Mechanism of the Steam Reforming of Methane over aCoprecipitated Nickel-Alumina CatalystBY J. R. H. Ross AND M. C. F. STEELSchool of Chemistry, University of Bradford, Bradford BD7 I DPReceived 30 June, 1972The kinetics of the steam reforming of methane over a coprecipitated Ni/A1203 catalyst have beenexamined in the temperature range 773 to 953 K and in the pressure range 0-10 Torr. Examinationof the stoichiometry of the reaction showed that a catalyst freshly reduced in hydrogen at 873 K wasfurther reduced by the reaction mixture. This is taken to imply that reduction of some phase such asNiA1204 was occurring. The rate determining step of the reaction on the fully reduced catalystunder reducing conditions was found to be the rate of adsorption of methane and competition for theadsorption sites by water occurred.The water-gas shift reaction did not proceed appreciably, andthis implies that significant CO and COz adsorption does not occur on the catalyst surface ; this is inagreement with the kinetic results. Experiments involving D,O or D, helped to confirm theseconclusions.The catalysed reaction of methane with steam to produce hydrogen is of consider-able economic importance, as is the steam reforming of higher hydrocarbons. Themethane steam-reforming process is represented by eqn (1) :CH,+H20--+3H,+C0. (1)The CO formed may react with further water by the water-gas shift reaction :CO + H20-+C0, + H2. (2)In the steam reforming of higher hydrocarbons, it is thought that the hydrocarboninitially breaks down to form methane, and that reactions (1) and (2) subsequentlyreach equilibrium.1-6 In industrial use, the catalyst used for these reactions isalmost invariably nickel based; it may be of either the impregnated or the copreci-pitated type, and complex oxide supports are often used in order to reduce the lay-down of carbon and the formation of polymeric material on the surface.There have been several kinetic studies of the steam reforming of methane, eachof which indicates that the rate determining step for the reaction involves methaneadsorption.Akers and Camp,' using an integral flow reactor at atmospheric pres-sure, found that the kinetics of the reaction over a Ni/Kieselguhr catalyst at 911 Kwere first order in methane partial pressure, and independent of water and productpartial pressures.Bodrov et al., using a circulating flow reactor, also at atmosphericpressure, again found first order behaviour with respect to methane partial pressurewith nickel foil * and with two different nickel catalysts.g* lo In all three systems, adependence of the rate on PHI0, PH2 and Pco was observed, although this was greateston the Ni foil; inhibition by hydrogen was important at lower temperatures butdisappeared above 973 K.1°These limited results clearly indicate that the catalyst structure has a marked effecton the kinetics of the reaction. Similarly, it has been shown lq6 that the steamreforming of higher hydrocarbons is dependent on the structure of the catalyst ; e.g.the reaction of butane over nickel on alumina catalysts has very different kinetics1J.R . H. ROSS AND M. C. F. STEEL 11to that over a nickel on urania c a t a l y ~ t . ~ The present study was therefore undertakento examine the effect of a systematic variation of the catalyst composition on thekinetics of the steam reforming of methane. Attention has also been focussed onchanges of the catalyst composition and of the stoichiometry of the reaction withcatalyst use. The present paper reports the results obtained with a coprecipitatedNi/alumina catalyst similar to those used commercially l1 ; the changes in kineticsand stoichiometry of the reaction under a variety of conditions are related to changesoccurring in the catalyst. Subsequent papers will examine changes in the kinetics ofthe reaction when the catalyst formulation is systematically varied.EXPERIMENTALAPPARATUSThe apparatus used was constructed largely of Pyrex ; metal valves were used to isolatethe constant