A Systematic Study of the Interactions between Chemical Partners

A Systematic Study of the Interactions between Chemical Partners (Metal, Ligands,. Counterions, and Support) Involved in the Design of Al2O3-Supported...
1 downloads 0 Views 443KB Size
Download PDF
2836

J. Phys. Chem. B 2005, 109, 2836-2845

A Systematic Study of the Interactions between Chemical Partners (Metal, Ligands, Counterions, and Support) Involved in the Design of Al2O3-Supported Nickel Catalysts from Diamine-Ni(II) Chelates Fabien Ne´ grier, Eric Marceau,* and Michel Che† Laboratoire de Re´ actiVite´ de Surface (UMR 7609 CNRS), UniVersite´ Pierre et Marie Curie, 4 Place Jussieu, 75252 Paris Cedex 05, France

Jean-Marc Giraudon, Le´ on Gengembre, and Axel Lo1 fberg Laboratoire de Catalyse (UMR 8010 CNRS), UniVersite´ des Sciences et Techniques de Lille, Cite´ Scientifique, 59655 VilleneuVe d’Ascq Cedex, France ReceiVed: May 24, 2004

1.5 Ni wt %/Al2O3 catalysts have been prepared by incipient wetness impregnation using [Ni(diamine)x(H2O)6-2x]Y2 precursors (diamine ) 1,2-ethanediamine (en) and trans-1,2-cyclohexanediamine (tc); x ) 0, 1, and 2; Y ) NO3- and Cl-), to avoid the formation, during calcination, of difficult-to-reduce nickel aluminate. N2 was chosen for thermal treatment to help reveal and take advantage of the reactions occurring between Ni2+, ligands, counterions, and support. In the case of [Ni(en)2(H2O)2]Y2 salts used as precursors, in situ UV-vis and DRIFT spectroscopies show that after treatment at 230 °C Ni(II) ions are grafted to alumina via two OAl bonds and that the diamine ligands still remain coordinated to grafted nickel ions but in a monodentate way, bridging the cation with the alumina surface. With Y ) Cl-, the chloride counterions desorb as hydrogen chloride, and hydrogen released upon decomposition of the en ligands is able to reduce a fraction of nickel ions into metal as evidenced by XPS. In contrast, with Y ) NO3-, compounds such as CO or NO are formed during thermal treatment, indicating that nitrate ions burn the en ligands. After thermal treatment at 500°C, a surface phase containing Ni(II) ions forms, characterized by XPS and UV-vis spectroscopy. Temperatureprogrammed reduction shows that these ions can be quantitatively reduced to the metallic state at 500 °C, in contrast with the aluminate obtained when the preparation is carried out from [Ni(H2O)6]2+, which is reduced only partly at 950 °C. On the other hand, a total self-reduction of nickel complexes leading to 2-5-nm metal particles is obtained upon thermal treatment via the hydrogen released by a hydrogen-rich ligand such as tc, whatever the Y counterion. An appropriate choice of the ligand and the counterion allows then to obtain selectively Ni(II) ions or a dispersed reduced nickel phase after treatment in N2, as a result of the reactions occurring between the chemical partners present on alumina.

Introduction NiOx/Al2O3 catalysts catalyze various types of reactions, oxidation or hydrogenolysis/hydrogenation, for which an oxidic1 or metallic2,3 nickel phase, respectively, is required. A common way to obtain these phases is via deposition of nickel(II) nitrate on alumina, followed by a single thermal treatment in O2, or by successive treatments in O2 and H2, respectively. A major problem encountered during calcination in O2 is the migration of nickel ions into alumina to form nickel aluminate.1,4,5 Unfortunately, the latter is difficult to reduce to the metallic statesonly at temperatures at which nickel sintering occurss and they are poorly active in oxidation and hydrogenation processes.6,7 For catalysts with a low nickel loading (e2 Ni wt %), the distribution of nickel ions between octahedral and tetrahedral sites of alumina depends on several preparation parameters: support hydration, calcination atmosphere, and temperature4,5,8-12 or use of additives such as Na, Zn, Ge, or Ga,13,14 but for all cases, the problem of low reducibility remains. * Corresponding author. Tel: + 33 1 44 27 60 04; fax: + 33 1 44 27 60 33; e-mail: [email protected]. † Institut Universitaire de France.

Figure 1. Scheme of the six possible interactions between cation, ligand, counterion, and support during the preparation of supported catalysts by incipient wetness impregnation.

Incipient wetness impregnation is the most common method to prepare catalysts and consists of filling the void volume of alumina with a solution of nickel salt. It offers an interesting feature in that the next step of drying forces the counterions to remain on the support with the cation and its ligands, potentially giving way to a variety of interactions and reactions between these species at the solid-liquid or solid-gas interfaces during the successive preparation steps of the catalyst (Figure 1). It can then be attempted to evaluate to what extent an adequate choice of the reactants, that is, ligands and anions (since cation

10.1021/jp0403745 CCC: $30.25 © 2005 American Chemical Society Published on Web 01/26/2005

