14588
J. Phys. Chem. B 2005, 109, 14588-14594
Correlations between Acidity, Surface Structure, and Catalytic Activity of Niobium Oxide Supported on Zirconia Thomas Onfroy, Guillaume Clet, and Marwan Houalla* Laboratoire Catalyse et Spectrochimie (UMR CNRS 6506), ENSICAEN-UniVersite´ de Caen, 6 Bd. du Mare´ chal Juin, 14050 Caen (Cedex), France ReceiVed: April 5, 2005; In Final Form: June 8, 2005
The development of the acidity and the relationship between acidity, catalytic activity, and the surface structure for niobium oxide supported on zirconia were investigated for a series of solids. The catalysts were active for 2-propanol dehydration only above a threshold in Nb loading. The acidity was studied by infrared spectroscopy of adsorbed 2,6-dimethylpyridine as a probe molecule, and the onset of activity was correlated with that of the formation of relatively strong Brønsted acid sites. The variation in the abundance of these sites also correlated with the catalytic activity. Raman, IR, and UV spectroscopy results indicated that the active sites were related to polymeric Nb surface species. These results were compared to those previously reported for the WOx/ZrO2 catalysts.
Introduction Solid acids catalyze a large number of industrially important reactions.1 Optimization of the performance of these catalysts requires a detailed knowledge of the surface structure. In particular, it is of interest to investigate for a given system the development of acid sites with loading, in relation to the molecular structure of the surface species and their catalytic performance. Previous study of the WOx/ZrO2 system obtained by deposition on monoclinic zirconia indicated the presence of a threshold of W surface density for the appearance of Brønsted acid sites.2 A correlation between the abundance of Brønsted acid sites and the 2-propanol decomposition activity was established. The study of the molecular structure of surface WOx species showed that the presence and the development of these relatively strong Brønsted acid sites are associated with the formation of an infrared band tentatively assigned to polymeric WOx species. It is thus of interest to see to what extent the observed behavior is characteristic of the zirconia support. Supported niobia catalysts have been investigated as solid acids.3-13 The NbOx/ZrO2 system has been examined by Wachs and co-workers.8,9,13-15 The study of the acidity of this system, by pyridine adsorption, showed a decrease in the number of Lewis acid sites of the support with Nb addition and the formation of new sites attributed to NbOx.7,8 The presence of Brønsted acid sites for the NbOx/ZrO2 system has been reported only in a minor amount for Nb loadings exceeding monolayer coverage.8,9 Preliminary results from our group, using a different probe molecule, showed that Brønsted acid sites were present on NbOx/ZrO2 for lower Nb loadings.11 The study of the structure indicated the presence of polymeric Nb species for a loading of 1 wt % Nb2O5.15 However, the development of these species with Nb loading has not been investigated. Moreover, the relation between acidity, surface structure, and catalytic activity has not been fully addressed. The objective of the present paper is to extend our previous study of WOx/ZrO2 to the NbOx/ZrO2 system. Toward this * To whom correspondence
[email protected].
should
be
addressed.
E-mail:
purpose, a series of NbOx/ZrO2 catalysts containing up to 3.8 wt % Nb2O5 was obtained by pore volume impregnation. The surface structure was examined by Raman, infrared, and UV spectroscopy. The development of the acidity was monitored by adsorption of probe molecules followed by infrared spectroscopy. The catalytic activity was assayed for the reaction of 2-propanol decomposition. The results were utilized to investigate the relation between surface structure, acidity, and catalytic activity. Experimental Section 1. NbOx/ZrO2 Catalysts Synthesis. The zirconia support (Degussa) was first calcined at 873 K for 24 h. NbOx/ZrO2 samples were prepared by incipient wetness impregnation of ZrO2 with niobium oxalate solution [93 wt % oxalic acid and 7 wt % niobium(V) oxalate (Niobium Products)]. The solids were then dried and calcined in air at 723 K for 16 h. Samples will be designated as Nbx Z where x refers to the surface density in Nb atom/nm2. 2. Brunauer-Emmett-Teller (BET) Surface Area. Nitrogen adsorption was measured at 77 K with an automatic adsorptiometer (Micromeritics ASAP 2000). The samples were pretreated at 573K for 2 h under vacuum. The surface areas were determined from adsorption values for five relative pressures (P/P0) ranging from 0.05 to 0.2 using the BET method. The pore volumes were determined from the total amount of N2 adsorbed between P/P0 ) 0.05 and P/P0 ) 0.98. 3. X-ray Diffraction. X-ray powder diffraction spectra were recorded using a Philips X’pert diffractometer with a copper anode (KR1 ) 0.15405 nm) and a scanning rate of 0.025 deg s-1. The results were used to estimate the relative amount of monoclinic and tetragonal zirconia. The intensities of the (111h) and (111) reflections of the monoclinic phase and the (111) reflection of the tetragonal phase were used for quantification. The intensity ratio (Xm) was calculated using eq 1, and the fraction of monoclinic zirconia (Vm) was estimated from an empirical nonlinear relationship (eq 2).16
