Journal of Catalysis 248 (2007) 143157 www.elsevier.

com/locate/jcat

Cobalt dispersion, reducibility, and surface sites in promoted silica-supported FischerTropsch catalysts
J.-S. Girardon a,1 , E. Quinet a , A. Griboval-Constant a , P.A. Chernavskii b , L. Gengembre a , A.Y. Khodakov a,
a Unit de Catalyse et de Chimie du Solide, UMR 8181 CNRS, Universit des Sciences et Technologies de Lille, Bat C3, Cit scientique,

59655 Villeneuve dAscq, France

b Department of Chemistry, Moscow State University, 119992 Moscow, Russia

Received 12 October 2006; revised 3 March 2007; accepted 5 March 2007 Available online 18 April 2007

Abstract Cobalt particle size, cobalt reducibility, and metal surface sites in a series of ruthenium- and rhenium-promoted cobalt silica-supported Fischer Tropsch catalysts were studied by X-ray diffraction, UVvis spectroscopy, in situ X-ray absorption, in situ magnetic method, X-ray photoelectron spectroscopy, DSC-TGA thermal analysis, and propene chemisorption. The catalysts were prepared by co-impregnation; in several catalyst syntheses, sucrose was added to the impregnating solutions. Mononuclear octahedral cobalt complexes were observed in the catalysts after impregnation and drying. Cobalt repartition on silica in the impregnated and dried catalysts depended primarily on the pH of the impregnating solution. Cobalt repartition was uniform on the silica surface if the pH of the impregnating solution was higher than the point of zero charge (PZC) of silica, but was less uniform at pH below that of the PZC of silica. Cobalt dispersion proceeded during catalyst calcination in air. Decomposition of cobalt nitrate and crystallization of cobalt oxide seemed to be the crucial steps in the preparation of highly dispersed cobalt catalysts. Promotion with noble metals resulted in greater cobalt dispersion, probably due to higher concentrations of cobalt oxide crystallization sites. Addition of sucrose modied the structure of supported cobalt complexes and led to higher temperatures of crystallization of cobalt oxide and to catalysts with extremely high cobalt dispersion. In situ magnetization measurements show that promotion with Ru moderated the temperature of reduction of cobalt oxide to metal phases, whereas the effect was less signicant for Re-promoted catalysts. The addition of sucrose during impregnation, although signicantly enhancing cobalt dispersion, did not diminish cobalt reducibility. Due to a combination of high cobalt dispersion and reducibility, the ruthenium- and rhenium-promoted catalysts prepared using sucrose had the highest number of cobalt metal surface sites. FischerTropsch reaction rates were determined principally by the number of cobalt surface sites, with high cobalt dispersion and easy reducibility resulting in more active FischerTropsch catalysts. 2007 Elsevier Inc. All rights reserved.
Keywords: Clean fuels; FischerTropsch synthesis; Nanoparticles; Catalyst preparation; Point of zero charge (PZC); EXAFS; Cobalt catalysts; Promotion; Dispersion; Reducibility

1. Introduction FischerTropsch (FT) synthesis produces clean fuels from syngas, which can be generated from natural gas, coal, and biomass. Most VIII group metals have measurable activity in
* Corresponding author. Fax: +33 3 20 43 65 61.

E-mail address: andrei.khodakov@univ-lille1.fr (A.Y. Khodakov).

1 Current address: Max-Planck-Institut for Coal Research, Kaiser-Wihelm-

platz 1, 45470 Mlheim an der Rhur, Germany. 0021-9517/$ see front matter 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.jcat.2007.03.002

carbon monoxide hydrogenation but yield different products, including hydrocarbons, alcohols, acids, and esters. Cobaltsupported catalysts are particularly suited for the production of long-chain hydrocarbons [13]. Preparation of cobalt FT catalysts involves several essential steps. The catalysts for FT synthesis are commonly synthesized using aqueous impregnation or co-impregnation with cobalt nitrate and promoters, followed by oxidative and reductive pretreatments. The catalysts are then exposed to syngas in an appropriate reactor to conduct FT reaction.

144

J.-S. Girardon et al. / Journal of Catalysis 248 (2007) 143157

It has been shown [36] that for cobalt particles larger than 78 nm, FT activity is a function of the number of surface metal sites. Thus, cobalt dispersion and reducibility seem to be important parameters affecting the number of cobalt surface sites and thus the overall catalytic performance. Our previous reports [712] have shown that the support texture and catalyst calcination could provide efcient tools for controlling cobalt particle size, reducibility, and catalytic properties of monometallic supported cobalt catalysts. FT catalysts are often promoted with noble metals (e.g., Ru, Pt, Re). Previous work suggests that introduction of noble metals could lead to several phenomena, including much easier reduction of cobalt oxide particles [1215], formation of bimetallic particles and alloys [16,17], a lower fraction of barely reducible mixed oxides, enhanced cobalt dispersion [18], inhibition of catalyst deactivation [19,20], appearance of additional hydrogen activation sites [21], and increased site intrinsic reactivity [22]. Culross and Mauldin showed [2325] that the addition of different chelating molecules during catalyst impregnation could affect deposition of cobalt phases, cobalt dispersion, and the number of cobalt active sites in the reduced catalysts. This paper addresses the impact of different preparation steps, the use of metallic promoters and a chelating agent on cobalt dispersion, reducibility, and number of cobalt surface sites in silica-supported FT catalysts. The catalysts have been studied using a comprehensive set of characterization techniques. The catalytic performance in FT synthesis was measured in a xed-bed microreactor. The catalytic results, as well as the characterization data, are discussed. 2. Experimental 2.1. Catalyst preparation Cobalt catalysts were synthesized via incipient-wetness impregnation or co-impregnation using aqueous solutions of cobalt nitrate and promoters. Cab-o-sil M-5 fumed silica (SBET = 214 m2 /g, Cabot) was used as a catalytic support in all catalyst preparations. Before impregnation, Cab-o-sil M5 was agglomerated by wetting and dried at 373 K. The precursors of noble metal promoters were either a commercial solution of ruthenium nitrosyl nitrate in HNO3 (Ru = 1.5 wt%, Sigma Aldrich) or an aqueous solution of perrhenic acid (HReO4 , SigmaAldrich). In several preparations, the impregnating solutions also contained sucrose with the Co/sucrose molar ratio of 10. The pH of the relevant impregnating solutions is presented in Table 1. The pH of the solutions containing ruthenium was lower than that of the solutions containing rhenium because of the presence of nitric acid in the ruthenium precursor. The contents of cobalt and promoting noble metal (Ru or Re) in the catalysts were respectively 5.58.2 wt% and 0.10.2 wt% (Table 1). The impregnated catalysts were dried in an oven, calcined in a ow of air at different temperatures (373673 K), and then reduced in hydrogen at 673 K for 5 h. The temperature ramp rates during calcination and reduction were 1 and 3.3 K/min, respectively. The cobalt content in the catalysts was measured by atomic absorption at the Service Central

Table 1 Chemical composition and surface area of promoted cobalt silica supported catalysts Catalysts Cobalt content (wt%) 5.5 7.7 6.3 7.7 5.5 6.5 8.2 5.6 8.1 7.0 pH of impregnating solution 0.3 1.3 Co/Ru or Co/Re molar ratio 76 75 73 72 36 102 354 58 111 160 Relative carbon content (nC /nCo ) 1.1 <0.2 1.8 <0.2 BET surface area (m2 /g) 112 191 113 174 118 174 160 124 154 162