Design of Alumina-Supported Nickel Catalysts and support are determined by the system to prepare), can prevent the migration of Ni2+ ions into the support and lead to supported ionic or metallic systems well-dispersed and prepared in mild conditions. Previous studies on impregnation have established that chelating ligands (such as acetylacetonate,15,16 EDTA,17 or citrate18) are good candidates for the preparation of easy-toreduce nickel catalysts. In systems prepared by equilibrium adsorption of complexes (where the counterions of the complex are eliminated by washing), hydrosoluble diamine chelates lead to highly dispersed nickel species suitable for catalytic applications.19-24 These results led us to investigate the effect of 1,2-ethanediamine and trans-1,2-cyclohexanediamine, differing by their atomic H/C content, on the properties of NiOx/ Al2O3 catalysts prepared from nickel complexes containing these ligands, whereas nitrate and chloride were selected as anions because of their different redox properties. In and ex situ characterization techniques were used in combination, to identify the different reactions occurring on the whole temperature range of the catalysts preparation. In this molecular approach, the cation, its ligands, and its counterions are taken as spectroscopic probes of their own interactions with their chemical partners and the support. [Preliminary results have been presented in Ne´grier, F.; Marceau, E.; Che, M. Chem. Commun. 2002, 1194.] Experimental Section Preparation of Ni(II) Complexes. 1,2-Ethanediamine (en, NH2(CH2)2NH2), trans-1,2-cyclohexanediamine (tc, C6H10(NH2)2) (99+%, Aldrich), hexaaquanickel(II) nitrate (99+%, Fluka), and chloride and bromide (98+%, Pro Analysis) were used as reactants to synthesize the nickel complexes following preparation modes adapted from the literature.25-29 The precursor [Ni(en)(H2O)4]2+, 2Y (Y ) NO3- or Cl-), was obtained in solution by adding 1,2-ethanediamine to an aqueous solution of nickel nitrate or chloride with a molar ratio en/Ni ) 1. The salts [Ni(en)2(H2O)2](NO3)2, dimeric [Ni(en)2(µ-Cl)]2Cl227 (in solution, [Ni(en)2(H2O)2]2+, 2 Cl-), and [Ni(en)3](NO3)2 were obtained by slowly adding 1,2-ethanediamine to 50 mL of a 2 M aqueous solution of nickel nitrate or chloride up to the desired en/Ni molar ratio; after 10 min of magnetic stirring, the solvent was eliminated under vacuum at 70°C to obtain crystals, which were washed twice with 3 mL of 1/1 ethanol/water mixture and dried under vacuum at 70 °C. The salts [Ni(tc)2(H2O)2]Cl2, [Ni(tc)2(H2O)2](NO3)2, [Ni(tc)3](NO3)2‚3H2O, and [Ni(tc)3]Br2‚H2O were synthesized using trans-1,2-cyclohexanediamine by precipitation in a 1/1 mixture of ethanol/water. Dehydrated [Ni(tc)2(NO3)2] and [Ni(tc)3](NO3)2 were obtained by heating the precursor salts for 2 h in nitrogen at 100 °C;28 exposure of these compounds to air humidity for 2 h led back to the starting hydrated salts and their characterization was consequently performed in a dry nitrogen atmosphere. [Ni(en)Cl2] was obtained similarly by heating [Ni(en)2(µ-Cl)]2Cl2 at 250 °C under nitrogen.29 [Ni(en)(H2O)4]2+, 2Y (Y ) NO3- or Cl-), [Ni(en)2(H2O)2](NO3)2, [Ni(en)2Cl]2Cl2, [Ni(tc)2(H2O)2](NO3)2, and [Ni(tc)2(H2O)2]Cl2 were used as precursors to prepare the catalysts. The other salts were used as references for the characterization of supported complexes. Caution. The addition of diamines in a nickel salt solution is exothermic. Solid nitrates of diamine complexes decompose explosively when heated above 200 °C. Preparation of Supported Catalysts and Reference Samples. The supported catalysts will be referred to as follows: Ni diaminex Y, with diamine ) en or tc, x ) molar ratio diamine/

J. Phys. Chem. B, Vol. 109, No. 7, 2005 2837 Ni(II) (x ) 1 or 2), and Y ) N or C for nitrate or chloride, respectively. Six samples will be studied: NienN, NienC, Nien2N, Nien2C, Nitc2N, and Nitc2C. As support, 150-250 µm grains of γ-Al2O3 (IFP EC1285, specific surface area 200 m2/g, pore volume 0.68 cm3/g) were used. The catalysts were prepared by incipient wetness impregnation, using aqueous solutions of the nickel complexes synthesized previously. A 0.68-mL portion of a 0.38 M nickel aqueous solution (pH ) 8) was added dropwise to 1 g of the support at room temperature for en and at 50 °C for tc, because of the low solubility of tc Ni(II) complexes at ambient temperature (deposition step). The solution concentration was adjusted to obtain 1.5 Ni wt % reduced catalysts. After 10 min of homogeneization obtained by stirring, the humid solid was dried in static air for 30 min at 20 °C in a desiccator or for 30 min at 50 or 100 °C in an oven (drying step). The dried samples were then treated for 2 h in air or N2 at 500 °C, with a heating rate of 7.5 °C/min (thermal treatment). In addition, thermal treatments in N2 of two catalysts (NienC and Nien2N) were performed with heating rates of 20 and 2 °C/min, respectively. Reduction was performed for 2 h in pure H2 at 500 °C, with a heating rate of 7.5 °C/min. Gas flows of 100 cm3/min were used for all preparations and characterization experiments in air, nitrogen, argon, or hydrogen. Some experiments were performed on reference catalysts prepared in the same way, using [Ni(H2O)6](NO3)2 as a precursor salt. Finally, two reference supported systems, KNO3/ Al2O3 and en/Al2O3, were prepared by impregnation, using titrated solutions of NaOH and HCl to adjust the desired pH of the impregnating solution at pH ) 8. Characterization. Chemical analyses were performed by ICP (Ni, Cl) or catharometry after fast calcination (C, H, N) at the Vernaison Center of Chemical Analysis of the CNRS. X-ray diffraction (XRD) analyses were carried out on a Siemens D500 diffractometer, using Cu KR radiation (1.5418 Å). After drying, no other peaks were detected apart from those of the support (for instance, no peaks from hydrotalcites were detected). XRD experiments were also carried out after thermal treatments in N2 and H2. Combined differential thermal and thermogravimetric analyses (DT-TGA) were obtained on a Seiko DT-TGA 320 module operated by a Seiko SSC5200 disk station, with a heating rate of 7.5 °C/min in a nitrogen atmosphere. Temperature-programmed reductions (TPR) under 5% H2 in argon (25 cm3/min) were performed with air- or N2-thermally treated catalysts. The hydrogen consumption was measured by catharometry, from room temperature to 1000 °C with a heating rate of 7.5 °C/min. A HPR20/DSMS mass spectrometer (MS) was used to analyze the gases produced upon thermal treatment in Ar or TPR up to 1000 °C with a heating rate of 7.5 °C/min. A 2-m capillary line connected to the reactor outflow was used to collect the gases at atmospheric pressure. The gases were analyzed after fragmentation in an ionic source. UV-visible-near-infrared (UV-vis-NIR) spectra of the nickel salt solutions were recorded with a resolution of 1 nm in the transmission mode on a Cary 5 spectrometer (Varian). UVvis-NIR spectra of the solids were recorded in the reflectance mode (1-nm resolution) on the same spectrometer equipped with an integration sphere, using Teflon as a reference, or BaSO4 powder for in situ experiments. The spectra were collected in situ in the range 20-320 °C using a high-temperature environmental cell (Spectratech). After drying at 20 °C for 30 min, the samples were placed in the cell and purged for 10 min in He at

