J . Chem. Soc., Faraday Trans. 1, 1985,81, 1357-1367 Influence of Support Acidity and Ce3+ Additives on the Reactivity of Nickel Particles Highly Dispersed on Various Oxide Supports BY GUY-NOEL SAUVION, JEAN-FRANCOIS TEMPERE,* MARIE-FRANCE GUILLEUX, GERALDJEGA-MARIADASSOU~ AND DENISE DELAFOSSE Laboratoire de Chimie des Solides, E.R. 133 ‘Reactivite de Surface et Structure’, Universite P. et M. Curie, 4 Place Jussieu, 75230 Paris Cedex 05, France Received 2nd July, 1984 Important modifications to the reactivity of metallic nickel highly dispersed on various oxide supports have been observed. These vary according to the acidity of the support and the presence of Ce3+ additives. The loss or attenuation of hydrogen chemisorption properties and catalytic activity in butane hydrogenolysis seems to be correlated to the acidity of the support, which leads to a total or partial coverage of the metallic surface during the reduction process by hydrogen, which remains strongly chemisorbed up to high temperatures. The presence of Ce3+ additives causes an increase in the catalytic activity in the CO/H, reaction and a significant shift of the selectivities in favour of the formation of heavier hydrocarbons from both the hydrogenation of CO and the hydrogenolysis of butane.These results are discussed in terms of strong metal-support interactions and concomitant modifications of the electronic environment of the metal. Numerous studies have recently been devoted to metal-support interactions, which can strongly modify the reactivity of Group VIII metals. In the case of reducible oxide supports like titania,l-* a strong metal-support interaction, occurring when the reduction of the catalyst is performed at a relatively high temperature, reduces the ability of the metal to adsorb hydrogen or CO and decreases or suppresses its catalytic activity in hydrogenolysis reactions.However, it leads to an increase in methanation activity and a shift in product distribution towards higher hydrocarbons. These modifications of the catalytic and chemisorption properties have been explained as the result of an electronic interaction between the support and the metal crystallites rather than as the effect of crystallite size or surface structure. In Nio/TiO, systems5 spectroscopic measurements have shown an electron enrichment of the metal particle when the cations of the support were reduced to lower valence (Ti3+).In acidic supports like zeolites, metal-support interactions may also occur for well dispersed metallic systems.6* Lewis-acid sites interact strongly with the metal crystallites causing them to become electron deficient.6 The increased activities of these systems in hydrogenation and hydrogenolysis reactions, when observed, cannot unambiguously be interpreted in terms of electron deficiency of the metal. Other factors can account for these modifications, such as the shape selectivity of zeolites, their electrostatic field and their acidity. More generally, the nature of the support, its reducibility but particularly its acidity 7 Present address: Laboratoire de Cinetique Chimique, 1 Rue Guy de la Brosse, 75230 Pans CCdex 05, France.13571358 REACTIVITY OF NICKEL PARTICLES or basicity and the presence of additives can induce some electron transfer from the support towards the metal or, conversely, from the metal towards the support, especially in well dispersed systems. These changes in electron density of the metal can significantly modify its chemisorptive and catalytic properties. In an earlier works we studied the reactivity of well dispersed metallic nickel (0.7-1.2 nm diameter) on Ce3+ zeolites differing mainly in their acidity, the presence of the Ce3+ cation leading to stabilization of nickel particles inside the zeolitic cavities. The preliminary experimental results have shown that nickel supported on the more acidic supports does not chemisorb hydrogen and has no activity in butane hydrogenolysis.These results have been explained by the presence on the sample surface of large quantities of hydrogen species remaining strongly chemisorbed until desorbed by heating to 923 K. However, after similar treatment the dispersed nickel is not catalytically active in