CoRu(S) CoRu673(S) CoRe(S) CoRe673(S) CoRu CoRu373 CoRu673 CoRe CoRe373 CoRe673

0.1

0.9

dAnalyse du CNRS, (Vernaison, France). Specic surface areas and porosity of the calcined catalysts were measured by low-temperature nitrogen adsorption using the BET method (Table 1). The catalysts are designated CoMT(S), where M denotes the promoting metal (Ru or Re) and T and (S) denote the temperatures of catalyst calcination and eventual addition of sucrose during impregnation, respectively. The impregnated and dried cobalt catalysts (before calcination at T 373 K) are designated CoM(S). 2.2. X-ray diffraction Powder X-ray diffraction patterns were recorded by a Siemens D5000 diffractometer using Cu(K) radiation. The average crystallite size of Co3 O4 was calculated according to Scherrers equation [26] using a (440) peak at 2 = 65.344. 2.3. UVvis spectroscopy Diffuse reectance UVvis spectra of catalysts and catalyst precursors were obtained at ambient conditions with a VarianCary 4 spectrophotometer using BaSO4 as a reference. 2.4. DSC-TGA Simultaneous differential scanning calorimetry and thermogravimetric analysis were carried out in a ow of air at heating rate of 1 K/min using a DSC-TGA SDT 2960 thermal analyzer. The sample loading was typically 1015 mg. 2.5. X-ray absorption The X-ray absorption spectra at the Co K-edge were measured at the European Synchrotron Radiation Facility (DUBBLE-CRG, Grenoble, France) and the Elettra Synchrotron Light Laboratory (beamline 1.1, Trieste, Italy). In situ decomposition of catalyst precursors and reduction in hydrogen were performed at ESRF using our in situ X-ray absorption cell described elsewhere [27], while ex situ characterization of

J.-S. Girardon et al. / Journal of Catalysis 248 (2007) 143157

145

to air. The pressure in the XPS preparation chamber during the transfer was about 106 Pa. The XPS of cobalt catalysts were systematically compared with the XPS spectra of the reference CoO, Co3 O4 , and amorphous cobalt silicate using relevant literature references. The 2p1/2 2p3/2 spinorbital splittings and intensities of satellite structures were used in addition to the absolute Co2p binding energies for the analysis of XPS data. 2.7. In situ magnetic measurements In situ magnetic measurements were performed using a Foner vibrating-sample magnetometer as described previously [30,31]. Design of magnetometer allows recording curves of magnetization during temperature-programmed heating or under isothermal conditions at 280973 K. Temperature-programmed reduction experiments were carried out in pure hydrogen. The sample amount was about 30 mg in all experiments. The gas ow velocity was 60 mL/min. The appearance of cobalt metallic species in the catalysts was monitored in situ by a continuous increase in sample magnetization during the reduction. For evaluation of cobalt metal particle sizes from magnetic data, the catalyst was rst reduced in a ow of hydrogen at 773 K. Once a constant value of magnetization was attained at that temperature, the catalyst sample was cooled to 473 K. The hydrogen ow was replaced by argon ow at 473 K to desorb hydrogen species from the catalyst, after which the sample was cooled in argon to 280 K. The dependency of magnetization on the intensity of the magnetic eld (eld dependency) was measured at 280 K by scanning the intensity of the magnetic eld to 6.3 kOe. 2.8. Propene chemisorption Recently, [32] we proposed a new method for evaluating the number of cobalt active sites using propene chemisorption. We showed that propene chemisorption was irreversible on cobalt catalysts. The method has been tested for a number of cobalt-supported catalysts [10,12], and good agreement has been found between propene chemisorption and results of other characterization techniques. The measurements could be done in the pulse mode in the same xed-bed reactor used for measuring catalytic performance. In brief, in the experimental procedure used for propene chemisorption, after reduction in pure hydrogen at 673 K for 5 h, the catalyst sample (0.1 g) was cooled and purged with He at 323 K. Then 0.25-mL pulses of propene were introduced to the ow of He. The relative number of metal surface sites was estimated from the amount of chemisorbed propene. No propene chemisorption was observed on pure silica support. Analysis of the reaction products was done by gas chromatography with a packed column containing XOA 400 silica. The pulse experiments were completed when the detector showed no propene chemisorption. This method provides only relative information about cobalt metal sites. Note that no assumption is made about the stoichiometry of propene chemisorption or thus

Fig. 1. XANES spectra of the reference compounds: Co(NO3 )2 6H2 O (1), Co3 O4 (2), CoO (3), -cobalt silicate (4), cobalt foil (5).

calcined catalysts using XANES and EXAFS was done at Elettra. The measurements were performed in transmission mode, with two ionization chambers used for X-ray detection. The Si (111) double-crystal monochromator was calibrated by setting the rst inection point of the K-edge spectrum of Co foil at 7709 eV. Measuring an X-ray absorption spectrum (7600 8400 eV) took about 3040 min. The X-ray absorption data were analyzed using the conventional procedure. The XANES spectra after background correction were normalized by the edge height. After subtracting metal atomic absorption, the extracted EXAFS signal was transformed without phase correction from k space to r space to obtain the radial distribution function (RDF). Crystalline Co3 O4 , CoO, Co foil, and - and -cobalt silicate were used as reference compounds for XANES and EXAFS data analysis. The XANES spectra of the reference compounds are shown in Fig. 1. Additional information about coordination of cobalt in the reference compounds is available elsewhere [9,10,28,29]. 2.6. XPS Surface analyses were performed using the VG ESCALAB 220XL spectrometer. The Alk nonmonochromatized line (1486.6 eV) was used for excitation with a 300 W applied power. The analyzer was operated in a constant pass energy mode (Epass = 40 eV). Binding energies were referenced to the Si2p core level (103.6 eV) of SiO2 support. The reproducibility was 0.2 eV for Co2p binding energy. The vacuum level during the experiments was >107 Pa. The powdered catalyst was pressed as a thin pellet onto a steel block. In situ reduction was carried out in pure hydrogen at 673 K for 5 h in the reactor cell of the preparation chamber attached to the analysis chamber of the spectrometer. The temperature ramp rate was 3.3 K/min, with a hydrogen ow rate of 20 mL/min. The reduced sample was then transferred from the preparation to the analysis chamber of the spectrometer under vacuum without exposure