2838 J. Phys. Chem. B, Vol. 109, No. 7, 2005 room temperature before acquisition of the first spectrum. The other spectra were collected after heating of the catalyst at a rate of 7.5 °C/min to the desired temperature and 10 min at this temperature. The spectra of the 500 °C-treated samples and of the crystallized nickel salts (except dehydrated salts) were not recorded in situ but at room temperature and exposed to air. The diffuse reflectance spectra were submitted to the Kubelka-Munk transform. The diffuse reflectance infrared Fourier transform (DRIFT) spectra were recorded with a resolution of 4 cm-1 (128 scans averaging) on a Bruker IFS66V spectrometer equipped with an MCT (mercury cadmium telluride) detector. The DRIFT accessories (Collector, Spectratech) include a high-temperature environmental chamber for thermal treatments in Ar and H2. The supported samples (dried in air at 20 °C for 30 min) and crystallized nickel salts were dispersed into industrial singlecrystalline diamond powder (IR-transparent and diffusing matrix) with an average particle size of 6 µm (10 wt % of the supported compound in diamond).30 Fifty milligrams of the mechanical mixture was placed in the environmental chamber and exposed to the gas. Before acquisition of the spectra, the samples were heated for 10 min at the desired temperature (20500 °C, heating rate ) 7.5 °C/min). The final spectra were obtained by division of the absorption spectra intensity by the diamond reference absorption spectra intensity recorded in the same thermal conditions and submission to the Kubelka-Munk transform. X-ray photoelectron spectra (XPS) were recorded with a VG ESCALAB 220 XL spectrometer equipped with a monochromatized Al source (Al KR ) 1486.6 eV). The analyzer was operated in a constant pass energy mode (Epass ) 30 eV) using the electromagnetic mode for the lens. The resolution measured on the Ag 3d5/2 line was 0.75 eV. Energy correction was performed by using the Al 2p peak of alumina at 74.6 eV as a reference. Transmission electron micrographs (TEM) were collected on a JEOL 100 CXII UHR microscope. The average nickel particles diameter Ø was calculated from the following formula: Ø ) Σnidi/Σni, where ni is the number of particles of diameter di and Σni ∼ 300. Results The results below will concern the solutions of the precursor complexes and then the catalysts prepared from the latter, thermally treated in various temperature ranges. 1. pH of the Precursor Solutions. In aqueous solution, Ni2+ can undergo three successive complexation reactions with en and tc, characterized by the formation constants β1, β2, and β3. According to values reported in the literature (log βi ) 7.50, 13.84, 18.30 for en and 8.22, 15.30, 20.44 for tc),31 mono- or bis-complexes are the major species formed when 1 or 2 equivalents of ligand, respectively, are added to the solution, but the complexation is not complete and a small quantity of diamine remains free. This results experimentally in a slightly basic pH (∼8), close to the alumina zero-point charge32 and measured for all [Ni(diamine)x(H2O)6-2x]2+ solutions (diamine ) en and tc, x ) 1 or 2). Electrostatic interactions between complexes and alumina can be thus considered as negligible during deposition. 2. NienY Catalysts (Y ) NO3-, Cl-), T ) 20-500 °C. The results described in this section concern complexes with a ratio en/Ni ) 1 and are illustrated by schemes A-D of Figure 12 referred to in the Discussion section.

Ne´grier et al. TABLE 1: Experimental UV-Visible Absorption Band Positions (nm) for Nickel Reference Complexes and Corresponding Assignments complex, symmetry

λI

λII

λIII

λIV

1. [Ni(H2O)6]2+, Oha 2. [Ni(en)(H2O)4]2+, C2v 3. cis-[Ni(en)2(H2O)2]2+, C2v 4. cis-[Ni(tc)2(H2O)2]2+, C2v 5. [Ni(en)3]2+, D3 6. trans-[Ni(tc)2(NO3)2], D4hb 7. [Ni(en)Cl2], C2vc

395 370 354 351 344 345 406

650 620 574 568 542 540 672

1130 996 920 908 880 709 1030

1136 1280

a

Complexes 1-5 (interpreted on the basis of Oh symmetry):33 λI: 3 3 3 3 3 b 2g f T1g(P), λII: A2g f T1g(F), λIII: A2g f T2g. Complex 6 (interpreted on the basis of D4h symmetry):33 λI: 3B1g f 3A2g or Eg, λII: 3B1g f 3Eg, λIII: 3B1g f (3B2g, A2g), λIV: 3B1g f 3Eg. c Complex 7 (interpreted on the basis of Td symmetry):29,34 λII: 3T1(F) f 3T1(P), λIII and λIV: 3T1(F) f 3A2. 3A

Figure 2. UV-visible spectra of (a) NienN washed and dried at 20 °C; (b) NienN dried at 100 °C; (c) Nien2N dried at 20 °C; (d) Nien2N dried at 50 °C; (e) Nien2N treated at 230 °C in N2; (f) Nitc2N dried at 20 °C; (g) Nitc2C treated at 230 °C in N2; (h) Nitc2N treated at 230 °C in N2. The band in the 750-760-nm region corresponds to the spinforbidden 3A2g f 1Eg transition.

After deposition of [Ni(en)(H2O)4]2+ complexes, washing with water, and drying at 20 °C, 50% of the nickel ions initially present in solution remain on alumina (0.75 Ni wt %). Whatever the anion, a bathochromic shift of the three absorption bands of [Ni(en)(H2O)4]2+ is observed in the UV-vis region (compare values for complex 2 in Table 1, and Figure 2a, that show the spectrum recorded on NienN) without any significant change in their relative intensities. The alumina surface groups being classified as weak field ligands compared with water, this shift is interpreted as arising from grafting of [Ni(en)(H2O)4]2+ onto the support, by substitution of labile aqua ligands by AlOgroups.33 The application of Jørgensen’s average environment law suggests [Ni(en)(H2O)2(OAl)2] as the formula for the grafted complex, following the good agreement between the experimental ∆o value (1/λIII ) 9800 cm-1) and the value calculated from data from Table 1 (complexes 1 and 5) and ref 33: ∆o[Ni(en)(H2O)2(OAl)2] ) ∆o[Ni(en)2(OAl)2] + ∆o[Ni(H2O)6]2+/ 3 - ∆o[Ni(en)3]2+/3 ) 10440 + 8880/3-11400/3 ) 9600 cm-1. One-half of the complexes has thus been grafted onto the support during deposition. After deposition and drying at 100 °C of catalysts NienY without intermediate washing, the spectrum reveals four intense absorption bands at 394, 654, 1020, and 1220 nm, that is, at higher wavelengths than [Ni(en)(H2O)2(OAl)2] and close to those of the pseudo-tetrahedral compound [Ni(en)Cl2] (complex

Design of Alumina-Supported Nickel Catalysts

J. Phys. Chem. B, Vol. 109, No. 7, 2005 2839

Figure 5. Ni wt % of catalysts prepared from (a) [Ni(H2O)6]2+ and (b) [Ni(en)2(H2O)2]2+ (Y ) NO3-) remaining on the alumina surface after drying and subsequent washing, as a function of drying temperature.

Figure 3. UV-visible spectra (A) and temperature-programmed reductions (B) of (a) NienN treated at 500 °C in N2; (b) NienC treated at 500 °C in N2, heating rate ) 7.5 °C/min; (c) NienC treated at 500 °C in N2, heating rate ) 20 °C/min.

Figure 4. MS signal (a) of the production of CO (m/z 28) during the thermal treatment of NienN in N2; DTA curves of (b) NienN and (c) Nien2N during the same treatment.