butane hydrogenolysis. This behaviour of NiCeX zeolites reduced at a relatively low temperature (623 K) has to be compared with that of metals supported on reducible oxides reduced at high temperatures. In order to explain the origin of these profound changes in the catalytic and chemisorptive properties of nickel highly dispersed on non-reducible carriers, we have extended this work to the study of the various parameters capable of modifying the metallic reactivity, namely the nature of the support, its acidity and morphology and the effects of the presence of Ce3+ additive and metallic dispersion. The catalytic activity and chemisorption properties of nickel well dispersed on X and A zeolites and silica have been studied in butane hydrogenolysis, CO+H, reactions and hydrogen chemisorption.EXPERIMENTAL MATERIALS The zeolite samples were prepared by ion exchange, contacting dilute solutions of nickel and cerium(1n) nitrates with NaX zeolites (samples X,-X, ) or NaA molecular sieve (sample Al ). Silica sample S, was obtained by precipitation of nickel and cerium(1n) nitrates in ammonia medium. Sample S,, without cerium, was studied for comparison.1° Sample compositions and nomenclature are reported in table 1. Generally, zeolite samples were reduced at 623 K in a flow ( 5 dm3 h-l) of high-purity hydrogen for 16 h, while silica systems were pretreated under a helium flow up to 773 K before reduction at 923 K.METHODS The oxidation state of cerium before reduction was evaluated either by X-ray determinations or potentiometric titration of Ce3+ against Fe3+ ions. The metallic-particle-size distribution and the extent of reduction of the sample were determined by magnetic measurements using the Weiss extraction method.ll The experiments were generally carried out in the temperature range 77-300 K and magnetic fields up 18 kOe. In some cases a superconductive coil reaching 70 kOe (for measurements at 4.2 K) was used. A ferromagnetic resonance (f.m.r.) study was performed at 300 K using a Varian spectrometer (model C.S.E., 109-X band) in order to follow the modifications of the magnetic properties of the metallic particles during the various treatments to which the samples were subjected.Hydrogen chemisorption measurements on reduced samples degassed at the reduction temperature were carried out at room temperature and 100 Torrt gas pressure in a classical micromanometric apparatus equipped with a Texas Instruments pressure gauge. The temperature-programmed-desorption measurements (t.p.d.) were performed on reduced t 1 Torr = 101 325/760 Pa.SAUVION, TEMPERE, GUILLEUX, DJEGA-MARIADASSOU AND DELAFOSSE 1 3 59 Table 1. Sample compositions and characterizations chemical Ni ions Ce ions degree of nickel particle metallic sample composition (wt %) (wt %) reduction diameter/nm dispersion X, Ni,Ce,Na,,H,,X 2.53 4.09 0.28 X, Nil,Ce7Na,,H,,X 4.03 5.47 0.73 X, Ni,,Ce,,Na,H,X 3.52 13.3 0.54 X,a Ni,,Ce,~,Na,,H,X 6.83 4.25 0.95 Ni,.7Ce1.7Na1H0.5A 7.22 11.85 1 S, NiCe/SiO, 1.20 8.60 1 SZb Ni/SiO, 4.55 0 1 0.7 0.92 1 .o 0.85 1.2 0.78 87% 0'7 0.87 13% 3.0 2.5 0.49 2.5 0.49 2.5 0.49 a Sample reduced at 623 K after desorption at the same temperature. Sample prepared and characterized by Martin and Dalmon.lo catalysts cooled to room temperature still under hydrogen flow.The evolved gas was characterized by mass spectrometry to be a mixture of water and hydrogen. Quantitative measurements of these gases were made by calibration of the apparatus with known amounts of nickel sulphate hydrate and hydrogen gas, respectively.The activities of the supported nickel samples as catalysts of butane hydrogenolysis at atmospheric pressure were measured in a differential mode in a fixed-bed flow reactor. A ternary gas mixture (total flow 5 dm3 h-l) comprising pure, dry butane, hydrogen and helium, was contacted with 40 mg of catalyst at temperatures in the range 473-573 K. The reaction products were analysed using an Intersmat I.G.C. 112 F gas chromatograph fitted with an ionization detector and a 2 m Porapak N column. The areas of the product peaks were determined using an L.T.T. ICAP 5 integrator. Catalytic activities and selectivities were measured at 523 K (see table 4 later). The