J. Chem. Soc., Faraduy Trans. 1, 1987, 83, 1711-1718 Hydrogenation of CO, over Co/Cu/K Catalysts Herv6 Baussart, Rene' Delobel, Michel Le Bras and Jean-Marie Leroy* Laboratoire de Physicochimie des Solides, E.N.S.C.L., U.S.T.L., B.P. 108, 59652 Villeneuve D' Ascq Cedex, France The influence of the different elements Co, K, Cu as components of catalysts, has been investigated for the reaction of hydrogenation of CO,. Cobalt, without promoter, appears to be predominant in the methanation process. The presence of potassium in the matrix changes the nature of bonds between the other chemical species and favours the formation of CO. The addition of Cu reduces the specific role of potassium; it limits the quantity of products formed and decreases the selectivity in CO. In the absence of potassium, copper lowers the temperature of reduction of the catalysts, decreasing their activity, and affects the distribution of the active carboneous species formed during the transient regime.The development of techniques incorporating C, carbon sources to compete with or even supplant oil as a source of raw materials, explains the interest in the compounds CO and CO,. In this context, the catalytic hydrogenation of CO to CH, and hydrocarbons of high molecular weight has been particularly studied and de~eloped.l-~ On the other hand, despite the existence of many cheap sources, the development of the hydrogenation reactions of carbon dioxide is still limited. The catalysts used are generally the same as those used in the methanation of C0.49 In this paper, we have particularly studied the catalytic hydrogenation of carbon dioxide to methane with Co-Cu-K catalysts.Previously, only Russells has studied such a catalytic system on purpose to synthesize oils. The aim of the specific work is to bring about a contribution to the knowledge of the specific part played by each of the elements constituting the catalyst on the distribution of the reaction products. Experimental Materials Russell Catalysts ( R) Russell precursors were obtained according to the process described by the author :6 precipitation of basic carbonates of cobalt and copper by a solution of potassium carbonate. The precipitates obtained were filtered, washed and dried at 373 K, then pelletized in cylinders (diameter: 5 x m) under a pressure of 1.47 x los Pa.These pellets are treated as described in table 1. m, depth: 3 x C U , - ~ C O ~ + ~ O ~ Catalysts (C) All the Cu,-,Co,,,O, catalysts were obtained by the same procedure. The oxalates formed in the first step were decomposed into Co,O, and CuO. The mixture of the two oxides was then heated under oxygen at 450 "C using conditions described previ~usly.~ This temperature allows one to obtain defined compounds and consequently an optimal dispersion of the elements, as proved by electron probe microanalyses which showed the homogeneity of the catalysts. For Co/Cu > 3.92, X-ray analyses showed that these 171 11712 Co/ Cu/ K Hydrogenat ion Catalysts Table 1. Catalyst composition and conditions for the activation pretreatments catalyst precursors Co/Cu K/Co catalyst activation Co basic carbonate - - Co-K carbonate - :} Co-Cu-K carbonate 20 1.8 1 h under N, at 423 K, 24 h under CO,:H, = 1:4 at 723 K l a 8 1 R2 C, cobalt oxalate - - 24 - 12 - 6.9 - ' Co-Cu oxalates 5.5 - 3.9 - 2.3 - 2 - 360 h under 0, at 723 K, 1 h under N, at 423 K and 24 h under CO,: H, = 1 : 4 at different temperatures I compounds presented the spinel structure.For x < 3.92, a mixture of monoclinic CuO and a spinel phase was always observed. The pelletized oxides obtained were treated as described in table 1. Textural Characterization The B.E.T. surface areas were measured by a standard method using N, physisorption at 77 K (Carlo Erba Sorptomatic). They are in the ultimate range of the apparatus (< 1 m2 g-l). The low values obtained were corroborated by scanning electron micro- graphs using a Stereoscan mark 2A electron microscope. The observed images of the specimens show that the grains may be assimilated roughly with spheres without any apparent rugosity.As the surface areas could not be precisely determined, only the specific rates of the reactions or the fractions of CO, converted (x in %) were compared. Thermogravimetric Studies Thermogravimetric studies were carried out using a Setaram M.T.B. lo-* microbalance. This apparatus allowed the measurement of the temperature TR which corresponds to the beginning of the reductions of the samples. These reductions were carried out under continuous flow of the reactant gases (CO,:H, = 1 :4, flow rate: 1.2 dm3 h-l) and the relative losses of weight of the catalysts, A W% , were measured after working in the catalytic conditions.Catalytic Apparatus Measurements