7 in Table 1, spectrum in Figure 2b), suggesting that the dehydrated tetrahedral [Ni(en)(OAl)2] complex is formed. It appears to be stable up to 230 °C in air or N2. When air is chosen for thermal treatment of catalyst NienN up to 500 °C, the characteristic UV-vis bands of nickel ions in octahedral (418 nm) and tetrahedral (591, 632 nm) sites of an aluminate phase9 are observed (Figure 3Aa), with no difference compared with a system prepared from [Ni(H2O)6]2+. A TPR experiment performed on the green air calcined sample shows that this aluminate phase reduces at 950 °C with a ratio H2/Ni ) 0.6 (Figure 3Ba). The same result is obtained for all catalysts after calcination in air, whatever the precursor salt. This led us to experiment with the use of an inert gas, nitrogen or argon, for thermal treatments described from now on (except for drying, always carried out in air). During thermal treatment in N2 of catalyst NienN, CO (m/z 28) and NO (m/z 30) (result not shown) are detected from 230 to 370 °C by MS, suggesting that nitrates have oxidized the ligand (Figure 4a). This is accompanied by an exothermic peak in DTA (Figure 4b). A green aluminate phase that can be

reduced above 950 °C is formed, like when [Ni(H2O)6]2+ is used as precursor or air as gas for thermal treatment. When sample NienC is treated in N2, a new absorption band appears around 300 nm along with bands of the aluminate (Figure 3Ab). TPR experiments carried out on this sample (Figure 3Bb) exhibit a weak and broad hydrogen consumption peak at 530 °C and a strong one at 900 °C, with a total ratio H2/Ni ) 1. However, if the heating rate during thermal treatment is increased from 7.5 to 20 °C/min, the sample obtained is now yellow, the band at 320 nm grows in intensity (Figure 3Ac), the nickel aluminate absorption bands disappear, and the TPR consumption peak at 500 °C becomes prominent, while the reduction peak related to the aluminate has become minor (Figure 3Bc). The characteristic color of this sample, yellow, because of the strong absorption in the UV region, may be interpreted in terms of nickel coordination in distorted octahedral sites.35 Nickel ions in this coordination are easier to reduce than nickel aluminate. 3. Nien2Y Catalysts (Y ) NO3-, Cl-), T ) 20-230 °C. The results described in sections 3-6 concern complexes with a ratio en/Ni ) 2 and are illustrated by schemes E-J of Figure 12. Deposition on alumina of cis-[Ni(en)2(H2O)2]2+ (the stable form of [Ni(en)2(H2O)2]2+ in solution), washing of the sample with water, and drying at 100 °C lead to a system containing only 0.25 Ni wt % which is spectroscopically identical to NienN after washing and drying at 100 °C. Chemical analyses of Ni, C, and N give a ratio en/Ni ) 1 which confirms the existence of [Ni(en)(OAl)2]. After deposition of cis-[Ni(en)2(H2O)2]2+ and drying at 20 °C without washing, the three bands detected are attributed to cis-[Ni(en)2(H2O)2]2+ (complex 3 in Table 1, spectrum in Figure 2c). In addition, the small amount of the grafted mono(diamine) complex contributes to the broadening of the absorption bands of the spectrum at high wavelengths (indicated by ; on Figure 2c). Further drying at 50 °C leads to a bathochromic shift of the three absorption bands of the bis(en) complex (Figure 2d), consistent with the formation of grafted [Ni(en)2(OAl)2]. However, a washing of the sample after drying gives only 0.25 Ni wt % corresponding to [Ni(en)(OAl)2] grafted during deposition, indicating that the grafting of [Ni(en)2(H2O)2]2+ is reversible upon exposure to water. The amount of grafted [Ni(en)(OAl)2] is almost independent from the drying temperature and remains low, in contrast to what is observed in the case of [Ni(H2O)6]2+ (Figure 5). After drying at 100 °C, the NIR band of NH vibration has shifted from 2029 to 2048 nm on Nien2N and Nien2C. This shift is similar to the ones observed comparing nickel salts

2840 J. Phys. Chem. B, Vol. 109, No. 7, 2005

Ne´grier et al.

Figure 6. Diffuse reflectance NIR spectra in the 1830-2100-nm domain of the following salts: (a) [Ni(tc)2(H2O)2](NO3)2, (b) [Ni(tc)3](NO3)2‚3H2O, (c) [Ni(tc)3](NO3)2, and (d) [Ni(tc)3]Br2‚H2O; and of the following samples: (e) Nien2N dried at 50 °C, (f) Nien2N dried at 100 °C, (g) Nitc2N dried at 50 °C, and (h) Nitc2N treated at 230 °C in N2.

crystallizing without water, and their form crystallized with water, with H2O molecules interacting with N-H groups via hydrogen bonding (Figure 6a-f).36 Since elimination of physisorbed water takes place on alumina during drying, this shift is interpreted in the terms of hydrogen bonding between the ligand amine ends of grafted [Ni(en)2(OAl)2] and hydroxyl groups of alumina surface. After thermal treatment in N2 at 230°C, UV-vis data are similar to those related to the pseudo-tetrahedral [Ni(en)(OAl)2] grafted complex detected previously (Figure 2e). However, in this case elemental analyses of Ni, C, and N are consistent with a molar ratio en/Ni ) 2. The desorption of en, which can be detected on reference sample en/Al2O3 by TGA/MS in the 160240 °C range with a maximum at 200 °C, is not observed here. It can be concluded that free en and consequently mono(en) complexes are minor species on the surface and that for most complexes, the two en ligands remain bonded to nickel. In addition to the NH bands initially present and observed between 3200 and 3400 cm-1, DRIFT spectra (Figure 7) exhibit sharp bands at 3049 and 3137 cm-1, also observed for en/Al2O3 dried at 100 °C. These bands and another strong band at 1530 cm-1 are attributed to the protonated R-NH3+ function,37,38 leading to [Ni(enH+)2(OAl)2] as the likely formula for the grafted complex at 230 °C: in this hypothesis, each en is bonded in a monodentate way to Ni2+, bridging it to the alumina surface via its protonated end. 4. Nien2Y Catalysts (Y ) NO3-, Cl-), T ) 230-500 °C. For catalyst Nien2C, UV-vis spectroscopy shows that [Ni(enH+)2(OAl)2] is stable on the alumina surface up to 320 °C. At that temperature, the migration of Ni(II) ions into the octahedral sites of alumina has already begun when [Ni(H2O)6]2+ is used for impregnation (absorption band at 420 nm). A molar ratio Cl/Ni ) 2 shows that chlorides have remained present on the support. Hydrogen chloride desorption is detected by a silver nitrate test at T > 320 °C. Hydrogen (m/z 2) and ammonia (m/z 15 and 17) are produced at 350 °C (Figure 8a). This phenomenon is accompanied by the appearance of a C)N vibration at 1635 cm-1 on DRIFT spectra, previously reported for alumina-supported imine species.39 Acetonitrile is produced above 400 °C (m/z 41), followed by HCN (m/z 27) and a second desorption of hydrogen. After heating catalysts Nien2N at 230 °C in N2, the characteristic infrared band of NO3- at 1380 cm-1 splits into 1305

Figure 7. DRIFT spectra in the 1400-1700 and 2500-4000 cm-1 regions of (a) Nien2N dried at 20 °C; (b) Nien2N treated at 230 °C in N2; (c) en/Al2O3 treated at 100 °C in N2.

