J. CHEM. SOC. FARADAY TRANS., 1994, 90(13), 1993-1998 Addition of Manganese to Iron Catalysts supported on Silicalite-I and its Effect on CO Hydrogenation Gopal Ravichandran, Debasish Das and Dipak K. Chakrabarty* Solid State Laboratory, Department of Chemistry, Indian Institute of Technology, Bombay 400076, India The addition of manganese to iron catalysts supported on silicalite-1 increases the selectivity of the catalysts to alkenes in CO hydrogenation. The catalysts were prepared by impregnating the support with metal nitrates followed by calcination and in situ reduction in hydrogen. The Mossbauer spectra of the calcined samples showed that the addition of manganese reduced the size of the cr-Fe,O, particles. This was confirmed by XRD and TPR results. The active phase consists of Hagg carbide and some oxides of Fell', the latter being responsible for the increased alkene formation.The role of manganese is to reduce the particle size of the iron oxide precursor, making carburization unfavourable. This leads to an increase in the amount of the oxide phase in the catalyst, thereby increasing the selectivity to alkenes. Recently there has been renewed interest in Fischer-Tropsch (FT) synthesis. The focus has been on selective conversion of synthesis gas to feedstocks for chemical industry rather than to liquid fuel.',, One such important aspect is the selective synthesis of C2-C, alkenes. However, FT synthesis yields a wide spectrum of products from methane to waxes, and the design of an appropriate catalyst to obtain specific products is of vital importance.Zeolite-supported FT catalysts are known to show unusually high specificity for lower hydrocar- bons or aromatics. The formation of lower hydrocarbons can be achieved by termination of chain growth by containing the active metal particles inside the small pores of the zeolite^.^-^ However, the nature of the zeolite support has a strong influ- ence on the product selectivity. Thus, while acidic zeolite such as ZSM-5 show poor selectivity to alkenes and favour the formation of gasoline-range hydrocarbon^,^,^ catalysts sup- ported on silicalite yield alkenes selectively.' A comparison of silicalite, mordenite and ZSM-5 supported cobalt catalysts has shown that Co/silicalite has very low methanation activ- ity and very high selectivity for C2-C4 alkenes.' Very high selectivity to lower alkenes has also been reported on potassium-promoted Fe/silicalite catalysts.' It has been claimed that the addition of manganese oxide to precipitated iron catalysts yields a catalyst with a high selectivity to low-molecular-weight alkenes.'0-'2 Although there is considerable interest in the use of zeolites in the FT reaction, and zeolite-supported iron and cobalt catalysts have shown very good selectivity to lower alkenes, no study had been reported on the effect of manganese on the behaviour of zeolite-supported iron or cobalt FT catalysts until we report- ed the results on CO hydrogenation over iron-manganese catalysts supported on silicalite-1 and zeolite ZSM-5.' A comparison of our results with unsupported iron-manganese catalysts showed that the zeolite-supported catalysts have better selectivity to alkenes.The aim of this work is to study the influence of systematic manganese addition to iron catalysts supported on silicalite- 1 in an attempt to understand the role of manganese. Experimental