Studies in Surface Science and Catalysis 155 A. Gamba, C. Colella and S. Coluccia (Editors) 9 2005 Elsevier B.V. All rights reserved
Vanadium and niobium mixed-oxide catalysts obtained via solgel: preparation and catalytic behaviour in oxidative dehydrogenation of propane P. Moggi ~, S. Morselli ~, C. Lucarelli a, M. Sarzi-Amad~ a and M. Devillersb a Department of Organic and Industrial Chemistry, University of Parma, Parco Area delle Scienze 17/A, 43100 Parma, Italy b Unit6 de Chimie des Mat6riaux Inorganiques et Organiques, Universit6 Catholique de Louvain, Place Louis Pasteur 1/3, B-1348 Louvain-la-Neuve, Belgium 1. INTRODUCTION The development of efficient catalysts for the selective functionalization of alkanes is a major application area in the research on bulk and supported mixed oxides [ 1]. Among the various catalysts proposed for the oxidative dehydrogenation (ODH) of propane, mixed oxide catalysts containing vanadium are the most frequently studied. Extensive work is focused on mixed oxides of magnesium and vanadium [2-7], and on vanadium oxides supported on different materials such as silica, alumina, boria, titania [8-9]. The superior performance of vanadium oxide-based systems arises from the specific activity of V-O bonds in the C-H bond activation of alkanes, which is the rate-limiting step of the reaction. After the first C-H bond of alkane is broken, a surface alkyl species is formed. This intermediate can react either by breaking another C-H bond at the 13-position, thereby generating the dehydrogenation product, or by forming a C-O bond, resulting in oxygen containing products (including COx). The reactivity of lattice oxygen to form C-O bonds with the adsorbed hydrocarbon intermediate must be taken into great consideration, to avoid the complete oxidation of the alkane to carbon oxides. At this regard, it has been suggested that the reactivity of lattice oxygen is related to the strength of the metal-oxygen bond, which can also be viewed in terms of reducibility of the metal cations involved in the M-O-V bonds . Pure V205 is active in many partial oxidation reactions, because it is characterised by easily reducible V-O-V bonds; nevertheless, the selectivity to the desired product is often low. Mixed oxides of magnesium and vanadium lead to dominant propene selectivity, because they are characterised by less reducible lattice V-O-Mg bonds . Since the strength of the niobium-oxygen bond is greater than that of the vanadium-oxygen bond, niobium oxide was examined as catalyst in propane ODH, to see if it gave any improvement compared to the vanadium compounds. It was found to be a very selective catalyst for propane ODH, but with rather low activity [ 12]. An increase in the catalytic activity, without compromising the high selectivity to propene, was obtained by promoting niobium oxide with vanadium [ 13-14]. It was suggested that the activity increase could be related to the formation of surface V-O-V ensembles, which have more reactive bridging oxygen than the V-O-No ensembles (the former being easier to reduce than the latter). The high selectivity to propene was probably related to the presence of V-ONb bonds. Further experiments on niobia-supported vanadium oxide catalysts clearly
428 confirmed that active sites containing vanadium are definitely necessary, to provide a significant catalytic activity in propane ODH. Moreover, selectivity to propene strongly depended on the catalyst nature, being in relationship with the optimisation of the sorption properties of propene, to prevent total oxidation . The key point to reach high catalytic performances seemed to be the ability to control the nature and interdispersion of the mixedoxide phases. A preparation method, which causes the vanadium to be distributed homogeneously at the surface and in the bulk, was preferred over a method which deposits vanadium only at the surface, possibly in large clusters [ 13]. Sol-gel methods are known to be particularly powerful to achieve molecular scale dispersion in mixed oxides. As far as pure niobia is concerned, an