Synthesis of Nb-containing mesoporous silica molecular sieves

Synthesis of Nb-containing mesoporous silica molecular sieves

Applied Catalysis A: General 241 (2003) 39–50 Synthesis of Nb-containing mesoporous silica molecular sieves Analysis of its potential use in HDS cata...

498KB Sizes 0 Downloads 14 Views

Applied Catalysis A: General 241 (2003) 39–50

Synthesis of Nb-containing mesoporous silica molecular sieves Analysis of its potential use in HDS catalysts Luis Cedeño a,b , Diana Hernandez a,b , Tatiana Klimova a , Jorge Ramirez a,∗ a

UNICAT, Fac. de Qu´ımica, UNAM, Mexico City 04510, DF, Mexico Instituto Mexicano del Petroleo, Mexico City 07730, DF, Mexico

b

Received 14 June 2002; received in revised form 26 July 2002; accepted 27 July 2002

Abstract In the preparation of mesoporous materials of the MCM-41 type, the synthesis aging time is an important parameter that can alter the structural and chemical properties of the final product. This work analyzes the effect of synthesis aging time on the structural and chemical properties of niobium-containing mesoporous silica with two different Si/Nb atomic ratios, 52 and 31. Samples of Nb-modified mesoporous silica were prepared at aging times of 2, 3, 4 and 8 days. The samples were characterized by N2 adsorption measurements, XRD, differential thermogravimetric analysis, FT-IR, pyridine adsorption analyzed by FT-IR, HREM, SEM, DRS and Raman spectroscopy. The results show that synthesis aging time has a definitive influence in the textural, structural and chemical properties of niobium-modified mesoporous silica. Short synthesis aging times (2 days) do not allow the proper incorporation of niobium to the silica framework. With synthesis aging times of 3 and 4 days the niobium is well incorporated to the silica framework. Synthesis aging times up to 4 days promotes the formation of hexagonal structures while aging during 8 days induces the formation of cubic structures. Niobium sulfide has been reported to have good hydrodesulfurization (HDS) and hydrogenation (HYD) activities. The results show that niobium incorporated to the framework of Si-MCM-type materials undergoes partial sulfidation, as attested by the TPS experiments and its small catalytic activity in the thiophene hydrodesulfurization reaction. The solids studied here present therefore an interesting potential application as supports for HDS catalysts. © 2002 Elsevier Science B.V. All rights reserved. Keywords: Nb-modified MCM-41; Synthesis; Characterization; HDS

1. Introduction The recent success obtained in the synthesis of well-defined mesoporous molecular sieves of the M41S family [1–3] opened an interesting field of applied research due to the potential application of these materials as catalysts and catalytic supports in fine chemistry and the petroleum industry.

∗ Corresponding author. Fax: +52-56225366. E-mail address: [email protected] (J. Ramirez).

The catalytic properties of molecular sieves rely on the presence of active sites in their frameworks. In the case of MCM-41, incorporation of heteroatoms to the otherwise electrically neutral purely siliceous framework may generate active sites. Among the different transition elements incorporated to the MCM-41 framework, Nb is particularly interesting since Nb-containing materials have shown catalytic properties in several reactions such as hydrocracking [4], ethanol dehydration [5], ethanol dehydrogenation [6], dimerization and oligomerization of ethene [7], oxidation reactions [8] and hydrotreating reactions [9,10].

0926-860X/02/$ – see front matter © 2002 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 6 - 8 6 0 X ( 0 2 ) 0 0 4 2 9 - 5