10.1021/jp0517347 CCC: $30.25 © 2005 American Chemical Society Published on Web 07/13/2005
Niobium Oxide Supported on Zirconia
Xm )
Im(111h) + Im(111) Im(111h) + Im(111) + It(111) 1.31Xm Vm ) 1 + 0.31Xm
J. Phys. Chem. B, Vol. 109, No. 30, 2005 14589 TABLE 1: Characteristics of the NbOx/ZrO2 Catalysts
(1)
(2)
4. Raman Spectroscopy. Raman characterization was performed on samples exposed to ambient conditions or following dehydration. The samples were dehydrated in a N2-O2 (75%25%) flow at 723 K for 2 h. They were, then, transferred to the Raman cell, in a glovebox, under controlled atmosphere (N2). Raman spectra were recorded with a dispersive Raman (Kaiser) equipped with a diode laser source (λ ) 532 nm) and a CCD detector. The spectra were acquired after 20 10 s scans. The relative amount of monoclinic and tetragonal zirconia was determined from the intensity of the peaks characteristic of monoclinic (178 and 189 cm-1) and tetragonal (148 cm-1) phases using a nonlinear relationship obtained by Kim et al.17 from physical mixtures of monoclinic and tetragonal zirconia. The intensities of the ν(NbdO) bands at 996, 980, and 969 cm-1 were determined using a linear background for the overall spectral region. The relative contributions of various ν(NbdO) bands were estimated from their intensities after normalization of the spectra with respect to the zirconia band at 476 cm-1. The results were confirmed by curve fitting with Gaussian bands. For curve fittings, the positions were kept nearly constant ((1 cm-1) and the fwhm was maintained at 10.5 ( 0.5 cm-1. 5. Infrared Spectroscopy. IR spectra were recorded with a Nicolet 710 FT-IR spectrometer (resolution, 4 cm-1; 128 scans). All of the spectra presented here were normalized for 100 mg of the solid. Samples were pressed into pellets (ca. 20 mg for a 2 cm2 pellet) and activated at 723 K. The samples were first heated under vacuum at 723 K for 1 h. This was followed by a treatment in O2 (Pequilibrium ) 13.3 kPa) for 1 h and evacuation for 1 h at 723 K before cooling down to room temperature (RT). 2,6-Dimethylpyridine (lutidine) was introduced at RT (Pequilibrium ) 133 Pa) after activation. The spectra were then recorded following desorption from 323 to 573 K. The amount of Brønsted acid sites titrated by 2,6-dimethylpyridine was calculated using an integrated molar absorption coefficient value of ) 6.8 cm µmol-1 for the sum of the (ν8a + ν8b) vibrations of protonated lutidine (DMPH+) at ca. 1644 and 1628 cm-1.18 6. UV-Visible Spectroscopy. The samples were subjected to the same pretreatment used for Raman measurements (i.e., dehydration in a 75% N2-25% O2 flow at 723 K for 2 h); transfer of the sample to the UV cell in the glovebox under N2 atmosphere). A UV-visible source (Avantes) was directed toward the quartz cell via an optical fiber. The reflected radiation was collected with a second optical fiber connected to a UV spectrometer. A Halon disk was used for white reference measurements. Absorbance spectra thus obtained were then transformed into transmittance spectra and finally transformed with Kubelka-Munk function. The value of the absorption edge energy was determined for each spectrum by a linear extrapolation of the curve to zero absorption. 7. Catalytic Activity. The catalytic conversion of 2-propanol was measured in a fixed bed flow reactor. A 40 mg amount of sample was pretreated at 723 K in a 10% O2-90% He mixture (60 mL/min) for 1 h. The reaction was performed at atmospheric pressure with N2 as a carrier gas (P2-propanol ) 1.23 kPa) at 473 K for various weight hourly space velocity (WHSV) (8.7-53.3 mmol h-1 g-1). Reactants and products were analyzed with an
sample surface area (m2/g) pore volume (cm3/g) wt % Nb wt % Nb2O5 Nb atoms/nm2 % of monolayera % of monoclinic phase
ZrO2 34 0.17 0 0 0 0 95
Nb0.6 Z ND ND 0.35 0.50 0.6 10 93
Nb1.2 Z 34
Nb2.5 Z 34
Nb3.6 Z 35
Nb4.8 Z 35
0.17
0.17
0.18
0.18
0.69 0.99 1.2 20 89
1.37 1.96 2.5 40 88
2.02 2.89 3.6 60 94
2.67 3.82 4.8 80 90
a Theoretical value assuming that each Nb2O5 unit occupies a surface of 0.32 nm2. ND, not determined.
