J. Chem. SOC. Faraday Trans. 1, 1984,80, 3245-3255 Silica-supported Binuclear Copper Oxide Catalysts Derived from Cupric Acetate Monohydrate Their Spectroscopic Characterization and Catalytic Nature in Carbon Monoxide Isotope Equilibriation or Carbon Monoxide Oxidation with Nitrous Oxide BY NORIYOSHI KAKUTA, AKIO KAZUSAKA,* AKIKO YAMAZAKI AND KOSHIRO MIYAHARA Research Institute for Catalysis, Hokkaido University, Sapporo 060, Japan Received 3rd January, 1984 Novel silica-supported copper oxide catalysts have been prepared by impregnating silica with an aqueous solution of cupric acetate monohydrate, a typical binuclear copper complex. Binuclear copper(I1) ions, mononuclear copper(I1) ions and highly dispersed cupric oxide on the calcined samples have been characterized using e.s.r.and/or i.r. spectroscopy using acetic acid as a probe. Significant amounts of binuclear copper(r1) ions were estimated on silica from the spin concentration of the e.s.r. signal. The rate of CO isotope equilibriation or CO oxidation with N,O is a function of the number of binuclear copper ions, indicating that this binuclear structure is responsible for both reactions. The importance of the hydroxyl group in anchoring the cupric acetate complex on silica is stressed, and a reaction mechanism for the oxidation of CO with N,O has been suggested on the basis of the binuclear structure of the copper ions. The states of copper on silica-supported copper catalysts prepared from a cupric ammine complex or cupric nitrate have been studied by various physical and chemical techniques.l-s Highly dispersed CuO and cupric silicate species have been detected on those catalysts.Moreover, the technique of temperature-programmed reduction has revealed that these two kinds of copper species differ from each other in their reactivity for hydrogen; dispersed CuO is reduced by hydrogen at ca. 280 "C, whereas the copper silicate species are reduced at higher temperatures depending on the calcination temperat~re.~q We have attempted to prepare a novel silica-supported copper catalyst from cupric acetate m~nohydrate.~ The latter is known to be a typical binuclear copper complex with a copper-copper bond of 2.61 A,a so that we can expect to prepare a silica-supported binuclear copper catalyst different from those reported previously.In this contribution we report details of the characterization of binuclear copper ions on catalysts derived from the cupric acetate complex and their catalytic behaviour in CO isotope equilibriation and CO oxidation with N,O. EXPERIMENTAL The cupric acetate monohydrate and cupric nitrate employed in this work were of certified grade and purchased from Wako Pure Chemicals. Silica gel was Kiesel Gel 60 from Merck, having a mean pore diameter of 60 A. The gases CO, 0,, N,O and CO, were of high-purity grade and were used without further purification. Carbon monoxide labelled with oxygen- 18 (96.8%) and carbon-13 (90.7%) was purchased from Prochem B.O.C. Ltd. Acetic acid was purified carefully by freezepumpthaw techniques. The silica-supported catalysts derived from the cupric acetate complex were prepared by impregnating silica gel with an aqueous solution of cupric acetate monohydrate and then dried 32453246 SILICA-SUPPORTED COPPER OXIDE CATALYSTS at 100 "C overnight.The concentration of copper was determined by atomic absorption analysis to be 0.5, 1.6,2.8,3.0,3.2,3.7,4.7 or 6.2 wt% . For comparison, a conventional silica-supported copper catalyst was prepared from the cupric ammine complex by ion exchange4v6 and contained 3.4 wt% of copper. Catalysts prepared in this way were activated in a vacuum for 1 h stages at room temperature, 100, 200 and 300 "C and subsequently calcined with 100 Torrt of oxygen at 300 "C. CO,, acetic acid and methyl acetate were observed by mass spectrometry through the activation