J . Chern. Soc., Favaday Trans. I, 1988, 84(5), 1311-1328 Hydrocarbon Formation from Methanol and Dimethyl Ether using WO,/Al,O, and H-ZSM-5 Catalysts A Mechanistic Investigation using Model Reagents Graham J. Hutchings,* Lawrence Jansen van Rensburg, Wolfgang Pick1 and Roger Hunter" Department of Chemistry, University of the Witwatersrand, PO Wits, 2050 Johannesburg, South Africa The methanol conversion catalysts H-ZSM-5 and 10 YO WO,/A1,0, show significant differences in their activities as well as product selectivities towards a range of reagents Me-X [X =-OH, -OMe, -1, -OSO,Me, -OP(OMe),O]. WO,/Al,O, is typically two orders of magnitude less active and gives significant yields of methane and hydrogen compared to the zeolite catalyst. Formation of C , , hydrocarbons over WO,/Al,O, with Me,SO, and (MeO),PO provides evidence that the trimethyloxonium ion is an unlikely intermediate in the methanol to hydrocarbon conversion reaction, since these reagents cannot form this type of intermediate.Reactions in the presence of added hydrogen do not show significant differences in the product selectivities, particularly methane, indicating that the methane does not originate from reaction of an intermediate, e.g. methylene carbene, with molecular hydrogen. Significant differences in catalyst behaviour in the presence of added NO and 0, indicate different reaction mechanisms for initial C-C bond formation with both catalysts. The conversion of methanol, or dimethyl ether, into gasoline range hydrocarbons has received considerable attention since 1973 as it represents a process by which liquid hydrocarbon fuels can be formed independently of crude oil, and the process has recently been commercialised in New Zealand.The formation of hydrocarbons from methanol has been known for more than a century, and zinc halides were used as the early catalyst^.^.^ Supported aluminium sulphate4 and phosphorus pentoxide5 have also been cited as catalysts for this reaction. However, the catalyst which has received the most attention in this context is the highly siliceous zeolite of intermediate pore size denoted ZSM-5,6*7 which has been shown to produce premium-grade gasoline from methanol by virtue of its high acidity and shape selectivity. Initially it was Chang and Silvestri' who demonstrated that ZSM-5 converted a wide range of small organic molecules into higher hydrocarbons, and the effect of reaction conditions using ZSM-5 has now been well studied,' producing a considerable amount of literature data.'-'' More recently, Olahl2, l3 and Ip14 have shown that bifunctional acid-base catalysts, such as W0,/A1,03, can also convert methanol into ethene and lower alkenes.Whereas supported tungsten oxide and tungstate catalysts have been extensively investigated for alkene metathe~is'~ and methanol oxidation reactions,l' such catalysts have not been well studied for the methanol conversion reaction. Of particular interest for both W03/A1203 and ZSM-5 catalysts is the mechanism by which the initial carbon-carbon t Present address : Leverhulme Centre for Innovative Catalysis, Department of Inorganic, Physical and Industrial Chemistry, The University of Liverpooll Liverpool L69 3BX.131 11312 Hydrocarbon Conversion using W0,/A120, and H-ZSM-5 bond is formed. To date there has been no detailed comparative mechanistic study of the C , -+ C, transformation step using these two catalysts, whilst it has been previously ~onsidered'~ that the same mechanism operates for this step with both. In this paper we extend our previous mechanistic and report on a comparative study of WO,/Al,O, and ZSM-5 as catalysts. In particular, by using model reagents and probe reactions we demonstrate that the mechanisms of CH, and C-C bond formation are significantly different for the two catalysts. Experimental The acid form of zeolite ZSM-5 (H-ZSM-5) was prepared as previously described.lo Tungsten oxide (10 YO) supported on y-alumina was prepared according to the method of Olah et aZ.13 y-Al,O, (Strem Chemicals, surface area 240 m2 g-') was impregnated with an aqueous solution of ammonium metatungstate using the incipient wetness technique (ammonium paratungstate was found to give inferior catalysts). The catalyst was dried in air at 120 "C for 8 h and calcined at 550 "C for 5-16 h. Catalysts prepared using the longer calcination period gave superior results. Following calcination the catalyst was found to be amorphous by X-ray diffractometry, which is in agreement with previous structure studies of this catalyst. The calcined W0,/A120, was pelleted, without addition of binder, and sieved to give particles (0.5-1.0 mm).10% W03/y~-A120, was also prepared according to the method of Maitra et aZ.22 y-Al,O, was stirred in an excess of aqueous ammonium metatungstate at pH 6.5, 25 "C for 48 h. The catalyst was recovered by filtration, dried at 120 "C for 2 h and calcined at 500 "C for 16 h. H-ZSM- 5 and W03/A1,0, were reacted with methanol and methylating agents using the procedures previously described. 2" Results Reaction of Methanol and Dimethyl Ether over WO,/AI,O, Methanol and dimethyl ether were reacted over WO,/Al,O, using a range of experimental conditions, and the results are given in table 1. For the W0,/A120, catalyst the principal hydrocarbon product under most conditions was methane, as has been previously noted by Olah.'2,13 In addition, hydrogen was formed in comparable yield to that of methane and in general both methane and hydrogen yields increase with increasing reaction temperature and conversion.It is also apparent that methane selectivities were high for methanol as a reagent compared to dimethyl ether at comparable reagent feed rates. These product selectivites are in direct contrast to those observed for methanol conversion using H-ZSM-5 as catalyst when both methane and hydrogen are minor reaction products. Previous studies have shown that methane only