volume reaction system from the pumps and gas-handling line, and the reactionvessels were made of quartz. Analysis was carried out using a low resolution mass spectro-meter, attached to the reaction system by a capillary leak and continuously pumped byseparate pumps.Both the reaction system and the mass spectrometer were bakeable to550 K, and background pressures of < lo-* Torr (1 Ton = 133 N m-,) could be attainedafter bakeout and suitable trapping. Oil diffusion pumps and oil manometers were used inthe system in order to avoid the need for trapping between the reaction system and the gashandling line ; the vapour pressure of the oil at 298 K was - 5 x lo-' Torr.The catalyst was pIaced in one of two matched quartz reaction vessels.Both vesselswere situated in a furnace whose temperature was regulated by a thermocouple and a temper-ature controller. A magnetically operated sliding valve arrangement was used to admit thereaction mixture to either vessel. A magnetically operated centrifugal pump was alsoincorporated to give more efficient circulation of the reaction mixture at higher pressures(> 5 Torr). The total volume of the reaction system, including the reaction vessel, was about4.3 x m3.The whole system was maintained in an oven at 400 K to minimise adsorption of watervapour on the walls and hence to increase the accuracy of the measurement of water partialpressure. Even under these conditions, the response of the mass spectrometer to changes inwater partial pressure was not rapid, and for this reason, the rate of change of water partialpressure was never completely representative of its behaviour in the reaction.Although itwould have been desirable to correct for this, no correction was applied because the effect iscomplex (depending on various factors such as total pressure, water partial pressure, emissioncurrent, etc.). However, it is small and unimportant in the present circumstances, as quotedrates of reaction are based on methane partial pressures.MATERIALSA coprecipitated Ni/alumina catalyst, supplied by Laporte Industries Ltd., containing75 % Ni in its reduced form, was used in this work.This was prepared by a proceduresimilar to that outlined in ref. (1 l), and involved treatment of an aqueous solution of alumin-ium and nickel nitrates with Na2C03, followed by calcining in air at 723 K. The reducedcatalyst contained 0.22 % Na and 0.73 % K. The pelletted catalyst was powdered beforeuse and particles in the range 250-355 pm were used for the experiments reported here.X-ray powder diffraction photographs showed that the catalyst in its unreduced formconsisted mainly of NiO and poorly crystallised y-alumina. The total area of the reducedcatalyst was 175 mZ g-l (N, adsorption at 78 K, B.E.T. method) and the active metal areawas 37 m2 g-' (H2 adsorption at 298 K).Two 0.2 g samples of the catalyst were used andboth gave similar results; one was used mainly for the investigation of the behaviour of afreshly reduced catalyst, while the other was used for the kinetic results on a thoroughlyreduced catalyst, and had been used for preliminary work.CH4, Hl, CO and C 0 2 were supplied in sealed ampoules by the British Oxygen Co12 STEAM REFORMING OF METHANE(Grade X). Freshly deionised water was used after several freezing, pumping and thawingcycles in the gas handling line. D20 was supplied by Prochem Ltd. (99.7 %) and Dz wassupplied by Matheson Inc. (c.P. grade).P I t 0 CE D U R EThe apparatus was calibrated for both reactants and products ; calibrations of the massspectronieter were repeated at frequent intervals.Before each experiment or series of experiments, the catalyst sample was reduced in Hz(-25 Torr) at a temperature of 873 K for several hours.The criterion taken for completereduction was that no water should be produced when the sample was exposed to a freshdose of hydrogen (- 1020 molecules of H2). The reaction system was then pumped, thecatalyst vessel was isolated from the remainder of the system, and a known