146

J.-S. Girardon et al. / Journal of Catalysis 248 (2007) 143157

about the absolute number of cobalt metal sites in the reduced cobalt catalysts. 2.9. Catalytic measurements Carbon monoxide hydrogenation was carried out in a xedbed stainless steel tubular microreactor (dint = 9 mm) operating at 463 K under atmospheric pressure. The thermocouple was in direct contact with the catalyst. The reactor design allowed measurement of temperature along the catalyst bed; no hot spot was detected in the catalyst bed during the FT reaction. The carbon monoxide conversion was <10% in all experiments. The catalyst was crushed and sieved to obtain catalyst grains of 0.050.2 mm diameter. The catalyst loading was typically 0.5 g. The samples were reduced in hydrogen ow at 673 K for 5 h. After the reduction, a ow of premixed synthesis gas with molar ratio H2 /CO = 2 was introduced to the catalysts. The carbon monoxide contained 5% nitrogen, which was used as an internal standard for calculating carbon monoxide conversion. To avoid possible condensation of the reaction products, the gas transfer lines were constantly heated at 423 K. Gaseous reaction products were analyzed online by gas chromatography. Analysis of H2 , CO, CO2 , and CH4 was performed using a 13X molecular sieve column and a thermal conductivity detector. Hydrocarbons (C1 C18 ) were separated in a 10% CP-Sil5 on Chromosorb WHP packed column and analyzed by a ameionization detector. The hydrocarbon selectivities were calculated on a carbon basis. The AndersonSchulzFlory (ASF) chain growth probabilities in the C4 C16 hydrocarbon range were calculated from the slope of the curve ln(Sn /n) versus n, where n is the carbon number and Sn is the selectivity to the Cn hydrocarbon. The FT reaction rate is expressed as cobalt-time yield (in mol of converted CO/s divided by the total amount of cobalt [in mol] loaded into the reactor). The total amount of cobalt in the catalysts was determined by elemental analysis. 3. Results 3.1. Impregnation and drying The UVvis spectra of the cobalt nitrate solutions containing Ru and Re promoters with and without the addition of sucrose (not shown) exhibited two broad bands at 470 and 510 nm characteristic of high-spin Co2+ ions in octahedral coordination [3335]. Cobalt octahedral coordination in the impregnating solutions prepared without a promoter (Re or Ru) was previously observed by UVvis spectroscopy [10]. This suggests that adding Ru, Re, or sucrose to the impregnating solutions of cobalt nitrate did not signicantly modify cobalt local coordination. The UVvis spectra of the impregnated and dried samples are shown in Fig. 2. The maxima of major bands in the spectra attributed to cobalt species are located at 520 and 470 nm. These bands indicate that cobalt is predominately in the octahedral environment [3335] in the dried catalysts. The sharp band seen at 295300 nm in the UVvis spectra is related to nitrate ions. More specically, this band is due to the NO ions 3

Fig. 2. UVvisible spectra of impregnated and dried cobalt catalysts.

present in cobalt nitrate complexes and to the NO ions from 3 the commercial solution of ruthenium nitrosyl nitrate diluted in nitric acid used for promotion with ruthenium. For the catalysts prepared with the addition of sucrose, a shoulder also could be seen at 360 nm. This feature cannot be attributed unambiguously using our UVvis data. For the dry CoRu catalyst, the UVvis data suggest some slight decomposition of cobalt nitrate. Cobalt nitrate decomposition results in a broad UVvis band at 730 nm. This band is characteristic of Co3 O4 [34,36]. Slight decomposition of cobalt nitrate after impregnation and drying of the unpromoted silica-supported catalysts was also observed in our previous study [10]. The XANES and EXAFS results for impregnated and dried catalysts (Figs. 3 and 4) are consistent with UVvis data. The XANES of both impregnated and dried CoRu and CoRu(S) catalysts are almost identical to that of cobalt nitrate (Fig. 3, curves Troom , air). The EXAFS Fourier transform moduli (Fig. 4, curves Troom , air) show an intense peak at 1.6 , which is related to a CoO coordination shell in cobalt nitrate. No cobalt atoms can be seen in the second cobalt coordination shell. The XPS spectra of all impregnated and dried cobalt catalysts (Fig. 5) exhibit an intense satellite structure [37,38]. High-energy XPS Co2p3/2 peaks (782.3 eV) indicate the presence of Co2+ ions in the impregnated and dried catalysts. The ICo /ISi ratio of the XPS intensities provides information about the repartition of cobalt ions on silica surface after drying. Fig. 6 represents the relationship between the ICo /ISi ratio and pH of the impregnating solutions. Higher ICo /ISi ratios are characteristic of higher cobalt dispersion in the dried cobalt catalysts; they are observed when the catalyst is impregnated by solutions with high pH. This effect could be interpreted in terms of

J.-S. Girardon et al. / Journal of Catalysis 248 (2007) 143157

147

Fig. 3. In situ XANES spectra measured for CoRu catalysts prepared without (a) and with (b) sucrose addition under ow of air and hydrogen at different temperatures. The spectra are offset for clarity. The XANES of reference Co3 O4 is displayed in (a).

Fig. 4. In situ EXAFS Fourier transform moduli measured for CoRu catalysts prepared without (a) and with (b) sucrose addition under ow of air and hydrogen at different temperatures. The moduli are offset for clarity. The EXAFS Fourier transform modulus of reference Co3 O4 is displayed in (a).

electrostatic attraction or repulsion between Co2+ ions and the silica surface. At pH higher than that of the PZC (pH 3 [39]), the silica surface is negatively charged and there is a certain attraction between cobalt ions and the surface. XPS shows uniform repartition of cobalt ions on the surface at these conditions. At pH below that of the silica PZC, the surface is charged positively and Co2+ ions are repulsed from the silica surface. This results in nonuniform cobalt repartition and lower cobalt dispersion. Thus, cobalt dispersion in the impregnated and dried cobalt catalysts is signicantly affected by the pH of the impregnating solution. Table 1 shows that catalyst impregnation results in some decrease in BET surface areas of the samples (from 214 to 110125 m2 /g in Cab-o-sil). BET surface areas increase (from 110125 to 150190 m2 /g) after calcination. Catalyst calcination results in decomposition of cobalt precursor and sucrose. The increase in the apparent surface area after calcination suggests [40] that cobalt nitrate and eventually sucrose introduced on impregnation were preferentially located inside the silica pores.

Fig. 5. Co2p XPS spectra of impregnated and dried catalysts.

148

J.-S. Girardon et al. / Journal of Catalysis 248 (2007) 143157

3.2. Decomposition of cobalt precursors in the presence of noble metals and sucrose The DSC-TGA curves of the catalysts prepared from cobalt nitrate promoted with ruthenium and rhenium and with or without the addition of sucrose are shown in Fig. 7. All of the DSC-TGA curves were measured in air ow. The decomposition proles for the CoRu and CoRe catalysts are similar and resemble the DSC-TGA curves of unpromoted cobalt silicasupported catalysts [10]. Two endothermic weight losses (6.7 7.6%) are observed at 323328 and 363364 K. These losses can be attributed to the dehydration of cobalt nitrate and silica. An additional weight loss (10.411.7%) was detected at 431432 K, attributed to the decomposition of NO groups. As 3 with monometallic cobalt catalysts [10], for all promoted catalysts, the total experimental weight loss is much smaller that the theoretical one calculated from decomposition of supported bulk cobalt nitrate to Co3 O4 (29%). This difference can be explained in terms of lower extent of hydration and partial decomposition of cobalt nitrate before calcination (during catalyst

Fig. 6. ICo /ISi ratio measured by XPS in the impregnated and dried cobalt catalysts versus pH of the impregnating solutions. Co1.6, Co2.6, and Co samples were obtained using impregnation of silica with solutions of cobalt nitrate; pH of these solutions was adjusted respectively to 1.6, 2.6, and 4.3 using nitric acid.