becomes a significant product at low conversions'1~20~23 and it is considered to be the primary hydrocarbon product at such conditions. l3 In addition to the differences observed in product selectivities, W0,/A120, and H- ZSM-5 demonstrate considerably different catalytic activities.In general, low reactant feed rates are required to achieve complete conversion for W0,/Al,0,,12-14 and this has been confirmed in this study. With H-ZSM-5 high conversions can be achieved at comparable temperatures with increased reactant flow rates,' and H-ZSM-5 is typically two orders of magnitude more active than WO,/Al,O, for the methanol conversion reaction. In addition H-ZSM-5 exhibits a longer lifetime compared to that of WO,/ A120,. Under the conditions cited in this study the W0,/A120, became rapidly discoloured owing to the laydown of carbonaceous deposits, and a significant irreversible decrease in the catalyst activity was observed from ca. 10 h reaction time with MeOH as reactant. For MeOMe, by virtue of the higher feed rates used, the lifetime was onlyG.J . Hutchings et al. 1313 Table 1. Reaction of MeOH over 10% W03/A1203 hydrocarbon selectivity (YO by mass) conversion H Z b H2/CH4 WHSV" T I T (YO) CH, C,H, C2H6 C3H6 C,H, C, C,+ (mol%) molar ratio 0.008 0.008 0.008 0.008 0.002 0.017 0.30 1.48 14.8 275 0.8 29.2 26.4 300 17.2 22.8 13.2 350 38.4 28.3 13.6 400 97.5 31.7 16.8 400 100 71.6 12.0 400 100 51.4 17.8 400 81.0 20.4 19.6 400 68.5 18.6 22.1 400 18.1 25.4 16.0 MeOHc 0.2 35.3 0.2 23.9 0.4 24.4 1.9 17.4 0.8 7.8 5.9 12.1 MeOMed 5.7 21.9 5.4 25.4 5.6 20.5 tr 2.6 6.3 0.05 0.7 0.1 19.5 20.3 0.14 0.6 tr 13.3 20.0 0.70 0.7 tr 7.3 24.9 4.5 1.8 tr 5.4 2.4 - tr 6.2 6.6 - e - tr 11.8 - tr 15.8 8.4 0.8 0.3 tr 15.4 8.7 - a WHSV = g(Me0H) g(cata1yst)-' h-l. according to Olah.13 Based on total exit gas analysis.Catalyst prepared Catalyst prepared according to Maitra et aZ.22 Not determined. Table 2. Reaction of methylating agents over W03/A120, a hydrocarbon selectivity (YO by mass) conversion WHSV/h-l T/"C (Yo) Cl c2 c3 c, c,+ Me2S0, 0.05 300 2 92 8 tr - - 0.05 370 7 37 19 11 33 tr 0.03 400 21 82 11 4 3 tr 0.05 440 15 74 14 5 7 tr Me1 0.9 300 0.2 95 5 tr - - 0.9 400 8 97 2 1 tr - (MeO),PO 0.05 300 1 1 12 35 52 tr 0.05 350 4 81 5 5 9 tr 0.9 300 1 9 2 3 4 1 - " Test experiments showed the absence of blank thermal reaction for all reaction conditions given. a few hours. In contrast, H-ZSM-5 showed no decrease in activity over a similar reaction period. However, it has not been established that 10%-WO,/Al,O, is the optimum tormulation, and hence significant improvements may be possible with respect to both catalyst activity and lifetime.Reaction of Methylating Agents over WO,/Al,O, Me,SO, and Me1 were individually reacted over WO,/A1,0, using a range of reaction conditions, and the results are shown in table 2. These reagents were found to be much less reactive than MeOH or MeOMe, and the order of reactivity was similar to that observed for H-ZSM-5.l' At the low reagent conversions observed the hydrogen yield was low in all experiments,'l and for Me,S04 ranged from trace levels at 300 "C to 0.03 mol % at 440 "C. With both reagents the major product was methane, and methane1314 Hydrocarbon Conversion using WO,/Al,O, and H-ZSM-5 selectivity increased with increasing conversion. The formation of ethene and higher alkenes with Me,SO, is considered to be significant with respect to the mechanism of the C, -+ C , transformation, since under these reaction conditions formation of a trimethyloxonium ion from Me,SO, is not considered possible.'" To exemplify further that reagents incapable of trimethyloxonium formation can give rise to C,, hydrocarbons, (MeO),PO was reacted over WO,/Al,O,.The results (table 2) show that although a low conversion was effected a significant yield of C,, products were observed in addition to methane. Olah12* l3 has proposed that carbon-carbon bond formation with the WO,/Al,O, bifunctional acid-base catalyst occurs via a trimethyloxonium intermediate, and this mechanism has been termed the oxonium-ion ylide mechanism. Such an intermediate has also been proposed for super-acid catalysts, e.g.TaF,,,, and it has been extrapolated" that it is also involved in carbon-carbon bond formation with H-ZSM-5. However, in this study we have shown that carbon-carbon bond formation can occur with reagents, Me,SO, and (MeO),PO, which are unlikely to give a tri- methyloxonium ion intermediate. Whilst we have shown this previously for H-ZSM-5,Othe results of this study confirm that the onium-ion ylide mechanism, proposed by Olah12713 and van der Berg et aZ.25 is an unlikely possibility with WO,/Al,O,. The product selectivities for the formation of hydrocarbons from methylating reagents of the type Me-X (X = OH, I, OS0,Me) are considerably different for WO,/A1,0, and H-ZSM-5, particularly for C,, hydrocarbons.Fig. 1 shows the selectivities observed with these three reagents at comparable conversions. These vast differences in product selectivity may be indicative of different reaction mechanisms for these two catalysts. However, differences in C,, hydrocarbons can also be accounted for by the relative differences in Bronsted acidity for the two catalysts. For ZSM-5 the high Bronsted acidity gives rise to rapid secondary 27 that occur via carbocationic intermediates. Such mechanistic pathways are not so favoured with WO,/Al,O, and hence the C,, products tend to be the alkenes C,H, and C,H,. Reaction of Methanol over Model Catalyst Systems Methanol was separately reacted over y-Al,O, and SiO, and the results are shown in table 3. These two materials were chosen to model the separate behaviour of [AlO,] and [SiO,] tetrahedra in the ZSM-5 structure.Under all reaction conditions the products comprised almost exclusively methane and dimethyl ether, whilst only limited C-C bond formation was observed. These systems can be considered as models for the