rates are expressed in moles of butane decomposed per second per gram of metallic nickel (rl) or per second per metal surface area ( r , ) .The metallic surface area was calculated as being 6 x 103/8.9 dnm, where d,, is the average particle diameter determined by magnetic measurements. The selectivity is defined by either S,,, = rclt/rt or SC1 = rcl/rt, where rt is the total butane hydrogenolysis rate, rclt the overall rate of production of methane formed during the reaction and rcl the rate of production of methane due to the complete cracking of the butane molecule (C,Hlo+3H, + 4CH,). For the CO + H, reaction, kinetic experiments were carried out at atmospheric pressure using a fixed-bed flow reactor in a differential mode. The total gas flow rate was equal to 3.61 dm3 h-' and the catalyst weight was 100 mg.The standard conditions were a CO:H, ratio of 1 :4. Gas analysis was performed by on-line gilts chromatography with catharometric and flame ionization detectors. Apart from methane, ethane was the only product detected. Catalytic activities and selectivities are given for the reference temperature of 573 K. The rates are expressed in moles of CO hydrogenated per second per gram of metal ( r , ) or per second per metal surface area (r,). The selectivity is relative to the C,H,/CH, + C,H, molar ratio. RESULTS SAMPLE CHARACTERIZATION As shown by X-ray determinations and potentiometric titration, the cerium ions in X-zeolite samples are all located in the crystallographic sites of the zeolite framework and are always in the trivalent state. In samples A, and S, cerium is present partly in the + 4 oxidation state.For sample A, a CeO, phase was detected by electron microdiffraction. An X-ray determination1360 REACTIVITY OF NICKEL PARTICLES of the location of the Ce3+ ions in cationic sites could not be carried out for this sample, owing to the partial damage of the zeolitic structure. For sample S,, no CeO, phase could be detected by microdiffraction, probably because of its high dispersion or low concentration. The reduced samples exhibit superparamagnetic or paramagnetic behaviour, as shown by the magnetic study. In such cases saturation magnetization, os, is difficult to attain. Thus after magnetic measurements had been made on the reduced samples, these samples were sintered under flowing helium at 1073 K. The large ferromagnetic crystallites thus formed are easily saturated and the os value was measured precisely.Nevertheless, some samples were oxidized during this treatment. In this case the reduction degree, which leads to the o, value, was obtained from volumetric measurements under static conditions of hydrogen consumption by unreduced Ni2+ on samples reduced under dynamic conditions. This measurement was performed at 100 Torr hydrogen pressure and 923 K, the temperature necessary to achieve a complete reduction of the samples. The difference between the total nickel content and the value obtained by this measurement gives the degree of reduction. Values of the extent of reduction, particle size distribution and metallic dispersion are listed in table 1. The metallic dispersion was estimated from values of the metallic particle diameter.HYDROGEN CHEMISORPTION The results, expressed by the ratio H/Ni$ where Ni; is the total number of metallic nickel atoms, are summarized in table 2, as are the values of the metallic dispersion. As shown previously,8 no hydrogen chemisorption occurs for samples X, and X,. These samples possess the greater metallic dispersion. Sample X,, in the same range of metallic dispersion but containing a small percentage of nickel particles of 3 nm diameter, shows a weak but detectable hydrogen chemisorption. It is tempting to attribute this chemisorption uniquely to the 3 nm particles. In all cases the H/Ni$ ratio of samples containing cerium is less than that corresponding to the metallic dispersion, especially for the X-zeolite samples.These results led us to characterize the surface state of the X samples after the reduction treatment by qualitative t.p.d. analysis. These samples retain large quantities of strongly chemisorbed hydrogen up to 923 K. The chemisorbed hydrogen was released during t.p.d. as water (in the temperature range 273-623 K) and as hydrogen (at temperatures above 700 K). The spectra are given in fig. 1. The present work affords further quantitative measurements of desorbed water and hydrogen below 923 K. TEMPERATURE-PROGRAMMED DESORPTION The results, expressed as the number of H,O and H, molecules desorbed per nickel atom, are summarized in table 3. For samples X, and X, the number of H atoms desorbing as H, molecules per nickel atom is greater than unity.Moreover, the amount of water desorbed is large, with the major fraction desorbing between 273 and 623 K. Blank t.p.d. experiments performed on non-reduced samples previously degassed under helium flow at the reduction temperature (623 K) showed that no water desorption occurs below 623 K. Water desorbed above this temperature arises from the dehydroxylation of OH groups chemisorbed on the zeolite. On the reduced samples, the amounts of water desorbing below 623 K cannot be attributed only to the contribution of OH groups generated during the reduction process, which would lead to an H,O/Ni$ ratio equal to or, more probably, lower than unity. The present results strongly suggest that, during the reduction process, hydrogen is dissociatively adsorbed on the nickel with part of it diffusing onto the zeolite.This adsorbate is desorbed during t.p.d. first as water1, and secondly as hydrogen. The loss of adsorbedSAUVION, TEMPERE, GUILLEUX, DJEGA-MARIADASSOU AND DELAFOSSE 1 36 1 Table 2. Hydrogen chemisorption hydrogen chemical metallic chemisorption,a sample composition dispersion H/Ni$ X, Ni,Ce5Na,,H,,X 0.92 0 X, Ni12Ce,Na,,H,,X 0.85 0 X, Ni,,Ce,,Na,H4X 0.78 0.15 x4 Ni21Ce6. 5Na20H4X 0.87 0.04 Ni2.7Ce1.7Na1H0.5A 0.49 0.24 Sl NiCe/SiO, 0.49 0.39 s2 Ni / Si 0, 0.49 a Hydrogen chemisorption expressed by the H/Ni& ratio, where Ni; is the total number of metallic nickel atoms. 400 500 600 700 800 900 TI K Fig. 1. Temperature-programmed-desorption spectra of samples X, (a) and X, (b) : (-) H,O and (---) H,.hydrogen as water is corroborated by i.r. and f.m.r. results. Indeed, the i.r. spectra of the reduced samples show a band at 1990cm-l which may be ascribed to a hydrogen-metal bond.13 This band disappears, during t.p.d., in the same temperature range over which water is desorbed. Furthermore, the f.m.r. spectra of all the samples, except sample XI, remain unchanged during this treatment, indicating no change in the magnetization of the nickel particles. This result suggests that the hydrogen species dissociating on the metal donates an electron and is removed as H+ without any change in the electronic state of the metal previously covered by the chemisorbed hydrogen during the reduction process. At higher desorption temperatures, when hydrogen molecules are evolved, the magnetization of the samples increases significantly as the hydrogen species leaves the metallic particle in the atomic form.In the case of sample X,, f.m.r. measurements recorded during the desorption treatment have shown that at high temperature this sample is progressively oxidized to total oxidation. The metallic nickel supported on the more acidic supports is stable only in presence of adsorbed hydrogen.l4? l51362 REACTIVITY OF NICKEL PARTICLES Table 3. Temperature-programmed-desorption measurements chemical nickel particle sample composition diameter/nm H,O/Ni$a H/Ni$* Xl Ni,Ce5Na,,H3,X 0.7 5.66 0.53 x2 Ni,,Ce,Na,,H,,X 1 .o 6.68 0.40 2.91 0.82 x4 Ni21Ce6.5Na20H4X 13% 3.0 x3 Nil ,Ce,,Na,H,X 1.2 3.35 0.21 Number of water and hydrogen molecules evolved up to 923 K per metallic nickel atom, Ni$ being the total number of metallic atoms.All the above results explain why, according to the hydrogen coverage of nickel crystallites, the hydrogen chemisorption capacities are small or suppressed. Similar observations have been reported by Menon and Froment16 with Pt/Al,O, systems. The attenuation of the hydrogen-adsorption properties of these solids, when reduced above 723 K, is ascribed to the presence of strongly chemisorbed hydrogen on the metal in this temperature range. BUTANE HYDROGENOLYSIS In table 4 are reported the catalytic activities of the various samples studied and their selectivities, defined as the degree of fragmentation of the butane molecules, towards me thane formation. From these results the following points may be deduced.