Applied Catalysis B: Environmental 29 (2001) 207–215 Highly effective conversion of CO2 to methanol over supported and promoted copper-based catalysts: influence of support and promoter Jamil Toyir a , Pilar Ramı́rez de la Piscina a , José Luis G. Fierro b , Narcı́s Homs a,∗ a Departament de Quı́mica Inorgànica, Facultat de Quı́mica, Universitat de Barcelona, Diagonal 647, 08028 Barcelona, Spain b Instituto de Catálisis y Petroleoquı́mica, C.S.I.C., Cantoblanco, 28049 Madrid, Spain Received 30 January 2000; received in revised form 5 July 2000; accepted 5 July 2000 Abstract In this study, gallium-promoted copper-based catalysts prepared by impregnation methods on silica and ZnO supports, were examined for the hydrogenation of CO2 to methanol. The surface characteristics of catalysts depended on the support and were related to their catalytic performance. Silica-supported catalysts tested at reaction temperatures between 523 and 543 K were highly selective and stable. The selectivity to methanol was around 99%, the conversion to CO was very low and negligible amounts of hydrocarbons were formed. The use of hydrophobic silica enhanced the performance of the catalyst in terms of activity, selectivity and stability. The modification of properties of copper particles is related to the presence of very small particles of Ga2 O3 on the surface. © 2001 Elsevier Science B.V. All rights reserved. Keywords: CO2 hydrogenation; Methanol synthesis; Copper-based supported catalysts; Gallium-oxide promoter; Hydrophobic silica; X-ray photoelectron spectroscopy (XPS) 1. Introduction Greenhouse effect is regarded as a serious threat to the global environment. The mitigation of carbon dioxide, a major greenhouse gas, is an urgent issue. As a countermeasure, through catalytic hydrogenation of CO2 a significant amount of CO2 can be effectively converted into valuable products, such as methanol, which is considered an alternative energy source, a medium for the storage and transportation of hydrogen and a starting material for several chemicals as well. Methanol is made from syngas (a mixture of CO/CO2 /H2 ) over Cu/ZnO/Al2 O3 ∗ Corresponding author. Tel.: +34-93-402-1235; fax: +34-93-490-7725. E-mail address: nhoms@kripto.qui.ub.es (N. Homs). catalyst. Recent studies have shown that methanol can be produced either from a mixture CO/CO2 /H2 or a mixture containing only CO2 and H2 over Cu/ZnO-based multicomponent catalysts [1]. It has been widely reported that Cu/ZnO-based catalysts are among the most useful systems for the catalytic hydrogenation of CO2 into methanol [2–8]. Saito et al. developed a high performance multicomponent catalyst Cu/ZnO/ZrO2 /Al2 O3 /Ga2 O3 following a co-precipitation method using metal nitrates and sodium carbonate [9,10]. The activity of the multicomponent catalyst was proportional to the surface area of copper, which was considered to be the main active site for the hydrogenation of CO2 . The addition of Al2 O3 or ZrO2 to Cu/ZnO increased the total surface area as well as the dispersion of Cu particles on the surface, while the association of Ga2 O3 0926-3373/01/$ – see front matter © 2001 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 6 - 3 3 7 3 ( 0 0 ) 0 0 2 0 5 - 8 208 J. Toyir et al. / Applied Catalysis B: Environmental 29 (2001) 207–215 to the binary catalyst increased the activity per unit of Cu surface area. The addition of small amounts of silica increased the stability of the catalyst [11]. Both Cu+ and Cu0 species were essential to catalyze the reaction and Ga2 O3 regulated the optimal ratio of Cu+ /Cu0 throughout the reaction. When a co-precipitated Cu/ZnO-based catalyst is used in the synthesis of methanol, ZnO has the following functions: (i) it favours the formation of appropriate precursors during the catalyst