methane formation reaction over ZSM-5. Reaction of SiO, or y-Al,O, with methanol would lead to the formation of a surface methoxyl species. These species would preferentially react with the high localised concentration of methanol uia hydride abstraction to form methane, and via methylation to form dimethyl ether. Deprotonation via an adjacent oxygen site to form the methylide species considered responsible for C-C bond formation cannot occur to any appreciable extent. The surface coverages of WO, on a high surface area y-Al,O, for the 10%-W03/ Al,O, catalyst has previously been shown to be less than a mono1ayer.28 Hence in addition to y-Al,O,-supported WO, an appreciable surface area of y-Al,O, remains exposed.The results shown in table 3 indicate that any exposed y-Al,O, surface could contribute to the high methane yield observed with these catalysts. Additionally, unsupported WO, (Merck, P. A., calcined 550 "C in air, 16 h) was also found to be active for methane formation under reaction conditions comparable to those used for the WO,/Al,O, catalyst. The catalytic effect of pure WO, was short-lived, but that of y-Al,O, was observed for an extended reaction period. Comparison of the results for WO,/y-Al,O, with the pure components indicates that in combination the WO, and y- A1,0, exhibit a synergistic effect since the methane yield is significantly decreased and carbon-arbon bond formation becomes the dominant reaction. It is well k n ~ w n ~ ~ - ~ OG.J. Hutchings et al. 100 80 60 1315 - - ( b ) - 100 80 60 40 20 c, c2 c3 c4 c5+ 100 80 60 40 20 Cl c2 c3 c4 1 c5 40 20 100 80 60 40 20 i? Cl c2 c3 c5 100 80 60 40 20 100 80 60 40 20 Fig. 1. Comparison of product selectivities for reaction of reagents MeX (X = OH, I, OSO,, Me) over H-ZSM-5 and WO,/AI,O, at comparable conversions. catalyst reagent T/"C WHSV/ h-l conversion (%) (a) H-ZSM-5 MeOH 250 0.005 87 (b) H-ZSM-5 MeSO, 250 0.075 21 ( c ) H-ZSM-5 Me1 250 0.6 0.1 ( d ) W0,-AI,O, MeOH 400 0.008 97.5 (e) W0,-Al,O, MeSO, 400 0.05 21 cf) W0,-AI,O, Me1 350 0.09 1 that the structure of WO, is modified by y-Al,O,, and the results of this study demonstrate the catalytic significance of this with respect to methanol conversion.It is also apparent that the WO,/Al,O, catalyst requires further optimisation to eliminate the unwanted methane yield generated via reaction of methanol with the uncoated y-Al,O, surface. 44 F A R 11316 Hydrocarbon Conversion using WO,/Al,O, and H-ZSM-5 Table 3. Reaction of MeOH over WO,, AI,O, and SiO, at 400 "C hydrocarbon product reaction conversion to selectivity (mass YO) time /min MeOMe (CH,), c, c2 c3 c4+ 25 75 150 210 270 20 75 150 225 100 180 120 170 210 295 330 570 20 375 475 100 180 68.6 34.2 22.6 13.4 12.9 24.5 27.0 46.7 11.1 17.2 15.4 86.3 79.2 78.8 77.8 76.8 70.6 62.9 72.6 74.9 3.8 3.5 WO,," WHSV = 0.20 14.5 90.6 4.3 5.4 79.4 5.4 0.4 45.9 20.7 0.1 49.0 17.3 0.06 56.5 9.3 69.0 93.9 5.2 69.7 92.2 5.1 40.7 91.1 6.8 3.3 94.5 2.4 1 .o 88.6 4.8 0.1 71.5 10.3 3.4 95.1 2.1 3.6 96.6 1.6 3.2 96.7 1.8 3.4 97.4 1.8 4.6 98.1 1.9 2.7 97.1 2.2 22.5 96.6 0.9 17.3 97.7 1.0 14.6 98.1 0.8 0.06 68.5 11.4 0.04 74.7 10.0 WO,," WHSV = 0.075 WO,,b WHSV = 0.18 A1,0,, WHSV = 0.15 A1,0,, WHSV = 0.075 SiO,, WHSV = 0.22 1.6 3.9 30.8 33.7 34.2 0.5 0.4 0.6 2.2 6.6 18.2 I .8 1.8 1.5 0.9 tr 0.7 2.1 1 .o 0.9 20.1 15.3 3.5 11.3 2.6 - - 0.4 2.3 1.5 0.9 - - 1 .o - - - - - 0.4 0.3 0.2 tr tr " Merck, P.A. Prepared by calcination of ammonium metatungstate. Model Reaction for Methane Formation involving Hydride Transfer In our previous studiesz0 we have proposed that methane formation from methanol over the zeolite catalyst H-ZSM-5 involves transfer of a hydride from methanol to a surface methoxyl group.To model this proposal we have carried out the methanol conversion reaction in the presence of an excess of cycloheptatriene, a known hydride donor and a more potent hydride source than methanol. The results for both H-ZSM-5 and WO,/Al,O, are shown in table 4. Cycloheptatriene is not stable within the H-ZSM-5 catalyst; it preferentially reacts to give the isomerisation product toluene and is also methylated via reaction with methanol. However, in the presence of cycloheptatriene with H-ZSM-5 the methane yield is increased by ca. 50%, which is highly significant. Blank thermal reaction of cycloheptatriene did not yield any methane, and it is therefore concluded that the increase in methane observed can be ascribed to hydride donation from cycloheptatriene. These data therefore support the proposa120 for reaction of methoxyl with a hydride source.G.J . Hutchings et al. 1317 Table 4. Reaction of MeOH and cycloheptatriene (CHT) over W0,/A1203 and H-ZSM-5 at 400 "C product selectivity (mol %) time conversion hydro- toluene hydrocarbon product selectivity reaction total (% by mass) /min (%) MeOMe carbons CH4 'ZH4 C2H6 '3 '4 ' 5 , 35 225 350 460 35 170 205 300 20 60 100 155 30 85 130 180 96.0 94.8 92.3 87.1 98.5 100 100 100 100 100 100 100 100 100 100 100 49.8 53.3 60.6 63.6 17.1 13.2 9.3 15.8 - - - - - - - - WO,/Al,O,, MeOH" 50.2 - 81.1 3.4 46.7 - 87.7 2.4 39.4 - 88.6 2.2 36.4 - 88.0 2.4 24.6 58.3 70.1 3.3 29.0 57.8 77.0 2.5 28.1 62.6 78.7 2.4 27.1 57.1 79.3 2.4 100 - 0.76 17.8 100 - 0.76 16.8 100 - 0.78 17.4 100 - 0.79 15.9 H-ZSM-5, MeOH/CHTb9" 65.4 34.6 1.10 21.7 64.3 35.7 1.08 23.1 63.1 36.9 1.13 22.3 64.5 35.5 1.24 22.2 WO,/AI,O,, MeOH/CHT'vb H-ZSM-5, MeOH" 0.1 1.6 5.8 8.0 0.4 1.1 6.1 2.3 0.5 1.0 6.7 1.0 0.6 1.1 6.2 1.8 1.0 2.5 6.3 16.8 0.9 1.8 5.0 12.8 0.9 1.6 3.3 13.1 0.9 1.7 3.6 12.2 0.3 31.7 24.6 24.8 0.3 30.9 24.3 26.9 0.3 31.3 24.0 26.2 0.3 30.1 24.1 28.8 0.5 25.4 14.8 36.5 0.3 21.1 10.8 43.6 0.3 22.7 12.8 40.8 0.3 24.4 11.0 59.1 a WHSV = 0.2 h-l.CHT/MeOH = 1 : 1. WHSV = 0.1 h-l. With the WO,/Al,O, catalyst different results are obtained and a slight decrease in methane formation is observed in the presence of cycloheptatriene (table 4).These results are therefore indicative