processes.E.s.r. spectra were obtained with a Varian E4 spectrometer at a frequency of 9 GHz. The g-values were calculated relative to DPPH as standard (g = 2.0036). The spin concentrations of the cupric ions were estimated by comparing the integrated sample spectra with those of CuSO, - 5H,O or Cu,(CH,COO), - 2H,Q. The spin concentration due to the cupric ions in cupric acetate monohydrate was confirmed to be proportional to the sample weight in the range 0-30 mg. The error was estimated to be within 40%. A Jasco IR-G spectrometer was used to obtain the i.r. spectra. Samples were prepared by grinding the catalysts into a powder and pressing into wafers of typically 10-20 mg ern-,. They were pretreated individually in a Pyrex cell fitted with KBr windows. A conventional closed recirculating reactor (ca. 700 cm3) was employed for the kinetic studies of CO isotope equilibriation or CO oxidation with N,O.1 g of sample was placed in a cylindrical reactor and activated in the manner described previously. The isotope composition in carbon monoxide was determined using a Hitachi RMU-6 mass spectrometer and the partial pressures of CO, CO,, N,O and N, were followed using a gas chromatograph connected directly to the reactor. RESULTS AND DISCUSSION E.S.R. SPECTROSCOPIC STUDIES Fig. 1 shows the e.s.r. spectra of a sample of the catalyst derived from the acetate complex containing 3.0 wt% copper under the processes of activation from room temperature to 300 "C and subsequent calcination with oxygen at 300 "C. As shown in fig. 1 (a), the freshly prepared sample exhibits two kinds of e.s.r.signal: a signal characteristic of powdered cupric acetate monohydrate having absorptions at low and high magnetic fields (referred to as signal A), and a familiar axially symmetric signal having the respective e.s.r. parameters of gll = 2.34, gl = 2.06 and All = 138 G (referred to as signal B). The e.s.r. spectrum of powdered cupric acetate monohydrate has been identified by three absorptions at low and high magnetic fields and interpreted by the triplet state arising from the coupling of pairs of CuII i~ns.~-ll The three absorptions in this spectrum are given in fig. 1 (a) and the e.s.r. parameters are estimated to be gll = 2.36, gl = 2.11 and D = 0.32 cm-l, where D is the tetrahedral zero-field splitting parameter.Values of g,, = 2.35, gl = 2.08 and D = 0.34 cm-l have been reported for powdered cupric acetate monohydrate by Lewis et aZ.l0 Signal B, on the other hand, can be attributed to isolated cupric ions coordinated with acetate ligands in distorted octahedral symmetry, suggesting that the binuclear cupric acetate complex is decomposed into a mononuclear structure during its impregnation onto silica. When the catalyst was activated under vacuum stepwise from room temperature to 300 "C, signal A disappeared completely at 100 "C [fig. 1 (c)]. Calcination of the sample with oxygen at 300 "C changed signal B into a new anisotropic signal, giving the parameters g, = 2.50, g, = 2.09 and g, = 2.04. The same signal has been reported for a silica-supported copper catalyst prepared by conventional methods and assigned to isolated cupric ions in distorted tetrahedral ~ymrnetry.~ t 1 Torr = 101325/760 Pa.N.KAKUTA, A. KAZUSAKA, A. YAMAZAKI AND K. MIYAHARA 3247 signal A < J signal B x4 0 1.0 2.5 3.0 3.5 4.0 5.0 6.0 H/kG Fig. 1. E.s.r. spectra of catalyst derived from the cupric acetate complex (3.0 wt% Cu) under processes of activation, calcination and adsorption of acetic acid. Arrows in signal A show position of axial resonance field. The frequency was 9.48 GHz. Temperature shows activation temperature under vacuum and calcination was carried out with 100 Torr of oxygen at 300 "C. (a) Fresh sample, (b) 25 "C, (c) 100 "C, ( d ) oxidation and (e) adsorption of CH,COOH. When the calcined catalyst was exposed to