preparation, which leads to higher dispersion; (ii) in the presence of Al2 O3 , it shows a refractory behaviour and attenuates the unavoidable agglomeration of Cu particles which takes place during a long-term operation; (iii) it improves the resistance of Cu particles to poisoning by feed gas impurities, such as sulfides and chlorides and (iv) ZnO, as a basic oxide, partially neutralizes the acidity of Al2 O3 , preventing the transformation of methanol to dimethyl ether. However, in these catalysts the effect of the zinc oxide texture on the interaction with Cu and on the dispersion of Cu particles is difficult to determine. Moreover, the crystallisation of Cu particles by the sodium remaining in the catalyst (as sodium nitrate) hinders the obtention of a stable co-precipitated Cu/ZnO-based catalysts [10]. As an attempt of avoiding these problems, here we report the use of alternative Cu-Ga2 O3 supported catalysts prepared by co-impregnation using zinc oxides of different surface areas. On the other hand, since silica, as ZrO2 and Al2 O3 , allows higher dispersion and stabilizes the catalyst [10], methanol synthesis on silica-supported Cu-ZnO-Ga2 O3 catalysts was investigated. Taking into account that water produced during the hydrogenation of CO2 inhibits the rate of methanol [8] and the related crystallisation of ZnO in a long-term operation leads to a significant deactivation of the catalyst [10], the behaviour and structural properties of a catalyst prepared using hydrophobic silica as support was also examined. 2. Experimental The catalysts were prepared using the incipient wetness technique. The precursors were Cu(NO3 )2 ·3H2 O (Panreac, >99%), Zn(NO3 )2 ·6H2 O (Avocado, >98%) and Ga(NO3 )2 ·xH2 O (Aldrich, >99%). SiO2 (Aerosil 200, 200 m2 /g) from Degussa (SiO2 ), SiO2 (hydro- Table 1 ICP-AES analysis and Cu crystal size (determined from XRD) of catalysts after testing in methanol synthesis Catalyst Cu (wt.%) ZnO (wt.%) Ga2 O3 (wt.%) Size (nm) Cu-Ga/ZnO(LS) Cu-Ga/ZnO(HS) Cu-Zn/SiO2 Cu-Zn-Ga/SiO2 Cu-Zn-Ga/SiO2 (HD) 6.0 6.8 5.0a 4.7 4.7 – – 5.0a 2.6 2.3 1.9 2.5 – 1.7 1.5 67 67 43 37 35 a Initial loading. phobic silica HDKH20, 170 m2 /g) from Wacker (SiO2 (HD)), ZnO (Kadox, 11 m2 /g) (ZnO(LS)) and ZnO (100 m2 /g) (ZnO(HS)) were used as supports. ZnO(HS) was prepared by decomposition of 3ZnO·2ZnCO3 ·3H2 O at 573 K. Initially, the appropriate weight of metal nitrates precursors required to prepare a catalyst having a nominal weight loading was dissolved respectively in deionized water for ZnO and SiO2 supported catalysts and in methanol for SiO2 (HD)-supported catalysts. The amount of water or methanol was calculated on the basis of the pore volume and the weight of the support. The solution containing the metals was added to the support in a dropwise manner to give a thick slurry. The resulting slurry was kept for 1 h in ambient conditions and dried at 373 K during 24 h. Thereafter, the catalysts were calcined in air at 873 K for 2 h. The catalysts were reduced at atmospheric pressure in flowing hydrogen at 573 K for 6 h and were stored in argon before use. The designations and compositions of all the catalysts used in the present study are listed in Table 1. A comparative standard sample Cu (5 wt.%)-ZnO (5 wt.