extended sol-gel method based on the hydrolysis of a modified Nbalkoxide precursor obtained from Nb(OEt)5 and acetylacetone was reported to be an interesting method to produce nanocrystalline mesoporous films of the T-phase of Nb205 [ 16]. In addition, the implementation of a classical sol-gel route based on metal alkoxides to prepare Nb-rich (Nb/V = 6 or 10) amorphous V-Nb oxides, displaying high surface areas, was reported [ 17]. In a previous paper, we already reported preliminary results dealing with the development of a non-hydrolytic sol-gel route to synthesise Nb-V and Nb-V-Si catalysts [ 18]. In a subsequent work, the more commonly used hydrolytic sol-gel method was adopted as a viable alternative to the non-hydrolytic one . Very encouraging results were obtained in the ODH of propane to propene with Nb/V catalysts prepared by both sol-gel ways . The present contribution focuses on the optimisation of the hydrolytic sol-gel preparation method, to achieve the highest possible control of the molecular scale interdispersion of the mixed-oxide phases. More particularly, the effects of various promoters (HCI, HNO3, oxalic acid and citric acid) and higher Nb/V ratio on the hydrolytic sol-gel preparation are investigated. The aim is to study the possible relationship between the nature and interdispersion of the oxide phases present in the Nb-V systems, and the resulting catalytic performances in the ODH of propane. 2. EXPERIMENTAL Nb/V gels with an atomic ratio Nb/V of 1:1 were prepared starting from Nb(OPf)5 and WO(OPr~)3 as metal precursors. The syntheses were conducted by adding.dropwise a mixture of water, sol-gel promoter and 2-propanol to a stirred mixture of Nb(OPr')5 and VO(OPri)3 in 2-propanol, leading each time to a stable sol, which turned into a gel or a gelatinous precipitate within few days. HCI, HNO3, oxalic acid and citric acid as sol-gel promoters were adopted. In one case, no sol-gel promoter was added. Two samples with Nb/V ratio of 1:1 and 9:1 were prepared starting from NbC15 and VO(OPri)3 as precursors. A mixture of water and 2-propanol was added dropwise to a stirred mixture of precursors in 2-propanol, leading each time to a stable sol, which turned into a gel or a gelatinous precipitate within few days. A sample with a Nb/V ratio of 4.5:1 was prepared by adding a mixture of water and 2-propanol to a stirred mixture of Nb(OPrJ)5 and VO(OPr~)3 in 2-propanol, leading to a stable sol, which turned into a gel in few minutes. The xerogels were all activated in air at 350~ for 15 h, then at 400~ for 2 h, finally at 550~ for 4 h. They were characterised by surface area determinations, X-ray diffraction (XRD), RAMAN spectroscopy and scanning electron microscopy (SEM). Powder X-ray diffraction patterns were measured on a Philips PW 3710 diffractometer using the Cu K~ radiation (~ =1.54178/~). The crystalline phases were identified by reference to the powder diffraction data files (JCPDS-ICDD). The BET specific surface area measurements were carried out on a Micromeritics Pulse Chemisorb 2705 analyser using nitrogen at 77 K. Samples were previously outgassed under helium at 473 K. Raman spectroscopy was
429 performed on a DILOR-JOBIN YVON-SPEX spectrometer, model Olympus DX-40, equipped with a He-Ne (Z = 632.8 nm) laser. SEM micrographs were taken with a Philips XL 30 ESEM instrument equipped with BSE and SE detectors. The ODH catalytic experiments were generally performed with 0.3 g of catalyst, at atmospheric pressure, in the temperature range 400-500~ at a space velocity of 90 ml min "1 g~-~. The gas feed composition was 10% C3I-Is, 10% 02 and 80% He (total flow rate 30 ml minl). Some experiments were also performed at the higher space velocity of 300 ml min "t gc~t"l. 3. RESULTS AND DISCUSSION