Fig. 7. DSC-TGA curves of cobalt silica-supported catalysts promoted with Re and Ru and prepared without and with addition of sucrose. Temperature ramp 1 K/min.

J.-S. Girardon et al. / Journal of Catalysis 248 (2007) 143157

149

Fig. 7. (continued)

impregnation and drying). UVvis spectroscopy (Fig. 2) conrms partial decomposition of cobalt nitrate in the impregnated and dried CoRu catalysts. The addition of sucrose to the impregnating solution modies the DSC-TGA curves signicantly. The decomposition of cobalt precursors in CoRu(S) and CoRe(S) samples turns from endothermic into exothermic (Figs. 7b and 7d). The relevant exothermic peaks are situated at slightly higher temperatures than the endothermic peak of decomposition of cobalt nitrate in the catalysts prepared without sucrose. This suggests that in CoRu(S) and CoRe(S) decomposition of supported cobalt complexes proceeds at higher temperatures than decomposition of cobalt nitrate in both promoted and unpromoted catalysts prepared without sucrose. The in situ X-ray absorption data conrm the DSC-TGA results. Figs. 3 and 4 show that calcination of CoRu catalyst at 423 K results in decomposition of cobalt nitrate. The XANES and EXAFS Fourier transform modulus of CoRu catalysts after heating in air at 423 K are almost identical to those of Co3 O4 , whereas calcination of the CoRu(S) sample even at 473 K does not result in any noticeable modication of the Co adsorption edge and EXAFS Fourier transform modulus. This supposes

higher temperatures of decomposition of cobalt complexes and crystallization of Co3 O4 phase in the catalysts prepared with sucrose addition than in the catalysts prepared without sucrose. In other words, sucrose addition to cobalt silica-supported FT catalysts seems to alter the mechanism of decomposition of cobalt nitrate and to affect nucleation and crystallization of the resulting cobalt oxide species. 3.3. Calcined catalysts The chemical compositions of the catalysts calcined at 373 K (100 h) and 673 K (5 h) under air are presented in Table 1. Calcination of the catalysts prepared without sucrose leads to a signicantly decreased ruthenium content. The Co/Ru molar ratio is multiplied by 3 after calcination at 373 K and by 10 after calcination at 673 K. These data seem to conrm earlier results on the sublimation of ruthenium under oxidizing atmosphere in supported catalysts [41]. The rhenium content varies less signicantly after calcinations at 373 and 673 K (Table 1). The presence of carbon atoms is detected in several catalysts prepared with sucrose addition. The presence of carbon atoms in the dried and calcined catalysts indicates some

150

J.-S. Girardon et al. / Journal of Catalysis 248 (2007) 143157

Fig. 9. XANES spectra of calcined cobalt silica-supported catalysts. Fig. 8. UVvisible spectra of calcined cobalt silica supported catalysts.

concentration of sucrose and other organic compounds, which can originate from sucrose. The nC /nCo atomic ratios detected in dried CoRu(S) and CoRe(S) (Table 1) are relatively close to those expected from the chemical composition of impregnating solutions. Indeed, the molar ratio of nCo /nSucrose (C12 H22 O11 ) = 10 in the impregnating solutions gives a theoretic atomic ratio of nC /nCo = 1.2 in the dried catalysts. Calcination at 673 K results in a signicantly decreased concentration of carbon (nC /nCo < 0.2), suggesting that the sucrose and sucrose-originated organic compounds disappear during calcination at 673 K. In contrast to the catalysts prepared without sucrose, in sucrose-promoted samples, the concentrations of Ru and Re do not drop after calcination. The Co/Ru and Co/Re ratios remain almost unchanged (Table 1). This suggests that sucrose helps maintain a high ruthenium content and prevents sublimation of Ru from the surface of silica-supported catalysts. The structure of cobalt species in the calcined catalysts was characterized by UVvis spectroscopy, XAS, XPS, and XRD. The relevant UVvis spectra are shown in Fig. 8. All of the catalysts except CoRu373(S) and CoRe373(S) exhibit two broad absorption bands at 410 nm and 710730 nm, which are related to Co3 O4 species [34,36]. This means that the decomposition of cobalt nitrate under air leads primarily to Co3 O4 . CoRu373(S) and CoRe373(S) maintain a sharp pink color. Note that these catalysts were calcined in air at 373 K for 100 h under exactly the same conditions as CoRu373 and CoRe373. The relevant UVvis spectra also remain unchanged for these samples. This suggests that in contrast with the samples prepared without sucrose, cobalt oxides (Co3 O4 or CoO) are not generated after calcination of CoRu(S) and CoRe(S) catalysts

at 373 K. Thus, local coordination of cobalt in CoRu373(S) and CoRe373(S) remains similar to that in the impregnated and dried catalysts. The calcined catalysts were analyzed by X-ray absorption spectroscopy at cobalt K absorption edge. The XANES spectra of the catalysts are presented Fig. 9. Whereas the XANES spectra of CoRu373(S) and CoRe373(S) (not shown) are almost identical to the XANES spectra of cobalt nitrate, the XANES spectra for the CoRu673, CoRe673, CoRu673(S), and CoRe673(S) catalysts suggest the presence of Co3 O4 . With the CoRe373 and CoRu373 catalysts, XANES is more complex and probably suggests the presence of several cobalt species. Fitting the XANES of CoRe373 and CoRu373 using linear combination of XANES spectra of reference compounds is indicative of both cobalt nitrate and Co3 O4 . The contributions of Co3 O4 and cobalt nitrate to the XANES spectrum of CoRe373 are estimated as 27% and 73%, respectively. These contributions are equal to 44 and 56% for the XANES spectrum of CoRu373. Thus, the XANES data suggest that decomposition of cobalt nitrate in both CoRu373 and CoRe373 is not complete after calcination at 373 K. The extent of decomposition of cobalt nitrate is higher in CoRu373 than in CoRe373, however. The surface analysis of the calcined catalysts was carried out by XPS (Fig. 10). For CoRu673, CoRe673, CoRu673(S), and CoRe673(S), the spectra show a Co2p3/2 binding energy at 780.1780.8 eV and a Co2p1/2 binding energy at 795.1 795.8 eV. Very weak shake-up satellites are observed at approximately 789.5 and 804.5 eV. The binding energies and low intensity of satellites suggest the presence of Co3 O4 on the catalyst surface [38,42,43]. In CoRe373 and CoRu373, which were precalcined at low temperatures, the Co2p3/2 and Co2p1/2 binding energies are 781.8782.3 eV and 797.4798.1 eV, re-

J.-S. Girardon et al. / Journal of Catalysis 248 (2007) 143157

151

Table 2 Characterization of cobalt silica-supported catalysts and their catalytic performance in FT synthesis Catalysts ICo2p /ISi2p ratio in calcined catalysts (XPS) 1.24 1.11 1.60 0.69 1.70 0.76 Size of Co3 O4 cryst., XRD (nm) 6 5 9 12 9 13 Fraction of metal cobalt in reduced catalysts, XPS (%) 58 43 31 54 24 45 Propene chemisorption (mol/g) Cobalt time yield (104 s1 ) SCH4 (%) SC5+ (%) Alpha

CoRu673(S) CoRe673(S) CoRu373 CoRu673 CoRe373 CoRe673

93 76 60.0 50.1 37.6 35

10.1 9.9 8.6 6.1 6.8 4.1

12 13 15 15 11 12

67 67 62 72 71 69

0.76 0.74 0.75 0.82 0.80 0.77

The FT reaction rate is expressed as cobalt-time yield (in moles of converted CO per second divided by the total amount of cobalt (in moles) loaded into the reactor).