%)/SiO2 was also prepared. The catalytic tests were performed in a fixed-bed continuous flow reactor equipped with a thermocouple allowing the measure of the temperature inside the catalytic bed. The catalyst (0.50 g) was introduced between wool plugs in a tube reactor (i.d.: 11.3 mm, height: 24 cm) made of stainless steel which did not affect the reaction. The catalyst was re-reduced in situ with a mixture H2 /He (10/90) at 573 K and total pressure 2 MPa for 2 h. Thereafter, the temperature was decreased to 523 K in pure hydrogen and the catalyst was exposed to a feeding gas mixture CO2 : H2 = 1 : 3. After periods ca. 20 h reaction, temperature was increased consecutively to 533 and 543 K. After each temperature change the system was J. Toyir et al. / Applied Catalysis B: Environmental 29 (2001) 207–215 stabilized for 3 h, then the sample corresponding to the initial activity of catalyst at a given temperature was taken (noted ai in Table 5). The reduction and reaction were carried out at a total flow rate of 9 l/h. The gas products were analyzed using a suitable gas chromatogragh (Varian Star) equipped with TCD and FID detectors connected on-line to the reactor. The total specific surface area, pore volume and pore size of the catalyst after reaction were determined by nitrogen adsorption at 77 K. The overall chemical compositions of catalysts were measured using inductively coupled plasma with atomic emission spectroscopy (ICP-AES). All the catalysts were characterized by X-ray diffraction (XRD) after calcination and reaction. XRD diffractograms were recorded on a Siemens D-500 powder X-ray diffractometer using Cu Ka (λ = 0.15406 nm) equipped with a graphite monochromator and a Cu target. All the samples were scanned in the range between 2θ = 20 and 80◦ . Diffractograms were recorded in step mode (3.0 s, 0.05◦ ) for the samples after calcination and (9.0 s, 0.025◦ ) for the post-reaction samples. The Cu crystal size was determined from the full width at half maximum (FWHM) of the highest peak of metallic Cu using the Scherrer equation, D = λ/b cos θ, where θ is the Bragg angle, D the crystallite mean size, and b the corrected FWHM (a-Al2 O3 standard). Photoelectron spectra (XPS) were acquired with a VG ESCALAB 200 R spectrometer equipped with a Mg Ka (hν = 1253.6 eV, 1 eV = 1.6302 × 10−19 J) X-ray exciting source and a hemispherical electron analyser. The calcined samples were evacuated at 383 K for 1 h, while the post-reaction samples were re-reduced at 523 and/or 573 K for 1 h in the pretreatment chamber before being transferred to the analysis chamber. The residual pressure in the ion-pumped analysis chamber was maintained below 209 4.2×10−9 mbar (1 mbar = 101.33 Pa) during data acquisition. The binding energies (BE) were referred to the adventitious C 1 s peak at 284.9 eV which gave BE values with an accuracy of ±0.1 eV. For quantification, the intensities of the peaks were estimated by calculating the integral of each peak after smoothing and subtraction of the ‘S-shaped’ background and fitting the experimental curve to a sum of Gaussian and Lorentzian lines of variable proportion. 3. Results and discussion 3.1. Characterisation of catalysts ICP-AES analytical data of the catalysts used in the present study are listed in Table 1. Specific surface areas, pore volume and pore size of the post-reaction catalysts are presented in Table 2, which shows that the type of support greatly influenced the resulting physical properties of the catalysts. For each catalyst tested, the BET surface area was smaller than that of the initial supporting material, especially when ZnO(HS) was used as support. The silica-supported catalysts showed higher surface areas and pore volumes than the ZnO-supported catalysts. The pore size depended on the nature of the support. ZnO with low surface area led to a smaller pore size than ZnO with high surface area. The silica-supported catalysts showed smaller pore volume and pore size when the hydrophobic silica was used. The XRD diffraction patterns of calcined catalysts are shown in Fig. 1. For all the samples, peaks corresponding to CuO are clearly visible. Moreover, ZnO-supported catalysts showed characteristic peaks of ZnO. For silica-supported catalysts, only the Cu-Zn-Ga/SiO2 (HD) showed a visible small peak Table 2 Specific surface area and pore volume of the post-reaction catalysts Catalyst Surface area (m2 /g)a Pore volume (mm3 /g) Average diameter of pore (nm) Cu-Ga/ZnO(LS) Cu-Ga/ZnO(HS) Cu-Zn/SiO2 Cu-Zn-Ga/SiO2 Cu-Zn-Ga/SiO2 (HD) 6 15 139 157 137 10 38 691 619 343 19 26 19 28 11 a (11) (100) (200) (200) (170) Values in parenthesis correspond to the initial supports. 