(1) Samples X, and X, are totally inactive in the reaction. In no case was metal sintering observed during the course of the catalytic runs. Taking into account the assumption that the hydrogenolysis mechanism requires the existence of a sufficient number of contiguous free metallic sites accessible to the reactant molecules,17 it seems obvious that samples X, and X,, with high hydrogen coverage after reduction, cannot hydrogenolize butane. (2) Sample X,, with a bidispersed metallic distribution, shows weak but detectable activity. It seems likely that the activity of this sample can be ascribed, as suggested for hydrogen chemisorption, uniquely to the 3 nm nickel particles. (3) Higher catalytic activities can be observed with nickel particles dispersed on silica in the presence or absence of cerium additives.(4) For X, and A, samples, which display lower activities than nickel/silica systems, it may be noted that these activities, expressed per total metallic surface area, are obviously underevaluated for samples partly covered by hydrogen species. ( 5 ) The catalytic selectivities depend strongly on the presence of cerium additives. Thus the selectivity values are greatly lowered in the presence of cerium-containing nickel catalystscompared with thecerium-free nickel/silica reference sample. lo These results concerning butane hydrogenolysis show that, when the nickel catalysts present an accessible metallic surface to the reactants, the activities seem to be of the same order of magnitude for all samples whether or not they contain Ce3+ ions, whereas the selectivities are modified in the presence of cerium in the 3 + state.The partial or total loss of reactivity of the reduced catalysts, when observed, may be attributed to the coverage of the metallic surface by hydrogen entities. In the case of samples XI, X, and X,, the particularly large amount of hydrogen species retained after the reduction process seems to imply the existence of a hydrogen-spillover effect.SAUVION, TEMPERE, GUILLEUX, DJEGA-MARIADASSOU AND DELAFOSSE 1 363 Table 4. Activity and selectivity for butane hydrogenolysis at 523 K butane hydrogenolysis rate hydrogen selectivitya sample composition H/Ni% C4Hl0 s-' g& C4H10 s-' m& SClt SC, chemical chemisorption, rl/ mol r z / lop8 mol - - - X, Ni,Ce,Na,,H,,X 0 0 0 X, Ni,,Ce7Na2,Hl,X 0 0 0 - - X, NillCe17Na,H,X 0.15 62.3 11.0 1.43 0.09 x.4 Ni21Ce,.,Na20H4X 0.04 5.2 0.6 - 0.24 47.0 17.4 1.57 0.12 S, NiCe/SiO, 0.39 121.0 44.8 1.57 0.12 S, Ni/SiO, 102.7 38.0 2.28 0.35 - A 1 Ni2.7Ce1. 7Na1 HO. SA a For definition see text. It was of interest to study the reactivity of these samples towards a reaction which, unlike hydrogenolysis, would not be inhibited by the presence of hydrogen preadsorbed on the metallic sites and would consume these species. Two reactions seem to meet these conditions : the hydrogenation of ethylene and the hydrogen-deuterium exchange reaction. Both reactions might occur only by the introduction of ethylene or deuterium onto the reduced samples. REACTIVITY OF HYDROGEN SPECIES IN THE PRESENCE OF ETHYLENE OR DEUTERIUM Of the X-zeolite samples, X, was chosen for this study on account of its homogeneous particle-size distribution, its stability and its high coverage by adsorbed hydrogen.After reduction at 623 K and evacuation of the hydrogen gas phase at room temperature, the sample was submitted for 2 h to ethylene or deuterium at 433 or 273 K, respectively, in a static system under a pressure of 100 Torr of the considered gas. Afterwards, the gas phase of the reactor was analysed by on-line mass spectroscopy and a t.p.d. spectrum of the catalyst was obtained and compared with the reference t.p.d. spectrum of sample X, reduced at 623 K. Quantitative experimental results are reported in table 5. After introduction of ethylene into the reactor according to the preceding conditions the following results are obtained.