40

L. Cedeño et al. / Applied Catalysis A: General 241 (2003) 39–50

The incorporation of niobium to the framework of high-surface area mesoporous silica has been recently reported [16]. It was shown that niobium can be successfully incorporated to the framework of mesoporous silica through a micellar templating route, and that the porous structure of this system can be tailored to a certain extent by controlling the synthesis parameters such as dopant concentration, sol pH, and aging temperature. However, because the rates of polymerization of the silicium and niobium species are different, the synthesis aging time must be also an important parameter that can change not only the textural but also the surface chemical properties of the final solid. In the first part of this study, we report the synthesis and characterization of Nb-containing mesoporous silica, synthesized with different aging times (2, 3, 4 and 8 days), and Si/Nb atomic ratios (31 and 52). Regarding the use of supported niobium sulfide in hydrotreatment reactions, there are only a few studies about niobium sulfide supported on carbon [11–13] and SiO2 [14,15]. Allali et al. [13] studied the problem of transposition of niobium sulfide from the bulk to the supported state, using alumina and carbon as supports in order to study the effect of using supports with high and low interaction with the active phase. Their results show that the nature of the support influences drastically the HDS activity as well as the nature of the precursor niobium oxide species and the difficulty of their sulfidation. Niobium sulfide as been reported as a good catalyst for HDS and HYD reactions [17] and the addition of niobia as promoter to the typical hydroprocessing catalysts resulted in an increase of the HDS and HDN activities [9]. Although it appears interesting to use Nb mesoporous silica as support for HDS catalysts, it is of interest to explore also if these high-surface area Nb-containing mesoporous materials present by themselves some HDS activity. To this end, in the final part of this study, the Nb mesoporous materials synthesized here were characterized by temperature programmed reduction in the sulfided state (TPR-S), and tested in the thiophene hydrodesulfurization reaction.

2. Experimental Niobium-containing mesoporous silicas with Si/Nb atomic ratios of 52 and 31 were synthesized using ag-

ing times of 2, 3, 4 and 8 days for the former and 4 days for the later. The samples were prepared according to the following procedure: Ludox was used as the silicium source, niobium ethoxide served as the metal precursor and cetyltrimethyl ammonium bromide as the surfactant. Using a Si to surfactant ratio of 10.25, the adequate amounts of reactants to yield Si/Nb ratios of 31 or 52 were loaded into a teflon-lined autoclave and heated at 373 K for the selected aging time. The product was filtered and washed with water and ethanol. The solid products were calcined in static air at 873 K for 22 h. Hereafter, the samples will be referred as y-Nb-xd-MCM, where y is the Si/Nb ratio and x is the aging time expressed in days. The mesoporous materials were characterized by several techniques. X-ray diffraction patterns were recorded using a Siemens D500 powder diffractometer with Cu K␣ radiation. Surface areas and N2 adsorption–desorption isotherms of the samples were measured with an ASAP 2000 Micromeritics apparatus. Nitrogen physisorption isotherms were analyzed using the BJH method. Prior to the textural characterization the samples were outgassed for 8 h in vacuum at 623 K. For HREM analysis the samples were ultrasonically dispersed in ethanol, and a drop of the suspension was deposited on a holey carbon coated copper grid. Micrographs were recorded with a 2010 Jeol microscope operated at 200 kV with point to point resolution of 1.9 Å. Elemental composition was determined by SEM-EDX in a Stereoscan 440 Leica microscope equipped with an energy dispersive X-ray (EDX) elemental analysis system. Infrared spectral data were gathered using a Nicolet Magna-IR 760 at 4 cm−1 resolution, using the KBr pastille method. The acid properties of the samples were evaluated by analysis of the IR spectral features of adsorbed pyridine. The adsorption of pyridine was performed on self-supported wafers using a special IR cell connected to high vacuum and a gas-manipulating manifold. Before pyridine adsorption at room temperature (RT), the catalysts were pretreated under flow of oxygen at 673 K during 20 h, followed by evacuation (2 h) and cooling. The IR spectra of the samples were taken, at RT, 423, 473, 573 and 673 K, after evacuating during 30 min. For differential thermogravimetric (DTG) analysis of the calcined samples, a heating rate of 10 K/min from RT to 1073 K, under flow of N2 (40 cm3 /min) was used. UV-Vis diffuse reflectance

L. Cedeño et al. / Applied Catalysis A: General 241 (2003) 39–50

spectra were recorded with a Cary 5E UV-Vis-NIR spectrometer in the 250–2500 nm wavelength range. Raman spectra were recorded in a Nicolet FT-Raman 950 spectrometer. The samples were pretreated at 373 K during 12 h in static air before taking the spectrum. For the catalytic evaluations in the thiophene HDS reaction, 0.25 g of catalyst were used in all cases. Prior to the reaction the catalysts were sulfided in situ for 4 h at 673 K under a stream of H2 S (15% v/v)/H2 . The reaction products were analyzed using a Hewlett-Packard 5850 chromatograph equipped with a FID detector and an Ultra I-30 m column. Temperature programmed reduction of the sulfided catalysts (TPR-S), recorded immediately after the reaction test, were performed using a hydrogen/argon (30% H2 v/v, 25 cm3 /min) mixture at atmospheric pressure, 0.25 g of sample and a heating rate of 10 K/min. The outlet

41

reactor gas stream was analyzed sequentially with TC and UV detectors in order to quantify the hydrogen consumed and the evolution of H2 S, respectively.