210 J. Toyir et al. / Applied Catalysis B: Environmental 29 (2001) 207–215 Fig. 1. X-ray diffractograms of the catalysts after calcination. (A) Cu-Ga/ZnO(LS); (B) Cu-Ga/ZnO(HS); (C) Cu-Zn/SiO2 ; (D) Cu-Zn-Ga/SiO2 ; (E) Cu-Zn-Ga/SiO2 (HD). (m) CuO; (h) ZnO. located at 2θ = 36.2◦ , corresponding to the most intense peak of the ZnO zincite phase. Neither ZnO-supported nor silica-supported catalysts showed other peaks, and so no gallium oxides were detected by XRD. Fig. 2 shows the X-ray diffraction pattern of catalysts after reaction, which reveal the presence of metallic Cu. Only in the pattern of Cu-Zn-Ga/SiO2 (HD) remained two very small peaks at 2θ = 35.4◦ and 39.8◦ which correspond to the most intense peaks of the CuO phase. None of these patterns shows peaks of Ga2 O3 phase. The Cu crystal sizes for each catalyst studied are listed in Table 1. Both types of silica led to similar Cu crystal sizes, which were smaller than those resulting from ZnO-supported catalysts. Tables 3 and 4 summarise the XPS results of the calcined catalysts and the catalysts tested in methanol synthesis and subsequently re-reduced in situ at 523 and/or 573 K. For all the catalysts, the binding energy (BE) of the Cu 2p3/2 peak in the post-reaction samples was ranged from 932.8 to 933.0 eV. Calcined cat- alysts showed a similar BE for the most intense peak (932.8–933.3 eV). This hinders the identification of Cu species in the samples. However, the differentiation between different valencies of copper can be obtained through the examination of the modified Auger parameter. The modified Auger parameter (αA′ ) is defined by the equation: αA′ = hν + (CuLMM − Cu 2p3/2 ), where αA′ represents the difference between the kinetic energy of the CuLMM Auger electron and the Cu 2p3/2 photoelectron. Addition of the energy of the incident photon allows the modified Auger parameter to be independent of the excitation energy. The αA′ values of the calcined samples at 1852.0–1852.1 eV show the presence of Cu2+ species. After reaction, Cu-Ga/ZnO(HS) and Cu-Zn/SiO2 showed an Auger parameter of 1851.1 and 1851.3 eV, respectively, which are attributable to Cu0 [12]. The post-reaction catalyst Cu-Zn-Ga/SiO2 (HD) gave a modified Auger parameter ranging from 1849.1 to 1850.1 eV, which can be attributed to an intermediary chemical state between Cu0 and Cu2+ or a coexistence of metallic J. Toyir et al. / Applied Catalysis B: Environmental 29 (2001) 207–215 211 Fig. 2. X-ray diffractograms of the catalysts after ca. 60 h in methanol synthesis. (A) Cu-Ga/ZnO(LS); (B) Cu-Ga/ZnO(HS); (C) Cu-Zn/SiO2 ; (D) Cu-Zn-Ga/SiO2 ; (E) Cu-Zn-Ga/SiO2 (HD). (m) CuO; (n) Cu; (h) ZnO. copper and Cu+ , as has been reported in an in-situ surface study on Cu/ZnO-based catalysts [2]. For all the calcined and post-reaction samples, the binding energy of the Zn 2p3/2 peak was between 1021.9 and 1022.5 eV, which indicates that, in all cases, Zn is present as ZnO. The Ga 3d5/2 peak was between 20.7 and 20.8 eV for ZnO-supported catalysts and between 22.0 and 22.5 eV for silica-supported catalysts after reaction. In both cases, the BE corresponds to Ga3+ species. Higher energies are probably due to smaller Table 3 Binding energies (eV) of core electrons, Auger parameter for copper (αA′ ) (eV) and XPS surface atomic ratios of ZnO-supported catalystsa Catalyst Ga 3d5/2 Cu-Ga/ZnO(LS) red, 523 K 20.8 Cu-Ga/ZnO(HS) red, 523 K 20.8 Red, 573 K 20.8 a αA ′ Cu/Zn Ga/Zn – 0.040 – 0.060 1851.2 