Silicalite-1 was synthesized according to a method described in the 1iterat~re.l~ Impregnation was carried out by adding appropriate amounts of 0.45 mol dm-3 metal nitrate solu- tions to 10 g of the freshly calcined support. The slurry was stirred continuously on a water bath and the excess of solvent was removed slowly under vacuum. The solid material was then crushed, dried at 120 "C for 12 h and finally calcined at 450°C in air for 8 h to convert the metal nitrates into oxides.The catalysts were designated as Fe(x)/Sil and Fe-Mn(x, y)/ Sil, where x and y indicate the wt.% of Fe and Mn, respec- tively. The X-ray diffraction (XRD) patterns of the catalysts were recorded using a Philips PW-1820 diffractometer with nickel-filtered Cu-Kcr radiation at a scanning rate of 2" min-'. Temperature-programmed reduction (TPR) of the catalysts was studied in a conventional flow apparatus, as described previously.' The transmission Mossbauer spectra of the catalysts in their calcined form and after use were obtained using a PC-based multichannel analyser. A 5 mCi 57Co(Rh) source was used in the constant-acceleration mode.The spectrometer was calibrated with a standard a-Fe absorber. The spectra were fitted using a computer program. Mossbauer spectra of the reduced catalysts could not be recorded as they were pyrophoric. CO hydrogenation was carried out in a high-pressure flow reactor (BTRS-Jr., Autoclave Engineers, USA) at 21 atm. The detailed experimental set-up for the synthesis gas conversion and the analytical procedure were described elsewhere.' The calcined catalysts (1 g, particle size 180-300 pm)were loaded in the reactor and reduced in situ in a flow of hydrogen (20 cm3 min -') at 450 "C for 10 h before CO hydrogenation was carried out. The conversion of CO on the various catalysts did not change much after ca. 1 h of reaction.The data reported in the tables are those obtained after 6 h on stream. Results and Discussion XRD Studies The XRD pattern of the silicalite-1 support showed that it was highly crystalline and free from impurities. The XRD patterns show the presence of a-Fe,O,. The intensity of the M-F~,O, peaks increased with the iron content from 5 to 20 wt.%. In contrast, the manganese-promoted Fe-Mn(x, y)/Sil catalysts showed only peaks that are characteristic of the support. The absence of the peaks for oxides of iron and manganese may be due to their very small particle size. For unsupported iron and manganese oxides, which were pre- pared by calcining the corresponding nitrates at 450 "C, strong peaks for a-Fe203 and fi-MnO, respectively, were noted.The unsupported Fe-Mn (1 : 2) catalyst showed only weak lines due to a-Fe,O, and p-(Mnl -%FeJ203 indicating poor crystallinity.' 5*16 This indicates that the addition of Mn 1994 J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 Table 1 Mossbauer parameters of calcined Fe(x)/Sil catalysts reduces the particle size of iron oxide in unsupported Fe-Mn catalysts. A similar effect appears to be responsible for the 6 A Heff areaabsence of the XRD lines of the metal oxides for the Fe- catalyst component /mm s-' /mm s-' /kG (Yo)Mn(x, y)/Sil catalysts. It may be concluded that the addition of Mn increases the dispersion of iron oxide. Fe(S)/Sil a-Fe,O, 0.44 0 503 77.4 The catalysts used did not show any XRD lines other than SPO" 0.26 0.87 -22.6 those due to the support.There was also some