mixture of reac-tants was admitted to the reaction system and the blank reaction vessel. Complete mixingof the reaction mixture was confirmed and the mass spectrometer calibration was checkedonce a steady state had been achieved. The reaction mixture was then admitted to thereaction vessel containing the catalyst at the desired reaction temperature and the massspectrum was scanned periodically.From these results, graphs showing the partial pressuresof reactants and products as a function of time were constructed.Blank experiments showed that no reaction between CH4 and HzO occurred on the wallsof the quartz reaction vessel or on a sample of y-alumina in the temperature range of theexperiments carried out (773 to 953 K).RESULTSSTOICHIOMETRY OF THE CH4+H20 REACTION OVER A REDUCEDCATALYSTPreliminary work showed that although measurable rates of reaction occurredbelow 773 K, these were not reproducible and depended markedly on the catalystpretreatment. However, above 773 K, reproducible rates were obtained. Under theconditions used, i.e.T = 773 to 953 K and P = 0-10 Torr, the equilibrium in eqn (1)is well to the right, whereas CO and C02 may be expected in roughly equal propor-tions (eqn (2)).Fig. l(a) shows a typical experiment for the reaction of CH4 with H20 at 873 Kon a freshly reduced catalyst ; the reaction mixture consisted of -4.7 x 1019 moleculesor about three molecules of each reactant/ten surface Ni atoms. The rate of disap-pearance of methane after 30 s was - 5 x 1015 molecule s-l. Little water reactedand CO and CO, were produced in comparable proportions. The amounts of COand CO, are compatible with the loss of methane, but not with that of water, and thisindicates that oxygen must be produced from the catalyst. In subsequent experi-ments, carried out without further reduction of the catalyst in hydrogen, the consump-tion of water increased and the proportion of C 0 2 in the reaction products decreasedsteadily. Fig.l(b) shows the results obtained after an extended series of experiments.C02 is no longer a significant product of the reaction, and the stoichiometry of eqn (1)is approximately obeyed. At this stage, 1021 oxygen atoms had been produced fromthe catalyst, which corresponds to - 10 atoms/surface Ni atom. Once established,this stoichiometry of reaction was maintained reproducibly throughout subsequentexperiments, and the results reported below were obtained under these conditions.Fig. l(b) shows that the partial pressure of water does not appear to fall as rapidlyas that of methane; careful consideration of the mass balance of the reactants andproducts shows that this is due to a slow response of the mass spectrometer to changesin water partial pressure.Similar considerations do not apply to the other reactantsand products, and hence the stoichiometry of eqn (1) appliesJ. R. H . ROSS AND M. C. F . STEEL 13100 2 0 0 3 0 0 4 0 0 5 0 0timeis(a)100 2 0 0 3 0 0 420 5 0 0time/s(b)FIG. 1.-Reaction of CH4 with H20 at 873 K over (a) a freshly reduced catalyst and (b) a catalyst forwhich eqn (1) holds. Note that the water partial pressure is subject to some error due to slowresponse of the mass spectrometer.FIG. 2.-Plots (a) of log (rate) against log PCH~ at various values of PH20 and (6) of log (rate) againstlogPH20 at various values of P C H ~ (T = 873 K)14 STEAM REFORMING OF METHANEDEPENDENCE OF THE REACTION RATE ON THE PARTIAL PRESSURES OFREACTANTSIf the kinetics of the reforming reaction adhere to eqn (3),and if we assunie for the present that products do not affect the rate of reaction, thenn and m may be obtained from plots of log (rate) against log PCH4 and log (rate) againstlog PHz0 respectively, with PHLO and PCH4 held constant in turn.The rates used wereobtained after reaction for - 30 s from plots such as those shown in fig. 1(b) ; asdiscussed below, reproducible results were not achieved until that point. Fig. 2(a)and (b) show plots of the data, from which the following values of n and m were found :1.2 = 1.0, 112 = -0.5.The possible error in both values was about 5 0.1.eqn (1) holds throughout the experiment, then eqn (3) becomesIf the initial pressures of methane and water are equal, and the stoichiometry ofand this may be integrated to givePg$4 = const.