acetic acid vapour at room temperature, signal A reappeared and the anisotropic signal was changed into signal B, as seen in fig.1 (e). The catalyst derived from the ammine complex, on the other hand, gave an anisotropic signal having the parameters g , = 2.47, g , = 2.10 and g , = 2.04, after activation and subsequent calcination. Exposure of the sample to acetic acid vapour showed only the faint signal A with the anisotropic signal becoming axially symmetric. Thus, the catalyst derived from the acetate complex preserves a significant number of binuclear cupric ions on the silica in comparison with the conventional catalyst. In addition, acetic acid is a good probe molecule for detecting binuclear cupric ions supported on silica. The concentrations of isolated or binuclear cupric ions were estimated by doubly integrating signal A or B and comparing the result with that of cupric sulphate or cupric acetate monohydrate. The high-field component of signal A was used to estimate the spin concentration of the binuclear cupric ions, because the low field component is obscure near 0 G.Fig. 2 shows the concentration of binuclear cupric ions for the calcined samples plotted against total copper loading. The numbers of isolated cupric ions are almost constant on all the catalysts in the3248 8 2.0 s a a '"1 O O h 3 z 3.0 Q) 0 .d I .o ' 0 1.0 2.0 3.0 4.0 5.0 6.0 copper loading (wt%) Fig. 2. Concentrations of copper species on catalysts derived from the cupric acetate complex: 0, binuclear Cu" ions; a, highly dispersed CuO; x , isolated CuII ions.range 0.5-6.2 wt% copper, while those of the binuclear cupric ions increased with copper loading. On the other hand, the conventional catalyst containing 3.4 wt% copper gave only one-tenth the amount of binuclear cupric ions as the catalyst derived from the acetate complex at nearly the same copper content (not shown in fig. 2). The sums of the amounts of these two copper species are less than the total copper loading, indicating that the CuII ions, which cannot be detected using e.s.r. spectroscopy with the acetic acid probe molecule, exist on silica. Two kinds of e.s.r.-inactive CuII ions have been reported in the literature:12 one is present in CuO and is diamagnetic and the other is CuII in sites of trigonal symmetry which have relaxation times too short to give e.s.r.signals. The e.s.r.-inactive Curl ions on these samples must be present as highly dispersed CuO on silica, because the latter should become an e.s.r.-active species by adsorbing acetic acid on exposure to acetic acid vapour. CuO with no silica support was confirmed to give no signal on exposure to acetic acid vapour in a separate experiment. Fig. 2 shows the amounts of highly dispersed CuO present, which were obtained by subtracting those of the isolated or binuclear CuII ions from the total loading. I.R. SPECTROSCOPIC STUDIES Fig. 3 shows the i.r. spectra of the catalyst derived from the acetate complex containing 3.2 wt% copper under the processes of activation and subsequent calcination. In the 1300-2500 cm-l range two bands due to overtones of the Si-0 stretching mode were observed at 1650 and I880 cm-l, as for most silica-supported metal catalysts.The fresh sample gave two strong bands at 1565 and 1440 cm-l with a shoulder at 1630 cm-l. On evacuation at room temperature the shoulder at 1630 cm-l dis- appeared along with a decrease in the intensities of the other bands. In a separate experiment a fresh sample of silica support was observed to give a band at 1630 cm-l which was removed on evacuation at room temperature. The 1630 cm-l shoulder can consequently be assigned to water adsorbed on silica. On the other hand, pressed cupric acetate monohydrate without silica was found to give two strong bands at 1595 and 1440 cm-l, which have been assigned, respectively, to the antisymmetric andN.KAKUTA, A. KAZUSAKA, A. YAMAZAKI AND K. MIYAHARA 3249 wavenumberlcm-' Fig. 3. 