Fig. 10. Co2p XPS spectra of calcined cobalt silica-supported catalysts.

Fig. 11. Magnetization of cobalt catalysts promoted with Ru (a) and Re (b) measured during in situ reduction in pure hydrogen. Temperature ramp 28.2 K/min.

spectively. An intense satellite structure is also observed at 787 and 804 eV. All of these XPS features are characteristic of Co2+ ions [37,38,44]. These results are consistent with XANES and EXAFS data, which show almost complete decomposition of cobalt nitrate to Co3 O4 in CoRu673, CoRe673, CoRu673(S), and CoRe673(S) and partial decomposition and the presence of residual cobalt nitrate species in CoRu373 and CoRe373. Co3 O4 crystalline phase is detected in all calcined catalysts by XRD. The sizes of Co3 O4 particles, calculated according to Scherrers equation, are shown in Table 2. Co3 O4 crystallites of CoRe673 and CoRu673 have an average size of 1213 nm. The Co3 O4 particles were slightly smaller (9 nm) in CoRe373 and CoRu373. The addition of sucrose during impregnation results in a further decrease in the size of Co3 O4 crystal-

lites. The Co3 O4 average particle size of the CoRu673(S) and CoRe673(S) calculated from XRD patterns drops to 56 nm. The XRD results are consistent with ICo /ISi ratios in the calcined catalysts measured by XPS (Table 2). Higher ICo /ISi ratios and correspondingly high cobalt dispersion are observed in the catalysts prepared using sucrose addition. 3.4. Reduction of cobalt catalysts The results of in situ temperature-programmed magnetization measurements in pure hydrogen for promoted cobalt silicasupported catalysts are shown in Fig. 11. A magnetization curve for unpromoted cobalt silica-supported catalyst precalcined at

152

J.-S. Girardon et al. / Journal of Catalysis 248 (2007) 143157

673 K (Co673 [10]) is shown in Fig. 11a for comparison. Note that only cobalt metallic particles have noticeable magnetization under these conditions. Thus, the increase in magnetization directly indicates the growth of cobalt metallic phases in the catalysts. Fig. 11 shows that promotion with Re and Ru results in a signicant decrease in the temperature of emergence of the cobalt metal phase. This is consistent with previous reports indicating that promotion with noble metals results in signicant decrease in the catalyst reduction temperature [1315]. The presence of Ru seems to affect catalyst reduction to a much greater extent than promotion with Re. The metallic particles emerge at 451 K for CoRu373, at 473 K for CoRu673 and CoRe373, and at 523 K for CoRe673. Interestingly, sucrose addition during catalyst impregnation affects cobalt reducibility only slightly. The magnetic data on cobalt reducibility in the presence of sucrose are consistent with XANES and EXAFS ndings. CoRu(S) and CoRu catalysts precalcined and then reduced in hydrogen at 673 K have almost identical XANES spectra and EXAFS Fourier transform moduli (Figs. 3 and 4); both are characteristic of metallic cobalt. Thus, X-ray absorption data for these catalysts indicate a high extent of cobalt reduction. An important observation is that addition of sucrose during impregnation, although dramatically enhancing cobalt dispersion, does not diminish cobalt reducibility (Table 2). The magnetic eld dependence curves (magnetization vs intensity of the magnetic eld) for CoRe673 and CoRe673(S) catalysts are shown in Fig. 12. This curve for CoRe673 exhibits a hysteresis loop, indicating the presence of cobalt metallic ferromagnetic particles (>7 nm diameter), whereas the eld dependence curve for CoRe673(S) shows no hysteresis. The absence of a hysteresis loop is characteristic of cobalt superparamagnetic particles, the diameter of which is <7 nm at ambient temperatures [30,45,46]. Similar to CoRe673(S), only small cobalt superparamagnetic particles are detected in the CoRu673(S) sample. Fitting the eld dependence curves using the Langevin function yields cobalt metal particle sizes of 4 nm in CoRe673(S) and 6 nm in CoRu673(S) [31]. Note that the cobalt metal particle sizes calculated from the magnetic measurements are consistent with XRD, taking into account some decrease in molar volume after reduction of Co3 O4 to metallic cobalt. The reduced catalysts were also studied by XPS (Fig. 13). For XPS measurements, the catalysts were reduced for 5 h at 673 K (ramp rate, 3.3 K/min; hydrogen ow rate, 20 mL/min) and then transferred to the XPS vacuum chamber for measurements without exposure to air. Both cobalt metallic-phase and oxidized Co2+ species can be observed in the XPS spectra. The Co2p3/2 peak at 778779 eV and the Co2p1/2 peak at 793 794 eV are characteristic of cobalt metal species [47]. The presence of oxidized Co2+ species can be identied by XPS peaks at 782 and 797 eV and an intense satellite structure [37,38]. The fractions of Co2+ ions and cobalt metal species in the reduced catalysts were evaluated using decomposition of XPS spectra as described previously [10]. The corresponding extents of reduction are presented in Table 2. The extent of reduction is slightly higher in the catalysts promoted with Ru prepared with-

out and with sucrose addition compared with the Re-promoted catalysts. Propene chemisorption data, presented in Table 2, are consistent with the results on cobalt dispersion and reducibility obtained by other characterization techniques. Higher numbers of cobalt surface sites were detected in CoRu673(S) and CoRe673(S), which were prepared with the addition of sucrose. These catalysts have high cobalt dispersion along with good cobalt reducibility. Lower concentrations of cobalt surface sites are seen in CoRe673 and CoRe373. The decreased number of cobalt surface sites is probably due principally to lower cobalt dispersion in CoRe673 and to poor cobalt reducibility in CoRe373. 3.5. Catalytic performance The catalytic performance of cobalt silica-supported catalysts was evaluated in a differential catalytic reactor at 463 K under atmospheric pressure. To discount any transient behavior of the xed-bed reactor, the conversion and selectivities were measured after 24 h on stream. The results of catalyst evaluation, summarized in Table 2, show that FT reaction rates (cobalt-time yields) follow the same trend as the number of cobalt surface sites measured by propene chemisorption. The FT rates were much higher with CoRu673(S) and CoRe673(S). CH4 selectivity was 1115%, whereas C5+ selectivity was 61 72%. The apparent AndersonSchulzFlory parameter () calculated for C4 C16 hydrocarbons was 0.740.82. 4. Discussion 4.1. Repartition and structure of cobalt species in the impregnated and dried catalysts Preparation of cobalt-supported FT catalysts involves several important stages. Impregnation is one of the initial important stages in catalyst preparation. Our results suggest that impregnation of silica with cobalt nitrate and promoters leads to the formation of mostly cobalt mononuclear octahedrally coordinated cobalt complexes. Fourier transform modulus of EXAFS of impregnated and dried catalysts (Fig. 4) reveals no cobalt atoms in the second cobalt coordination sphere. UVvis spectra of impregnated and dried catalysts also conrm octahedral coordination of cobalt ions (Fig. 2). The repartitioning of Co2+ ions after impregnation and drying depends primarily on the pH of the impregnating solution. Contact between the silica and impregnating solution results in formation of an electric double layer. For silica, the PZC occurs at pH 23 [39]. When impregnation occurs at a pH above the PZC, the surface of silica is negatively charged. The negatively charged silica surface attracts the positively charged cobalt ions. Below this PZC value, the surface of silica is positively charged, which involves an electrostatic repulsion of the positively cobalt-charged ions. Consequently, below the PZC, the silica support attracts negatively charged nitrate and hydroxyl groups and repulses the cobalt complexes. This results