3. Results and discussion Fig. 1 shows the diffractograms of the samples prepared with Si/Nb ratio of 52 at different aging times. All the samples present the intense diffraction peak at low angles (2θ = 2.45–2.65◦ ), characteristic of MCM-type materials, which corresponds to reflections of the (1 0 0) planes. Two small peaks between 4 and 5◦ , which correspond to the reflections of the (1 1 0) and (2 0 0) planes, were also observed. These reflections are better resolved for the 52-Nb-3d-MCM41 and 52-Nb-4d-MCM41 samples, indicating that synthesis

Fig. 1. X-ray powder diffraction patterns of 52-Nb-xd-MCM samples.

42

L. Cedeño et al. / Applied Catalysis A: General 241 (2003) 39–50

aging times between 3 and 4 days lead to more ordered structures. However, synthesis aging times beyond 4 days appear to produce disorder in the structure, evidenced by the broadening of the main peak. These results are in agreement with previous reports by Gallis and Landry [18] and Zhao et al. [19], showing that the crystallinity of MCM-type materials reaches a maximum with synthesis aging time. In agreement with literature data on MCM-41 samples [20], the TG and DTA analysis of our samples only show the loss of water (RT–373 K) and the desorption and decomposition of the template agent associated to SiO (373–573 K) and Brönsted sites (573–773 K). No phase transitions were observed for any of the samples, indicating the absence of formation of niobium silicates or any other phase.

Table 1 Results of XRD and N2 adsorption measurements Sample

Nb (wt.%)a

Sg (m2 /g)

V (cm3 /g)

dp (Å)

ξ (Å)

52-Nb-2d-MCM 52-Nb-3d-MCM 52-Nb-4d-MCM 52-Nb-8d-MCM 31-Nb-4d-MCM Si-MCM-41

2.52 2.44 2.92 2.52 4.82 0.0

1263 683 1009 953 1351 850

0.81 0.43 0.71 0.65 0.83 0.80

19.9 24.5 23.2 20.0 19.9 26.4

18.5 16.4 16.6 18.8 15.4 15.4

a

Niobium content obtained by SEM-EDX; Sg: BET surface area; V: pore volume; dp: pore diameter; ξ : wall thickness.

The results from the chemical analysis of the different samples obtained by SEM-EDX are shown in Table 1, where we also present some of the results from the textural and structural (XRD). The surface area

Fig. 2. Nitrogen adsorption–desorption isotherms of (䉬) 31-Nb-4d-MCM, (×) 52-Nb-2d-MCM, (䊐) 52-Nb-3d-MCM, (䊊) 52-Nb-4d-MCM and (䉱) 52-Nb-8d-MCM.

L. Cedeño et al. / Applied Catalysis A: General 241 (2003) 39–50

reaches a minimum for the sample 52-Nb-3d-MCM. It is not clear why this sample exhibits a smaller surface area (683 m2 /g) while the other samples show areas close to 1000 m2 /g. However, this sample exhibits the same type of nitrogen adsorption–desorption hysteresis curve characteristic of mesoporous materials (see Fig. 2). All the samples with Si/Nb ratio of 52 show a unimodal pore size distribution. In contrast, the 31-Nb-4d-MCM sample, with higher niobium content, shows a small contribution of macropores around 350 Å. This is in agreement with the idea that as the Si/metal ratio decreases, the deterioration of the mesoporous structure increases. In this case, it is possible that the larger pores are partially blocked due to the existence of some extra-framework niobium oxide and/or to some destruction of the ordered tubular

43

porous structure. As will be seen later from the DRS results, if niobium oxide is present in our samples, it must be highly dispersed [21] and should not affect the shape of the pore size distribution curve. The pore diameter of the samples reaches a maximum for the sample with 3 days synthesis aging time and this coincides, as expected, with a minimum of the wall thickness of this sample. Accordingly, the variations in the interplanar distance (d0 ) with synthesis aging time show a maximum for the same sample. The difference between the wall thickness obtained for a purely Si-MCM sample (15.4 Å) and the ones modified with niobium indicates that Nb has been incorporated to the siliceous framework. The HREM micrographs of the different samples are shown in Fig. 3. In all the samples, except the

Fig. 3. HREM micrographs of (a) 31-Nb-4d-MCM, (b) 52-Nb-4d-MCM and (c) 52-Nb-8d-MCM.