(1) The gas phase evacuated at room temperature consists of 90% ethane. (2) During the course of the t.p.d. experiment heavier hydrocarbons are evolved in large quantities up to high temperatures. (3) The amounts of water and hydrogen molecules desorbed during the t.p.d. treatment are negligible. (4) In addition, after the ethylene reaction the i.r. band located at 1990cm-l, previously ascribed to an NiO-H bond, can no longer be observed, while the f.m.r. signal of the sample increases. All these results show that ethylene consumes the chemisorbed hydrogen, yielding ethane and higher hydrocarbons. With regard to deuterium reactivity at 273 K we can make the following observations.(1) The gas evolved at room temperature is composed of D, and HD molecules. (2) By temperature-programmed desorption and as compared with the reference sample (table 5 ) the solid yields an identical number of water molecules, while the amount of hydrogen molecules evolved is diminished significantly. (3) The f.m.r. signal of the sample is increased.1364 REACTIVITY OF NICKEL PARTICLES Table 5. Temperature-programmed-desorption measurements after reaction of sample X, with ethylene and deuterium treatment H,O/Ni$" H,/Nigb X, taken as reference X, after reaction with C2H4 at 433 K X, after reaction with D, at 273 K 6.7 0.4 0 0 6.8 0.2 Number of water and hydrogen molecules evolved up to 923 K per metallic nickel atom, Ni; being the total number of metallic atoms.Table 6. Activity and selectivity for CO hydrogenation at 573 K CO hydrogenation rate hydrogen chemical chemisorption rl/ lop6 mol,, r 2 / mol,, selectivity, sample composition H/Ni$ s-' g,;o s-1 S" X, Ni,Ce,Na,,H,,X 0 8.4 9 0.04 X, Ni,,Ce,Na,,H,,X 0 10.1 15 0.06 X, Ni, Ce, Na,H,X 0.15 203 361 0.04 Ni2. 7Na1H0.,A 0.24 157 585 0.04 Sl NiCe/SiO, 0.39 94 1 3.500 0.07 s2 Ni/SiO, 44.6 166 0.0 1 a S is the selectivity defined by the C,H,/CH, + C2H6 molar ratio at 573 K. Thus in contrast to ethylene the deuterium molecule only reacts with part of the hydrogen species which left the surface as molecular hydrogen during the thermo- desorption of the freshly reduced sample. It follows that sample X,, inactive for butane hydrogenolysis, presents a degree of catalytic activity in hydrogenation reactions and for H,-D, isotopic exchange.The hydrogen species present on the reduced sample surface are completely consumed in the presence of ethylene, but only partially in the presence of deuterium. CARBON MONOXIDE HYDROGENATION We have shown that highly dispersed nickel catalysts displayed little or no activity with regard to butane hydrogenolysis. This does not imply that these solids cannot present good activities for other reactions, and particularly for the hydrogenation of CO as mentioned above. The results of this study are summarized in table 6 , which lists the catalytic activities and selectivities of the various samples. The reactivity of these catalysts differs widely according to the sample considered.The activities of cerium-containing catalysts, expressed per metallic surface area, do not seem dependent on the particle size or cerium content, but rather on the hydrogen surface coverage of the reduced samples; this is shown for all the catalysts by the difference observed between the values of the metallic dispersion and the H/Ni& ratio. Thus samples X, and X, are by far the less active.SAUVION, TEMPERE, GUILLEUX, DJEGA-MARIADASSOU AND DELAFOSSE 1 365 In addition, sample S,, with the same metallic dispersion as sample S,, is twenty times more active than the latter. This greater activity must be attributed to the presence, on sample S,, of cerium additives. Furthermore, it appears that samples X, and A,, with a metallic surface which is mostly inaccessible to hydrogen at room temperature, possess higher activities than sample S,, with a fully accessible metallic surface.The presence in sample X, of cerium only in the trivalent state seems to show that the Ce3+ ion is responsible for the increase in activities observed. DISCUSSION The above results clearly show how the catalytic properties of nickel metal dispersed on various oxides can be modified as a function of the support acidity, the metallic dispersion and the presence of additives. For X-zeolite samples it has been shown that correlations exist between the support acidity, the average particle diameter, the metallic surface accessible to