J.-S. Girardon et al. / Journal of Catalysis 248 (2007) 143157

153

Fig. 12. Field dependences measured for reduced CoRe673 (a) and CoRe673(S) (b) cobalt catalysts.

in nonuniform Co2+ repartition on the silica surface in the impregnated and dried catalysts. XPS data (Fig. 6) show higher ICo /ISi ratios at higher pH of the impregnating solutions. This suggests more uniform distribution of cobalt precursor on the catalyst surface at pH above the PZC of silica. The addition of sucrose does not affect cobalt repartition on the silica surface. Thus, the pH of the impregnating solution seems to be a major parameter controlling cobalt dispersion in the impregnated and dried catalysts. Adsorption of cobalt species on the surface of silica and other oxides has been recently reviewed by Bourikas et al. [48], who reported that an increase in the impregnation pH (up to a critical value above which cobalt phases precipitate) brings about an increased extent of deposition and thus the Co2+ surface concentration. In fact, cobalt adsorption on the silica surface is maximal at pH 7.38.0. These results are consistent with earlier data of Ming et al. [49], who predicted that Co2+ deposited on the surface as a cation at pH < 5. At pH < 2, cobalt dispersion could be low due to repulsion between the positively charged silica surface

and Co2+ ions. Ming et al. [49] also suggested that cobalt repartition on silica immediately after impregnation could affect the catalytic performance of the nal FT catalysts. However, our results demonstrate that repartition of cobalt in the impregnated and dried catalysts is very volatile and evolves signicantly during further catalyst preparation steps. 4.2. Cobalt species in the catalysts after decomposition of cobalt precursor and calcination After impregnation and drying, cobalt ions seem to be anchored only weakly to the silica surface. Cobalt repartition changes dramatically after decomposition of cobalt precursors. Indeed, in the present work, co-impregnation involves a solution of ruthenium nitrosyl nitrate and perrhenic acid. Adding promoters to the impregnating solution reduces the pH to values below the pH of silica PZC. Due to the repulsion between Co2+ ions and positively charged silica surface, cobalt repartition is nonuniform in the impregnated and dried catalysts. Thus,

154

J.-S. Girardon et al. / Journal of Catalysis 248 (2007) 143157

Fig. 13. Co2p XPS spectra of cobalt catalysts reduced in situ in hydrogen at 673 K for 5 h.

silica impregnation with acidic solutions of cobalt and promoters is expected to yield less dispersed cobalt species. However, XRD of calcined catalysts shows that Co3 O4 crystallites are smaller after calcination of cobalt catalysts prepared by co-impregnation with more acid solutions containing Ru, Re, and sucrose (Table 2). This suggests that the genesis of cobalt oxide crystallites essentially occurs at the stage of decomposition of cobalt precursor and catalyst calcination, but not at the stage of impregnation and drying. Our results also show that cobalt dispersion in silicasupported catalysts is affected by both the temperature of decomposition of cobalt precursors and catalyst promotion. The temperature of cobalt decomposition affects the size of cobalt oxide crystallites. In agreement with previous results for unpromoted cobalt catalysts [10], smaller Co3 O4 crystallites are observed when the catalysts are pretreated at 373 K instead of 673 K (Table 2). Decomposition of cobalt precursor and cobalt dispersion are also strongly affected by promotion. Decomposition of cobalt nitrate in the presence of noble metals leads to smaller crystallites of cobalt oxide than those obtained in unpromoted silicasupported catalysts ([10], Table 2). Similar results were obtained earlier by Schanke et al. [18] with Pt-promoted silicasupported catalysts. One reason for the higher cobalt dispersion in the catalysts promoted with noble metals could be related to a higher concentration of cobalt oxide crystallization sites, which at similar cobalt content would result in an increased number of cobalt particles and, consequently, higher cobalt dispersion.

The addition of sucrose during co-impregnation enhances cobalt dispersion considerably. This nding is consistent with earlier patents by Mauldin et al. [24,25], who claimed that for titania-supported cobalt catalysts, impregnation with cobalt nitrate solutions containing monosaccharide or disaccharide could improve cobalt dispersion without any negative effect on cobalt reducibility. DSC-TGA (Fig. 7) and X-ray absorption measurements (Figs. 3 and 4) show that decomposition of cobalt precursor proceeds at much higher temperatures in the catalysts prepared using sucrose. A high temperature of decomposition probably suggests the presence of cobalt complexes, which differ from those in the catalysts prepared without the addition of sucrose. It is known that in the presence of nitric ions, sucrose is oxidized to saccharic acid, which can polymerize on silica surface into polysaccharic acid, an excellent chelating agent. The synthesis of supported catalysts using metal chelate complexes has been reviewed by van Dillen et al. [50]. A gel-like phase formed after decomposition of organic precursor generally favors high metal dispersion. Highly dispersed Co3 O4 crystallites are probably produced after decomposition of complexes of polysaccharic acid. Thus, as with other promotors, the addition of sucrose affects the mechanism of cobalt precursor decomposition and thus the genesis of Co3 O4 in silica-supported catalysts. The presence of sucrose also reduces sublimation of ruthenium suboxides. Table 2 shows that calcination of silica Rupromoted catalysts prepared without sucrose results in a significant loss of ruthenium. The presence of sucrose prevents the formation of volatile ruthenium suboxide and maintains a high Ru content in the calcined catalysts. Previously [10], we found that high exothermicity of cobalt acetate decomposition results in signicant concentrations of amorphous cobalt silicate. Note that there are some important differences in the decomposition of cobalt acetate and cobalt complexes with sucrose. TGA-DSC curves [10] show that decomposition of cobalt acetate is an autocatalytic combustion and proceeds very fast in a narrow temperature range (473 510 K). Decomposition of cobalt complexes with sucrose involves several steps (e.g., saccharic, polysaccharic acids, resin) and proceeds in a much wider temperature range (403495 K; Fig. 7). Multistep decomposition of cobalt-sucrose complexes leads to much lower heat ow and thus less signicant formation of amorphous cobalt silicate species. 4.3. Cobalt species in the reduced cobalt catalysts and their catalytic performance Reduction of cobalt species in silica-supported catalysts is an important step in catalyst preparation, because catalyst reduction generates active cobalt metal sites for FT synthesis. Reducibility of silica-supported FT catalysts is a function of the fraction of Co3 O4 crystalline phase, sizes of Co3 O4 crystallites, and promotion with noble metals. In agreement with several previous reports [1315], the results of in situ magnetic measurements (Fig. 11) show much easier reducibility of cobalt oxides in the presence of noble metals. Cobalt particle size is also an essential parameter affecting