Figure 8. MS thermographs during treatment in N2 for catalysts: (a) Nien2C (m/z 2, 15, 17, 18, 27, 41); (b) Nitc2C (m/z 2, 15, 27, 41, 78, 93).

and 1518 cm-1 (∆ν ) 213 cm-1), as observed for the reference sample KNO3/Al2O3 or for [Ni(en)2(NO3)]I where nitrate coordinates in a bidentate way to nickel.25 It is supposed that nitrates bridge two coordination sites of alumina. Heating the catalyst at T > 230 °C leads to the desorption of oxidation products, CO and NO, detected by MS (Figure 9a). Nitrate ions burn the en ligands, but with an exothermicity lower for Nien2N than for NienN (Figure 4c). Bands at 1492 and 1560 cm-1 detected by DRIFT are attributed to organic species remaining from en oxidation and adsorbed on the surface.15 5. Characteristics and Reducibility of N2-Thermally Treated Nien2N. After heating at 500 °C (heating rate ) 7.5 °C/min), the sample is yellow-beige and gives a UV-vis spectrum similar to that of NienC prepared with a heating rate of 20 °C/min (Figure 3Ac): a broad band at 320 nm and no bands of aluminate. The latter compound appears only for a slower heating rate (2 °C/min), that is, for a longer time of heating. XPS measurements (Figure 10a and b) evidence a phase

Design of Alumina-Supported Nickel Catalysts

Figure 9. MS thermographs during treatment in N2 for catalysts: (a) Nien2N (m/z 2, 28, 30); (b) Nitc2N (m/z 2, 28, 30).

Figure 10. Ni 2p3/2 XP spectra of (a) aluminate prepared from [Ni(H2O)6]2+; (b) Nien2N treated at 500 °C in N2; (c) Nien2C treated at 500 °C in N2; (d) Nien2C treated at 500 °C in N2 and then at 500 °C in H2; (e) Nitc2N treated at 500 °C in N2 and then at 500 °C in H2. The Ni0 2p1/2 peak is seen at 869.5 eV on spectra (d) and (e).

Figure 11. Thermograms of Nien2N treated at 500 °C in N2: (a) MS thermograph (m/z 16) and (b) TPR; Nien2C treated at 500 °C in N2: (c) TPR; Nitc2N treated at 500 °C in N2: (d) MS thermograph (m/z 16) and (e) TPR; Nitc2C treated at 500 °C in N2: (f) TPR.

constituted only of Ni2+ ions12,40 (Ni(II) 2p3/2 signal at 856.4 eV, |∆E| between the main peak (m) and its satellite (s) ) 6 eV, Is/Im ) 0.5), but with a Ni/Al atomic surface ratio higher than for an aluminate prepared from [Ni(H2O)6]2+ (2.6 × 10-2 compared to 1.8 × 10-2). The ionic nickel phase reduces at 500 °C with a ratio H2/Ni ) 0.85, and a hydrogen consumption observed at 850 °C correlated with the desorption of methane (Figure 11a, b) is attributed to the reduction of the residual organic species. The reduction in hydrogen leads to nickel particles (Ø ) 2-5 nm, TEM and XRD). In comparison, larger nickel particles (Ø ) 15-60 nm) are observed when a supported aluminate obtained

J. Phys. Chem. B, Vol. 109, No. 7, 2005 2841 from [Ni(H2O)6]2+ is reduced at 950 °C in H2 (micrographs of the two samples have been presented in ref 41). 6. Characteristics and Reducibility of N2-Thermally Treated Nien2C. The UV-vis spectrum of Nien2C treated in N2 at 500 °C exhibits both a band at 320 nm, attributed to ionic nickel according to results from §.2 and 5, and a large absorption band in the whole visible region attributed to Ni0. In TPR performed with dilute H2 (Figure 11c), a consumption of hydrogen is detected at 500 °C with a ratio H2/Ni ) 0.5: about one-half of nickel has been reduced during the N2-treatment. A minor H2 consumption at 200 °C may be attributed to the reduction of the oxidized surface of nickel particles, and a small peak corresponding to methanation (mass spectrometry) is present at 730 °C. However, a surprisingly low Ni/Al atomic surface ratio is measured by XPS (1.2 × 10-2, close to the bulk ratio 1.4 × 10-2) (Figure 10c), with a major Ni2+ 2p3/2 signal at 856.4 eV and a very small peak at 853.5 eV attributed to Ni0, in contradiction with TPR results (50% of nickel reduced). Carbon detected both by elemental analysis (0.4 wt %) and XPS (285.6 eV) is probably the polluting species covering Ni0. It is eliminated by a treatment with pure H2 in the XPS chamber at 500 °C; at that temperature, if we refer to TPR results, one expects to reduce at least a part of nickel ions as well. After this treatment, a major XPS peak from metallic nickel is indeed detected at 852.6 eV (Figure 10d). A minor component from Ni2+ at 856.3 eV is superimposed to Ni0 satellite (Is/Im ) 0.6 is too high for the peak at 856.3 eV being only Ni0 satellite)40 and is accompanied by its own satellite at 862.1 eV. The Ni/Al atomic surface ratio is now 3 × 10-2. Finally, TEM micrographs suggest that after reduction of sample Nien2C in H2 at 500 °C, particles are detected with characteristics identical to those observed for Nien2N. 7. Nitc2Y Catalysts (Y ) NO3-, Cl-), T ) 20-500 °C. After deposition of cis-[Ni(tc)2(H2O)2]2+ on alumina and drying at 20 °C, three broad bands at 351, 561, and 913 nm show that the complex is deposited as such, alike their supported en analogues (complex 4 in Table 1, spectrum in Figure 2f). The band found at high wavelength can be attributed to grafted mono(tc) complexes (indicated by ; on Figure 2f). After drying at 100 °C and N2-treatment at 230 °C of catalyst Nitc2C, four absorption bands are observed at 390, 674, 1020, and 1200 nm (Figure 2g). The analogy with [Ni(enH+)2(OAl)2] leads to the formula [Ni(tcH+)2(OAl)2]. Hydrogen chloride is detected at T > 320°C; hydrogen (m/z 2) and ammonia (m/z 15 and 17) are produced at 400 °C (Figure 8b). It is concomitant with the desorption of aromatic compounds, benzene (m/z 78), and aniline (m/z 93). As in the case of Nien2C (§.4), acetonitrile, HCN, and H2 productions are detected above 400 °C (m/z 41), as is the imine infrared band. The case of Nitc2N catalyst treated in N2 at 230 °C is different. The four absorption bands (348, 541, 712, and 1050 nm) can be explained by the presence of the D4h dehydrated [Ni(tc)2(NO3)2]complex (complex 6 in Table 1, spectrum in Figure 2h). The two infrared bands at 1310 and 1420 cm-1 (∆ν ) 110 cm-1) confirm the bonding of nitrates to nickel via one oxygen. No direct interaction between nickel and alumina is thus detected. However, a bathochromic shift of the NH NIR band of the complex occurs during heating (Figure 6g and h): at 2051 nm, the position of the band is close to that of the salt crystallized with water. This is interpreted in terms of hydrogen bonding between the ligands and the hydroxylated surface. The desorption of oxidation products (CO, NO) at 280 °C and hydrogen at 400 °C shows that tc is only partially burned by