0.051 0.077 0.081 0.093 XPS analysis has been performed after reaction and in-situ re-reduction at 523 and/or 573 K under pure hydrogen. particles of Ga2 O3 . In addition, it can be pointed out that a shift to higher BE of Ga 3d5/2 was associated with a shift to lower Auger parameters (αA′ ). Table 4 Binding energies (eV) of core electrons, Auger parameter for copper (αA′ ) (eV) and XPS surface atomic ratios of silica-supported catalystsa Catalyst Ga 3d5/2 αA′ Cu-Zn/SiO2 Calcined Red, 523 K Red, 573 K – – – 1852.1 0.0065 0.0286 – 1851.1 0.0049 0.0330 – 1851.3 0.0069 0.0320 – Cu-Zn-Ga/SiO2 Calcined Red, 523 K 21.3 22.0 1852.0 0.0080 0.0153 0.0060 – 0.0045 0.0120 0.0180 Cu-Zn-Ga/SiO2 (HD) Red, 523 K 22.5 Red, 573 K 22.4 1850.1 0.0028 0.0135 0.0153 1849.1 0.0029 0.0112 0.0156 Cu/Si Zn/Si Ga/Si a XPS analysis has been performed on calcined catalysts and after reaction and in-situ re-reduction at 523 and/or 573 K under pure hydrogen. 212 J. Toyir et al. / Applied Catalysis B: Environmental 29 (2001) 207–215 22.0 eV for the post-reaction catalyst). This assesses the concordance between the dispersion of Ga2 O3 and the values of Ga 3d5/2 BE. 3.2. Catalytic activity results Fig. 3. Cu/M (M = Zn, Ga) bulk atomic ratios vs. XPS atomic ratios for the catalysts after reaction. The catalyst Cu-Zn-Ga/SiO2 (HD), which showed the highest BE for Ga 3d5/2 (22.4–22.5 eV), also showed the lowest αA′ (1849.1–1850.1 eV). The surface atomic ratios compiled in Tables 3 and 4 were calculated from peak areas and atomic sensitivity factors [13]. Fig. 3 shows a plot of Cu/Zn and Cu/Ga bulk atomic ratios versus the surface ratios determined by XPS for the catalysts after reaction. For all the catalysts tested, the concentration of surface Zn and Ga was notably higher than that of Cu, with respect to the bulk compositions obtained by elemental analysis. Only the ZnO-supported catalysts have Cu/Zn surface ratios close to Cu/Zn bulk ratios. On the other hand, the surface Cu/Ga ratios were higher in ZnO-supported catalysts than in silica-supported counterparts. Taking into account that the particle size of metallic copper (determined by XRD) in ZnO-supported catalysts is higher than in SiO2 -supported catalysts, the Cu/Ga surface ratios are compatible with an also higher particle size of the Ga2 O3 agglomerates on ZnO-supported catalysts than those on SiO2 counterparts. This agrees with the BE values of Ga 3d5/2 , which, as stated above, were lower for ZnO- than for SiO2 -supported catalysts. For Cu-Zn-Ga/SiO2 catalyst, a redispersion of gallium during the reduction and/or reaction took place. A significant increase in the Ga/Si surface ratio and a decrease in the Cu/Si ratio led to a Ga/Cu surface ratio for the post-reaction catalyst at least five-fold higher than that of the catalyst after calcination. In this case, the BE associated with Ga 3d5/2 increased in about 0.7 eV (from 21.3 for the calcined catalyst to The catalytic performances of the catalysts tested for CO2 hydrogenation to methanol are summarised in Table 5 and Figs. 4 and 5. For gallium-promoted catalysts, the results depended on the support used. Silica-supported catalysts were more active and selective for methanol synthesis than ZnO-supported counterparts. When a mixture of CO2 : H2 = 1 : 3 passed over ZnO-supported catalysts at temperatures between 523 and 543 K at total pressure 2 MPa, the main products detected were methanol and carbon monoxide. Selectivities for CO were between 31 and 54%, depending on the reaction conditions and the catalyst used. Traces of the by-products methane, ethane and ethylene were also detected. Only slight differences between the performance of Cu-Ga/ZnO(HS) and Cu-Ga/ZnO(LS) can be appreciated, the most significative was at the highest temperature (543 K). Initially, the