- kt/2. ( 5 )Fig. 3(a) shows a plot of this relationship for the data of fig. l(b) and also for 14additional experiments carried out under identical conditions at various stagesthroughout the whole experimental programme ; the plot shows satisfactory repro-ducibility from experiment to experiment. All the plots of this type which we haveobtained show a deviation from linearity in the first 30 s of reaction. This deviationdepends largely on reaction conditions, and will be discussed below ; it is because ofthis deviation that the values of 12 and in reported above were obtained from ratesafter - 30 s.DEPENDENCE OF T H E RATE O F REACTION ON THE PARTIAL PRESSURESOF PRODUCTS(a) HYDROGEN.-Initial rate measurements show that the rate of the reaction isapparently retarded by addition of H2 to the reaction mixture.However, when thedata is plotted according to eqn (5), it is seen that the rate is affected only for the first30 s, and thereafter the value of Ic is constant (see fig. 3(b)).(b) CARBON MONOXIDE.-Fig. 3(c) shows similar plots when CO is added to thereaction mixture. Variation of Pco not only affects the initial rate of the reactionbut also causes a slight change in k ; this dependence is given byand is sufficiently small not to affect the general applicability of eqn (3) and hence ofeqn (5).(c) CARBON DIOXIDE.-Fig.3(d) shows the results obtained when C 0 2 was added,and these were similar to those for CO addition ; it was found thatiC cc p;p4 (6)(7) kKp-0.05 co15 J . R . H . ROSS A N D M . C. F. STEEL2.50 . 5 I 1 I I Itime/s(42.510 5 3 100 150 2 0 0 2!time/s(c)time/s@)FIG. 3.-Plots of P$z4 against time for the CH4+ H20 reaction at 872 K for equal initial pressures ofCH4 and H20 of 2.36 Torr showing : (a) reproducibility of the data ; (b) effect of Hz addition : V,4.72; 0, 2.83; A, 0.65; 0, 0.00Torr; (c) effect of CO addition: A, 4.72; 0, 2.36; a, 0.00Torr ; (d) effect of COz addition : V, 4.72 ; 0, 2.36 ; A, 0.65 ; 0, 0.00 Torr.TEMPERATURE DEPENDENCE OF THE CH,+H20 REACTION RATEValues of k (eqn (3)) were determined for temperatures in the range 773 to 953 Kand the results were plotted as log k against 1/T.A value of the activation energy Efor the reaction of 29.0 kJ mol-1 was obtained. Evaluating the pre-exponentialterm, the rate of reaction of methane is given by :where dnldt is expressed in molecules of methane reacted per m2 of surface per secondand the pressures are expressed in Torr (cf. eqn (3))16 STEAM REFORMING OF METHANEREACTION OF METHANE ALONE OVER THE NICKEL CATALYSTWhen a well reacted catalyst, such as that which gives the behaviour shown infig. I@), was exposed to methane alone at a temperature of 873 K, hydrogen wasproduced. The rate of the Ni/CH4 reaction (see fig. 4) was greater than that of theCH4 + H20 reaction.The rate of the carburisation reaction was directly dependenton P&&, as shown by the plot of log PCH4 against time, also shown in fig. 4. Incorpor-ation of carbon from the methane was not restricted to one dose (2.4 x l O I 9 molecules),but continued with a gradual decrease of rate for many doses, equivalent to -30monolayers of “ carbide ”.I 1FIG. 4.-Comparison of the rate of reaction ofCH4 with the reduced catalyst (0) with that ofthe CH4 + H20 reaction (a). The first orderdependence of the CH4 reaction is also shown(A) (T = 873 K).0 . 0 00 100 2 0 0 3 0 0 400timelsREACTION OF WATER WITH A “CARBIDED” CATALYSTCO and H2 were both produced when water vapour was reacted with a “ car-bided ” catalyst at 873 K but no methane was observed; nor was there any evidencefor the loss of oxygen to the catalyst.The reaction may therefore be representedby eqn (8) :“ carbide ” + H,O-+Ni + H2 + CO.Fig. 5 shows that the rate of reaction of water with the “ carbide ” was considerablygreater than that of the CH,+H,O reaction on the “ carbided ” catalyst and wasalso more rapid than the latter reaction on a reduced catalyst.