2842 J. Phys. Chem. B, Vol. 109, No. 7, 2005 nitrates and hydrocarbon residues can thereafter decompose to liberate hydrogen (Figure 9b). 8. Characteristics and Reducibility of N2-Thermally Treated Catalysts Nitc2Y. For samples Nitc2N and Nitc2C treated in N2 at 500 °C, a broad absorption band is observed in the whole visible region, attributed to the presence of nickel in the metallic state. 2-5 nm nickel particles can be detected by XRD and TEM, but XPS shows a broad signal between 850 and 860 eV that can be explained by the residual presence of Ni2+ ions. TPR/MS experiments show hydrogen consumptions due to the methanation of carbon in the 100-600 °C range (Figure 11d-f) and probably to the reduction of nickel ions. After treatment in pure H2 in the chamber at 500 °C that eliminates the carbon species initially present (0.4 wt %), XP spectra (Figure 10e) exhibit only peaks due to Ni0 (unique peak at 852.6 eV, Ni/Al atomic surface ratio ) 3 × 10-2). The distribution in size of nickel particles is not modified by H2 treatment. Discussion While the results section was devoted to the solutions of the precursor complexes and the samples derived from the latter and thermally treated in various temperature ranges, the discussion will highlight the interactions identified between the chemical partners involved in the design of Al2O3-supported nickel catalysts from diamine-Ni(II) precursor complexes. The interaction between cation and ligands will not be singled out as being obvious. Catalysts Prepared from [Ni(H2O)6](NO3)2: Migration of Cations into the Support. When catalysts are prepared from nickel nitrate with H2O only as ligand, the formation of aluminates cannot be avoided in the conditions of our study (§.2). The use of an inert gas for thermal treatment also leads to aluminates. The fate of nickel in this case seems to be determined as soon as the catalyst is dried: nearly all the nickel ions are fixed on the surface and are not removed by washing after drying (§.3). Higher temperatures drive the ions to migrate deeper into the alumina lattice. As indicated by Rynkowski et al.,4 the reducibility of nickel is made easier if the migration of the ions in the support is hindered. This is the case upon increasing the heating rate and shortening the duration of calcination: the formation of aluminates depends kinetically on the diffusion of Ni2+ ions. Another possibility is the use of a diamine ligand which delays the time of nickel migration into alumina. Catalysts NienY: Interactions Cation-Support and Ligand-Anion. When one equivalent of chelating ethanediamine per nickel is used, grafting of half of the [Ni(en)(H2O)4]2+ complexes is observed on the support during deposition (§.2, Figure 12A). The other half becomes grafted during drying, leading to a tetrahedral complex (Figure 12B); the species in direct chemical interaction with alumina is the dehydrated [Ni(en)(OAl)2], similar to the speciation detected on exchanged CuOx/SiO2 catalysts.42 No interaction is observed between the ligand and the support surface. When the thermal treatment is performed in air or the anion is nitrate, that is, when an oxidizing compound is present at some stage of preparation, for example, the counterion, heating results in the total oxidation of the ligand. Aluminate is formed, identical to that obtained in absence of ligand (Figure 12C): the diamine cannot prevent the formation of aluminate. On the contrary, when no oxidant is present (when the anion is chloride

Ne´grier et al. and N2 is the gas used), the formation of aluminates after destruction of the ligand is slowed provided that the time of heating is kept short (i.e., the heating rate high): in these conditions, the simple coordination of the diamine to Ni(II) ions can prevent the latter to migrate into alumina. XPS shows that the distorted octahedral sites which accommodate nickel ions are located in the upper layers of γ-alumina (Figure 12D). Catalysts Nien2Y: Interactions Cation-Support, LigandSupport, Anion-Support, and Ligand-Anion. When two equivalents of chelating ethanediamine per nickel are used, grafting of complexes on the alumina surface is also observed, but in this case three steps can be distinguished (§.3): (1) During deposition, a small quantity of mono(diamine) complex, a minor species in equilibrium in solution, is grafted onto the surface (Figure 12E), like in the NienY samples. The quantity of grafted mono(diamine) complexes remaining after washing is the same whatever the drying temperature. Their strong grafting compared to that of the rest of the complexes possibly depends on factors such as steric constraints or interaction with specific configurations of hydroxyl groups on the alumina surface.43 (2) During drying, bis(diamine) complexes graft onto alumina (Figure 12F-G). Second-sphere interactions are now observed between the ligands and the surface, probably via hydrogen bonding, as postulated by Bonneviot et al. for ammine nickel complexes on silica.44 Nevertheless, this grafting is reversible upon contact with water; the interaction between nickel and the support is not as strong as in the cases examined so far. (3) At 230 °C in N2, these second-sphere interactions become stronger and lead to the dissociation of the ligand from Ni2+ (Figure 12H). One extremity of the ligand is now bonded to the cation (Lewis acid), and the other one has reacted with the alumina hydroxyls (Brønsted acid). Such an unusual coordination mode has been identified by X-ray diffraction on single crystals, for complexes retained in low-dimensional or microporous phosphates and vanadates.45 The cation can bear two protonated ligands46 and the -NH3+ end forms hydrogen bonds with the oxidic framework. The ligand that was chelating nickel before is now bridging nickel ion and alumina, thanks to its basic properties. EDTA is also a chelating ligand able to bridge metal ion and support, as in [Cd(EDTA)]2- bonded to Al(OH)3.47 For thermal treatment below 230 °C in N2, nitrates also grafted on alumina have not chemically reacted. At higher temperatures, they react with the ligand and destroy it by oxidation, like for NienY systems (§.4). XPS (§.5) shows that the nickel species giving the yellow color to the N2-treated sample are located in the upper layers of the support (Figure 12I) and, consequently, are more easily reducible than the aluminate, in line with the work of Molina and Poncelet.16 The migration of nickel into alumina lattice has been avoided. It can be questioned though why two equivalents of diamine per Ni give a better result than one, given the fact that the time elapsed between the destruction of the ligand and the end of the calcination is the same, which brushes aside a purely kinetic argument. A possible answer is that two nitrates are excedentary to oxidize one ligand, but not two ligands. When nitrates are excedentary, some of them decompose exothermically upon contact with nickel ions,48,49 which may contribute to trigger their migration into the lattice and bury the cations into the alumina. Another possibility is that the organic residues resulting from the ligands oxidation by nitrate (those that are hydrogenated at high temperature in TPR) are still able to bridge Ni(II) and the alumina surface in Nien2N.

Design of Alumina-Supported Nickel Catalysts

J. Phys. Chem. B, Vol. 109, No. 7, 2005 2843

Figure 12. Speciation of nickel on alumina during the history of the preparation of catalysts NienY (A-D) and Nien2Y (E-J).