J.-S. Girardon et al. / Journal of Catalysis 248 (2007) 143157

155

cobalt reducibility. Cobalt catalysts with larger cobalt particles (CoRu673 and CoRe673) have a higher extent of cobalt reduction (Table 2). Promotion with Ru has a greater impact on cobalt reducibility than promotion with Re. Re can be reduced to metallic state at higher temperature than Ru. It is known that reduction of Co3 O4 to metallic cobalt proceeds via intermediate formation of CoO. Thus, the promoting effect of Re on cobalt reducibility can be observed only at higher temperatures than the promoting effect of Ru [5153]. The addition of sucrose during impregnation leads to extremely high dispersion of supported cobalt oxide. Reduction is usually more difcult for smaller cobalt particles than larger ones [29]; thus, the small cobalt particles in the samples prepared with sucrose might be expected to have decreased reducibility in hydrogen. However, as shown in Table 2, the addition of sucrose does not lead to poor reducibility. Thus, it can be suggested that the genesis of Co3 O4 particles in the catalysts prepared with the addition of sucrose likely proceeds via intermediate formation of chelate complexes of cobalt and a noble metal. Strong interaction between Co and noble metal promoters in such complexes and possible formation of bimetallic metal oxide particles could be among the reasons for the relatively high cobalt reducibility in the sucrose-promoted catalysts. The reduction of silica-supported cobalt catalysts promoted with sucrose leads to superparamagnetic cobalt particles, which are identied by the magnetic method. These superparamagnetic cobalt particles were smaller than 7 nm. Propene chemisorption shows that the number of cobalt surface sites in the reduced catalysts depends on both cobalt dispersion and reducibility (see Table 2). As expected, higher numbers of cobalt metal sites are detected in ruthenium- and rhenium-promoted samples prepared with the addition of sucrose. Note that these catalysts have both highly dispersed and easily reducible cobalt species. There is controversy in the literature regarding the effect of catalyst preparation parameters on cobalt dispersion. Several authors have suggested that cobalt dispersion is affected

principally by the method of cobalt deposition on the catalytic support. Ming et al. [49] observed that smaller cobalt oxide particles supported on silica could be obtained at higher pH of impregnating solutions. A correlation between the particle size of cobalt oxide and the pH of the impregnating solution of cobalt nitrate also was observed in titania-supported catalysts [54]. Lok et al. [5557] obtained highly dispersed cobalt catalysts by optimizing the deposition of cobalt precursors on the support surface. In all of these works, the pH of the impregnating solutions was adjusted by adding various organic and inorganic compounds. However, the effect of the added compounds on decomposition of cobalt precursors and catalyst calcination was often underestimated in these reports. Drying also can affect cobalt repartition in the catalyst grain. It has been shown that solvent with higher viscosity can prevent migration of active phase during catalyst drying and enhance cobalt dispersion [50,58]. Several reports have emphasized the role of oxidative pretreatments in the design of metal catalysts. Joyner et al. [5962] showed that the temperature of calcination has a signicant affect on the Pt metal particle size in zeolites. De Jong et al. [63] found that a very low heating rate during calcination was essential in the preparation of highly dispersed Pt particles in Y zeolite. Our earlier report [10] and current results extend this approach to the preparation of cobalt silicasupported catalysts for FT synthesis. We nd that cobalt dispersion in silica-supported FT catalysts depends principally on the conditions of cobalt precursor decomposition and catalyst calcination. The catalysts with high number of cobalt surface sites display higher FT reaction rates (see Table 2). Fig. 14 shows the relationship between the number of cobalt metal sites and cobalttime yield for the series of unpromoted and promoted cobalt silica-supported catalysts. A higher concentration of cobalt surface sites leads to higher FT reaction rates, while affecting hydrocarbon and methane selectivities only slightly. Note that the catalytic results were obtained in a differential reactor under rel-

Fig. 14. Cobalt time yield versus propene chemisorption on different cobalt catalysts. Note: Co373 and Co673 are monometallic cobalt silica-supported catalysts precalcined respectively at 373 and 673 K. Additional information about the monometallic cobalt catalysts is available in Ref. [10].

156

J.-S. Girardon et al. / Journal of Catalysis 248 (2007) 143157

atively mild conditions, when catalyst deactivation is not very signicant. 5. Conclusion Cobalt repartition in silica-supported impregnated and dried catalysts is a function primarily of the pH of the impregnating solution; a higher pH results in more uniform cobalt repartition on the silica surface. Although all preparation steps (i.e., impregnation, drying, calcination, and reduction) are signicant in the design of efcient cobalt FT catalysts, cobalt dispersion is affected primarily by cobalt precursor decomposition and catalyst calcination. Both a lower temperature of catalyst calcination and promotion with ruthenium, rhenium, and sucrose lead to smaller cobalt particles. Promotion with Re and particularly with Ru enhances cobalt reducibility. Extremely small, easily reducible cobalt particles are seen in the cobalt catalysts prepared using sucrose addition. The overall number of cobalt surface sites in the reduced catalysts is a function of both cobalt dispersion and reducibility. The FT reaction rates are strongly affected by the number of cobalt surface sites. In differential reactor and at atmospheric pressure, FT reaction rates vary as a function of cobalt surface sites, with hydrocarbon selectivities affected by cobalt dispersion only slightly. Acknowledgments The authors thank L. Burylo, M. Frre, O. Gardoll, and G. Cambien for the X-ray diffraction, XPS, DSC-TGA, and BET surface measurements, respectively. The help of V. Briois, S. Nikitenko, and G.V. Pankina during X-ray absorption and magnetic measurements is particularly appreciated. ESRF and Elettra are acknowledged for the use of synchrotron beamtime. P.A.C. acknowledges nancial support from the Russian Foundation for Fundamental Research (Grant 06-03-32500-a). References
[1] A. Steynberg, M. Dry (Eds.), FischerTropsch Technology, in: Studies in Surface Sciences and Catalysis, vol. 152, 2004. [2] B.H. Davis, Top. Catal. 32 (2005) 143. [3] S.L. Soled, E. Iglesia, R.A. Fiato, J.E. Baumgartner, H. Vroman, S. Miseo, Top. Catal. 26 (2003) 101. [4] E. Iglesia, S.C. Reyes, R.J. Madon, S.L. Soled, Adv. Catal. 39 (1993) 221. [5] E. Iglesia, Appl. Catal. A 161 (1997) 59. [6] G.L. Bezemer, J.H. Bitter, H.P.C.E. Kuipers, H. Oosterbeek, J.E. Holewijn, X. Xu, F. Kapteijn, A.J. van Dillen, K.P. de Jong, J. Am. Chem. Soc. 128 (2006) 3956. [7] A.Y. Khodakov, R. Bechara, A. Griboval-Constant, Appl. Catal. A 254 (2003) 273. [8] A.Y. Khodakov, A. Griboval-Constant, R. Bechara, V.L. Zholobenko, J. Catal. 206 (2002) 230. [9] A.Y. Khodakov, A. Griboval-Constant, R. Bechara, F. Villain, J. Phys. Chem. B 105 (2001) 9805. [10] J.-S. Girardon, A.S. Lermontov, L. Gengembre, P.A. Chernavskii, A. Griboval-Constant, A.Y. Khodakov, J. Catal. 230 (2005) 339. [11] A.Y. Khodakov, J.-S. Girardon, A. Griboval-Constant, A.S. Lermontov, P.A. Chernavskii, Stud. Surf. Sci. Catal. 147 (2004) 295.