hydrogen and the catalytic activity for butane hydrogenolysis. It thus appears that the greater the acidity of the support, the smaller the nickel particles and the lower the hydrogen chemisorption and hydrogenolysis rates.Similar relations are observed with the other catalytic systems having the same metallic dispersion, but differing in the acidity of the support (A zeolite or SiO,). The loss or decrease in activity observed for the more acidic samples could be ascribed to the total or partial coverage of the metallic surface by strongly chemisorbed hydrogen species. The coverage of the metal by hydrogen could be correlated with the presence of Lewis-acid sites on the zeolitic supports. During the reduction process these electron-acceptor centres can induce some electron depletion of the small nickel particles in their formation, leading to the establishment of strong metal-support interactions and to the enhancement of metallic-surface affinity for the chemisorption of hydrogen molecules, which can donate their electrons to the nickel atoms.In butane hydrogenolysis the extensive coverage of the metallic surface of the more acidic samples by hydrogen species hinders the dissociative chemisorption of the hydrocarbons. After temperature-programmed desorption of the X-zeolite samples up to 923 K, these hydrogen species are removed from the metallic surface as water and as molecular hydrogen. The nickel particles supported on the more acidic sample are oxidized in the absence of hydrogen. In contrast, the less acidic samples retain their metallic stability without sintering. An attempt to observe activity after the removal of these hydrogen species has been carried out on sample X,, which is not oxidized during t.p.d.treatment and whose metallic-particle size remains unchanged during this treatment. Again, no hydro- genolysis activity can be observed. An additional experiment has been performed on this catalyst thermodesorbed at 923 K, by contacting it with a hydrogen flow at 523 K, the reference temperature for the hydrogenolysis reaction. It appears from the thermodesorption curve of this sample that during this treatment the hydrogen uptake is negligible. The inactivity observed for the bare metallic nickel surface in H, chemisorption and hydrogenolysis could originate from superficial metal oxidation during this treatment. However, taking into account that the chemisorbed hydrogen leaves the metallic surface mainly as H+, one could also propose that the metal remains electron-enriched and consequently does not chemisorb hydrogen or hydrogenolize butane, as do the Nio/TiO, systems when reduced at high temperatures. Further investigation by X-ray spectroscopy might afford valuable information on the electronic state of the supported nickel.1366 REACTIVITY OF NICKEL PARTICLES With regard to CO hydrogenation, the presence of chemisorbed hydrogen species on the metallic surface could prevent the adsorption of the reactant molecules as in butane hydrogenolysis.However, the temperature at which the CO + H, reaction activity is measured (573 K) is higher than that taken as reference in butane hydrogenolysis and refers to a more important desorption of the hydrogen entities, rendering the surface more accessible to the reactants and thus justifying the weak but non-negligible activities observed.For the active samples, which have part of the metallic surface accessible to the reactants, it appears that another factor can also strongly modify the catalytic properties of nickel metal, namely the presence of Ce3+ ion additives. For X-zeolite samples, cerium cations are always present in the trivalent state, whereas for samples A, and S, it has been shown that, prior to reduction, cerium is partly present as a well dispersed CeO, phase. After reduction CeO, could no longer be detected by electron microdiffraction, and it is likely that during the reduction process the CeO, phase is much dispersed or reduced.Consequently, on all cerium- containing samples cerium is present, for the most part, in the trivalent state. In butane hydrogenolysis the low values of hydrocarbon fragmentation obtained on these samples, which generally occurs when the dehydrogenated