ca. 2115 cm-' to CO bonded to Cu ions.' 9-2+ In the present study, the position of the v(C0) bands, i.e. 2110, 2120 and 2105 cm-' for CuO/MgAl,O,, CuO/ZnAl,O, and CuO/CaAl,O, , respectively, is the spec- tral range previously assigned for Cu2+ and Cu+ species. However, according to the preparation method of the cata- lysts (calcination in air at 773 K) and the oxygen pretreat- ment at 773 K before IR measurements, it is unlikely that the surface copper ions could be in a reduced state (Cu' species).A reduction of Cu2+ into Cu' by CO at room temperature might occur as pointed out in ref. 14. Nevertheless such a reduction is expected to lead to changes in v(C0) of adsorbed CO, and to the formation of CO, . In the present study, such changes were not observed. Finally, in a previous study per- formed on CuO/Al,O, catalysts," CO adsorption led to a v(C0) band at 2120 cm-', whereas UV-VIS spectroscopy measurements evidenced the presence of Cu2+ ions. As a consequence, we assume that the v(C0) bands observed in the present work can be attributed to carbon monoxide adsorbed onto surface Cu2+ ions.Correlation between Catalytic Activity and F TIR Spectroscopy As in a previous study,I8 a correlation between the catalytic activities in methane combustion (low conversion levels, i.e. ~30%conversion) and the absorbance of the v(C0) band (after 15 min evacuation) was established. The conversion levels as well as the v(C0) absorbance were referred to 1 g catalyst and normalized to 1 wt.% CuO (Fig. 4). The linear relationship shown in Fig. 4 suggests strongly that the active sites in methane combustion are the same as those responsible for CO adsorption. The more active the catalyst, the more intense the v(C0) bands. For the three aluminate supports prepared by the sol-gel process, the catalytic activity does not follow the specific area of the carrier: CuO/ZnAl,O, and CuO/CaAl,O, have similar BET areas (108-103 m2 g-') whereas they exhibit strong differences in catalytic behaviour. It must be postu- J.CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 oxide followed by IR spectroscopy. Nevertheless, it seems that the dispersion of the active sites is more dependent on 6-the nature of the support than on its BET area. The nature and number of superficial groups present on the carrier 5-appears to be a key factor for the active-phase dispersion, and therefore for the catalytic activity. 4-Finally, it must be stressed that aluminate-based aerogel supports with high specific area are rather sensitive to 3-thermal sintering, probably because the small crystallites sizes favour sintering at high temperatures.In that case also, the 2-sensitivity towards sintering depends strongly on the alumin- ate type and seems not to be related to the specific area of the starting material. 0.0 1.o 2.0 3.0 References absorbance (arb. units) 1 M-C Marion, E. Garbowski and M. Primet, J. Chem. SOC.,Fig. 4 Correlation between absorbance of v(C0) bands (after 15 Faraday Trans., 1991,87,1795..,min evacuation) and catalytic activity of CuO/spinels. 2 S. J. Teichner, in Aerogels, ed. J. Fricke, Springer-Verlag, Berlin, CuO/MgAl,O, aerogel; A, CuO/ZnAl,O, aerogel; @, 1986, p. 22. CuO/CaAl 0, aerogel. lated that surface properties of the support, for instance the presence of functional groups, are very important for the dis- persion of the active phase.Further investigations of the surface properties of these aerogel supports should give infor- mation about the surface functional groups available for the anchoring of the active phase. Comparison of the two CuO/ZnAl,O, catalysts prepared by either the sol-gel process (present study) or by the solid- solid reaction’ shows that the support preparation by a sol-gel method leads to a drastic improvement of the cata- lytic activity. For instance, at 723 K, the methane conversion (normalized to 1 wt.% CuO) increases from 1.9 to 6% on going from the ZnAl,O, on alumina support to the same carrier prepared by sol-gel. At the same time, the BET area of the support increases by a factor of 10 (Table 2).In that case, the catalytic activity does not strictly follow the BET area of the support, suggesting that surface properties of the carrier are strongly dependent on the preparation procedure. Conclusion The sol-gel process coupled with supercritical drying is an efficient method for preparation of high specific area spinels at low temperatures. Copper oxide deposition onto these aerogel aluminates leads to catalysts active in methane com- bustion. As far as ZnAl,O, is concerned, the catalytic activity is strongly improved by the use of a carrier prepared by a sol-gel process and exhibiting a large specific area. Catalytic activity is directly governed by the dispersion state of the active phase as shown by specific adsorption of carbon mon- 3 0.Varnier, P. Bergez, N. Hovnanian and L. Cot, in Comptes-Rendus de I’Ecole d’itt! Sol-Gel 91, Oliron, 15-20 Septembre 1991, vol. 2, p. 569. 4 S. D. Peter, E. Garbowski, V. Perrichon and M. Primet, unpub- lished results. 5 W. R. Moser and J D. Lennhoff, Chem. Eng. Commun., 1989,83, 241. 6 M-C. Marion, PhD Thesis, Universite Claude Bernard Lyon 1, France, 1990, no. 02.90. 7 D. C. Bradley, R. C. Mehrotra and D. P. Gaur, in Metal Alkox- ides, Academic Press, London, 1978, p. 300. 8 C. J. Brinker and G. W. Scherer, in Sol-Gel Science, 1990, Aca-demic Press, London, p. 501. 9 N. Sheppard and T. T. Nguyen, in Advances in Infrared and Raman Spectroscopy, ed. R. J. H. Clark and R. E. Chester, Heyden, London, 1978, vol. 5, p. 67. 10 A. A. Davydov, Infrared Spectroscopy of Adsorbed Species on the Surj-ace of Transition Metal Oxides, Wiley, New York, 1984. 11 M. C. Kung and H. H. Kung, Catal. Rev., 1985,27,425. 12 J. Sarkany, J. L. d’Itri and W. M. N. Sachtler, Catal. Lett., 1992, 16, 241. 13 J. A. Dalmon, M. Primet, G. A. Martin and B. Imelik, Surf. Sci., 1975,50,95. 14 G. M. Millar, C. H. Rochester and K. C. Waugh, J. Chem. SOC., Faraday Trans., 1991,87,1467. 15 J. W. London and A. T. Bell, J. Catal., 1973,31, 32. 16 G. M. Millar, C. H. Rochester and K. C. Waugh, J. Chem. Soc., Faraday Trans., 1991,87,1477. 17 M. A. Kohler, N. W. Cant, M. S. Wainwright and D. L. Trimm, J. Catal., 1982, 117, 188. 18 M-C. Marion, E. Garbowski and M. Primet, J. Chem. SOC., Faraday Trans., 1990,86,3027. 19 Y. A. Lokhov, V. I. Zaikovskii and A. A. Solomennikov, Kinet. Catal., 1982,23,348. 20 A. A. Efremov and A. A. Davydov, Kinet. Catal., 1983,24, 1005. 21 Z. Chajar, M. Primet, H. Praliaud, M. Chevrier, C. Gauthier and F. Mathis, Catal. Lett., in the press. Paper 4/00728J; Received 7th February, 1994