(8)KINETICS OF THE CH,+H,O REACTION ON THE “ CARBIDED ” CATALYSTThe stoichiometry of the reforming reaction on the “ carbided ” catalyst wassimilar to that shown in fig. l(b).molecule m-2 s-l at 873 K, as opposed to - 5 x 1015 molecule s-l under similarconditions for the reduced catalyst (see fig. (5)). The following values of n and m(eqn (3)) were obtained :However, the rate was appreciably less (-3 xn = 1.0 m = 1.0.As before, the values of n and m were considered to be accurate to about kO.1J .R. H. ROSS AND M. C . F . STEEL 17EXCHANGE EXPERIMENTSA series of experiments was carried out with CH4+D20 and CH4+H20+D2mixtures, in order to examine any exchange that might occur in the methane con-current with the steam reforming reaction.Negligible exchange of the methane was observed during its reaction with D20,the reforming reaction being at least ten times faster than any exchange reaction.Both D2 and H2 were formed and complete equilibration giving HD was also ob-served. The unreacted D20 at any stage of the reaction was also fully in equilibriumwith product H2 to form HDO and H20.Similar results were obtained in the CH4 + H20 + D2 experiments : the exchangeof the methane was negligible, but complete equilibration of the H20 and D2 occurred.Complete exchange of H20+D2 mixtures was also found on the blank reactionvessel at 873 K, but the exchange occurred more slowly.2.5FIG.5.4omparison of the rate of the reactionof HzO with a carbided catalyst (0) with thoseof the CH4+H20 reaction on a reduced (a)and a carbided surface (V) (T= 873K). 1.5 -Note that although the water partial pressures --are subject to some error (see experimental s e ~tion), the comparison is still valid, 22.01.0 -Oo5 t0.00 100 2 0 0 3 0 0 4 0 0 5 0 0timelsDISCUSSIONCHANGES I N THE CATALYST COMPOSITION AND REACTION STOICHIOMETRYNickel oxide supported on alumina is notoriously difficult to reduce,l particularlyin the coprecipitated form l3 ; unsupported NiO, however, can be completelyreduced under carefully controlled conditions.l4 After using the criterion for com-plete reduction mentioned above, we still find the equivalent of N 10 monolayers ofoxygen produced from the catalyst during the CH4+H20 reaction, as evidenced bythe production of CO and C02 without loss of water (fig. l(a)). The standard freeenergy change for the reduction of nickel oxide by hydrogen :is -34.81 kJ mol-1 at 873 K, and hence the equilibrium position is well over to theright-hand side. In addition, the reduction of nickel oxide by CO may occur :NiO+H,-+Ni+H20 (9)NiO+CO-+Ni+CO, (10)as the value of AG& for this reaction is -42.05 kJ mol-l18 STEAM REFORMING OF METHANEHowever, we must also consider the possibility of the formation of nickel alum-inate, NiA1204, when the catalyst is calcined in air at 723 K during preparation,because its presence has been reported at temperatures as low as 673 K in similarl6 For the reduction of NiA1204 by hydrogen :NiA1204+H2-+Ni+A1203 +H20 (1 1)NiA1204+CO+Ni+A1203 +C02 (12)AG0873 is - 15.73 kJ mol-l, and for the reduction by CO :AGg73 is -22.97 kJ mol-l.These values are calculated from the data of Lenev.I7The lower values of AG& for the reduction of nickel aluminate compared to nickeloxide show that the presence of the spinel or similar oxide phases makes the reductionof the catalyst less energetically feasible, but at the same time does not precludecomplete reduction by either hydrogen or carbon monoxide.It is of interest to notethat Delgass has recently obtained evidence from photoelectron spectroscopy datafor the presence of an intermediate compound in the NiOjalumina system, althoughits composition is not specified.The values of AGS73 given for the reactions shown in eqn (9)-(12) apply to stoichio-metric oxides, and may be smaller if non-stoichiometric oxide phases are present inthe region