of activities and selectivities were performed in a continuous stirred reactor at ca. 101 325 Pa. This stirred gas-solid reactor (s.g.s.r.) has been described elsewhere8 and was placed in a furnace whose temperature could be controlled to within f 1 K. Products were analysed by gas chromatography (Intersmat I.G.C. 15 gas chromatograph). CH,, C2H4, C3H,, n-C,H,,, n-C5H12, n-C,H,,, CH30H and C2H50H were separated in a 2 m packed column containing Porapak Q ; an active carbon column was used for the separation of H,, CO, CH4 and C,H,. Determination of the Kinetic Range The CO,: H, (1 :4) mixture at ca. 101 325 Pa total pressure reacted in the absence of catalyst in the s.g.s.r.at temperatures up to 623 K. This homogeneous process limitsH . Baussart et al. 1713 contact time/g h dm-3 Fig. 1. Amount of CO, reduced in CH, us. contact time ( T = 433 K, CO: H, = 1 :4). the temperature range below 623 K. To ensure the mixing in the gas space is adequate, the evolution of the conversion rate was studied at a series of different agitation speeds in the presence of the catalysts. Under the conditions of reaction, satisfactory mixing tests were run in the range 500-5000 r.p.m., for which the external mass-transfer resistance was negligible. The curves in fig. 1 show the linear relation at 433 K between the percentage of CO, reduced in CH, and the contact time 6 = m / d (g h dm-3), where m is the catalyst mass and d the volumetric flow of reactant gases into the reactor.The relation, corroborated when d is high for a given m, proves the absence of diffusional effects. Such a relation was always observed in the temperature range used. In the kinetic regime, x remained below 2%, Throughout this study the constant experimental conditions were: catalytic charge, 4 g; total flow, 6 dm3 h-l; reactant mixture, CO,:H, = 1 :4; rotation speed, 3500 r.p.m.; pressure, ca. 101 325 Pa. Results Russell-type catalysts were taken as a reference; an activation process of these R catalysts was carried out using the conditions defined by the author: pretreatment under reactant mixture (flow rate, 6 dm3 h-l) in the temperature range 423-503 K. This procedure does not allow observation of any catalytic activity with R,.To obtain R catalysts with significant catalytic performances, it is necessary to activate the specimens at 723 K, as described in table 1. On the other hand, C-type catalysts do not require such a high temperature for activation. Table 2 shows the evolution of the catalytic performances of C, as a function of the temperature of the activation process. Note that the increase of this temperature involves a drop in the selectivity for methane as well as a decrease of the conversion of CO,. In this paper, the activities of the catalysts are compared using the temperature T, for which x is 1 % (specific activity: 1.34 x mol g-l h-l). The chosen convention allows the comparison of the selectivities at a defined level of x in the kinetic regime.This comparison between R, and C, activated at 433 K shows that C, is slightly more active and selective than R,. Owing to these performances, every C catalyst was activated at 433 K. Among the C catalysts, samples containing copper have activation profiles markedly 57 F A R 11714 C o / C u / K Hydrogenation Catalysts Table 2. The influence of the temperature of activation on the activities and the selectivities for C, selectivity (%) activation sample TIK T,/K CH.4 C,H, C,H* C*KO 433 424 97.7 1.6 0.6 0.04 Cl 533 428 96.6 1.9 1 . 1 0.09 633 434 92.5 4.7 2.5 0.3 723 447 90 5.9 3.5 0.5 RO 723 433 95.5 2.9 0.9 0.7 I I . c7 5 10 15 20 25 time/h Fig. 2. Activation of the C catalysts at 433 K. different from that of C, (fig. 2). For C , the activity increases regularly before reaching a.steady state. With the other C compounds, the initial sharp rise in activity is followed by a slow decline. The steady state is reached after ca. 30 h of activation. It is obvious that a comparison with the activation profiles of the R-type catalysts is not possible because the process is, in that case, carried out in the temperature range where the homogeneous reaction takes place. The CH, production is shown in fig. 3, in Arrhenius form. Apparent activation energies E,, calculated from the slopes of the lines, are summarized in tables 3 and 4. They are virtually the same for all the catalysts (83 < E,/kJ mol-1 < 102), except for the catalyst containing Co, Cu and K. In that case, the value of E, is comparatively low (64.6 kJ mol-l).A drop in total activity of R-type catalysts is observed when they contain K. The addition of copper slightly reduces this evolution. Furthermore, R, is selective for the formation of methane. The presence of potassium favours the synthesis