methanol activity for Cu-Ga/ZnO(HS) was higher than that of Cu-Ga/ZnO(LS). However, after 20 h on stream of CO2 /H2 at this temperature, the catalytic behaviour of the two ZnO-supported catalysts was almost the same (see Fig. 4). These results are consistent with the characterisation data presented above. Even if ZnO(HS) showed higher initial surface area than ZnO(LS), no significant differences between the structural and surface characteristics of Cu-Ga/ZnO(HS) and Cu-Ga/ZnO(LS) were found after reaction. The behaviour of silica-supported samples depended on the silica used as support and on the promoters present in the catalyst. Over Cu-Zn-Ga/SiO2 , the hydrogenation of CO2 produced mainly methanol (99.8% at 543 K after 20 h on stream of CO2 /H2 ). CO and dimethyl ether (DME) were also produced at the initial stage of reaction but after some time their selectivities soon fell to zero. The initial selectivity of the catalyst for DME was favoured by high temperature (70.1% at 543 K). It is well known that DME is produced from dehydration of methanol at high temperature [14]. On the other hand, dimethyl ether requires acidic sites and methanol synthesis sites. Here, it seems that, initially, the distribution of sites J. Toyir et al. / Applied Catalysis B: Environmental 29 (2001) 207–215 213 Table 5 Performances of catalysts for CO2 hydrogenation Catalyst Temperature of reaction (K) Methanol activitya (mmol/kg cat. h) ai b Cu-Ga/ZnO(LS) af b Initial and final selectivities (% mol) CH3 OH CO MF HCc DME 523 948 1415 47.8 55.3 51.4 44.3 – – – – 0.8 0.4 533 543 1767 2392 1620 2018d 52.1 50.4 49.0 51.0 47.5 49.0 50.8 48.8 – – – – – – – – 0.4 0.6 0.2 0.2 Cu-Ga/ZnO(HS) 523 533 543 1177 1398 3539 1033 1758 1884d 46.0 52.6 67.0 54.5 56.6 55.5 53.8 47.1 31.0 45.4 41.2 44.4 – – – – – – – – – – – – 0.2 0.3 <0.1 0.1 0.2 0.1 Cu-Zn-Ga/SiO2 523 533 543 1056 3112 3488 2689 2622 4152d 13.3 74.5 29.6 99.7 99.6 99.8 44.0 25.3 – – – – – – – – – – 42.7 – 70.1 – – – <0.1 <0.2 0.3 0.3 0.4 0.2 Cu-Zn-Ga/SiO2 (HD) 523 533 543 6126 8525 10293 7019 7696 10914e 99.8 99.3 99.1 99.8 99.7 99.5 – – – – – – – 0.4 0.5 – <0.1 0.2 – – – – – – 0.2 0.3 0.4 0.2 0.3 0.3 Cu-Zn/SiO2 523 533 543 2048 2594 2132 2072 2125 2042d 65.7 60.7 58.8 55.6 49.3 47.2 34.3 36.4 38.8 44.4 49.3 51.8 – 2.7 2.1 – 1.2 0.8 – – – – – – – 0.2 0.2 – 0.2 0.2 a Reaction conditions: P t = 2 MPa; F /W = 18, 000 l/kg cat. h; CO2 /H2 = 1/3. Initial (ai ) and final (af ) activities (cf. Section 2). c Hydrocarbons detected were methane, ethane and ethylene. d Total CO conversions: around 2.0%. 2 e Total CO conversion: 5.6%. 2 b Fig. 4. Methanol synthesis activity of ZnO-supported catalysts as a function of time on-stream at different temperatures. Square, triangle and circle symbols correspond to reaction temperatures of 523, 533 and 543 K, respectively. Filled symbols: Cu-Ga/ZnO(LS); empty symbols: Cu-Ga/ZnO(HS). Reaction conditions: P t = 2 MPa; F /W = 18, 000 l/kg cat. h. Fig. 5. Methanol synthesis activity of SiO2 -supported catalysts as function of time-on-stream at different reaction temperatures. Square, triangle and circle symbols correspond to reaction temperatures of 523, 533 and 543 K, respectively. Filled symbols: Cu-Zn/SiO2 ; empty symbols: Cu-Zn-Ga/SiO2 ; spotted symbols: Cu-Zn-Ga/SiO2 (HD). Reaction conditions: P t = 2 MPa; F /W = 18, 000 l/kg cat. h. 214 J. Toyir et al. / Applied Catalysis B: Environmental 29 (2001) 207–215 favours the DME production, but the catalyst changed in the reaction conditions and showed another distribution of active sites, more favourable and stable for methanol synthesis. The reverse water gas shift reaction (RWGS), which competes with methanol synthesis, also seems to be limited in this case. Independently of temperature and time-on-stream, traces of methane were