[12] J.-S. Girardon, A. Constant-Griboval, L. Gengembre, P.A. Chernavskii, A.Y. Khodakov, Catal. Today 106 (2005) 161. [13] L. Guczi, D. Bazin, I. Kovacs, L. Borko, Z. Schay, J. Lynch, P. Parent, C. Lafon, G. Steer, Z. Koppany, I. Sajo, Top. Catal. 20 (2002) 129. [14] G. Jacobs, T.K. Das, P.M. Patterson, J. Li, L. Sanchez, B.H. Davis, Appl. Catal. A 247 (2003) 335. [15] A. Kogelbauer, J.G. Goodwin Jr., R. Oukaci, J. Catal. 160 (1996) 125. [16] F.B. Noronha, M. Schmal, R. Frty, G. Bergeret, B. Moraweck, J. Catal. 186 (1999) 20. [17] B. Mierzwa, Z. Kaszkur, B. Moraweck, J. Pielaszek, J. Alloys Compounds 286 (1999) 93. [18] D. Schanke, S. Vada, E.A. Blekkan, A.M. Hilmen, A. Hoff, A. Holmen, J. Catal. 156 (1995) 85. [19] E. Iglesia, S.L. Soled, R. Fiato, G.H. Via, J. Catal. 143 (1993) 345. [20] B. Jongsomjit, J. Panpranot, J.G. Goodwin Jr., J. Catal. 204 (2001) 98. [21] L. Guczi, Z. Schay, G. Steer, F. Mizukami, J. Mol. Catal. A 141 (1999) 177. [22] U. Bardi, B.C. Beard, P.N. Roos, J. Catal. 124 (1990) 22. [23] C.C. Culross, US Patent 5 928 983 (1999). [24] C.H. Mauldin, US Patent 5 968 991 (1999). [25] C.H. Mauldin, US Patent 6 331 575 (2001). [26] B.D. Cullity, Elements of X-Ray Diffraction, AddisonWesley, London, 1978. [27] J.-S. Girardon, A.Y. Khodakov, M. Capron, S. Cristol, C. Dujardin, F. Dhainaut, S. Nikitenko, F. Meneau, W. Bras, E. Payen, J. Synchrotron Radiat. 12 (2005) 680. [28] R.W.G. Wyckoff, Crystal Structures, Interscience, New York, 1960. [29] A.Yu. Khodakov, J. Lynch, D. Bazin, B. Rebours, N. Zanier, B. Moisson, P. Chaumette, J. Catal. 168 (1997) 16. [30] V.V. Kiselev, P.A. Chernavskii, V.V. Lunin, Russ. J. Phys. Chem. 61 (1987) 151. [31] P.A. Chernavskii, A.Y. Khodakov, G.V. Pankina, J.-S. Girardon, E. Quinet, Appl. Catal. A 306 (2006) 108. [32] A.S. Lermontov, J.-S. Girardon, A. Griboval-Constant, S. Pietrzyk, A.Y. Khodakov, Catal. Lett. 101 (2005) 117. [33] A.A. Verberkckmoes, B.M. Weckhuysen, R.A. Schoonheydt, Microporous Mesoporous Mater. 22 (1998) 165. [34] Y. Okamoto, K. Nagata, T. Adachi, T. Imanaka, K. Inamura, T. Takyu, J. Phys. Chem. 95 (1995) 310. [35] M.G. Ferreira de Silva, Mater. Res. Bull. 34 (1999) 2061. [36] J.H. Aschley, P.C.H. Mitchell, J. Chem. Soc. A (1968) 2821. [37] S.W. Ho, M. Horialla, D.M. Hercules, J. Phys. Chem. 94 (1999) 6396. [38] T.J. Chaung, C.R. Brundle, D.W. Rice, Surf. Sci. 59 (1976) 413. [39] G.A. Park, Chem. Rev. 65 (1965) 177. [40] A.Y. Khodakov, V.L. Zholobenko, R. Bechara, D. Durand, Microporous Mesoporous Mater. 79 (2005) 29. [41] W. Zou, R.D. Gonzales, Catal. Today 15 (1992) 443. [42] J.P. Bonnelle, J. Grimblot, A. Dhuysser, J. Electron Spectrosc. 7 (1975) 151. [43] D.G. Castner, P.R. Watson, I.Y. Chan, J. Phys. Chem. 93 (1989) 3188. [44] J. Haber, J. Stoch, L. Ungier, J. Electron Spectrosc. Relat. Phenom. 9 (1976) 459. [45] D.L. Leslie-Pelecky, R.D. Rieke, Chem. Mater. 8 (1996) 1770. [46] P.A. Chernavskii, Mendeleev Chem. J. XLVI (2002) 19. [47] D.G. Castner, P.R. Watson, I.Y. Chan, J. Phys. Chem. 94 (1990) 819. [48] K. Bourikas, Ch. Kordulis, J. Vakros, A. Lycourghiotis, Adv. Colloid Interface Sci. 110 (2004) 97. [49] H. Ming, B.G. Baker, Appl. Catal. 123 (1995) 23. [50] A.J. Van Dillen, R.J.A.M. Terorde, D.J. Lensveld, J.W. Geus, K.P. De Jong, J. Catal. 216 (2003) 257. [51] A.M. Hilmen, D. Schanke, A. Holmen, Catal. Lett. 38 (1996) 143. [52] A.M. Hilmen, D. Schanke, K.F. Hanssen, A. Holmen, Appl. Catal. A 186 (1999) 169. [53] G. Jacobs, T.K. Das, Y. Zhang, J. Li, G. Racoillet, B.H. Davis, Appl. Catal. A 233 (2002) 263. [54] Z.-Q. Zhu, K.-G. Fang, J.-G. Chen, Y.-H. Sun, J. Fuel Chem. Technol. 33 (4) (2005) 506. [55] C.M. Lok, Stud. Surf. Sci. Catal. 147 (2004) 283.

J.-S. Girardon et al. / Journal of Catalysis 248 (2007) 143157

157

[56] [57] [58] [59]

C.M. Lok, G.J. Kelly, G. Gray, US Patent 6,927,190 B2, 2005. C.M. Lok, S. Bailey, G. Gray, US Patent 6,534,436 B2, 2003. N. Kotter, L. Riekert, Stud. Surf. Sci. Catal. 3 (1978) 51. P. Jonhson, R.W. Joyner, P.D.A. Pudney, E.S. Shpiro, B.P. Williams, Faraday Discuss. Chem. Soc. 89 (1990) 1. [60] R. Joyner, E. Shpiro, P. Johnson, G. Tuleouva, J. Catal. 141 (1993) 250.

[61] E.S. Shpiro, R.W. Joyner, K.M. Minachev, P.D.A. Pudney, J. Catal. 127 (1991) 336. [62] R.W. Joyner, K.M. Minachev, P.D.A. Pudney, E.S. Shpiro, G. Tuleouva, Catal. Lett. 9 (1991) 183. [63] J. De Graaf, A.J. Van Dillen, K.P. De Jong, D.C. Koningsberger, J. Catal. 203 (2) (2001) 307.