on line Gas Chromatograph (Perkin-Elmer Sigma 200) equipped with a capillary column (Porapak-T) and a flame ionization detector. Results 1. Preparation of the Catalysts. A series of solids with Nb contents ranging from 0.35 to 2.67 wt % was prepared by incipient wetness impregnation on zirconia. Table 1 reports the characteristics of the catalysts. The surface area and pore volume of the solids were little affected by Nb deposition. Therefore, Nb surface densities were calculated on the basis of the surface area of the zirconia support. The nominal monolayer coverage was calculated assuming that the cross-section of a Nb2O5 unit occupies a surface of 0.32 nm2.12 2. X-ray Diffraction. The XRD patterns of the zirconia support showed intense peaks for 2θ ) 28.2° (d ) 3.16 Å) and 31.5° (d ) 2.83 Å) and a weak one at 30.2° (d ) 2.95 Å). The first two peaks are characteristic of the (111h) and (111) planes of the monoclinic phase. The peak at 30.2° corresponds to the (111) plane of the tetragonal phase. From the relative intensities of these peaks, it was estimated that zirconia was essentially present in the monoclinic form (more than 88%). The Nb content had no significant effect on the composition (Table 1). No additional peaks that can be attributed to monoclinic Nb2O5 (2θ ) 24.4, 26.8, 29.0, and 29.9°) or hexagonal Nb2O5 (2θ ) 22.7 and 28.6°) were detected. 3. Raman Spectroscopy. The spectra of the samples in ambient conditions did not show any modifications of the relative intensities of the bands characteristic of monoclinic (176 and 188 cm-1) and tetragonal zirconia (144 and 263 cm-1).17,19 This indicates that, in accordance with XRD results, the composition of the support was not modified by Nb deposition. No bands characteristic of niobic acid (Nb2O5,nH2O) at about 655-675 cm-1 or of crystallized niobium oxide at 675-690 cm-1 were observed.12,20 However, the eventual formation of minor amounts of these phases cannot be ruled out because of the potential overlap of these bands with a relatively intense peak at 640 cm-1 due to the monoclinic form of the zirconia support. In the region 700-1100 cm-1, the spectra show a weak and broad band at 875 cm-1 attributed to a NbOx surface phase,13 which shifts to 920 cm-1 with Nb content. After dehydration, the 700-1100 cm-1 region gives more information on the surface species. Figure 1a shows the Raman spectra of the solids after dehydration. The spectra of NbOx/ ZrO2 indicate the presence of four bands at ∼980, ∼940, ∼820, and ∼760 cm-1. The band located at ∼760 cm-1 is ascribed to the zirconia support. The band at ∼980 cm-1 shifts with the Nb loading from 969 to 996 cm-1 and is attributed to NbdO bond vibration of monooxoniobate species.14,15,21 The overall band area increases almost linearly with Nb content (Figure 1b). The band at ∼940 cm-1 appears for loadings g 2.5 Nb at/nm2
14590 J. Phys. Chem. B, Vol. 109, No. 30, 2005
Onfroy et al.
Figure 2. Evolution of the intensity of the ν(NbdO) band measured at 996 (9) and 969 (0) cm-1.