that methane formation with the WO,/Al,O, catalyst may not involve a hydride transfer process. Reaction of Methanol and Dimethyl Ether with Gas-phase Additives Hydrogen For both H-ZSM-5 and WO,/Al,O, hydrogen and methane are produced in comparable amounts, but whereas for H-ZSM-5 yields of methane and hydrogen are low, these products are dominant for WO,/Al,O,. To test whether the methane with both catalysts is formed via reaction of an intermediate with hydrogen, reaction of methanol over WO,/Al,O, and H-ZSM-5 was carried out in the presence of added hydrogen. The results, shown in fig. 2 and 3, indicate that addition of hydrogen to the reactant feed stream had no significant effect either on product selectivity or on conversion for a wide range of experimental conditions and methanol conversion.In particular, there is no increase in methane in these experiments. These results are strong evidence against the involvement of free methylene carbene, a mechanistic proposal cited in a number of studie~,~7 319 32 since an increase in methane yield would have been expected based on the known reaction between methylene carbene and hydrogen.33 44-21318 Hydrocarbon Conversion using WO,/AI,O, and H-ZSM-5 40 30 30 I e 20 20 10 10 conv. C, C, C, C, LO Y conv. C, C, C, ( d ) 40 1 . - 30 30 1 E 20 20 10 20 m c 4 conv. C, C, C, C4 Cl c, c, c 4 Fig 2. Reaction of MeOH over WO3/A1,O3 with (unshaded) and without (shaded) co-fed hydrogen (9.8 mol% added to carrier gas): (a) 275, (b) 300, (c) 350 and (d) 400 "C; WHSV = 0.08 h-'.100 80 60 2 40 20 E rn 100 80 60 40 20 conv. C, C, C, C4 conv. C, C, C, C4 100 100 80 80 3 60 60 P 40 40 20 20 v rn conv. C1 C, C3 C4 conv. C1 C2 C3 C4 Fig. 3. Reaction of MeOH over H-ZSM-5 with (unshaded) and without (shaded) co-fed hydrogen (9.8 mol YO added to carrier gas): (a) 250 "C, WHSV = 0.05 h-l; (b) 300 "C, WHSV = 0.05 h-l; (c) 370 "C, WHSV = 0.05 h-'; ( d ) 250 "C, WHSV = 1.7 h-'.G. J. Hutchings et al. 1319 Table 5. Reaction of MeOH and MeOMe in presence of NO reaction product selectivity (mass YO) WHSV NO time conversion reactant /h-' (mol YO) /min (%) CH4 C2H4 'zHfj '3 '4 ' 5 , Me,O 50 0 0 0 0 Me,O 50 1 .o 1 .o 1 .o 1 .o 0 Me,O 14.8 0 0 Me,O 14.8 3.0 3 .O Ob MeOH 0.24 0 5.0d 30 60 120 180 30 60 120 180 200 6 120 30 60 90 235 300 H-ZSM-5" 28.8 7.7 6.8 5.5 H-ZSM-5" 27.4 11.3 6.9 6.6 5.5 H-ZSM-5" 95.4 23.9 86.6 73.7 18.8 H-ZSM-5' 100 100 0.8 12.4 0.1 24.7 21.5 40.5 0.3 9.2 0.1 16.4 34.5 42.5 0.5 10.7 0.1 19.8 38.9 30.0 0.5 10.2 0.1 20.0 45.4 23.8 0.9 12.9 0.3 23.6 26.2 36.1 0.5 14.1 0.1 20.1 27.8 37.4 0.5 13.5 0.1 20.8 34.3 30.8 0.5 13.1 0.1 20.8 37.1 28.4 0.4 8.4 0.1 18.9 40.6 31.6 1.3 11.7 0.5 28.2 34.5 23.8 0.5 19.9 0.1 26.8 21.9 30.8 0.8 14.0 0.4 24.0 30.6 30.2 0.7 15.0 0.2 22.9 29.7 31.5 0.5 12.6 0.1 20.0 17.9 48.9 1.0 17.3 0.3 32.2 23.5 8.2 1.0 22.5 0.1 36.0 10.8 5.0 a 300 "C.NO addition stopped after 61 min. 400 "C. NO addition started 236 min. Nitric Oxide Recently a number of studies3,, 3 4 9 3 5 on the methanol conversion reaction have proposed that a radical pathway is the dominant mechanism for initial carbon<arbon bond formation.We have recently that the specific radical pathways proposed by Clarke et aZ.,32 i.e. involving rearrangements and reaction of the methoxymethyl radical, are not suitable pathways to ethene formation. However, to investigate the possible involvement of a general radical pathway, reaction of methanol and dimethyl ether over H-ZSM-5 and W03/Al,03 was studied in the presence of nitric oxide (table 5 and fig. 4). Nitric oxide, a monoradical, is well known as a radical scavenger3' at temperatures of up to 600 0C,38939 and for H-ZSM-5 it is observed that addition of 1 mol % NO had no effect either on catalyst activity or on product selectivity. Addition of 3 mol% NO only slightly enhanced catalyst deactivation, and no marked changes in product selectivity were observed.For the W0,/Al,03 catalyst, reaction in the presence of 5% NO did affect the overall catalyst activity and enhanced the methane selectivity significantly (fig. 5). The results of this investigation provide evidence that, for H-ZSM-5, a radical pathway is not involved in initial carbon-carbon bond formation. In addition, the production of methane relative to C,, products was also not affected and hence the study further confirms that methane is not generated via a radical methanism distinctly separate from the main carbon-carbon bond formation reaction. It must therefore be concluded that the radicals observed by Clarke et aZ32 in the reaction of dimethyl ether over H-ZSM-5 play no major role in the mechanism of methanol conversion.For1320 Hydrocarbon Conversion using W03/A1203 and H-ZSM-5 100 80 n E C .z 60 8 5 40 n 4 z 20 I 0 0 50 100 150 200 250 time on line/min Fig. 4. Reaction of MeOMe (A) and MeOMe/NO (0) over WO,/Al,O, at 400 "C and GHSV = 360 h-l: Effect on conversion. (Dashed line indicates time of no addition.) 80 n E 20 I I I - 0 I 1 I 1 1 I I I I 0 20 40 60 80 100 120 140 160 time on line/min Fig. 5. Reaction of MeOMe (A) and MeOMe/NO (0) over WO,/Al,O, at 400 "C and GHSV = 360 h-I: Effect on CH, selectivity. (Dashed line indicates time of no addition.) W03/A1203, however, the results indicate that a radical mechanism may play a significant role in carbon-carbon bond formation with this catalyst since addition of NO significantly reduces the formation of C,, hydrocarbons relative to methane.Oxygen The reaction of dimethyl ether and methanol in the presence of added gas-phase oxygen was investigated for both H-ZSM-5 and W0,/A1203. For W03/A1203 (fig. 6) addition of oxygen at concentrations of up to 20 mol% had no marked effect on activity or selectivity. Conversely, as published previously4o we have shown that for H-ZSM-5G . J . Hutchings et al. 1321 20 I 1 I I I 50 100 150 20 0 2 50 time on line/min Fig. 6. Reaction of MeOMe (A) and MeOMe/O, ( x ) over WO,/Al,O,. oxygen addition (1 and 3 mol %) causes immediate and irreversible deactivation of the zeolite catalyst, the effect being more pronounced at the higher oxygen concentratior,. These experimental results using oxygen addition are considered to be highly significant mechanistically and clearly demonstrate that the mechanistic pathways for C-C and CH, formation on the two catalysts are considerably different. Although oxygen is a diradical, such a species would not be expected to be a more effective radical scavenger than NO under these conditions.Triplet oxygen as well as nitric oxide, both stable free n-radicals, do readily react with organic radicals. The addition reaction between a radical and oxygen is usually followed by hydrogen abstraction to give a hydr~peroxide.