3.1. Structural, spectroscopic and morphological characterisations The sol-gel preparation of the 1:1 Nb/V systems led to the formation of gelatinous precipitates with the only exception of the sample obtained by adding citric acid as promoter, which turned into a gel. The sol-gel preparation of 4.5:1 and 9:1 Nb/V systems led in both cases to the formation of yellow coloured gels. From these preliminary results, it was concluded that a high Nb amount or the presence of a V complexing agent such as citric acid are necessary conditions for preparing Nb/V mixed gels, otherwise the solution of Nb and V precursors is not stable upon water addition, and a yellow gelatinous precipitate readily forms, independently of the water amount and/or the precursors concentration adopted. Table 1 Results of XRD analysis and surface area (m 2 gl) determination Sample Crystalline phases BET Specific Surface Area 1:1 Nb/V NbVO5, ]~blgW4055;V205 5.1 1:1 Nb/V (HCI)
NbVOs; NblsV4055; V205
1:1 Nb/V (citric acid)
1:1 Nb/V (oxalic acid)
NbVOs; NblsV4055; V205
9:1 Nb/V (NbCIs)
1:1 Nb/V (HNO3)
1:1 Nb/V (NbCIs)
The results of X R measurements and surface area determinations on the prepared samples after the calcination treatments are reported in Table 1. As it can be seen, the systems were characterised by low surface areas whatever the additive used, the 4.5:1 Nb/V sample showing the highest value of 10.8 m2 g-~. The crystalline phases NbVO5 [46-0046] and Nb~sV4055 [46-0087] were recognised in the XRD patterns of all the precipitated 1:1 N b N systems; moreover, in agreement with the overall stoichiometry, crystalline V205 [09-0387] was detected as a third minor phase in the samples: (i) 1:1 N o N prepared without adding solgel promoter, (ii) 1:1 Nb/V (HCI) (Fig. l a) and (iii) 1:1 Nb/V (NbCIs). In contrast, NbVO5 [46-0046] containing only small amount of Nb~gV4055 [46-0087] (shoulder at 20 ~ 22 ~ was detected in the gel-derived 11 Nb/V (citric acid) system (Fig. 1b). The presence of crystalline V2Os in the 1:1 Nb/V (HCI) sample is evidenced in Fig. 2, where the XRD patterns of this
430 sample and of the gel-derived 11 Nb/V (citric acid) system are superimposed to the XRD pattern of VzO5 [09-0387], in the ranges of 20 14-24 ~ and 26-34 ~ 2000
1400 1200 1000
600 4O0 20O
2500 20O0 o 1500 ,2,o 1000 !
20 (deg) Fig. 1. XRD pattern of: (a) the 11 NoN system prepared by adding HCI as sol-gel promoter; (b) the 11 No/V system prepared by adding citric acid as sol-gel promoter. (x) NbVO5 [46-0046]; (-) NolsV4Oss [46-0087]
Co) !........................................................................ !
Fig. 2. Comparison of XRD patterns: (a) 11 NoFV (HCI" (b) 11 NoN (citric acid); (c) crystalline V205 [09-0387]
431 The Rarnan data are reported in Table 2. The heterogeneity of precipitated 11 Nb/V systems was evidenced by recording Raman spectra at different positions on the surface of each sample. Several phase compositions were detected, in agreement with XRD analysis. Table 2 Raman results ,,S~ple 1"1 N b N
1-1 Nb/V (HCI)
11 Nb/V (HNO3)
l'lNb/V(citricacid) 1:1 Nb/V (oxalic acid)
11 Nb/V (NbCIs) 4.5:1 Nb/V
9:1 Nb/V (NbCIs) .........................
" .....~ m a n shift (cm, 1) ...i ...........:..:................. .......... Phases ......._.:=:.:~i:i 995; 701,528; 481; 404: 302; 283 V20~ 779; 732; 631; 374; 342; 316; 229 NbVO5 970 NblsV4055 994; 700: 529; 479; 404; 304; 284 V2Os 1018; 779; 734; 630; 375; 341; 230 NbVO5 970 NblsV4055 996; 700; 529; 479; 404; 304; 284 V2()5 1018; 986; 946; 778; 734; 630; NbVO5 375; 341; 316; 230 1019; 988; 946; 778; 731; 698 NbVO5 (sh); 628; 374; 341; 317; 230 994; 701" 528; 481" 403: 303; 284 V20s 1016; 987; 946; 781; 727; 632; NbVO5 374; 340; 316; 228 970 NDIsV4055 996; 703; 528; 479" 404; 303:284 V2()s 1020; 984; 340 NbVO5 987; 898; 630; 533; 470; 313 (sh); t!t-Nbz()s  259; 233 1016; 970; 732; 322 NblsV4055 993- 899; 835; 667; 622;542; 470; H-Nb~.O~ 350 (sh); 310 (sh)j.259; 235 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Microcrystalline V205 was present in all the precipitated samples. The Raman bands at 284, 303, 483, 528, 703 and 996 cm l coincided with the main features of Raman spectra of crystalline V2Os assigned in the literature [21 ]. The Raman bands at 1020, 988, 946, 778, 731, 698 (sh), 628, 374, 341, 317, 290 and 230 cm-I of these systems coincided with the main features of Raman spectra of the 11 Nb/V (citric