loss in crys- Fe( lO)/Sil a-Fe,O, 0.44 0 502 86.9 SPO" 0.15 0.68 -13.1tallinity of the support. Fe(20)/Sil a-Fe,O, 0.39 0 506 97.9 SPO" 0.13 0.66 -2.1 Miissbauer Studies " Superparamagnetic oxide of iron. The room-temperature Mossbauer spectra of the calcined catalysts containing various amounts of iron are shown in Fig. 1. All three samples show superposition of a six-line size is found to be 15 nm on Fe(S)/Sil, 17 nm on Fe(lO)/Sil pattern of large particles of a-Fe,O, and a central doublet and >18 nm on Fe(20)/Sil samples. due to supermagnetic iron oxide (SPO).The area ratios of the The incorporation of manganese was found to decrease the six-line pattern and the superparamagnetic doublet are given average particle size of the iron oxide particles as the area of in Table 1.The area fraction of the six-line pattern increased the six-line pattern due to large a-Fe,O, particles was found with the iron content. Thus, the average particle size of iron to decrease with the addition of manganese (Fig. 2). The area oxide increased with iron content, as expected. According to ratios of the six-line pattern and the doublet are given in Kiindig et d.,"the average particle size of iron oxide can be Table 2. Thus, with 20 wt.% manganese added to the cata- estimated from the area ratio of the six-line pattern to the lyst, the sextet spectra completely disappeared, leaving only superparamagnetic doublet. The estimated average particle ..1oo.c 100.0 99.8 99.8 99.6 99.6 99.4 99.4 .-99.2 99.2 1 . *. 99.0 99.0 100.0 100.0 99.8 h h$ 99.8 S 99.6 v v c C .-0 .o 99.4v) v).% 99.6 .-E 6 99.2 c 4-z 2 99.099.4 98.8 98.699.2 ..* :. s.-.. 100.0 -100.0 -99.8 -99.6 -99.6 -99.2 -It99.4 -98.8 -99.2 -98.4 -99.0 -98.0 -98.8 -97.6 -98.6 -97.2 I I I I f I I I 98.4 I I I I I I I I 1 -12 -9 -6 -3 0 3 6 91;-12 -9 -6 -3 O 3 6 9 12 velocity/rnrn s-' velocity/rnm s-' Fig. 1 Mossbauer spectra of calcined Fe(x)/Sil catalysts: (a) Fe(5)/ Fig. 2 Mossbauer spectra of calcined Fe-Mn(x, y)/Sil catalysts: (a) Sil; (b) Fe(lO)/Sil and (c) Fe(2O)/Sil Fe-Mn(l0, 5)/Sil; (b) Fe-Mn(l0, lO)/Sil and (c) Fe-Mn(l0, 20)/Sil J.CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 Table 2 Mossbauer parameters of calcined Fe-Mn(x, y)/Sil cata- lysts 6 A Heff area catalyst component /mm s-' /mm s-l /kG (%) Fe-Mn(l0, 5)/Sil a-Fe,03 0.39 0 498 68.0 SPO" 0.28 0.85 -32.0 Fe-Mn(l0, lO)/Sil a-Fe,03 0.44 0 503 64.0 SPO" 0.28 0.87 -36.0 Fe-Mn( 10, 20)/Sil SPO" 0.3 1 0.83 -100.0 " Superparamagnetic oxide of iron. the doublet. This clearly shows that addition of manganese increases the dispersion of iron oxide in the catalyst precur- sor. The Mossbauer results also show that the a-Fe203 phase is actually present as small particles in the calcined Fe- Mn(x, y)/Sil catalysts, although XRD could not detect this phase. TPR Studies Fig. 3 shows the TPR spectra of the Fe(x)/Sil and Fe-Mn(x, y)/Sil catalysts and that of bulk a-Fe,O, .a-Fe,O, was found to undergo complete reduction in a single step under the experimental conditions and the peak maximum was at 600°C.However, the TPR spectra of the supported catalysts showed two peaks and the reduction temperature was lower. We attribute this lowering of the reduction temperature in our catalysts to the smaller particle size of the iron oxide on the silicalite support. This is corroborated by the fact that in manganese-promoted catalysts, where the iron oxide particle