~' Alternatively the peroxy radical intermediate might react with another radical to form a dialkyl Under the high temperatures used in the MTG process, however, peroxides and hydroperoxides will readily undergo homolytic cleavage,42 and thus the number of reactive open-shell species present in the system will not be reduced by 0,.NO, on the other hand, forms nitrosoalkanes on reaction with alkyl radical^^,.^^ which can tautomerise to an xim me.^^^^^ If a second a-hydrogen is present, the oxime might, under the acidic conditions within the zeolite, further undergo elimination of H20, forming a n i t ~ i l e . ~ ~ ' ~ ~ Therefore, NO is expected effectively to reduce the number of free radicals even at these high temperatures. Furthermore, the deactivation observed with oxygen addition was irreversible, and if oxygen were active as a radical scavenger such deactivation would have been expected to be (at least in part) reversible, and catalyst activity would be restored on elimination of the oxygen; this was not observed. We consider that the most likely explanation for this effect is that the crucial reaction intermediate in the C-C bond formation reaction with H-ZSM-5 is oxidised by molecular oxygen far more preferentially than by NO (only partial deactivation was observed with NO).However, for the WO,/Al,O, catalyst the crucial intermediate in the C-C bond formation is apparently not susceptible to oxygen attack. These results therefore present clear experimental evidence against the mechanistic proposals of Olah,127139 l7 in which dimethyloxonium methylide has been cited as the crucial intermediate with the catalyst WO,/Al,O,, since the ylide intermediate would be expected to be highly reactive towards molecular oxygen.471322 Hydrocarbon Conversion using WO,/Al,O, and H-ZSM-5 Table 6.Reaction of Me,O in presence of CH,O and HCO,H over H-ZSM-5 at 300 "C WHSV/h-l reaction Me,O product selectivity (% by mass) reactant Me,O CH,O CHO,H /min-' (YO) CH, C,H, C,H, C, C, C,, time conversion Me,O/CH,O/ 14.8 0.015 - 6 H20a 60 120 6 60 120 Me,O/HCO,Hc 14.8 - 1.55 6 60 120 Me,O/HCO,Hc 50 - 1.94 6 30 60 Me,0/H20b 14.8 - - 97.7 1.4 10.6 0.9 25.3 31.7 30.1 80.4 1.2 15.2 0.3 23.2 26.6 33.5 8.7 0.3 7.5 0.1 12.3 28.5 51.5 88.8 1.2 11.8 0.7 24.5 35.2 26.6 85.8 1.1 14.5 0.3 23.1 28.5 32.5 19.8 0.6 18.4 0.1 25.1 21.7 34.1 96.9 1.4 10.0 0.9 29.7 35.1 26.9 49.3 0.9 13.2 0.2 23.6 23.9 38.2 22.1 0.5 10.7 0.1 16.3 34.2 38.2 99.2 1.6 14.2 0.8 28.2 36.7 18.7 39.8 0.9 12.6 0.2 24.5 25.5 36.3 15.3 0.6 14.1 0.1 19.7 26.9 38.6 a Me,O bubbled through formalin.vapour pressure. Me,O bubbled through water to give equivalent water Me,O bubbled through formic acid, for comparable data see table 5. 100 80 n E 2 60 .* & > 0 40 01 I I I 1 1 0 50 100 150 200 250 time on line/min Fig. 7. Reaction of MeOMe (A) and MeOMe/HCHO (0) over WO,/AI,O,, 400°C and GHSV = 360 h-'. Reaction of Dimethyl Ether with Added Formaldehyde and Formic Acid Dimethyl ether was reacted over H-ZSM-5 in the presence of added formaldehyde and formic acid, and the results are shown in table 6. Formaldehyde addition was found to enhance catalyst deactivation even at low feed concentrations (ca. 0.1 YO by mass of dimethyl ether fed). Conversely, addition of formic acid did not demonstrate such severe deactivation even at feed levels of up to 10% by mass HCOOH relative to dimethyl ether.Dimethyl ether was reacted over WO,/Al,O in the presence of added formaldehyde and rapid catalyst deactivation was similarly observed (fig. 7).G. J . Hutchings et al. 1323 Discussion Comparison of Catalytic Activity of H-ZSM-5 and 10 % W03/A1,03 The principal difference in the reaction of methanol over H-ZSM-5 and W03/A1,03 is the relative yields of methane and hydrogen. For both catalysts these products are formed in roughly equal molar quantities, but for H-ZSM-5 these products constitute 1 % of the total product whereas for W03/A1203 methane and hydrogen are the dominant gaseous products. Experiments with the addition of hydrogen did not show any marked change in methane selectivity, demonstrating that hydrogen is not a precursor of methane formation with either catalyst.Since methane and hydrogen both increase by roughly the same factor with increasing temperature with W03/A1203 it is possible that they are formed via a common reaction mechanism. An additional major difference observed between W03/A1,03 and H-ZSM-5 is with respect to the conversion of the reactants. In all cases the W03/Al,03 is considerably less active than H-ZSM-5. This may be due to the high density of active sites within the zeolite catalyst4' compared to the oxide, or due to sequential autocatalytic reactions26, 27 that occur within H-ZSM-5, giving products of higher carbon number. Mechanism of Primary Product Formation H-ZSM-5 A large number of proposals have been made concerning the mechanism of formation of ethene,49 whereas only relatively 21 have considered the mechanism of formation of methane, which is a major primary product at low c o n ~ e r s i o n .