acid) sample (Fig. 3), in which the practically pure NbVO5 phase had been detected by XRD. These bands were systematically evidenced in the 1"1 Nb/V (citric acid) sample by recording spectra at different positions, thus confirming the homogeneity of the catalyst. NbVO5 is expected to contain NbO6 octahedral units sharing comers with VO4 tetrahedral units . In particular, in the (100) plane the structure can be described as zigzag chains of corner-shared NbO6 octahedra sharing four oxygens with the VO4 tetrahedra, which point up and down alternatively. On the basis of the spectroscopic data previously reported for the structurally-related NbPO5 compound , Raman bands due to the VO4 units were assigned as follows: 1020 cm l (weak), 988 cm "l (strong), 374 cm ~ (medium), 230 cm -I (medium). These bands are associated respectively to the stretching (antisymmetrical and symmetrical) and bending (in and out-of-plane) modes of terminal V-O bonds . The Raman bands at 778 and 340 cm l were associated to the stretching and bending modes of V-O bonds in VO4 tetrahedra, whereas the Raman bands at 946, 731,628, 317 and 290 cm -1 were attributed to the stretching and bending modes of Nb-O bonds in slightly and highly distorted NbO6 octahedral units [22-25], the bands at 946, 630
and 290 cm ~ being characteristic of distorted NbO6 octahedral structures connected by sharing comers [24,25]. Of a difficult interpretation was the shoulder at about 700 cm -~. According to spectroscopic studies on amorphous niobium and vanadium single oxides , it could be assigned to stretching vibrations of a residual polymeric amorphous mixed oxide deriving from the xerogel precursor, not destroyed by calcination.
Raman shift (cm -1) Fig. 3. Raman speetaun of the 1 1 Nb/V system prepared by adding citric acid as sol-gel promoter The Raman spectra of the 1:1 Nb/V (citric acid) sample are in quite good agreement with those recently reported in the literature by Briand et al. and Ballarini et al. [26,27]. The main differences are found in the region between 400 and 600 cm 1, in which the authors report the presence of two bands at about 530 and 475 cm -~ that are absent in our spectra. Since the broad band at 700 cm q is also reported in both studies, it is supposed that the bands at 530 and 475 cm -1 and the one at 700 cm -~ might be related to the polymeric functionality of a less ordered mixed oxide. Raman spectroscopy is particularly sensitive for detection of ordering in structures since ordering gives rise to strong Raman bands, while broad bands are seen in disordered structures. In contrast to Raman spectroscopy, XRD does not reflect such local disorder so much. The NblsV4055 phase was not easy to identify by Raman analysis owing to the superimposition of its Raman bands to those of NbVO5 and V205 phases. Nevertheless, the presence of a Raman band at 970 cm q in the spectra of samples in which the Nb~sV4055 phase had been detected by XRD, and not in the pattern of the 1:1 No/V (citric acid) sample in
which the practically pure NbVO5 phase had been detected, led to consider this band as indicative for this phase. SEM investigations confkrmed the results given by XRD and Raman analyses, by evidencing the presence of different crystalline phases in the 1:1 Nb/V precipitated system prepared without adding a sol-gel promoter. Fig. 4a shows the SEM image of the micrometersized crystallites of NbVOs, which was the predominant phase, uniformly distributed. Fig. 4b shows the SEM image of needle-shaped, thin crystallites of V205, less abundant and mainly localised on the surface of the catalyst particles. In Fig. 5, the SEM image of the sol-gel derived 1:1 Nb/V (citric acid) sample is reported. The system was homogeneous over the entire area examined. No segregation or surface decorations by extra phases were detected.
(a) (b) Fig. 4. SEM images of the 1:1 Nb/V system prepared without the addition of a sol-gel promoter: (a) NbVO5 crystallites; (b) V205 crystallites.
~ i .........
,IN i~W~ ..... ' : ~ j .......
Fig. 5. SEM image of the 1:1 Nb/V (citric acid) system.