produced. Cu-Zn/SiO2 showed lower activity and selectivity for methanol synthesis and much higher selectivity for CO. This indicates that gallium is essential to the catalyst in terms of activity and selectivity in synthesis of methanol. Cu-Zn-Ga/SiO2 (HD) displayed the highest performance among the catalysts tested in terms of activity, selectivity and stability in the methanol production (see Table 5 and Fig. 5). The maximum conversion of the catalyst was obtained at 543 K which is not very far from the equilibrium conversion of 6.2% for the synthesis of methanol from CO2 and H2 at the operating conditions. The experimental conditions used are different from those of a commercial process. However, the Cu-Zn-Ga/SiO2 showed a relatively good performance compared to the usual ternary systems CuO/ZnO/Al2 O3 of methanol synthesis from CO2 . For example, a system CuO(47%)/ZnO(47%)/Al2 O3 (6%) tested by Hattori et al. led to methanol synthesis activity of 293 g MeOH/Kg cat. h [15]. Cu-Zn-Ga/SiO2 (HD) tested in similar conditions showed an activity of about 224 g MeOH/Kg cat. h which is superior than Cu/ZnO/Al2 O3 catalyst in terms of activity per gram of copper. In all the conditions, the selectivity for methanol was above 99%. Methyl formate (MF) was formed as by-product and reached its maximum at 543 K (0.2–0.5%). Negligible amounts of methane were also produced. The thermal stability and the stability over time were highly remarkable. As stated above, this catalyst had similar composition and Cu particle size to those of Cu-Zn-Ga/SiO2 , but the Ga/Cu surface ratio determined by XPS was higher in the Cu-Zn-Ga/SiO2 (HD) than in the Cu-Zn-Ga/SiO2 . In an attempt to understand the influence of the Ga/Cu surface ratio on the production of methanol, a plot of both magnitudes for all the catalysts studied is shown in Fig. 6. Irrespective of the catalyst, at Ga/Cu ratios lower than 2.0, the activity showed no significant difference. However, for higher Ga/Cu ratios, a strong increase in the production of methanol Fig. 6. Methanol synthesis activity after ca. 60 h at 543 K of SiO2 and ZnO-supported catalysts as a function of Ga/Cu atomic XPS ratio. Values between brackets correspond to Ga 3d5/2 BE; values in parentheses correspond to Cu αA′ Auger parameter. Reaction conditions: P t = 2 MPa; F /W = 18, 000 l/kg cat. h. was observed. The promoting effect of gallium oxide seems to be crucial. Characterisation results showed that this promoting effect can be associated with the progressive formation of smaller particles of gallium oxide highly dispersed on the surface. In Fig. 6, the Ga 3d5/2 BE and the Cu Auger parameter are indicated for each catalyst, which can also be related to the catalytic behaviour. Supports like hydrophobic silica allowed the highest dispersion of Ga2 O3 , together with a probably better interaction between well-dispersed ZnO, Ga2 O3 and copper active sites. The role of ZnO in Cu-ZnO/SiO2 catalysts for methanol synthesis from CO2 hydrogenation has already been reported [16–19]. Arakawa et al. described the active species on the catalysts with high selectivity as large metallic Cu particle covered with a partially oxidised phase interacting with highly dispersed ZnO [16]. However, the formation of a well-dispersed zinc species is not sufficient for optimising the surface composition of different species of copper. On the other hand, zinc alone, even well-dispersed, cannot ensure a strong junction between the active species of copper and other components of catalysts. Despite many controversies on the role of copper in methanol synthesis [11,20,21], evidence from this study and other works leads to the conclusion that copper alone cannot be an independent active site for allowing a highly J. Toyir et al. / Applied Catalysis B: Environmental 