size is reduced with increasing manganese content, the reduction begins at an even lower temperature. The appearance of the two TPR peaks indicates that iron oxide either undergoes a two-step reduction process or it is located at two different sites with different reducibility.The first peak can be assigned to the reduction of a-Fe203 located outside the silicalite pores. This is clear from the 100 200 300 400 500 600 700 800 T/"C Fig. 3 TPR spectra of the calcined samples: (a) Fe-Mn(l0, O)/Sil, (b) Fe-Mn(20, lO)/Sil, (c) a-Fe,O,, (d) Fe-Mn(l0, 5)/Sil, (e) Fe-Mn(l0, lO)/Sil, cf)Fe-Mn(l0, 20)/Sil figure which shows that the intensity of this peak increases with increase in iron content from 10 to 20 wt.% as more oxide particles will be outside the silicalite pores at higher loading. The second peak we speculate as due to the reduction of iron oxide located inside the silicalite pores. In the Fe-Mn(x, y)/Sil system we also observed two TPR peaks. The first peak is assigned to the reduction of a-Fe,O, .The intensity of the first peak decreased with increasing man- ganese content from 5 to 20 wt.% in the catalysts. The second peak is shifted to an appreciably lower temperature than that in Fe(x)/Sil catalysts. Although we did not notice the pres- ence of any other phase in the XRD patterns, the possibility of the presence of very fine particles of fi-(Mnl-,Fe,),O, cannot be ruled out as such a phase has been reported in the calcined Fe-Mn sarnple~.'~-~~The intensity of the second TPR peak increased with manganese content, indicating the formation of the Fe-Mn mixed phase. CO Hydrogenation Eflect of Iron Oxide Loading The performances of the catalysts for CO hydrogenation was studied from 275 to 300°C at 21 atm.The results for the Fe(x)/Sil catalysts at 275°C are given in Table 3. The CO conversion per gram of iron increased with iron oxide par- ticle size in the calcined catalysts. This is possibly because the larger iron oxide particles are easily carburized in the reac- tion atmosphere. Mossbauer studies of the Fe(x)/Sil catalysts (Fig. 4, Table 4) showed increased carbide formation in the order Fe(S)/Sil < Fe( lO)/Sil < Fe(20)/Sil The increase in CO conversion with the increasing forma- tion of iron carbide is not unexpected as it was also observed Table 3 CO hydrogenation Fe(x)/Sil catalysts (T=275"C;P=21atm;CO:HZ= l;GHSV= 1200h-') Fe(S)/Sil" Fe( lO)/Sil Fe(20)/Sil CO conversion 22.7 8.7 14.3 CO converted to [pmol (g Fe)-'s-'] CO, (molYo) hydrocarbons (mol%) 2.0 10.6 1.2 5.9 4.4 17.0 hydrocarbon distribution (wt.?h) c, CZ c3 c4 C,(alkene) C,(alkene) C,(alkene) c5+ O/P* 14.2 2.5 12.8 1.5 23.2 2.0 17.9 26.1 9.0 11.2 2.7 11.1 1.1 19.6 1.5 13.9 39.0 8.4 12.4 8.9 6.4 4.7 20.3 8.9 8.4 30.2 1.6 " GHSV = 920 h-'.Alkene :alkane ratio in C,-C4 fraction. Table 4 Area fraction of Hagg carbide and superparamagnetic oxide in the used catalysts Hagg carbide superparamagnetic catalyst (X-Fe,C,) oxide Fe(S)/Sil 75 25 Fe( 1 O)/Sil 85 15 Fe(20)/Sil 89 11 Fe-Mn( 10, 5)/Sil 80 20 Fe-Mn( 10, lO)/Sil 79 21 Fe-Mn( 10, 20)/Sil 71 29 1996 100.0 99.8 99.6 99.4 J 100.0 h S 99.8 v C 0.-v) --99.66 !2+ gg.4199.2 *..--.... .. . -.... .... .. *....:-... ..