1000 x .El 500
c .... 10
20 (deg) 1400
Fig. 6. a) XRD pattern of the 9:1 NbN system; b) XRD pattern ofthe 4.5:1 Nb/V system. (x) Nb205 I32-07111 (-) Nb2Os [27-1003] The XRD spectra of the 4.5:1 Nb/V and 9:1 Nb/V systems after the thermal treatment evidenced the presence of crystalline Nb205 in both samples. In the 9:1 Nb/V system (Fig. 6a), two Nb205 phases were identified: JCPDS [32-0711], corresponding to the metastable M form of Nb205 and predominant in the pattern, and [27-1003], less abundant, corresponding to the low-temperature T orthorhombic form of Nb2Os. The M-form should be regarded as a less ordered precursor of the high temperature monoclinic H-form . In the 4.5:1 N b N system (Fig. 6b), the Nb205 [27-1003] phase was predominant in the pattern, while the Nb205 [320711 ] phase was detected in a minor amount. The peaks corresponding to the orthorhombic phase were slightly shifted towards higher values of theta. It was hypothesised that diffusion of some V 5+ ions had occurred into the structure of Nb2Os [27-1003], causing the shift of the XRD peaks. In both 4.5:1 Nb/V and 9:1 Nb/V systems, there was no evidence of Vcontaining crystalline phases. The presence of microcrystalline V205 was excluded in both samples by Raman analysis. The Raman spectra of the 9:1 Nb/V system showed the characteristic bands of the H-Nb205 phase (Fig. 7) : 993, 893, 667, 622, 542, 470, 309, 259, and 235 cm 1. These bands were related to the M-form detected in the sample by XRD. The Vl (A~) Raman active mode of the NbO4 tetrahedron present in the structure of H-Nb205 was also detected at 835 cm l. Nb s§ ions seldom occur in tetrahedral coordination. In the
9(q) pu~ (e) s-lmod [email protected]
o ~ re poDallOO tuo~,s.~sA/qN I :~;t, otOJo mtmaods umuelt "8 "i~!:l
(,_tua) ~qs uemea OOZ;
(~.ma) g.nts ueme,,d 0001
OOZ; 0017 009
"mols.~s A/qN I:6 ogljo ~ ~ I s
(~.tua) g.rqs ueme~ OOZ:
.. ~ .....
t~ ~ t::
"[~] oldm~s A/qN I :O ano u! eapo~.Ua~ aoj poui~!ss~~tttl~o~ OSOl~s! qa!q~ le sawadd~ p~q ueum~ ,^ oql ~apoqR.Uo~ ~oqN ~u.m.m~uoo 'punodmoo ~oqNA
436 In the 4.51 Nb/V system, the presence of two different phases was evidenced by recording Raman spectra at different positions. In one pattern (Fig. 8a), the main bands of a Nb205 phase at about 990, 630, 260 and 230 cm 1 were detected. In a second one (Fig. 8b), sharp bands at 1018, 970, 730 and 320 cm -l evidenced the presence of a microcrystaUine mixed phase containing vanadium and niobium, most probably NblsV4Oss. SEM investigations pointed out the homogeneous nature of the 4.5/1 Nb/V system, and evidenced the presence of two similar but different crystalline phases (Fig. 9).
Fig. 9. SEM image of the 4.51 Nb/V system.
3.2. Catalytic results The catalytic results obtained with the 11 Nb/V prepared systems at 450~ and space velocity of 90 ml min'lg,~t-1 are reported in terms of C3I-Is conversion and propene selectivity in Fig. 10. 60 50 40 30
HNO3 D C3H8 cony. (%)
1C3H6 Sel. (%)
Fig. 10. Catalytic performances of 11Nb/V catalysts in propane ODH. P = latm; T = 450~ = 30 ml min1 (10% C~Is, 10% 02 and 80% He); catalyst = 0.3g.