44

L. Cedeño et al. / Applied Catalysis A: General 241 (2003) 39–50

Fig. 4. SEM micrographs of (a) 52-Nb-4d-MCM and (b) 52-Nb-8d-MCM.

Fig. 3. (Continued)

52-Nb-2d-MCM, it was possible to observe the typical hexagonal pore arrangement of MCM-41 type materials. The samples with Si/Nb ratios of 31 (Fig 3a) and 52 (Fig. 3b) show long range order, indicating that at the Nb contents used here the hexagonal arrangement of the pore structure is still preserved. However, in the sample with 8 days synthesis aging time some parts of the sample presented a cubic arrangement similar to a MCM-48 type material (see Fig. 3c). It appears then that extended reaction times induce a change from hexagonal to cubic arrangement of the mesoporous structure. This result is in agreement with the SEM observations that show some cubic particles only in the sample with 8 days reaction time (see Fig. 4). A

niobium mapping performed by SEM-EDX on the different samples showed that in all cases the distribution of niobium was homogeneous. This observation supports the idea that the incorporation of niobium into the silica framework, at the concentrations used here, does not produce large particles of extra-framework niobium oxide species, typically Nb2 O5 . The absence of Nb2 O5 particles in the different samples was also supported by the FT-Raman results (not shown), which did not show the presence of the characteristic Nb2 O5 band at about 680 cm−1 [22]. Because the samples were analyzed fresh from the oven, and the observations were not made under vacuum to preserve the dehydrated state of the sample during the analysis, we observed only a small evidence of the band at 960 cm−1 , characteristic of niobium in interaction with the silica.

L. Cedeño et al. / Applied Catalysis A: General 241 (2003) 39–50

45

Fig. 5. Infrared spectra of (a) 52-Nb-2d-MCM, (b) 52-Nb-3d-MCM, (c) 52-Nb-4d-MCM and (d) 52-Nb-8d-MCM.

The FT-IR spectra of the samples are presented in Fig. 5. All samples present the characteristic vibrations at 1080–1096, 816–808 and 456–460 cm−1 , corresponding to the asymmetric and symmetric vibrations of the T–O–T structural units, and to the deformation modes of TO4 tetrahedra, respectively [23]. We also observe the presence of a band at 960 cm−1 , which has been previously taken in other studies as a proof of the incorporation of metals in the framework of MCM [24,25]. However, the origin and assignment of this band is still under debate and its presence cannot be taken as a conclusive proof of the incorporation of niobium into the silica framework. Nevertheless, we observe slight shifts in the bands corresponding to the internal vibrations of the TO4 units. These IR bands are sensitive to the composition and bond angles [26] and therefore, their shift and slight broaden-

ing, observed with synthesis aging time, may be taken as an indication of a change in the composition of the framework structure. The IR spectra of adsorbed pyridine shows slight changes in the acidity of the samples with synthesis aging time. As the synthesis aging time increases there is an increase in the Brönsted acidity of the samples, evidenced by the presence of the typical pyridinium band at 1550 cm−1 . However, under evacuation at different temperatures this band tends to disappear, indicating that the Brönsted sites are of a weak character. The bands corresponding to Lewis acid sites appear typically at 1610 and 1450 cm−1 . In niobium-doped samples the Lewis acidity has been attributed to the presence of coordinatively unsaturated surface exposed Nb5+ sites or to extra-framework niobium oxide species, most probably Nb2 O5 [27]. The

46

L. Cedeño et al. / Applied Catalysis A: General 241 (2003) 39–50

Fig. 6. Variations of the Lewis acid sites concentration vs. temperature of (䊊) 52-Nb-2d-MCM, (×) 52-Nb-3d-MCM, (䉱) 52-Nb-4d-MCM and (䊐) 52-Nb-8d-MCM.