-at. --:* *. f..... . J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 It is still uncertain whether iron carbide or iron oxide is responsible for alkene formation. According to Itoh et aL2 iron carbide, Fe2C, is the most favourable iron species for alkene production. Our catalysts did not have this carbide phase. Our results show that a large amount of z-Fe,C2 is formed and this favours the formation of alkanes. Although CO conversion increased with the amount of this carbide, it took place at the expense of alkene selectivity. This suggests that the formation of alkenes is favoured by the oxide phase containing Fe"'. Reymond and co-workers26-28 observed very high CO hydrogenation activity in the presence of Fe,O, but they did not report the alkene selectivity of their catalysts.Itoh et al.25 found MnxFe3-x04 to have good alkene selectivity . Effect of Addition of Manganese Conversion us. time on stream plots are shown in Fig. 5. The catalytic activity remains more or less constant after the first hour. The results of CO hydrogenation on Mn oxide pro- moted iron catalysts are shown in Table 5. Addition of Mn was found to improve both CO conversion and the selectivity of the catalysts to alkenes. However, methane selectivity remained almost unchanged by Mn addition and the amount of C, + fraction increased substantially. The alkene selectivity of Fe(x)/Sil catalysts rapidly reached a steady state whereas this took 3-4 h for the Fe-Mn(x, y)/Sil catalysts (Fig.6). If alkene formation is favoured by the Fe"' oxide phase, this phase must be formed more readily in the Fe(x)/Sil catalysts but slowly in the FeMn(x, y)/Sil catalysts. Perhaps the active phase in alkene synthesis is finely divided Fe,O, and 99.0 I -8 -6 -4 -2 0 2 4 6 8 velocity/mm s-' Fig. 4 Mossbauer spectra of used Fe(x)/Sil catalysts: (a) Fe(S)/Sil; (b)Fe(lO)/Sil and (c) Fe(20)/Sil i0, I I I I I I 0123456 by other workers. Raupp and Delgass18 noticed that the time on stream/h increase in activity of supported iron catalysts parallels their Fig. 5 CO conversion us. time on stream plots at 275 "C: D,increased carbide content and a similar observation was Fe( lO)/Sil; ., Fe-Mn(l0, lO)/Sil; Fe(20)/Sil; 0,Fe-Mn(l0, 5)/Sil; 0, 20)/Sil10,Fe-Mn( ,, made by other worker^.'^-'^ In the low-temperature synthesis gas reaction, the final form of catalysts are either Hagg (x) or hexagonal close- packed (E) carbides and only at temperatures >500°C is cementite phase (Fe,C) found.23 Callejea et al.also noted the presence of only the Hagg carbide phase in Fe/ZSM-5 cata- lyst~.~~However, in Mossbauer spectra we observed the pres- ence of some oxides of Fe"'. Although XRD of the catalysts did not show any carbide, Mossbauer spectra clearly showed Hagg carbide, suggesting that the latter is formed only on the surface layer. While an increase in iron loading in Fe(x)/Sil catalysts led to increased CO conversion, the alkene selectivity decreased.Thus, Fe(20)/Sil showed a drastic decrease in alkene selec- tivity, indicating the high hydrogenation activity of this cata- lyst. The selectivity of the catalysts to alkenes can be correlated with the particle size of the iron oxide precursor. Catalysts that initially contained larger iron oxide particles showed higher CO conversion and gave more alkane pro- ducts. 