29 (2001) 207–215 effective methanol synthesis. Electronic interactions and geometrical effects are exerted on copper as well as co-existent components in the catalysts. Chinchen et al. attributed the activity of Cu/ZnO-based catalysts to Cu surface area [21]. Saito et al. found out that higher activities in methanol synthesis over multicomponent Cu/ZnO-based catalyst were obtained when the ratio Cu+ /Cu was around 0.7 [11]. Frost introduced the hypothesis that the productivity of methanol depends on the junction between copper and the oxide and is related to oxygen vacancies [20]. In the case of the catalysts studied in this work, the Cu particle size did not always have a determinant effect. In silica-supported catalysts, gallium changed the chemical state of copper from metallic copper to an intermediary state of valency between 0 and 2. This study also showed that a surface rich in copper is not necessary the best surface for methanol synthesis. The nature of the copper present on the surface of the catalyst and the way in which it interacts with the promoters are essential for optimising catalytic performances. 4. Conclusions It can be pointed out that the promoting effect of Ga2 O3 on supported multicomponent copper-based catalysts for methanol synthesis from CO2 hydrogenation is associated with the Ga2 O3 particle size. High Ga 3d5/2 BE values are related to the presence of small Ga2 O3 particles, which favours the formation of an intermediate state of copper between Cu0 and Cu2+ , probably Cu+ . The silica-supported multicomponent catalysts prepared in this work, especially when hydrophobic silica is used, are effective and very stable in the production of methanol from CO2 at temperatures up to 543 K. Acknowledgements The authors acknowledge the financial support from the Spanish Government, under CICYT 215 project MAT99-0477 and research grant SB97BF0995051. References [1] T. Masami, T. Watanabe, J. Toyir, S. Luo, J. Wu, M. Saito, EP 0864 360 A1 (1998), to Research Institute of Innovative Technology for the Earth (RITE) and Agency of Industrial Science and Technology (AIST). [2] E. Ramaroson, R. Keiffer, A. Kiennemann, Appl. Catal. 4 (1982) 281. [3] T. Tagawa, G. Pleizier, Y. Amenomiya, Appl. Catal. 18 (1985) 285. [4] B. Denise, R.P.A. Sneeden, Appl. Catal. 28 (1986) 235. [5] T. Inui, T. Takeguchi, A. Kohama, K. Tanida, Energy Convers. Manage. 33 (1992) 513. [6] H. Arakawa, J.-L. Dubois, K. Sayama, Energy Convers. Manage. 33 (1992) 521. [7] M. Fujiwara, H. Ando, M. Tanaka, Y. Souma, Bull. Chem. Soc. Jpn. 67 (1994) 546. [8] M. Saito, T. Fujitani, M. Takeuchi, T. Watanabe, Appl. Catal. 138 (1996) 311. [9] M. Saito, M. Takeuchi, T. Watanabe, J. Toyir, S. Luo, J. Wu, Energy Convers. Manage. 38 (1997) 402. [10] J. Wu, S. Luo, J. Toyir, M. Saito, M. Takeuchi, T. Watanabe, Catal. Today 45 (1998) 215. [11] S. Luo, J. Wu, J. Toyir, M. Saito, M. Takeuchi, T. Watanabe, Stud. Surface Sci. Catal. 114 (1998) 549. [12] I. Melian-Cabrera, M. Lopez Granados, P. Terreros, J.L.G. Fierro, Catal. Today 45 (1998) 251. [13] D.A. Briggs, M.P. Seah (Eds.), Practical Surface Analysis by Auger and X-ray Photoelectron Spectroscopy, Wiley, Chichester, 1985. [14] H. Arakawa, Stud. Surface Sci. Catal. 114 (1998) 19. [15] Z. Xu, Z. Qian, L. Mao, K. Tanabe, H. Hattori, Bull. Chem. Soc. Jpn. 64 (1991) 1658. [16] K. Okabe, K. Sayama, N. Matsubayashi, K. Shimomura, H. Arakawa, Bull. Soc. Jpn. 65 (1992) 2520. [17] D.S. Brands, E.K. Poels, A. Blick, Stud. Surface Sci. Catal. 101 (1996) 1085. [18] H. Arakawa, K. Sayama, Stud. Surface Sci. Catal. 75 (1993) 2777. [19] Z. Xu, Z. Qian, H. Hattori, Bull. Chem. Soc. Jpn. 64 (1991) 3432. [20] J.C. Frost, Nature 334 (1988) 577. [21] G.C. Chinchen, P.J. Denny, D.G. Parker, M.S. Spencer, D.A. Whan, Appl. Catal. 96 (1991) 251.