437 The 1:1 Nb/V samples obtained respectively without adding a sol-gel promoter and by adding HCI as sol-gel promoter gave the highest conversions of C3I-Is. This high activity was counterbalanced by a low selectivity to propene. The samples prepared by adding HNO3 or citric acid as promoter showed rather poor activity, but they were more selective to propene. It was concluded that the presence of segregated crystalline V2Os in the former samples contributed to increase the propane conversion to mainly COx products. In Fig. 11, the catalytic results collected on the gel-derived 1:1 Nb/V (citric acid), 4.5:1 Nb/V and 9:1 Nb/V samples and on the precipitated 1"1 Nb/V sample at 450~ and space velocity of 90 ml m i n l g ~ l are compared. The 9:1 Nb/V sample containing the lowest amount of V behaved similarly to pure Nb2Os, showing rather poor activity and high selectivity to propene, about 50%. Both activity and selectivity increased only slightly when a higher amount of V was adopted in the 1:1 Nb/V (citric acid) system, containing V mainly as Nb-V mixed oxide. This evidenced that the V-O-Nb functionality in the NbVO5 compound is rather similar to the Nb-O-Nb one. The presence of small amount of V-O-V bonds as crystalline V205 was sufficient to increase significantly the catalyst activity in the precipitated 1:1 Nb/V sample; the system however was poorly selective to propene. The 4.5:1 Nb/V sample showed both higher activity and selectivity to propene than the precipitated 1:1 Nb/V one. This catalyst was also that displaying the highest specific surface area. The system contained an intermediate amount of V in solid solution with Nb oxide. Since no crystalline or microerystalline V205 was detected, it could be hypothesized that a peculiar V-O-Nb functionality existed in this sample, more reducible than the V-O-Nb functionality in the NbVO5 phase, and less reducible than the V-O-V one. The 4.5:1 N b N sample was tested also at the higher space velocity of 300 ml minl g,~t"l. The results are reported in table 3. A very good yield of propene, comparable to the best yields reported in the literature for various catalysts , was obtained at 550~ as the right compromise between activity and selectivity performances.
60 50 40 30
1:1NbN (Citric Acid)
El C3H8 cony. (%)
4.5:1 N b N
IIC3H6 Sel. (%)
Fig. 11. Catalytic performances of 1:1, 4.5:1 and 9:1 Nb/V catalysts in propane ODH. P = l atm; T = 450~ gas feed = 30 ml min-I (10% C~-h, 10% 02 and 80% He); catalyst = 0.3g.
438 Table 3 Catalytic performances of the 4.5:1 Nb/V sample in propane ODH. P = latm; gas feed = 30 ml min l (10% C3Hs, 10% O2 and 80% He); space velocity = 300 ml min-1 gc~tl. XC3H8 = propane conversion; Sc3n6 = selectivity to propene; YC3H6 = yield of propene T (~ X C388 (%) S C3I-I6(%) Y C386 (%) 400 3.9 72.6 2.8 450 8.6 55.7 4.8 500 12.7 55.9 7.1 550 19.4 47.4 9.2 4. CONCLUSIONS The hydrolytic sol-gel method was adopted in this work to prepare N b N mixed oxides systems as potential catalysts for propane ODH. Several 1:1 Nb/V systems were prepared by varying sol-gel promoter and Nb precursor. It was found that the addition of citric acid as sol-gel promoter stabilises the solution of Nb and V precursors after the addition of water, favouring the direct formation of a gel. As a result, a homogeneous and well-interdispersed 1:1 Nb-V system was obtained by this method. In all the other cases, precipitation of a yellow solid occurred, leading after calcination to heterogeneous systems, in which the presence of segregated V205 was detected. Two systems with higher amount of Nb, 9:1 Nb/V and 4.5:1 NbN, were also prepared by hydrolytic sol-gel method. In both cases, the addition of citric acid was not necessary. Sols containing higher amounts of Nb precursor were stable and turned rapidly into gels. In both 9:1 Nb/V and 4.5:1 Nb/V samples, there was no evidence of crystalline phases containing V. Only the presence of crystalline Nb205 was detected by XRD. In the 4.5:1 Nb/V sample, however, XRD pattern suggested that some diffusion of V 5+ ions had occurred into the structure of orthorhombic Nb20~ phase, and Raman patterns evidenced the presence of microcrystalline NblsV4055. A Nb-V 'synergistic effect' in the ODH of propane seemed to characterise the 4.5:1 Nb/V sample, which showed both higher activity and selectivity to propene than the l:l Nb/V samples. The 9:1 Nb/V sample containing the lowest amount of V was found to behave similarly to pure Nb~Os, showing rather poor activity and high selectivity to propene, about 50%. Both activity and selectivity were found to increase only slightly with the l:l Nb/V (citric acid) system, containing V mainly as Nb-V mixed oxide. The presence of small amount of V-O-V bonds as crystalline V2Os increased significantly the catalyst activity of precipitated 1:1 Nb/V samples.