16 , a--0-I I a 0 I I I I I I Fig. 6 Alkene :alkane ratio us. time on stream plots a 275 "C: D, Fe(lO)/Sil; ., Fe-Mn(l0, lO)/Sil; Fe(20)/Sil; 0,Fe-Mn(l0, 5)/Sil; 0,b,Fe-Mn( 10, 20)/Sil J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 Table 5 CO hydrogenation on Fe-Mn(x, y)/Sil catalysts (P = 21 atm; CO :H, = 1; GHSV = 1200 h-l) Fe-Mn(l0, 5)/Sil Fe-Mn(l0, lO)/Sil Fe-Mn( 10, 20)/Sil catalyst temperature 275 300 325 275 300 325 275 300 325 CO conversion/pmol (g Fe)-' s-' 6.7 28.4 47.7 13.9 32.8 69.6 23.8 64.3 102.3 CO converted to CO, (molyo) 0.7 4.4 10.8 2.7 5.5 13.1 3.5 15.9 29.1 hydrocarbons (mol%) 4.5 17.7 26.4 7.7 19.0 38.9 12.9 28.4 41.4 HC distribution (wt.Yo) c, 9.3 13.3 18.2 9.3 13.6 23.7 8.6 10.5 23.0 2.6 9.3 13.0 2.3 4.8 15.3 2.0 3.1 11.1c2 C,(alkene) 17.4 9.7 6.1 13.0 11.1 1.6 12.7 11.0 3.2 0.6 1.8 6.6 0.9 1.o 10.9 0.9 0.9 6.2c3 C,(alkene) 29.3 25.0 20.4 23.6 24.0 13.3 23.3 21.3 17.7 c4 2.9 14.1 12.9 1.1 2.9 12.5 1.2 1.2 11.7 C,(alkene) 18.6 4.5 4.0 17.0 15.5 2.6 17.9 16.1 4.6 c5 + 19.4 22.4 18.7 32.7 27.3 20.1 33.5 36.0 22.6 0 : P" 10.7 1.6 0.9 12.5 5.8 0.5 13.1 9.3 0.9 a Alkene : alkane ratio in C,&, fraction.MnxFe, -x04for the Fe(x)/Sil and Fe-Mn(x, y)/Sil catalysts, respectively, as these phases have been reported to be active -. .. * in alkene synthesis.25 These phases could be formed from r-._.-Fe,O, under the reducing conditions in the reactor, in which 100.0 case Fe,04 is formed readily and MnxFe3-x04 slowly. It is, however, not possible to confirm this speculation as the XRD results did not show the presence of these phases, possibly because of their extremely small size. 99.8 The Mossbauer spectra of the Fe-Mn(x, y)/Sil catalysts are shown in Fig. 7. These samples were collected after 6 h on stream at the reaction temperature.Catalyst precursors with 99.6larger iron oxide particles contained a higher ratio of the carbide phase. The amount of Hagg carbide present in these catalysts decreased with Mn loading (Table 4). This is because Mn reduces the particle size of iron oxide and small 99.4' %particles of iron oxide are difficult to carburize. Although the amount of Hagg carbide decreased continuously with Mn 100.0loading and CO conversion initially dropped at 5 wt.% Mn, further addition of Mn resulted in an increase of CO conver- 99.8 hsion. This behaviour is different from that of Fe(x)/Sil cata- $lysts where CO conversion was found to increase with the -99.6 amount of Hagg carbide in the catalysts. This indicates that .-0 in manganese-promoted iron catalysts, the amount of Hagg .-s 99.4 carbide is not the only phase responsible for CO conversion.5 It is possible that in addition to iron carbide, some mixed E 99.2 CIiron manganese phases may also be responsible for the increased conversion. 99.0 The alkene selectivity of the Mn-promoted iron catalysts supported on silicalite 1 was found to be higher than that for 98.8' .. .-, . .-.. . .unpromoted catalysts (Fig. 6). This increase in alkene selec- tivity with decreasing carbide formation further supports our view that alkene formation possibly takes place on the iron oxide. In Mn-promoted ultrafine iron particles the MnxFe3-x0, phase was found to be the preferable structure for high alkene ~electivity.'