1.r. spectra of catalyst derived from the cupric acetate complex (3.2 wt% Cu) under processes of activation and calcination. The conditions for each process were the same as those in fig. 1. (a) Fresh sample, (b) 25 "C, (c) 100 "C; (d) 200 "C, (e) 300 "C and (f) oxidation. symmetric stretching modes of -COO- in cupric acetate m0n0hydrate.l~ We therefore assign the 1565 and 1440 cm-l bands to v,(COO) and vs(COO), respectively, of the silica-supported cupric acetate complex. In addition, anchoring of the complex onto silica may shift the v,(COO) band from 1595 to 1565 cm-l. Activation at 100 "C gave rise to a further reduction in the intensities of these bands. However, the triplet e.s.r.signal A (due to the cupric acetate complex supported on silica) was removed completely by activation at this temperature [fig. l(c)], as described previously. These two spectroscopic results suggest that the paramagnetic properties of the cupric acetate complex supported on silica may change to being antiferromagnetic by the loss of some acetate ligands. Raising the activation temperature to 200 and 300 "C almost removed the 1565 and 1440 cm-l bands. Concomitantly new bands developed at 1735 and 1380 cm-l. They were obtained on exposing the silica support to acetic acid vapour and can be assigned to acetic acid physisorbed on silica, in agreement with the results of Kiselev et al.14 These bands were removed completely by calcination with 100 Torr of oxygen at 300 "C for 1 h.When the calcined sample was exposed to acetic acid vapour the 1565 and 1435 cm-l bands again developed, together with bands at 1750 (sh), 1715, 1615, 1415 (sh) and 1385 (sh) cm-l, as shown in fig. 4(a) (sh denotes a shoulder of the spectrum). Fig. 4(6), on the other hand, illustrates the spectrum obtained for the catalyst derived on the conventional ammine complex, no band being observed at 1565 cm-l. Thus it was confirmed from i.r. spectroscopic studies that only the catalyst derived from the acetate complex preserves binuclear cupric ions on silica. The band at 1615 cm-l is seen in the spectra of both fig. 4(a) and (b). It was not observed on the catalyst derived from the fresh acetate complex, as shown in fig. 3, or on both catalysts reduced with 100 Torr of CO at 300 "C for I h.15 E.s.r.studies3250 SILICA-SUPPORTED COPPER OXIDE CATALYSTS 2000 1800 1600 1400 wavenum ber/cm I 1 I I 1 1 I I 2000 1800 1600 1400 wavenum ber/cm -2 Fig. 4.1.r. spectra of adsorbed CH,COOH on (a) the catalyst derived from the acetate complex (3.2 wt % Cu) and (6) that derived from the ammine complex (3.4 wt % Cu) : (A) after calcination at 300 "C; (B) and (C) after exposure to 10-3-10-2 Torr of CH,COOH. have revealed that the isolated cupric ions on each sample are not reduced under these conditions.' Therefore, it is reasonable to attribute the 161 5 cm-l band to acetate ions adsorbed on dispersed Cu0.T The assignments of these bands are given in table 1. The coordination structures of acetate ligands to metals have been discussed from the viewpoint of their antisymmetric and symmetric stretching frequencies and the t The adsorption of acetic acid on another type of copper oxide catalyst, which was prepared by mixing commercial CuO and SO, (3.0 wt% Cu), was studied using i.r.techniques. No absorption band was observed in the range 1650-1 300 cm-l, indicating that the highly dispersed CuO on silica is different from crystalline CuO in the adsorption of acetic acid.N. KAKUTA, A. KAZUSAKA, A. YAMAZAKI AND K. MIYAHARA 325 1 Table 1. Infrared spectrum frequencies and assignments of acetate ions in the cupric acetate complex and adsorbed on catalystsa compound v(C=O) v(C-0) assignment Cu,(CH,COO), 2H,O(s) Cu,(CH,COO), anchored on SiO, CH,COOH adsorbed on catalyst derived