concentration of the Lewis acid sites in our samples, estimated from the intensity of the 1450 cm−1 band, is shown in Fig. 6. In this figure the samples with aging times of 3 and 4 days present the lowest concentration of Lewis sites. This can be rationalized by considering that at short synthesis aging times (2 days) part of the niobium has not yet been fully incorporated to the silica framework, giving rise to niobium species with a higher degree of unsaturation. Long synthesis aging times (8 days) cause a partial destruction of the structure and this could also create highly unsaturated niobium species giving rise to acid sites. Therefore, the results suggest that it is at intermediate synthesis aging times (3 and 4 days), when a greater amount of niobium is incorporated into the silica framework, that the structure has less defects, leading to smaller concentration of Lewis acid sites. In order to support the suggestion that niobium incorporates to the silica framework, the UV-Vis-DRS spectra of the samples were obtained. Fig. 7 shows the UV-Vis spectra of the catalysts and in addition,

for comparison purposes, the spectrum of Nb2 O5 . The DRS spectrum of Nb2 O5 presents an absorption edge at 410 nm. The spectrum of mesoporous silica has been reported earlier [16] and presents an absorption edge at 270 nm, in agreement with the more insulating character of this material. The incorporation of niobium to the mesoporous silica framework causes a shift in the absorption edge to higher wavelengths evidencing an increase in the semiconducting character of the material, but without reaching the value corresponding to Nb2 O5 . All the niobium-modified samples show the edge between 300 and 350 nm. This indicates that in all the samples niobium has been incorporated to the silica framework or that if it has been segregated to the surface it remains as isolated niobium oxide species in strong interaction with the silica framework without forming large Nb2 O5 particles. This is well in line with the Raman observations, where no band corresponding to Nb2 O5 was detected. As can be observed in Fig. 7, the magnitude of the absorption edge shift depends on the synthesis aging time. The sam-

L. Cedeño et al. / Applied Catalysis A: General 241 (2003) 39–50

47

Fig. 7. Diffuse-reflectance UV-Vis spectra of (䉬) Nb2 O5 , (䉱) 31-Nb-4d-MCM, (䊐) 52-Nb-2d-MCM, (×) 52-Nb-3d-MCM, (−) 52-Nb-4d-MCM and (䊊) 52-Nb-8d-MCM.

48

L. Cedeño et al. / Applied Catalysis A: General 241 (2003) 39–50

ple with 2 days synthesis aging time shows an additional absorption at higher wavelengths due, possibly, to the presence of some impurities. Because the position of the absorption edge is related to the size of the Nb clusters, the above results also indicate that the dispersion of niobium is higher at the synthesis aging times of 3 and 4 days. The dispersion of niobium in the 31-Nb-4d-MCM sample seems to be also high since the absorption edge is far from the one observed for bulk Nb2 O5 , suggesting that, in this sample with higher niobium content, niobium was also incorporated adequately to the silica framework and is not forming large Nb2 O5 particles. All the above results indicate that the synthesis aging time has a definitive influence in the textural, structural and chemical properties of niobium-modified mesoporous silica. Short synthesis aging times (2 days), do not allow the proper incorporation of niobium to the silica framework leading to samples with a less defined XRD pattern, a DRS absorption edge displaced to higher wavelength, and high Lewis and low Brönsted acidities. The samples with three and 4 days synthesis aging time seem to have the niobium well incorporated to the framework, as the calculations of wall thickness indicate, in spite that the DRS results show evidence of some surface polymeric species in the sample with 4 days synthesis aging time. Long synthesis aging times (8 days) appear to induce the formation of MCM-48 type structure. Evidence of this was found in the case of the 52-Nb-8d-MCM sample. The results from the catalytic activity measurements obtained with the sulfided samples indicate that the materials synthesized here have a small but significant catalytic activity in the thiophene HDS reaction (see Table 2). Therefore, the materials synthesized here cannot be used as HDS catalysts by themselves but have a promising use as supports for HDS catalysts. Table 2 Intrinsic activity in the thiophene HDS reaction (thiophene molecule per Nb atom s−1 × 106 ) Sample

Activity at 673 (K)

31-Nb-4d-MCM 52-Nb-4d-MCM 52-Nb-8d-MCM

16.73 28.63 27.94

Fig. 8. TPR-S patterns of (a) 52-Nb-4d-MCM and (b) 31-Nb4d-MCM, TCD signal.