of the surface. Hence, it is probable that the last stages of reduction willnot occur when the reducing agent is hydrogen, as in the initial reduction of thecatalyst, but that when CO is formed during the CH4+H,0 reaction, it is capable ofbringing about the complete reduction of the surface layers.The thermodynamic data also afford an explanation of the initial deviation fromlinearity of the P;g4 against time plots (fig.3). The spontaneous approach to equi-librium of the reactions given in eqn (9) and (12) is governed by :(13) PHlo' H 2AG873 = AG,",,+RT ln-.When AGg73 is negative, the reactions will proceed in the directions shown, but ifAG873 is positive, the reverse reaction will take place. NiO (eqn (9)) is reducedwhen PH20/PH2 < 117 but reduction of NiA1204 will not occur until PkIzo/PHz < 8.6.Until the hydrogen partial pressure is such that these ratios are achieved, oxidationwill occur, and so the initial stages in all the CH,+H,O reactions take place onpartially oxidised catalyst.In studies of the steam reforming of hydrocarbons, it has generally been as-sumed l-l0 that the water-gas shift reaction (eqn (2)) is at equilibrium. The decreasein the production of C02 as reduction of the catalyst proceeds indicates that this maynot be the case in the present work.Fig. 6 shows typical results for the disappearanceof methane as a function of time together with the calculated equilibrium values ofPco, Pco, and PHlo. Pco2 increases initially and then falls again as PHz builds up.Such behaviour was in no case observed in the present work (see figs l(a) and (b)).Initially, COz is formed by interaction of CO with the surface oxide, but Pcoz does notfall as the hydrogen partial pressure increases ; in later experiments, CO productionis predominant.We must therefore conclude that the shift reaction (eqn (2)) doesnot occur to an appreciable extent in the presence of methane. This implies that thesurface species resulting from the adsorption of methane are more strongly bondedto the surface than species arising from CO and C 0 2 adsorption, and this is in agree-ment with the kinetic results (see below).No evidence for loss of carbon to the catalyst was found during the CH, + H20 re-action. On the other hand, when methane was reacted with the catalyst, incorporatioJ . R . H. ROSS AND M. C. F . STEEL 19of carbon occurred, and this uptake of carbon was not limited to the surfacelayer.Nickel carbide (Ni3C) is known to be produced on heating nickel metal in thepresence of carbon containing compounds (e.g.CO, CH4),19 but it decomposesrapidly above 703 Kin vacuum to Ni and graphite.20 Galwey has shown 21 that Ni3Creacts with water at temperatures between 500 and 683 K to give COz, H2 and smalleramounts of CH4 ; the quantities of product formed decrease with increasing temper-ature and this may be due to slower reaction of the graphite formed from the thermaldecomposition of the Ni3C. Galwey's postulate that C 0 2 may oxidise the metal toNiO seems unlikely, however, in view of the thermodynamic data given above. Themain products of the reaction of the " carbide '' with water in the present work areCO and H2 and the reaction is rapid, which would be unlikely if graphite wereinvolved.The relative rates of the " carbide "/H20 reaction and the CH, + H20reaction on the reduced catalyst (fig. 5 ) explain why carbon deposition does not occurduring the reforming reaction, because, even if carbon atoms are formed, they areimmediately removed by reaction with water vapour.timeisFIG. 6.