from the acetate complex CH,COOH adsorbed on catalyst derived from the ammine complex 1 595b 144Ob 1 565b 1440b o-cuxl / \ / \ CH,-C 0-CU" 1 560b 1435b 0-CU" CH,-C o-cu'l 1615 1430 0 II 1715 1385 CH,COOH on SO, 1615 1430 0 II CH,--C-O-CuO CH,-C-O-CUO 1715 1385 CH,COOH on SO, a All frequencies in wavenumbers (cm-l).These correspond to v,(COO-) and v,(COO-) of the symmetrical COO- group. separation of these.ll Nakamoto,ls for instance, has classified them into three groups : unidentate (I), bidentate (11) or of the bridging (111) type: M-0, M-0, 04C-CH, M/O>C--CH, M-0 ,S-CH, ' 0 (1) (11) (111) and has indicated that these structures are related to the separations between the two stretching-vibration modes of -COO-. In the unidentate complex (I), v(C=O) is higher than v,(COO) of the free acetate ion (1 560 cm-l) and v(C-0) is much lower than v,(COO) (1416 cm-l).As a result the separation between the two v(C-0) bands is much larger in the case of the unidentate complex than for the free ion. The opposite trend is observed in the bidentate complex (11): the separation between the v(C0) is smaller than that of the free ion. In the bridging complex (111), however, the two v(C0) are close to those of the free ion. In agreement with this classification, the acetate ions adsorbed on the binuclear cupric ions gave absorption bands close to those of the free ion at 1565 and 1435 cm-l. On the other hand, those adsorbed on the dispersed CuO would be of unidentate structure, because the separation of the respective absorption bands of v,(COO) and v,(COO) at 1615 and 1440 cm-l is 175 cm-l, and larger than that of the free ions.Besides the sample containing 3.2 wt% copper the sample containing 6.2 wt% copper was used for i.r. spectroscopic studies; similar results were obtained for each process. On the basis of the results obtained in the present work and in the literature, we3252 SILICA-SUPPORTED COPPER OXIDE CATALYSTS suggest a plausible reaction sequence for each process. The water in cupric acetate monohydrate is known to be substituted easily by a weak base like amrnine.l7 The cupric acetate complex may consequently be anchored by substituting the water in the complex by the OH group of silica as follows: (H20)C~(CH,COO),C~(H,0) + 2-OH +(-OH)CU(CH,COO),CU(OH-) + 2H,O. The OH groups of the silica play an important role in this process.Since the i.r. absorption bands due to the acetate ligands were removed on evacuation at 200 "C, and acetic acid, methylacetate and carbon dioxide were detected by mass spectrometry through the activation process, the reaction (-OH)CU(CH~COO),CU(OH-)+-O - CUCU - 0- + 2CH3COOH + CH,COOCH, + CO, can be proposed for the activation process. Here the binuclear CuII ions are reduced to binuclear CuI ions. On calcining the catalyst with oxygen at 300 "C the CuI ions are oxidized to binuclear Cu*I ions bridged by oxygen as 0 / \ -o~CuCu-0-++0, + - 0 . c u Cu-0-. On exposure to acetic acid vapour they are changed into the original acetate complex as 0 / \ -0 * CU CU * 0- + 4CH,COOH + (-OH)CU(CH,COO),CU(OH-) + H20. CATALYTIC STUDIES The acetate-complex-derived catalysts containing 0.5-6.2 wt % copper or the ammine-complex-derived catalysts containing 3.4 wt % copper were employed in catalytic studies of CO isotope equilibriation or CO oxidation with N,O.The CO isotope equilibriation 13C180 + 12C160 ,13C160 + 12C180 was carried out with ca. 10 Torr of an equimolar mixture of 13C180 and l2Cl60 at room temperature. The rate was estimated by the equation 2.30pV (1 +X)(XO-X"O) y=- RTt log(l+XO)(X-Xm) where V is the system's volume p the pressure of CO, t the reaction time and X, Xm and X' the ratio of l2Cl80 and 13Cle0 at t = t , t = GO and t = 0, respectively.18 Fig. 5 illustrates the relationship between log [(XO - Xm) (1 + X)/(X- Xm) (1 + XO)] and the reaction time, indicating