Previous findings indicate that on conventional oxide supports like alumina or silica the sulfidation of supported niobium oxide species is rather difficult [13]. In fact, very poor HDS catalytic activity was detected for niobium supported on alumina when sulfided with H2 /H2 S, N2 /H2 S or H2 S at various temperatures. These results were related to the strong interaction of the niobium cations with the support that prevents the sulfidation of niobium [13]. In our case, it is expected that the difficulty of sulfidation will be greater in those materials where niobium was better incorporated to the silica framework. The results from the catalytic tests (Table 2) show that the HDS activity per niobium atom at 673 K is higher for the 52-Nb-4d-MCM sample. To corroborate the sulfidation of the catalysts and to enquire on the nature of active niobium species, characterization of the catalysts in the sulfided state by TPR (TPR-S) was performed in the samples immediately after the HDS reaction. In these characterizations the sulfided catalyst is subjected to a stream of hydrogen under a linear increasing temperature program and the consumption of hydrogen and production of H2 S are monitored. This technique allows the detection of the different sulfur species present in the catalyst and gives an indication of their relative amount. The results from TPRS, presented in Figs. 8 and 9, show some interesting differences between the samples with low and high niobium content. The samples show an opposite behavior regarding the consumption of hydrogen (Fig. 8) and the evolution of H2 S

L. Cedeño et al. / Applied Catalysis A: General 241 (2003) 39–50

49

persed, offering more active sites available for reaction. In contrast, in the 31-Nb-4d-MCM sample the niobium species segregate more from the silica framework and therefore they sulfide more readily but have a lower dispersion.

4. Conclusions From all the above results we can conclude that: Fig. 9. TPR-S patterns of (a) 52-Nb-4d-MCM and (b) 31-Nb-4d-MCM, UV signal.

Table 3 Quantitative results of TPR-S after reaction, H2 S evolution and H2 consumption per mol of Nb Sample

H2 S/Nb

H2 /Nb

31-Nb-4d-MCM 52-Nb-4d-MCM

0.149 0.076

0.150 2.942

(Fig. 9). For the 52-Nb-4d-MCM sample there is a high consumption of hydrogen compared to that in the 31-Nb-4d-MCM sample. In contrast, the evolution of H2 S is higher for the 31-Nb-4d-MCM sample (see Table 3). Furthermore, for the 31-Nb-4d-MCM sample the molar hydrogen consumption and the evolution of H2 S are similar whereas for the 52-Nb-4d-MCM sample there is a higher H2 consumption. This indicates that in the 52-Nb-4d-MCM sample in which niobium is in strong interaction with the siliceous framework, it is possible to reduce the niobium but is very difficult to sulfide it. When the niobium species have a higher degree of polymerization and also when they are segregated from the silica framework reduction and sulfidation can take place to the same extent, as it occurs in the 31-Nb-4d-MCM sample, which contains larger amounts of niobium. Since it is well known that the HDS active species is the sulfided form of niobium, the observed relative HDS catalytic activity per niobium atom (Table 2) can be rationalized as being the result of two opposite effects: in the 52-Nb-4d-MCM sample where niobium is better incorporated to the silica framework, sulfidation takes place to a lesser extent but the surface niobium species that are sulfided remain highly dis-

• The aging time in the synthesis of niobium-modified mesoporous silica is an important parameter that affects the textural, structural and acid (number and strength of Lewis and Brönsted sites) properties of the final solids. • The incorporation of niobium to the silica framework is successfully achieved at intermediate aging times (3 and 4 days). Long aging times (8 days) cause a transformation of the mesoporous structure from hexagonal to cubic form. Short synthesis aging times (<2 days) do not allow a proper incorporation of niobium to the silica framework. • The variations in the dispersion and type of surface niobium species induced by changes in synthesis aging time and Si/Nb ratio, affect the behavior of the solids under reducing and sulfiding conditions. It appears that incorporation of Nb to the silica framework leads to Nb species that are easily reducible but more difficult to sulfide than polymerized or agglomerated Nb2 O5 . These changes are reflected in the HDS activity of the different samples. Catalytic activity seems to be the result of a compromise between dispersion and sulfidability of the surface niobium species. • The niobium-containing materials synthesized here appear promising as catalytic HDS supports.