~ ~ . ~ ' ~ ~ ~ Of the mechanistic proposals previously cited for ethene formation, only a few have received experimental support : ( a ) formation of carbene from methanol by a e l i m i n a t i ~ n , ~ . ~ ~ ~ ~ (h) intermolecular reaction of dimethyloxonium methy1ide,l2-l4* 17g51 (c) involvement of a radical pathway32,34,35 and ( d ) deprotonation of a surface bonded methyloxonium ion to give a surface-bonded oxonium methylide.1s*20~ 21 For H-ZSM-5 the experimental data obtained using model reagents and catalyst systems in this and our previous ~tudies'*-~~~ 3 6 9 4 9 9 52 presents clear and strong evidence against the involvement of pathways (a), (6) and ( c ) noted above.In particular, our experiments with model catalysts5, and model reagent^'^?^^ showed that pathway (b) was not viable, and the arguments need not be reported in this discussion. We have previously shown2' that proposal (a) cannot account satisfactorily for methane formation. Experiments with co-feeding hydrogen further demonstrate that free gas- phase methylene carbene is not a central intermediate, since no increase in methane selectivity was involved. This conclusion is supported by the experimental data obtained for the co-feeding of oxygen. Reaction of a gas-phase methylene intermediate with molecular oxygen is known to produce formic acid via a Criegee inte~mediate.~~ Formic acid does not deactivate H-ZSM-5, and hence if gas-phase methylene carbene were to be the central intermediate in carbon-carbon bond formation, reaction with oxygen would not lead to permanent deactivation of the catalyst, and this is not experimentally observed.We have previously shown that radical pathways involving the methoxy methyl radical are not viable for C-C bond f ~ r m a t i o n . ~ ~ In addition, the present study using experimentation with the known radical scavenger NO demonstrates that for H-ZSM- 5 neither C-C bond formation nor CH, formation involve a radical pathway. Additionally Chang54 has recently proposed that a surface CH, radical could be generated via interaction of a radical initiator with a surface methyloxonium ion. However, the results obtained with NO would also mitigate against such a mechanistic proposal.1324 Hydrocarbon Conversion using WOJ A1,0, and H-ZSM-5 CH4 (b) C H j + H' - / (c) CH3' + H- Fig.8. Possible mechanisms for the formation of methane in the methanol conversion reaction. Based on our previous studies with methylating agents", 21 we proposed a methylation mechanism for H-ZSM-5 conversion of methanol in which a surface-bonded methyloxonium ion is the first key intermediate in the process. This species is formed oia nucleophilic attack of lattice zeolite oxygen at the carbon centre of protonated methanol which we have modelled by the reactions of LiAl(O-Pri),.52 Furthermore, methylation of H-ZSM-5 has been observed e~perimentally.~'~~~ Recently, Forester and Howe", 57 have provided evidence using in situ Fourier-transform infrared studies that formation of a surface bonded methyloxonium ion precedes carbon-carbon bond formation with H-ZSM-5.This methyloxonium ion is not responsible itself for initial carbon-carbon bond formation, but is the intermediate responsible for CH, formation and further methylation of the initial products.26v58 Formation of methane is considered to be highly significant mechanistically. Methane can only be formed via three mechanistic pathways during methanol conversion (fig. 8). In this study we have shown that methane formation from gas-phase carbene and methyl radical intermediates is not consistent with the experimental evidence. In addition, using a model study with the known hydride donor cycloheptatriene we have successfully modelled methane formation via H- donation, hence supporting the proposal that the surface-bonded methyl oxonium ion is the central intermediate in methane formation.We have p r o p o ~ e d l ' ~ ~ ~ that the key intermediate in the formation of the initial C-C bond is a surface bonded oxonium methylide formed via deprotonation of the surface methyloxonium ion. Recent experiments on methanol oxidation with molybdate catalystP have also postulated a deprotonation of a surface-bonded methoxyl to form a surface-bonded oxonium methylide as a key intermediate in formaldehyde formation. Evidence for this deprotonation step was obtained using kinetic-isotope data.5Y The experimental data obtained on the reaction of MeOH/O, over H-ZSM-5 provide evidence in favour of the existence of surface-bonded oxonium methylide as the key intermediate, and are therefore consistent with the oxidation of this intermediate to form formaldehyde, which we have subsequently shown to deactivate the catalyst at very low concentrations.From a mechanistic point of view it has always seemed attractive to us that the surface-bonded methyloxonium ion, in harmony with its adjacent conjugate base, should interact with methanol in a synergistically intimate fashion to effect C-C bond formation. Hence we envisage that the conceptually demanding prospect of a naked, surface-bonded methyloxonium-ion ylide acting as a nucleophile may be facilitated by considering that proton abstraction in the transition state by an adjacent alu- minium-oxygen bond renders the methanol carbon more accessible to nucleophilic attack by virtue of hydrogen bonding between the hydroxyl oxygen and the hydrogen being transferred (fig.9). Thus proton abstraction could well be enhancing activation of the poorly electrophilic methanol carbon in the tightly bound pocket of the zeolite channel. We consider that a radical hydrogen abstraction to generate an oxonium ion radical intermediate, although exciting as a possibility, is discounted in its extreme form by the NO co-feeding results. However, we recognise that serious thought as to the full implications of the two mechanistic extremes poses fascinating, if not daunting challenges on both the theoretical and synthetic fronts.In conclusion, the experimental data presented in this paper support the methylationG . J. Hutchings U et al. /*\ 1325 