Acknowledgments This work was performed within the frame of a Concerted Research Action of the 'Communaut6 Franfaise de Belgique'. The authors also acknowledge the financial support from the Ministero dell'Universita e della Ricerca Scientifica e Tecnologica (Rome) and the Belgian National Fund for Scientific Research (Brussels). M. Sarzi-Amad6 was recipient of a Socrates grant for a three-months stay in Louvain-la-Neuve.
REFERENCES  G. Centi and P. Mazzoli, Catal. Today 28 (1996) 351.  M. A. Chaar, D. Patel, M. C. Kung and H.H. Kung, J. Catal. 105 (1987) 483.  M. A. Chaar, D. Patel, M. C. Kung and H.H. Kung, J. Catal. 109 (1988) 463.
 M. J. Philips and M. Ternan, Proc. 9th Int. Congr. Catal. 4 (1988) 199B1.  D. Siew Hew Sam, V. Soenen and J. C. Volta, J. Catal. 123 (1990)417.  A. Corma, J. M. LSpez Nieto and N. Paredes, J. Catal. 144 (1993) 425.  X. Gao, P. Ruiz, Q. Xin and B. Delmon, J. Catal. 148 (1994) 56.  G. Colorio, J. C. V6drine, A. Auroux and B. Bonnetot, Appl. Catal. A 137 (1996) 55.  T. Blasco and J. M. Lopez Nieto, Appl. Catal. A 157 (1997) 117. [ 10] H. H. Kung, Adv. Catal. 40 (1994) 1. [ 11] F. Cavani and F. Trifiro, Catal. Today 24 (1995) 307.  R. H. Smiths, K. Seshan and J. R. H. Ross, J. Chem. Sot., Chem. Commun. (1991) 558. [ 13] R. H. Smiths, K. Seshan, H. Leemreize and J. R. H. Ross, Catal. Today 16 (1993) 513.  J. R. H. Ross, R. H. H. Smits and K. Seshan, Catal. Today 16 (1993) 503.  T.C. Watling, G. Deo, K. Seshan, I.E. Waehs and J.A. Lercher, Catal. Today 28 (1996) 139.  F. Lenzmann, V. Shklover, K. Brooks and M. Gr/itzel, J. Sol-Gel Sci. Technol. 19 (2000) 175.  M. Catauro, C. Pagliuca, L. Lisi and G. Ruoppolo, Thermochim. Acta 381 (2002) 65. [ 18] F. Barbieri, D. Cauzzi, F. De Smet, M. Devillers, P. Moggi, G. Predieri and P. Ruiz, Catal. Today 61 (2000) 353. [ 19] P. Moggi, G. Predieri, D. Cauzzi, M. Devillers, P. Ruiz, S. Morselli and O. Ligabue, Stud. Surf. Sci. Catal. 143 (2002) 149.  P. Moggi, M. Devillers, P. Ruiz, G. Predieri, D. Cauzzi, S. Morselli and O. Ligabue, Catal. Today, 81 (2003) 77.  T. R.Gilson, O. F. Bizri and N. Cheetham, J. Chem. Soc. Dalton Trans. 291 (1973)  J. M. Amarilla, B. Casal and E. Ruiz-Hitzky, Mater. Lett. 8 (1989) 132  A. A. McConnell, J. S. Anderson and C. N. R. Rao, Spectrochim. Acta 32A (1976) 1067.  I. R. Beattie and T. R. Gilson, J. Chem. Soc. A (1969) 2322.  M. A. Vuurman and I. E. Wachs, J. Phys. Chem. 96 (1992) 5008.  N. Ballarini, F. Cavani, C. Cortelli, C. Giunchi, P. Nobili, F. Trifirb, R. Catani and U. Comaro, Catal. Today 78 (2003) 353.  L. E. Briand, J-M Jehng, L. Comaglia, A. M. Hirt and I. E. Wachs, Catal. Today 78 (2003) 257.  H. Sch~fer, R. Gruehn and F. Schulte, Angew. Chem. Internat. Edit. 5, (1966) 40.  S. Albonetti, F. Cavani and F. Trifirb, Catal. Rev. Sci. Eng. 39 (1997) 413.