Acknowledgements The authors are grateful to Ivan Puente for the HREM microscopy work and to R. Cuevas and A. Gutiérrez-Alejandre for experimental assistance. Diana Hernandez wishes to acknowledge the Mexican Petroleum Institute for her scholarship. The financial support from the IMP-FIES program and DGAPA-UNAM is gratefully acknowledged.

50

L. Cedeño et al. / Applied Catalysis A: General 241 (2003) 39–50

References [1] A. Corma, A. Martinez, V. Martinez-Soria, J.B. Monton, J. Catal. 153 (1995) 25–31. [2] R. Mokaya, W. Jones, S. Moreno, G. Poncelet, Catal. Lett. 49 (1997) 87–94. [3] X.S. Zhao, G.Q. Lu, G.J. Millar, X.S. Li, Catal. Lett. 38 (1996) 33–37. [4] J. Zotin, A.C.B. Santos, G.T. Moure, D.L. Fonseca, A.A. Lira, T. Da Silva, Trabalhos Tecnicos do 9◦ Congresso Brasileiro De Catálise 1 (1998) 261–268. [5] N. Ichikuni, M. Shirai, Y. Iwasawa, Catal. Today 16 (1993) 427–434. [6] N. Ichikuni, M. Shirai, Y. Iwasawa, Catal. Today 28 (1996) 49–58. [7] A. Morikawa, A. Togashi, Catal. Today 16 (1993) 333–339. [8] J.C. Vedrine, G. Coudurier, A. Ouquor, P.G. Pries de Oliveira, J.C. Volta, Catal. Today 28 (1996) 3–15. [9] J.G. Weissman, Catal. Today 28 (1996) 159–166. [10] T. Da Silva, J. Zotin, A. Faro, M. Rocco, Trabalhos Tecnicos do 9◦ Congresso Brasileiro De Catálise 1 (1998) 253–260. [11] M.J. Ledoux, O. Michaux, G. Agostini, P. Panissod, J. Catal. 102 (1986) 275. [12] M. Breysse, T. des Couries, M. Danot, C. Geantet, J.L. Portefaix, US Patent 5,157,009 (1992). [13] N. Allali, A. Marie, M. Danot, C. Geantet, M. Breysse, J. Catal. 156 (1995) 279–289.

[14] V.N. Rodin, A.N. Starsev, V.I. Zaikovskii, Y.I. Yermakov, React. Kinet. Catal. Lett. 32 (2) (1986) 419. [15] E. Hillerova, J. Sedlacek, M. Zdrazil, Collect. Czech. Chem. Commun. 52 (1987) 1748. [16] L. Zhang, J.Y. Ying, AIChE J. Ceram. Process. 43 (1997) 2793–2801. [17] V. Gaborit, N. Allali, C. Geantet, M. Breysse, M. Vrinat, M. Danot, Catal. Today 57 (2000) 267–273. [18] K.W. Gallis, C.C. Landry, Chem. Mater. 9 (1997) 2035– 2038. [19] X.S. Zhao, G.Q. Lu, G.J. Millar, Ind. Eng. Chem. Res. 35 (1996) 2075–2090. [20] S. Hitz, R. Prins, J. Catal. 168 (1997) 194–206. [21] T. Tanaka, H. Nojiyama, H. Yoshida, H. Nakagawa, T. Funabiki, S. Yoshida, Catal. Today 16 (1993) 297–307. [22] J.M. Jehng, I.E. Wachs, J. Phys. Chem. 95 (1991) 7373– 7379. [23] H. Kosslick, G. Lischke, G. Walther, W. Storek, A. Martin, R. Fricke, Micropor. Mater. 9 (1997) 13–33. [24] A. Sayari, V.R. Karra, J.S. Reddy, I.L. Moudrakovski, Mater. Res. Soc. Symp. Proc. 371 (1995) 81–91. [25] A. Corma, M.T. Navarro, J. Perez-Pariente, J. Chem. Soc., Chem. Commun. (1994) 147–148. [26] X. Feng, J.S. Lee, J.W. Lee, J.Y. Lee, D. Wei, G.L. Haller, Chem. Eng. J. 64 (1996) 255–263. [27] M. Ziolek, I. Nowak, J.C. Lavalley, Catal. Lett. 45 (1997) 259–265.