that the experimental results fit the above equation well.In fig. 6 are shown the estimated rates and the concentration of binuclear cupric ions plotted against copper loading. The catalytic activity is related well to the concentration of the binuclear cupric ions. Furthermore the catalyst derived from the ammine complex was inactive in this reaction as shown in fig. 5 . Therefore it is concluded that the binuclear cupric ions are active sites for CO isotope equilibriation at room temperature. WinteP and Vorontsov et aZ.19 have studied this reaction on copper oxide, finding that it proceeds at temperatures as low as - 78 "C. Accordingly, the highly dispersed CuO on silica must be different from pure CuO in its catalytic nature for this reaction.N. KAKUTA, A.KAZUSAKA, A. YAMAZAKI AND K. MIYAHARA 3253 30.0 20.0 02 add (22 Torr) 10.0 5.00 I .oo time/min Fig. 5. CO isotope equilibriation at room temperature on catalysts derived from the acetate and ammine complexes. The numbers show the copper loading on the catalysts derived from the acetate complex while 'ion exchange' indicates that derived from the ammine complex (3.4 wt% CU). 1.0 2.0 3.0 4.0 5.0 6.0 Cu loading (wt%) Fig. 6. Variations in catalytic activities for CO isotope equilibriation or CO oxidation with N,O with concentrations of binuclear Cu'I ions as function of copper loading: 0, concentratiov of binuclear Cu" ions; 0, CO isotope equilibriation; x , CO oxidation with N,O. Voronsto~etal.~~ havediscussed thereactionmechanism of COisotopeequilibriation on the basis of kinetic results; at low temperatures it proceeds through the surface intermediate containing two or more CO molecules while at high temperatures the oxygen atom in copper oxide takes part in the reaction intermediate, and the mechanism of dissociation of CO into carbon and oxygen is excluded because of its high dissociation energy (255 kcal mol-l).The l80 concentration in CO was found to be constant throughout the reaction; on the catalyst containing 3.2 wt % of copper, for example, it changed from 47 to 45% . Also, only a small amount of CO, (< 0.1 % of the total pressure) was detected by mass3254 SILICA-SUPPORTED COPPER OXIDE CATALYSTS 2 0.c 5 b 4 1 2" 10.0 0 time/h Fig. 7. N, formation in CO oxidation with N,O at 150 "C. Initial pressure of CO or N,O was 30 Torr.Numbering of the curves is the same as in fig. 5. spectrometry. Therefore, we prefer the mechanism involving the associative CO intermediate. We have studied the state of adsorption of CO using i.r. spectroscopy in connection with the mechanism of this reaction. Upon exposing both types of catalyst (3.2 or 3.4 wt% Cu) to 30 Torr of CO, only a broad absorption band at 2135 cm-l was obtained. This has been reported on conventional silica-supported copper oxide catalysts in the literature. DeJong et al., for example, have ascribed it to adsorbed CO on dispersed CUO.~O Also, against our expectations, there were no absorption bands below 2000 cm-l attributable to associative CO, CO adsorbed on binuclear cupric ions of the bridged type or a carbonate species.The 2135 cm-l band was not affected by the addition of 30 Torr of oxygen. However, the isotope-equilibriation reaction was stopped completely by adding 22 Torr of oxygen, as shown in fig. 5, indicating that CO adsorbed on dispersed CuO is not responsible in this reaction. In a previous paper7 we have suggested that the binuclear CuI ions formed from binuclear CuII ions are active sites for CO oxcidation with N,O: CO+N,O -+ CO,+N, on the acetate-complex-derived catalyst. In order to confirm this suggestion we have studied the effect of binuclear cupric ion concentration on the catalytic activity of this reaction. The reaction was carried out with a mixture of 30 Torr of CO and 30 Torr of N,O at 150 "C on a series of catalysts derived from the acetate