of carbon monoxide ; primary alcohols, alkanes and ethylene are formed as by-products. Unlike Russell, the formation of oils has never been observed. The addition of copper decreases the selectivity for CO and limits the number of by-products. Table 4 shows that activities and selectivities for CH, of C-type catalysts decrease when the content of Cu increases, whereas their selectivities for ethane are in the opposite order. These catalysts are not active for the catalytic formation of CO.H. Baussart et al. 1715 - 8 - 9 - I s I ca 0 \ n d - c.E -10 a" F: - - 11 103 KIT Fig. 3. Arrhenius plot for CH4 conversion from CO, : H, = 1 : 4 on R- and C-type catalysts. Table 3. Catalytic performances of the R catalysts selectivity (% ) E a sample TJK /kJ mo1-I CH4 C2H6 c3H6 C4H10 RO 433 87.8 95.5 2.9 0.9 0.7 529 86.5 20.4 0.4 0.9 0.5 Rl R2 515 64.6 34.2 0.7 0.4 - selectivity (% ) sample C5Hl2 C13H14 C2H4 CH,OH C,H,OH co - - - - - - RO Rl 0.54 0.17 1.8 0.09 0.2 75 - - - R2 0.6 0.1 1 64 57-21716 C o / C u / K Hydrogenation Catalysts Table 4. Comparison of the catalytic performances of C-type catalysts selectivity (% ) Ea sample 7JK /kJ mol-l CH* C2H6 C3H8 C4HlO Cl 424 c3 44 1 c5 456 c2 437 c4 449 C6 477 c, 493 C8 500 85.5 86.9 83.6 88.6 84.0 95.3 83.6 102.6 97.7 97.2 97.9 96.7 96.8 96.8 96.7 95.2 1.8 2.0 1.7 2.5 2.5 2.5 2.6 4.1 0.4 0.7 0.3 0.7 0.6 0.6 0.6 0.6 0.04 0.09 0.06 0.06 0.07 0.08 0.07 0.06 Discussion The interpretation of the role in the reaction process of the different elements constituting the catalysts is particularly interesting.This task is not an easy one to accomplish, particularly in complex systems such as multicomponent catalysts. In spite of the existence of several techniques which have been developed to evaluate the active specie^,^ some investigators suggested that no standard method exists.1° This situation has led us to use surface reaction studies alone as probes of catalyst surfaces. The specific role of the element Co and the eventual synergisms which result from the association with potassium and/or copper have been investigated.A bibliographic study reveals that the presence of cobalt favours the formation of alkanes., Our work verifies this result: unsaturated species are never found in the reaction products when R, and C, are used. Two hypotheses can be proposed to explain the formation of the saturated compounds CH,, C,H,, C,H, and C4H1,. First, a classical mechanism of competition between chain-propagation and hydrogenating desorption : Secondly, it may be considered that methane is directly produced from CO,, whilst the other alkanes are formed by hydrogenation of C0.l1 It has been observed that this process occurs when CO is the major product, so this mechanism seems improbable. Our results show that the presence of potassium (R,, R,) favours the formation of CO.This compound is probably produced by the reverse water-gas shift reaction: CO, + H, CO + H,O. The formation of small quantities of ethylene is also observed. It is well known that the presence of potassium in the catalysts used for Fischer-Tropsch synthesis favours the formation of olefins. In our case, such a reaction cannot be rejected because CO and H, exist in the reaction medium; the presence of C2H4 can thus be explained. In a more general way, the reaction process may be explained taking into account the electron-donor character of potassium.12 This character favours the adsorp- tion of CO, increases the metal-C bond strength and, at the same time, weakens the C-0 bond. On the other hand, it does not favour the chemisorption of H, and decreases the hydrogenating properties of the catalyst, explaining the formation of unsaturated species.l3H. Baussart et al. 1717 Table 5. Comparison between temperature of beginning of reduction, TR, weight losses in the s.g.s.r. and temperatures T, catalyst c, 493 6.15 424 c5 443 8.36 456 c3 448 7.5 441 c, 43 8 12.16 493 The presence of copper in Russell catalysts increases the activity and the selectivity for CH, and reduces the number of products. On the other hand, in C catalysts copper leads to a loss of activity and minor changes in the distribution of the products. It may be proposed that copper, when introduced in a matrix containing potassium, presents a weakened electron-donating effect. Moreover, as suggested by Russell, if the alkali metal is a poison for the methanation sites, a competition between copper and potassium may be considered to be responsible for the increase of the carbon dioxide conversion to methane.If the matrix is free of potassium, the electron-donor character of copper may play its part: the temperature of reduction of the catalysts is lowered and, according to the literature,14 superficial carbides would