intermediates formed during the reaction are less strongly adsorbed on the metallic surface, may be correlated with the presence of trivalent cerium. The electron-donating power of Ce3+ might induce charge transfer towards nickel crystallites and, as a result, increase the electron density of the metal leading to weaker hydrocarbon-metal interactions. In addition, the substantial increase in methanation activity and the selectivity modification observed for cerium-containing catalysts seem to be correlated with the electron-donor properties of cerium in the + 3 state.An explanation of the role of Ce3+ can be proposed by again assuming that the trivalent cerium cations in the neighbourhood of the metallic nickel particles initiate electron transfer towards the metal, leading to an electron-density enrichment of the metallic surface. Using the (a, n) bond scheme for CO chemisorptionl* this electron transfer would decrease the carbon-oxygen bond strength and thus increase the C-0 bond-dissociation probability. In the same way Kao et aL5 have ascribed the increased methanation activities observed in Ni/TiO, systems, compared with Ni/SiO,, to an excess concentration of electrons on metallic nickel caused by electron transfer from the support to the metal. Concerning the selectivities of cerium-containing catalysts, in all cases these samples are four to seven times more selective towards ethane production than sample S,.This is again in good agreement with the preceding assumption, according to which Ce3+ cations could induce some electron transfer from the support towards the metal. The electron enrichment of the metal surface might increase the dissociative chemisorption of CO compared with that of H, and lead to the formation of hydrogen-poor hydrocarbons. The results of the present investigation emphasize the importance of two factors able to modify considerably the catalytic properties of nickel metal highly dispersed on non-reducible oxide supports. One is the acidity of the support, which can generate strong metal-support interactions leading to electron depletion of the metal and consequently to its coverage, during the reduction process, by strongly chemisorbed hydrogen.In this case, the loss of reactivity is to be correlated to the absence of free metallic surface sites. The second is the presence of Ce3+ additives, which could induce electron enrichment of the metallic surface, leading to enhancement of methanation activity and to a shift in product distribution, as observed for Ni/TiO, systems when reduced at high temperatures.SAUVION, TEMPERE, GUILLEUX, DJEGA-MARIADASSOU AND DELAFOSSE 1 367 We thank Dr G. A. Martin (Institut de Recherches sur la Catalyse, Villeurbanne) for the gift of the nickel-silica sample. We are also grateful to Mrs J. Jeanjean for assistance with the X-ray diffraction study and to Mr M. Lavergne, who recorded the electron microdiffraction spectra. S. J. Tauster, S. C. Fung, R. T. K. Baker and S. A. Horsley, Science, 1981, 211, 1121. C. H. Bartholomew, R. B. Pannell and J. L. Butler, J . Catal., 1980, 65, 335. M. A. Vannice and C. C. Twu, J. Catal., 1983, 82, 213. R. Burch and A. R. Flambard, J . Catal., 1982, 78, 389. Chia-Chieh Kao, Shou-Chin Tsai and Yip-Wam Chung, J. Catal., 1982, 73, 136. P. Gallezot, Catal. Rev. Sci. Eng., 1979, 20, 121. J. C. Vedrine, M. Dufaux, C. Naccache and B. Imelik, J. Chem. SOC., Faraday Trans. I, 1978,74,440. G. N. Sauvion, M. F. Guilleux, J. F. Tempkre and D. Delafosse, J . Chim. Phys., 1982, 79, 395. S. Djemel, M. F. Guilleux, J. Jeanjean, J. F. Tempere and D. Delafosse, J . Chem. SOC., Faraday Trans. I , 1982, 78, 835. P. Weiss and R. Forrer, Ann. Phys. (Paris), 1926, 5, 153. lo G. A. Martin and J. A. Dalmon, C.R. Acad. Sci., 1978, 286, 127. '* T. M. Apple, P. Gajardo and C. Dybowski, J . Catal., 1981, 68, 103. l3 T. Nakata, J . Chem. Phys., 1976, 65, 487. l4 P. Leroux, Th&e (Universitk de Lyon, 1977). Is G. Martino, Studies in Surface Science and Catalysis (Elsevier, Amsterdam, 1980), vol. 4, p. 399. l6 P. G. Menon and G. F. Froment, Appl. Catal., 1981, 1, 31. J. A. Dalmon and G. A. Martin, J . Catal., 1980, 66, 214. G. Blyholder, J. Phys. Chem., 1964, 68, 2722. (PAPER 41 1 1 33)