complex and from the ammine complex. Fig.7 illustrates the reaction curves describing the change of N, pressure with time. An induction period was observed in each run. The reduction of binuclear CuII ions to Cul ions should occur during the induction period. Logarithmic plots of N,O pressure revealed that the reaction proceeds by first-order kinetics withN. KAKUTA, A. KAZUSAKA, A. YAMAZAKI AND K. MIYAHARA 3255 respect to N,O after the induction period. In fig. 6 the rates of formation of N, under steady state, estimated from the slopes of the reaction curves at 20 Torr of N,O, are plotted together with the concentration of binuclear Cul* ions on the catalysts derived from the acetate complex. A good agreement can be seen between them, and accordingly the binuclear CuI ions are confirmed to be active sites for this reaction.Dell et aL21 have studied this reaction on cuprous oxide at ca. 250 "C and have proposed that it proceeds through a redox reaction involving the cuprous ions: cuprous oxide initially reacts with CO to form CO,, and then the reduced copper is reoxidized by N,O with the formation of N2. We found in our previous work that copper ions on the catalyst employed here could not be reduced to free copper with 100 Torr of CO even at 300 "C for 1 h. Furthermore, Scholten et al. have reported that free copper could not be oxidized to cupric oxide with N,0.22 Therefore, the redox mechanism cannot be valid in this case. One of the authors has studied this reaction kinetically on a partially reduced molybdenum oxide catalyst, proposing that it proceeds through a molecular complex of N,O and CO coordinated to molybdenum cation pairs.23 We may speculate a similar mechanism in the present case : (Co)b+ (N2O) N20+C0 I N, -0cucu0- - -0cu CUO- -b ( c o y + 0 8 - I I co2 -0cu CUO- + -0cucu0--. A detailed study of the reaction mechanisms is now in progress in our laboratory. We thank Prof.K. Tanaka of Tokyo University for valuable discussions and Prof. A. Aramata for help in estimating the spin concentrations of the e.s.r. signal. B. J. Hathaway and C. E. Lewis, J. Chem. SOC. A, 1969, 2295. V. A. Bogdanov, V. A. Shvets and V. B. Kazanskii, Kinet. Katal., 1974, 15, 176. S. A. Surin, B. N. Shelimov, I. D. Mikheikin and V. B. Kazanskii, Kinet. Katal., 1976, 17, 1569. H. Tominaga, Y. Ono and T. Keii, J. Catal., 1975, 40, 197. S. J. Gentry and P. T. Walsh, J. Chem. SOC., Faraday Trans. I , 1982, 78, 1515. M. Shimokawabe, N. Takezawa and H. Kobayashi, Appl. Catal., 1982,2, 379. ' N. Kakuta, A. Kazusaka and K. Miyahara, Chem. Lett., 1982, 913. G. M. Brown and R. Chidambaram, Acta Crystallogr., Sect. B., 1973, 29, 2393. J. R. Wasson, C. Shyr and C. Trapp, Znorg. Chem., 1968, 7 , 469. lo J. Lewis, F. E. Mabbs, L. K. Royston and W. R. Smail, J. Chem. SOC. A , 1969,291. l1 J. Catterick and P. Thornton, Adu. Inorg. Chem. Radiochem., 1977, 20, 291. l2 I. D. Mikheikin, V. A. Shvets and V. B. Kazanskii, Kinet. Katal., 1970,11,747; V. B. Kazanskii and I. D. Mikheikin, Izu. Otd. Khim. Nauk, 1973, 6, 361. l 3 K. Nakamoto, Y. Morimoto and A. E. Martell, J. Am. Chem. SOC., 1961,83,4528. l4 A. V. Kiselev and A. V. Uvarov, Surf: Sci., 1967, 6, 399. l5 A. Kazusaka, A. Yamazaki and K. Miyahara, to be published. l6 K. Nakamoto, Infrared and Raman Spectra of Inorganic and Coordination Compounh (Wiley- Interscience, New York, 3rd edn, 1978). E. Kotot and R. L. Martin, Inorg. Chem., 1964, 3, 1306. l8 E. R. S. Winter, J. Chem. SOC., 1964, 5781. l9 A. V. Vorontsov and L. A. Kasatkina, Kinet. Katal., 1980, 21, 1494. 2o K. P. DeJong, J. W. Geus and J. Joziasse, J. Catal., 1980, 65, 437. 21 R. M. Dell, F. S. Stone and P. F. Tiley, Trans. Faraday SOC., 1953, 49, 201. 22 J. J. F. Scholten and J. A. Konvalinka, Trans. Faraday SOC., 1969, 65, 2465. 23 A. Kazusaka and J. H. Lunsford, J. Catal., 1976, 45, 25. (PAPER 4/0 12)