H-ZSM-5 Fig. 9. Mechanistic proposal for C-C bond formation over H-ZSM-5. mechanism previously proposed by us1** 21 and provide clear evidence against alternative mechanistic proposals. Future studies are now required to model successfully the deprotonation step in this mechanism. The experimental results from reaction of Me,O/O, and Me,O/NO over WO,/Al,O, differ significantly from those obtained for H-ZSM-5, and it is proposed that a different mechanism exists for this catalyst system. Of particular note are the results using Me,O/O,, which suggest that the mechanistic pathway is insensitive to the addition of up to ca.20 mol % molecular oxygen. This presents clear evidence against the involvement of a non-surface-bonded dimethyloxonium methylide as proposed by Olahl29 13, l7 specifically for this catalyst, since such an ylide intermediate would react rapidly with oxygen47 to inhibit the C-C bond formation reaction. The result also mitigates against a surface-bonded oxonium methylide intermediate as proposed for H-ZSM-5. Again the origin of methane in this reaction is considered to be mechanistically significant. The evidence presented in this study indicates that methylene carbene [fig. S(a)] and hydride transfer [fig. 8 (c)] are not viable possibilities, since reaction with co-fed hydrogen (fig. 2) or co-fed cycloheptatriene (table 4) have no significant effect. In the light of the evidence presented the following mechanism is proposed for methanol conversion over WO,/Al,O, (fig.10). Methanol dissociatively adsorbs onto the tungsten surface to form surface methoxyl and hydroxyl groups. In this reaction methanol acts as a nucleophile towards the electrophilic WO, centre, so that the oxygen of the methoxyl group is retained from methanol. This is considerably different from the electrophilic behaviour of protonated methanol towards the conjugate base of H-ZSM- 5. Methane may be generated from the hydroxyl and methoxyl surface groups by a dissociative H/CH, recombination pathway. This possibility is not open to the ZSM-5 system, and is considered to account for the drastic difference in methane levels between the two catalysts. The reaction, in giving rise to surface-bonded radicals, consequently1326 Hydrocarbon Conversion using WO,/Al,O, and H-ZSM-5 H o/cH3 \O - H 2 (adjacent OH) H migration CH3OH * PNO 1 adjacent methyl "2'4 \ + CH2O deactivation p", \ /CH2-cH3 W - P/ O-3- /I Fig.10. Proposed mechanism for methanol, dimethyl ether conversion of form C,H, and CH, over WO,/Al,O,. allows mechanistic appraisal of two key processes, namely C-C bond formation and surface deactivation. The surface oxygen radicals may abstract hydrogen from methanol in a radical redox process to generate formaldehyde which results in irreversible deactivation as is experimentally observed. Alternatively they may abstract hydrogen from an adjacent neutral surface methoxyl to furnish a methyleneoxy radical which by radical attack on either another adjacent methoxyl, or less likely methanol itself, leads to C-C bond formation.Co-feeding NO, while inhibiting radical C-C bond formation, would not, according to the present rationale, be expected to affect CH, formation, since the latter is formed in a non-radical-dependent pathway. Indeed this is what is experimentally observed, with C,, product selectivity decreasing dramatically and CH, selectivity increasing under these conditions, Assuming that co-feedants interact with the surface as radicals, as discussed previously, the difference in reactivity between them for this catalyst may be rationalised in terms of considering the products of radical scavenging. Whilst 0, results in formation of a new radical which may sustain chain growth, NO results in chain termination.Such a drastic change in mechanistic behaviour for the two catalysts would be due primarily to the electronic nature of the oxygen bonded to the methyl group in each case. Hydrogen, a significant by-product in this reaction, may also be produced by a radical pathway, favoured at higher temperatures. The role of the exposed A1,0, surface known to be present in this catalyst system is unclear at this stage. However, its influence is obviously essential for C-C bond formation, which is virtually non-existent with either WO, or AI,O, alone.G. J. Hutchings et al. 1327 Deactivation of ZSM-5 during Methanol Conversion The mechanism of deactivation of zeolite ZSM-5 during methanol conversion has not received much attention.It is known that by virtue of its shape selectivity deactivation due to coke, a highly carbonaceous deposit, is restricted with ZSM-5 compared to other Solid state magic-angle spinning n.m.r. analysis of the coke indicates that it contains aromatic, aliphatic and oxygenated ether and ketonic carbon environments.61~ 62 In the present study reaction of MeOMe/HCHO mixtures over H-ZSM-5 has been shown to deactivate the catalyst rapidly, and additionally when MeOH/O, or MeOMe/ 0, are reacted rapid deactivation is proposed to result from formation of HCHO at the active site. It is therefore probable that some deactivation of H-ZSM-5 occurs during methanol conversion owing to HCHO formation and polymerisation reactions. The resulting formation of highly oxygenated coke residues would be consistent with the previously cited magic-angle spinning n.m.r.evidence. In our previous proposalsz1 we have concluded that HCHO would be formed via hydride transfer from methanol to the surface methyloxonium ion during CH, formation. Catalyst deactivation would therefore be proportional to CH, formation over H- ZSM-5; however, this situation would be complicated by the laydown of aromatic-type coke owing to the rapid secondary reactions occurring on this catalyst. The results of this study indicate that low levels of 0, could be significantly deleterious in the commercial operation of ZSM-5 for methanol conversion. Since zeolites are reactivated on a regular basis with 0,-N, mixtures it is particularly important that the effect of 0, in catalyst lifetime be more fully examined. We thank the University of the Witwatersrand and the FRD, CSIR Pretoria for Financial Support.References I G. J. Hutchings, New Scientist, 1986, (3rd July) 35. 