~ Thus it appears that in Mn- 99.2 -promoted iron catalysts, Mn plays an important role by reducing the iron oxide particle size and thereby making the 98.8 -particles difficult to carburize, which in turn increases the alkene selectivity.1Although using higher reaction temperatures increased CO 98.4 conversion, this has a negative effect on alkene selectivity 1 I 1 I I I I 1(Table 5). The use of higher temperatures increases the alkane 98.0 ! content and also gives a large amount of CO, in the product. -8 -6 -4 -2 0 2 4 6 8 velocity/mm s-'CO, is possibly formed by the parallel water gas shift reac- tion which is accelerated at higher temperature. Using a high Fig. 7 Mossbauer spectra of Fe-Mn(x, y)/Sil catalysts: (a) Fe-temperature also enhances the hydrogenation of the alkenes Mn(l0, 5)/Sil; (b)Fe-Mn(l0, lO)/Sil and (c) Fe-Mn(l0, 20)/Sil 1998 J.CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 initially formed, thus increasing the amount of alkanes in the product. 8 9 V. U. S. Rao and R.J. Gormley, Hydrocarbon Process, Int. Ed., 1980,59, 139. M. Peukert and G. Linden, Proc. 8th Int. Congr. Catal., Verlag Chemie, Weinheim, 1984, vol. 2, p. 135. Conclusions 10 B. Biissemeier, C. D. Frohning and B. Cornlis, Hydrocarhon 1. Addition of manganese to silicalite-supported iron cata- 11 Process, Int. Ed., 1976,55, 105. H. Kolbel and K. 0.Tillmetz, US Patent 4,177,203, 1979. lysts gives increased alkene selectivity in CO hydrogenation. 2. The active phase of the catalysts consists of Hagg carbide and oxides of Fe"'.3. The role of manganese is to reduce the particle size of the iron oxide precursor, making it difficult to undergo car- 12 13 14 J. Barrault, C. Forquy and V. Perrichon, Appl. Catal., 1983, 5, 119. D. Das, G. Ravichandran, D. K. Chakrabarty, S. N. Piramanay- agam and S. Shringi, Appl. Catal., 1994, 107, 73. C. V. V. Satyanarayana and D. K. Chakrabarty, Appl. Catal., 1990,66,1. burization. This increases the amount of the iron oxide phase 15 D. Das, G. Ravichandran, D. K. Chakrabarty, unpublished in the catalysts. 4. The oxide of iron and the mixed iron-manganese oxide present in the catalysts are possibly responsible for their increased alkene selectivity. 16 17 work. G. C. Maiti, R. MaIessa and M. Baerns, Appl. Catal., 1983, 5, 151.W. Kiindig, H. Bommel, G. Constabaris and R. H. Lindquist, Phys. Rev., 1966, 142, 327. This work was supported by a research grant from the 18 19 G. B. Raupp and W. N. Delgass, J. Catal., 1979,58,348; 361. H. P. Bonzel and H. J. Krebs, Surf: Sci., 1980,91,499. Department of Science and Technology, New Delhi. 20 21 H. P. Bonzel and H. J. Krebs, Surf: Sci., 1982,117,639. H. J. Krebs, H. P. Bonzel, W. Sebwarting and G. Gafaer, J. References 22 Catal., 1980,63, 226. D. J. Dwyer and H. Hardenbergh, J. Catal., 1984,87,66. R. Snell, Catal. Rev. Sci. Eng., 1987,29, 361. M. Janardanrao, Ind. Eng. Chem. Res., 1990,29,1735. H. H. Nijs, P. A. Jacobs and J. B. Uytterhoeven, J. Chem. SOC., Chem. Commun., 1979, 180. S. Kawi, J. R. Chang and B. C. Gates, J. Am. Chem. SOC., 1993, 115,4830. P. L. Zhou, S. D. Maloney and B. C. Gates, J. Catal., 1991, 129, 315. C. D. Chang, W. H. Lang and A. J. Silvestri, J. Catal., 1979, 56, 268. P. D. Caesar, J. A. Brennan, W. E. Garwood and J. Ciric, J. 23 24 25 26 27 28 K. B. Jensen and F. E. Massoth, J. Catal., 1985,92, 109. G. Callejea, A. de Lucas, R. van Grieken, J. L. Pefia, A. Guerrero-Ruiz and J. L. G. Fierro, Catal. Lett., 1993, 18, 65. H. Itoh, S. Nagano, K. Takeda and E. Kikuchi, Appl. Catal., 1993,%, 125. J. P. Reymond, P. Mkriandeau and S. J. Teichner, J. Catal., 1982, 75, 39. F. Blanchard, J. P. Reymond, B. Pommier and S. J. Teichner, J. Mol. Catal., 1982, 17, 171. B. Pommier, J. P. Reymond and S. J. Teichner, Stud. Surf: Sci. Catal., 1984, 19, 471. Catal., 1979,56, 274. Paper 3/07535D; Received 23rd December, 1993