be formed and would react as intermediates. When a comparison between activities and reducibilities of the samples is performed (table 5 ) an inverse relationship is observed between TR and the Cu content. The activation profiles which describe the evolution of the activity as a function of the time can explain this result. In the case of C,, the non-existence of an extremum during activation can be explained by considering that the catalyst is the least reducible, so the reduction process involves the lowest superficial change, thereby limiting the formation of active sites. For all the other catalysts which contain Cu we observe a maximum.This suggests a transient regime during which the passage through the maximum activity corresponds to the rate of production of active sites being equal to that of their spontaneous decay.15 More precisely, like Sachtler,14 we may propose the formation on these sites of active adsorbed carboneous species: Cads, CH,,,, CH,, ads. In this scheme, copper acts during the transient regime on the distribution of these species and determines their evolution either to products or to less-active intermediates which may go as far as graphitic carbon.The quasi-stability of the selectivity may be explained by the presence of active species defined above, but in a lower concentration when the amount of copper increases. This evolution may also be explained by a sintering phenomenon, the intensity of this process being a function of the copper content. It is then obvious that if TE decreases the sintering increases and the number of superficial active sites falls, these sites remaining unchanged. Finally, experimental results lead us to consider that the presence of Cu favours, in a first stage, the reduction of CO, to active carboneous species probably via formation of CO. In a second stage there is a competition between the synthesis of methane from active carboneous species and the reduction which leads to the formation of carboneous deposits on the surface of the catalyst.As pointed out by Somorjai,16 in the first stage (which corresponds in our pattern to the reduction of the oxide) we cannot exclude the formation of metallic clusters. The amount of these clusters and their coexistence with an oxidized form would condition the catalysis. Finally, it may be pointed out that, unlike Russell catalysts, the formation of oils on C-type catalysts has never been observed. This difference may be explained by differences in the hydrodynamic and thermal regimes of the reactor, the diffusional processes having a significant effect on the distribution of the products.1718 Co/Cu/K Hydrogenation Catalysts Conclusion The present work enables us to characterize the specific role played by Cu, K and Co in the hydrogenation of carbon dioxide at low pressure.The chemical nature of the elements plays a very important part in the orientation of the reaction. In a matrix containing cobalt, our results show the essential part played by potassium, which modifies the nature of the bonds between the other chemical species and favours the formation of Co. The presence of copper reduces the temperature of reduction of the catalyst, decreasing their activity, and affects the distribution of the active carboneous species, intermediates of the reaction, which are formed during the transient regime. References 1 G. A. Mills and F. W. Steffgen, Catal. Rev., 1975, 8, 159. 2 M. A. Yannice, Catal. Rev. Sci. Eng., 1976, 14, 153. 3 Y. Ponec, Catal. Rev. Sci. Eng., 1978, 18, 151. 4 W. W. Russell and G. H. Miller, J. Am. Chem. Soc., 1950, 72, 2446. 5 E. J. Gibson and C. C. Hall, J. Appl. Chem., 1954,4,464. 6 L. F. Cratty and W. W. Russell, J. Am. Chem. Soc., 1958, 80, 267. 7 H. Baussart, M. Lx Bras and J. M. Leroy. C . R. Acad, Sci. Paris, Ser. C, 1977, 281, 735. 8 M. L. Brissk, B. L. Day, M. Jones and J. B. Waarren, Trans. Znst. Chem. Eng., 1968, T3,46. 9 J. L. Lemaitre, P. Goving Menon and F. Delannay, in Chemical Industries, Characterization of Heterogeneous Catalysts, ed. F. Delannay (Marcel Dekker, New York, 1984), vol. 15, pp. 299-362. 10 E. G. Baglin, G. B. Atkinson and L. J. Nicks, Ing. Eng. Chem., Prod. Res. Den, 1981, 20, 87. 1 1 M. Pijolat and Y. Perrichon, C . R. Acad. Sci. Paris, Ser. IZ, 1982, 295, 343. 12 D. E. Pebbles, D. W. Goodman and J. M. White, J. Phys. Chem., 1983, 87, 4378; F. Solymosi, I. Tombacz and J. Koszta, J. Catal., 1985, 95, 578. 13 M. Papadopoulos, R. Kieffer and A. Deluzarche, B.S.C.F., 1982, no. 3-4,I-109. 14 P. Biloen and W. M. H. Sachthler, Adv. Catal., 1981, 30, 165. 15 A. Amariglio, M. Lakhdar and H. Amariglio, J. Catal., 1983, 81, 247. 16 D. J. Dwyer and G. A. Somorjai, J. Catal., 1978, 52, 291. Paper 6/ 1398 ; Received 14th July, 1986