2 A. V. Grosse and J. C. Snyder, US Patent 2492984 (1950). 3 L. Kim, M. M. Wald and S. Brandenberger, J. Org. Chem., 1978, 43, 3432. 4 D. C. Hargis and L. J. Kehoe, US Patent 4072732 (1978). 5 D. E. Pearson, J. Chem. SOC., Chem. Commun., 1974, 397. 6 C. A. Fyfe, G. C. Gobbi, J. Klinowski, J. M. Thomas and S . Ramdas, Nature (London), 1982, 296, 7 S. L. Meisel, J. P. McCullough, C. H. Lechthaler and P. B. Weisz, Chemtech., 1973, 3, 498. 8 C. D. Chang, and A. J. Silvestri, J. Catal., 1977, 47, 259. 9 C. D. Chang, Catal. Rev. Sci. Eng., 1985, 26, 323. 10 C.T-W. Chu and C. D. Chang, J. Catal., 1984,86, 297. 11 M. M. Wu and W. W. Kaeding, J. Catal., 1984, 88, 478. 12 G. A. Olah, US Patent, 4373109 (1983); G. A. Olah, Pure Appl. Chem., 1981, 53, 201. 13 G. A. Olah, H. Doggweiler, J. D. Felberg, S. Frohlich, M. J. Grdina, K. Karpeles, T. Keumi, S. Inaba, 14 W. M. Ip, Ph.D. Thesis (University of Southern California, Los Angeles, 1982). 15 R. L. Banks, in Catalysis, (Specialist Periodical Report, Royal Society of Chemistry, London, I98 I), 16 C. J. Machiels, U. Chowdhry, W. T. A. Harrison and A. W. Sleight, in Solid State Chemistry in 17 G . A. Olah, G. K. S. Prakash, R. W. Ellis and J. A. Olah, J . Chem. SOC., Chem. Commun., 1986, 9. 18 R. Hunter and G. J. Hutchings, J. Chem. SOC., Chem. Commun., 1985, 1643. 19 G. J.Hutchings, M. V. M. Hall and R. Hunter, J. Catal., 1986, 101, 224. 20 G. J. Hutchings, F. M. Gottschalk, M. V. M. Hall and R. Hunter, J. Chem. SOC., Faraday Trans. I , 1987, 83, 571. 21 R. Hunter and G. J. Hutchings, J. Chem. SOC., Chem. Commun., 1987, 377. 22 A. M. Maitra, N. W. Carst and D. L. Trimm, Appl. Catal., 1986, 27, 9. 23 W. 0. Haag, R. M. Lago and P. G. Rodewald, J. Mol. Catal., 1982, 17, 161. 24 G. Salem, Ph.D. Thesis (University of Southern California, Los Angeles, 1980). 530. W. M. Ip, K. Lammertsma, G. Salem and D. C. Tabor, J. Am. Chem. Soc., 1984, 106, 2143. vol. 4, p. 100. Catalysis, ACS Sym. Ser. 279, 1985, p. 103.1328 Hydrocarbon Conversion using WO,/Al,O, and H-ZSM-5 25 J. P. van den Berg, J. P. Wolthuizen and J. H. C. van Hoof, in Proc.5th Int. Conf. Zeolites, ed. 26 J. R. Anderson, T. Mole and V. Christov, J. Catal., 1980, 61, 477. 27 Y. Ono and T. Mori, J. Chem. SOC., Faraday Trans. 1, 1981, 77, 2209. 28 S. Soled, L. Morrell, I. E. Wachs, G. B. McVicker, L. G. Sherman, S. Chan, N. C. Dispenziere and 29 R. Thomas, F. P. J. M. Kerkhof, A. J. Moolijn, J. Medema and V. H. J. de Beer, J. Catal., 1980, 61, 30 P. Tittarelli, A. Iannibello and P. L. Villa, J. Solid State Chem., 1981, 37, 95. 31 P. B. Venuto and P. S . Landis, Adv. Catal., 1968, 18, 259. 32 J. K. A. Clarke, R. Darcy, B. F. Hegarty, E. O’Donoghue, V. Amir-Ebrahimi and J. J. Rooney, 33 W. Kirmse, in Carbene Chemistry (Academic Press, New York, 1971). 34 H. Chovkrovn, D. Brunel and A. Germain, J . Chem. SOC., Chem. Commun., 1986, 6. 35 P.Rimmelin, A. Brenner, K. Fischer and J. Somme, J . Chem. SOC., Chem. Commun., 1986, 1497. 36 R. Hunter, G. J. Hutchings and W. Pickl, J . Chem. Soc., Chem. Commun., 1987, 843. 37 R. A. Kenley and D. G . Henry, J . Am. Chem. SOC., 1982, 104, 220. 38 G. B. M. Eastmond and G. L. Pratt, J. Chem. SOC. A , 1970, 2329. 39 J. Esser and K. J. Laidler, Int. J. Chem. Kinet., 1970, 2, 37. 40 R. Hunter, G. J. Hutchings and W. Pickl, J . Chem. Sue., Chem Commun., 1987, 1369. 41 J. M. Hay, Reactive Free Radicals (Academic Press, London, York, 1974), p. 102. 42 0. L. Magelli, in Organic Peroxides, ed. D. Swarn (Wiley Interscience, New York, 1970), part 1, 43 G. L. Pratt and J. H. Purnell, Trans. Faraday Sot., 1964, 60, 371. 44 P. Kabasakalian and E. R. Townley, J. Am. Chem. SOC., 1962, 84, 271 1 ; M. Akhtar, D. H. R. Barton 45 D. H. R. Barton and J. M. Beaton, J . Am. Chem. Soc., 1961, 83, 4083. 46 K. Friedrich, in The Chemistry of Functional Groups, Supplement C, ed. S . Patai and Z. Rappoport 47 A. W. Johnson, Ylide Chemistry (Academic Press, New York, 1966), pp. 94-97. 48 J. Dwyer, Chem. Ind., 1984, 258. 49 C. D. Chang and C. T-W. Chu, J. Catal., 1982, 74, 203. 50 R. Hunter and G. J. Hutchings, J. Chem. SOC., Chem. Commun., 1986, 1006. 51 C. S. Lee and M. M. Wu, J . Chem. SOC., Chem. Commun., 1985, 250. 52 R. Hunter and G. J. Hutchings, J . Chem. SOC., Chem..Commun., 1985, 886. 53 R. I. Martinez, R. E. Huie and J. T. Herron, J . Chem. Phys., 1981, 75, 5975. 54 C. D. Chang, in Proc. Methane Conversion Symposium Auckland, New Zealand, April 1987, ed. D. M. 55 P. Salvador and W. Klading, J. Chem. SOC., Faraday Trans. 1, 1977, 73, 1153. 56 T. R. Forester, S. T. Wong and R. F. Howe, J . Chem. SOC., Chem. Commun., 1986, 1611. 57 T. F. Forester and R. F. Howe, J. Am. Chem. SOC., 1987, 109, 5076. 58 T. Mole, in ‘Proc. Methane Conversion Symposium’, Auckland, New Zealand, April 1987, ed. D. M. 59 C. J. Machiels and A. W. Sleight, J. Catal., 1982, 76, 238. 60 P. Dejaifre, A. Auroux, P. C. Grabelle, J. G. Vedrine, Z. Gatelica and E. G. Derouane, J . Card., 1985, 61 E. G. Derouane, J. P. Gilson and J. B. Nagy, Zeolites, 1982, 2, 42. 62 L. C. Carlton, R. Copperthwaite, G. J. Hutchings and E. C. Reynhardt, J. Chem. SOC., Chem. L. V. C. Rees (Heydon, London, 1980), p. 649. R. T. K. Baker, in Solid State Chemistry in Catalysis, ACS Symp. Ser. 279, 1985, p. 165. 559. J. Chem. SOC., Chem. Commun., 1986, 425. p. 1 . and P. G. Sammes, J. Am. Chem. SOC., 1965, 87, 4601. (Wiley, New York, 1983), part 2, pp. 1345-1390. Bibby, C. D. Chang, R. F. Howe and S . Yurchak (Elsevier, Amsterdam, in press). Bibby, C. D. Chang, R. F. Howe and S. Yurchak, (Elsevier, Amsterdam, in press). 70, 123. Commun., 1986, 1008. Paper 71023; Received 5th January, 1987