When chlorides are used as anions, they act as indifferent species toward nickel and its ligands (§.6). The ligand is allowed to decompose, following two mechanisms: (1) Departure of ammonia, we suppose from the protonated end of the diamine, possibly through a nucleophilic attack of the ligand by the alumina surface anions:

R-CH2-NH3+......-O-Al f R-CH2-O-Al + NH3

(2) Oxidative dehydrogenation of the -CH2-NH2 moiety, concerning the end of the diamine bonded to nickel (Lewis acid), with formation of imine -CHdNH and subsequently nitriles.50

R-CH2-NH2-NiII-O-Al f H2 + R-CHdNH-NiIIO-Al R-CHdNH-NiII-O-Al f H2 + R-CN + NiII-O-Al

2844 J. Phys. Chem. B, Vol. 109, No. 7, 2005 The diamine acts as a reservoir of hydrogen and thus as a reducing agent, as shown, for example, by Riahi et al.51 This hydrogen can reduce a large part of nickel ions to the metallic state, but the pressure does not seem to be high enough to reduce all of them (Figure 12J). Small nickel particles are revealed by a reducing treatment in hydrogen. Carbon covers Ni0 particles, but it can be eliminated by pure hydrogen and this is a second reason to use a further reducing treatment in H2. Catalysts Nitc2Y: Interactions Cation-Anion, LigandSupport, and Minor Ligand-Anion. When chloride is used, the grafting of nickel and the formation of second-sphere interactions between ligand and support are very similar to the results obtained with Nien2C (§.7). However, nitrate seems to behave as a better coordinating ligand to nickel than the alumina surface groups for this sterically hindered trans-complex. For Nitc2N, no first-sphere interaction between nickel and alumina is observed, though second-sphere interactions between ligands and support seem to exist. On the reverse, the anion plays only a minor role during thermal treatment (§.8). Whether the anion can oxidize the ligand (nitrate) or not (chloride), the C6 cycle of tc can dehydrogenate easily, giving aromatic compounds, as does cyclohexylamine bonded to nickel ions.52

C6H10(NH2)2 f C6H5-NH2 + NH3 + 2H2 C6H10(NH2)2 f C6H6 + 2NH3 + H2 Hydrogen is produced in sufficient quantity to reduce nickel ions to Ni0 particles detectable by XRD and TEM. However, like for Nien2C, a further reducing treatment is necessary for complete reduction and removal of adsorbed carbon. Advantages Given by the Proper Choice of Diamines and Counterions. The use of chloride as a counterion or tc as a ligand allows one to take advantage of the in situ production of hydrogen to obtain Ni0 particles during N2 thermal treatment. However, as far as catalysis is concerned, two major drawbacks must be mentioned concerning the choice of these compounds for the actual preparation of catalysts: chlorides are generally avoided in catalysis because of their poisoning effect53 and tc complexes are not very soluble, which limits the nickel content of catalysts prepared by impregnation. Finally, the presence of contaminating carbon over the reduced nickel makes a second reducing treatment compulsory. For these reasons, a better system to replace beneficially [Ni(H2O)6]2+/Al2O3 for catalysis is Nien2N, leading selectively to an ionic surface phase after the sample has been treated in nitrogen and to well-dispersed Ni0 particles if it is subsequently treated in hydrogen. Except for the addition of the ligand in the impregnation solution, the use of nitrogen for thermal treatment, and the lower temperature for reduction, the preparation of the catalysts follows exactly the same pattern as the classical procedure involving nickel nitrate and does not require any modification in the succession of the differents steps. Conclusion The combined deployment of various spectroscopies and physicochemical techniques has led to a better understanding of the interactions occurring between metal, ligands, counterions, and support during the successive preparation steps of a NiOx/ Al2O3 catalyst from Ni(II) precursor complexes. The use of a chelating diamine bonded to Ni(II) in the impregnating aqueous solution allows one to obtain easier-to-reduce Ni2+/Al2O3 systems and more dispersed nickel particles for Ni0/Al2O3

Ne´grier et al. catalysts without modifying the procedure of incipient wetness impregnation, provided that thermal treatments are carried out in an inert gas atmosphere, for example, N2. The chelating diamine ligand first appears to subdue the interaction between nickel and alumina during impregnation, by giving complexes weakly grafted or retained only by hydrogen bonding. Second, its basic and bridging properties stabilize nickel ions on the alumina surface during thermal treatment (T < 230 °C). If the thermal treatment is short enough after diamines have been burned by nitrates, Ni2+ migration inside alumina is limited by kinetic factors. Finally, the diamine ligand acts as a reducing agent toward Ni2+ upon its decomposition if no oxidizing counterion is present or if hydrogen-rich diamines such as tc are used. In summary, by choosing judiciously two components of the precursor salt, ligand and counterion, deposited with Ni2+ during impregnation of alumina, and an inert atmosphere for thermal treatments, it is possible to take advantage of the reactions (ligand substitution or dissociation, Brønsted acido-basic reaction, oxidation, reduction, nucleophilic substitution) taking place at different temperatures on the alumina surface, to selectively form supported systems in milder conditions than with nickel nitrate as a precursor salt and the thermal treatment in air generally used to prepare supported nickel catalysts. This study also shows that the earlier steps of catalysts preparation (deposition and drying) are important for they can lead, despite the same thermal treatment, to different final properties of the catalytic system. This appears obvious since the chemical partners at play differ depending upon the preparation method used: while counterions are eliminated by washing in the equilibrium adsorption method, their presence as nitrates or chlorides in the incipient wetness impregnation shows a different redox chemistry toward the precursor complex. Acknowledgment. The authors would like to thank JeanMarc Krafft and Patricia Beaunier for their help in carrying out DRIFT and TEM experiments, respectively, Prof. Franc¸ ois Bozon-Verduraz (Universite´ Paris VII - Denis Diderot) for providing access to the UV-Vis-NIR spectrometer and Prof. Lucien Leclercq (USTL) for the welcome of Fabien Ne´grier in his team. Fabien Ne´grier acknowledges a grant from the French Ministry of Education and Research. References and Notes (1) (a) Morikawa, K.; Shirasaki, T.; Okada, M. AdV. Catal. 1969, 20, 97. (b) Gravelle, P. C.; Teichner, S. J. AdV. Catal. 1969, 20, 167. (2) Pines, H. AdV. Catal. 1987, 35, 323. (3) Kochloefl, K. In Handbook of Heterogeneous Catalysis; Ertl, G., Kno¨zinger, H., Weitkamp, J., Eds.; Wiley-VCH: New York, 1997; p 1819. (4) Rynkowski, J. M.; Paryjczak, T.; Lenik, M. Appl. Catal., A 1993, 106, 73. (5) Zielinski, J. J. Catal. 1982, 76, 157. (6) Dissanayake, D.; Rosynek, M. P.; Kharas, K. C. C.; Lunsford, J. H. J. Catal. 1991, 132, 117. (7) Kester, K. B.; Falconer, J. L. J. Catal. 1984, 89, 380. (8) Rymer, G. T.; Bridges, J. M.; Tomlinson, J. R. J. Phys. Chem. 1961, 65, 2152. (9) Lo Jacono, M.; Schiavello, M.; Cimino, A. J. Phys. Chem. 1971, 75, 1044. (10) Gil, A.; Dı´az, A.; Gandı´a, L. M.; Montes, M. Appl. Catal., A 1994, 109, 167. (11) Greegor, R. B.; Lytle, F. W.; Chin, R. L.; Hercules, D. M. J. Phys. Chem. 1981, 85, 1232. (12) Wu, M.; Hercules, D. M. J. Phys. Chem. 1979, 83, 2003. (13) Houalla, M.; Delannay, F.; Delmon, B. J. Phys. Chem. 1981, 85, 1704. (14) Cimino, A.; Lo Jacono, M.; Schiavello, M. J. Phys. Chem. 1975, 79, 243.