-Calculated distribution of products for a typical rate of disappearance of CH4 at 873 K,assuming thermodynamic equilibrium.REACTION MECHANISMIn agreement with previous investigations of the methane steam-reforming reac-tion, the rate of reaction was found to be first order with respect to PCH4. This impliesthat the rate determining step is the dissociative adsorption of methane. Thedependence of the rate on implies that the water competes with the methanefor the active catalytic sites.This situation is shown in the scheme (see opposite).The rate of formation of CH3 surface species determines the rate of reaction (step 1).Reversible dissociative adsorption of water (steps 2 and 2') occurs competitively onthe same sites. Once surface CH3 species are formed, they may break down furtherto form CH2, CH or C surface entities which may form oxygenated surface species byinteraction with OH groups. These break down to give CO(g); if adsorbed C20 STEAM REFORMING OF METHANEgroups existed, equilibrium amounts of C 0 2 would be formed. Readsorption of COhardly occurs.Methane alone reacts with the catalyst more rapidly than does the CH, +H,Omixture, and this is further evidence that step 1 is inhibited by water.Also, in theabsence of water, step l a proceeds to form surface carbon atoms which becomeincorporated in the lattice. When the reforming reaction is carried out on thecarbided catalyst, the rate is directly proportional to both the water and methanepressures, and this implies that H 2 0 is reacting directly with carbon from the lattice,while CH4 is replenishing this carbon. The sites involved are different to those on thereduced catalyst, and the reaction is slower.SCHEMEThe exchange experiments reported above were carried out to help confirm thereaction scheme proposed above. For example, if any step other than 1 was ratedetermining and step 1 was reversible, then the mode of the exchange within theunreacted methane molecules would show which step was important.The lack ofexchange with either D,O or D2 confirms that step 1 must be rate determining, andthat the reverse reaction does not occur. Complete exchange of the hydrogen andwater species shows that steps 2 and 2’ and 3 and 3’ are rapid. It is of interest to notethat Morikawa et aLZ2 studying the CH4+D,0 reaction over a Ni/Kieselguhrcatalyst at 500 K, found slow exchange concurrent with an equally slow reformingreaction in which H2 and C02 were produced. However, their highest temperatureof reduction in pure hydrogen was 723 K, and our results suggest that this may not be asufficiently rigorous treatment to produce a catalyst of high reforming activity.The authors thank Professor M. W. Roberts for his interest and encouragement,and the General Chemicals Division of Laporte Industries Ltd. for generous support.M. C. F. S. thanks the S.R.C. for an award under the C.A.P.S. scheme.C. H. Riesz, H. A. Dirksen and W. J. Pleticka, Inst. Gas Tech. Res. Bull., 1952, 20.M. C. F. Rogers and W. M. Crooks, J. Appl. Chem., 1966, 16,133.K. S. M. Bhatta and G. M. Dixon, Trans. Faraday SOC., 1967, 63,2217.K. S. M. Bhatta and G. M. Dixon, Ind. Eng. Chem. Prod. Res. Deu., 1969, 8, 324.T. R. Phillips, T. A. Yarwood, J. Mulhall and G. E. Turner, J . Catalysis, 1970, 17, 28.C. R. Schnell, J. Chern. SOC. B, 1970, 158. ’ W. W. Akers and D. P. Camp, Amer. Inst. Chem. Eng. J., 1955, 1,471. * I. M. Bodrov, L. 0. Apel’baum and M. I. Temkin, Kinetika i Kataliz, 1964,5, 696.I. M. Bodrov, L, 0. Apel’baum and M. I. Temkin, Kinetika i Kataliz, 1967, 8, 821.lo I. M. Bodrov, L. 0. Apel’baum and M. I. Temkin, Kinetika i Kataliz, 1968, 9, 1065.l1 Brit. Pat. 969,637 1961.l2 D. Reinen and P. W. Selwood, J. Catalysis, 1963, 2, 109.l3 V. C. F. Holm and A. Clark, J. Catalysis, 1968, 11, 305.l4 M. W. Roberts and K. W. Sykes, Trans. Faraday SOC., 1958, 54, 548.l5 A. M. Rubinshtein, V. M. Akimov and L. D. Kretalova, Izuest. Akad. Nauk S.S.S.R., Otdel.khim. Nauk, 1958, 929J . R. H . ROSS AND M . C . F. STEEL 21l 6 M. Lo Jacono, M. Schiavello and A. Cimino, J. Phyx. Chem., 1971,75,1044.l7 L. M. Lenev and I. A. Novokhatskii, Zhur. neorg. Khim, 1965, 10,2400.W. N. Delgass, T. R. Hughes and C. S. Fadley, Catalysis Reu., 1971, 4, 179.l9 L. J. E. Hofer, E. M. Cohn and W. C. Peebles, J. Phys. Colloid Chem., 1950,54, 1161.2o S. Nagakura, J. Phys. SOC. Japan, 1957,12,482.'* K. Morikawa, W. S. Benedict and H. S. Taylor, J. Amer. Chem. SOC., 1936,58, 1445.A. K. Galway, J. Catalysis, 1963, 2, 176