Design of Alumina-Supported Nickel Catalysts (15) (a) Molina, R.; Centeno, M. A.; Poncelet, G. J. Phys. Chem. B 1999, 103, 6036. (b) Molina, R.; Poncelet, G. J. Phys. Chem. B 1999, 103, 11290. (16) Molina, R.; Poncelet, G. J. Catal. 1998, 173, 257; 2001, 199, 162. (17) Ryczkowski, J. React. Kinet. Catal. Lett. 1989, 40, 189. (18) van Dillen, A. J.; Tero¨rde, R. J. A. M.; Lensveld, D. J.; Geus, J. W.; de Jong, K. P. J. Catal. 2003, 216, 257. (19) Bonneviot, L.; Clause, O.; Che, M.; Manceau, A.; Dexpert, H. Catal. Today 1989, 6, 39. (20) Cheng, Z. X.; Louis, C.; Che, M. Stud. Surf. Sci. Catal. 1993, 75, 1785. (21) Che, M.; Cheng, Z. X.; Louis, C. J. Am. Chem. Soc. 1995, 117, 2008. (22) Cheng, Z. X.; Louis, C.; Che, M. Stud. Surf. Sci. Catal. 1995, 91, 1027. (23) Carriat, J. Y.; Che, M.; Kermarec, M.; Verdaguer, M.; Michalowicz, A. J. Am. Chem. Soc. 1998, 120, 2059. (24) Carriat, J. Y.; Lepetit, C.; Kermarec, M.; Che, M. J. Phys. Chem. B 1998, 102, 3742. (25) Curtis, N. F.; Curtis, Y. M. Inorg. Chem. 1965, 4, 804. (26) Saito, R.; Kidani, Y. Bull. Chem. Soc. Jpn. 1978, 51, 159. (27) Joung, K. O.; O’Connor, C. J.; Sinn, E.; Carlin, R. L. Inorg. Chem. 1979, 18, 804. (28) Saito, R.; Kidani, Y. Bull. Chem. Soc. Jpn. 1979, 52, 2320. (29) De, G.; Biswas, P. K.; Chaudhuri, N. R. Bull. Chem. Soc. Jpn. 1983, 56, 3145. (30) Te Vrucht, M. L. E.; Griffiths, P. R. Talanta 1991, 38, 839. (31) Bertsch, C. R.; Fernelius, W. C.; Block, B. P. J. Phys. Chem. 1958, 62, 444. (32) Olsbye, U.; Wendelbo, R.; Akporiaye, D. Appl. Catal., A 1997, 152, 127. (33) Lambert, J. F.; Hoogland, M.; Che, M. J. Phys. Chem. B 1997, 101, 10347. (34) Goodgame, D. M. L.; Goodgame, M.; Weeks, M. J. J. Chem. Soc. A 1967, 1676. (35) Rossman, G. R.; Shannon, R. D.; Waring, R. K. J. Solid State Chem. 1981, 39, 277.

J. Phys. Chem. B, Vol. 109, No. 7, 2005 2845 (36) Ishida, K.; Takashahi, T.; Sato, Y.; Yano, S.; Hidai, M.; Kobayashi, K.; Sakurai, T. Acta Crystallogr., Sect. C 1988, 44, 83. Interatomic distances were measured from the structures of the salts via the Cambridge Crystallographic Reference Database. (37) Jacobs, P. A.; Uytterhoven, J. B. J. Catal. 1972, 26, 175. (38) Rajic, N.; Stojakovic, D.; Kaucic, V. Zeolites 1991, 11, 612. (39) Sokoll, R.; Robert, H.; Schmuck, I. J. Catal. 1990, 121, 153; 125, 276. (40) Li, C. P.; Proctor, A.; Hercules, D. M. Appl. Spectrosc. 1984, 38, 880. (41) Ne´grier, F.; Marceau, E.; Che, M. Chem. Commun. 2002, 1194. (42) Trouillet, L.; Toupance, T.; Villain, F.; Louis, C. Phys. Chem. Chem. Phys. 2000, 2, 2005. (43) Boujday, S.; Lambert, J. F.; Che, M. J. Phys. Chem. B 2003, 107, 651. (44) Bonneviot, L.; Legendre, O.; Kermarec, M.; Olivier, D.; Che, M. J. Colloid Interface Sci. 1990, 134, 534. (45) (a) Helliwell, M.; Gallois, B.; Kariuki, B. M.; Kaucic, V.; Helliwell, J. R. Acta Crystallogr., Sect. B 1993, 49, 420. (b) Meyer, L. M.; Haushalter, R. C.; Zubieta, J. J. Solid State Chem. 1996, 125, 200. (c) Riou-Cavellec, M.; Serre, C.; Ferey, G. C. R. Acad. Sci. Ser. IIc 1999, 2, 147. (d) Zhao, Y.; Zhu, G.; Jiao, X.; Liu, W.; Pang, W. J. Mater. Chem. 2000, 10, 463. (46) Law, T. S. C.; Williams, I. D. Chem. Mater. 2000, 12, 2070. (47) Burnett, M. G.; Hardacre, C.; Mallon, J. M.; Mawhinney, H. J.; Ormerod, R. M. Phys. Chem. Chem. Phys. 2000, 2, 1273. (48) Cseri, T.; Be´ka´ssy, S.; Kenessey, G.; Liptay, G.; Figueras, F. Thermochim. Acta 1996, 288, 137. (49) Lalia-Kantouri, M.; Tzavellas, L. C. Thermochim. Acta 2001, 371, 65. (50) Constable, E. C. In Metals and Ligand ReactiVity; VCH: Weinheim, Germany, 1996; p 275. (51) Riahi, G.; Guillemot, D.; Polisset-Thfoin, M.; Fraissard, J. Stud. Surf. Sci. Catal. 2001, 135, 1594. (52) Breen, C.; Forsyth, J.; Yarwood, J.; Hugues, T. Phys. Chem. Chem. Phys. 2000, 2, 3887. (53) Richardson, J. T.; Ortego, J. D., Jr.; Coute, N.; Twigg, M. V. Catal. Lett. 1996, 41, 17.