Figure 1. (a) Raman spectra of zirconia and NbOx/ZrO2 catalysts pretreated in dry air at 723 K (spectra normalized to the 476 cm-1 zirconia band). (b) ν(NbdO) band area as a function of Nb content.
and is attributed to the symmetric vibration of the (O-NbO)n bond.15 The very broad band at about 820 cm-1, already observed for low Nb loadings, develops with Nb content and is attributed to the antisymmetric vibration of the Nb-O-Nb bond.21 For a given loading, the band at 969-996 cm-1 shows several shoulders indicating that the overall envelope consists of several components. Peaks or shoulders were detected at 996, 989, 980, 969, or 961 cm-1. Figure 2 shows the variation of the Raman intensity for high and low wavenumber ν(NbdO) peaks as a function of Nb loading. The results clearly show different evolutions for the two peaks. The low wavenumber peak at 969 cm-1 was detected with initial Nb addition. Its intensity appears to remain essentially constant. In contrast, the high wavenumber peak at 996 cm-1 appears, only, above a threshold of Nb content of 1.2 Nb atoms/nm2 and increases in intensity with further increases in Nb loading. Similar results were obtained by curve fitting. 4. Infrared Spectroscopy. 4.1. Structure. Infrared spectra relative to the ν(NbdO) region indicated the presence of a band centered around 970 cm-1, which shifts to higher values with Nb loading (Figure 3a). This band is characteristic of the stretching vibration of monooxoniobate species.15 Several bands or shoulders were also detected around 994, 980, and 960 cm-1. The positions of these bands are in accordance with those
Figure 3. (a) Infrared spectra of the NbOx/ZrO2 solids activated in a vacuum at 723 K. (b) Evolution of the 994 cm-1 peak intensity with Nb surface density.
observed by Raman spectroscopy. Figure 3b shows the evolution of the band at 994 cm-1 as a function of the Nb loading. As observed by Raman, this band is detected above 1.2 at Nb/nm2 and increases in intensity with further increases in Nb loading. 4.2. Acidity. Figure 4a shows the infrared spectra in the 16801590 cm-1 range following lutidine adsorption on the zirconia support. The spectra showed two bands at 1610 and 1580 cm-1 (the latter is not shown in the figure) attributed to lutidine
Niobium Oxide Supported on Zirconia
Figure 4. Lutidine adsorption followed by IR spectroscopy. (a) IR spectra for ZrO2 and NbOx/ZrO2 catalysts after lutidine desorption at 423 K. (b) Evolution of the concentration of Brønsted acid sites for ZrO2 and NbOx/ZrO2 catalysts following lutidine desorption at 423 (9) and 523 (0) K.
coordinated on Lewis acid sites.22,23 The intensity of these bands decreases with Nb loading. For Nb loadings higher than 0.6 Nb at/nm2, in addition to the bands attributed to lutidine coordinated on Lewis acid sites, a doublet appeared at 1644 and 1628 cm-1 due to ν8a and ν8b vibrations of protonated lutidine.23,24 Figure 4b presents the evolution of the number of Brønsted acid sites as a function of Nb loading for desorption temperatures of 423 and 523 K. Following desorption at these temperatures, a threshold of Nb loading for the formation of Brønsted acid sites is observed. After desorption at 523 K, only solids with Nb loadings higher than 1.2 at/nm2 exhibit Brønsted acid sites sufficiently strong to retain lutidine. The number of Brønsted acid sites able to retain lutidine at 523 K increased up to 5.0 µmol H+/g (∼0.018 H+/at Nb) with increasing Nb loading. Desorption at 423 K evidenced a higher number of acid sites and a lower threshold for their formation. However, a large number of these sites disappeared on evacuation at higher temperature indicating their lower strength. No bands characteristic of protonated lutidine were detected following desorption at 573 K. 5. UV-Visible Spectroscopy. A previous study by Fournier et al. indicated a relationship between the absorption edge energy and the degree of condensation of metal oxide species.25 Subsequent work used this approach to follow the evolution of the size of NbOx domains.26-28 The absorption edge energies
J. Phys. Chem. B, Vol. 109, No. 30, 2005 14591
Figure 5. (a) UV-visible spectra for ZrO2 and NbOx/ZrO2 catalysts after activation in air at 723 K. (b) Absorption edge energy for NbOx/ ZrO2 catalysts vs Nb density and for some reference Nb compounds.
of some Nb reference compounds (niobium oxalate, niobic acid, and niobium oxide) were thus determined, in the present study, from their UV-visible spectra. The estimated values (respectively, 4.3, 3.9, and 3.2 eV) were in agreement with those reported in the literature.20,26 Figure 5a shows UV-visible spectra of zirconia and NbOx/ ZrO2, after activation at 723 K in air. The zirconia absorption edge energy (5.10 eV) due to the charge transfer transition O2f Zr4+ is consistent with the reported values.29-31 The absorption edge energy for niobate species in NbOx/ZrO2 decreases from Eg ) 4.40 eV for Nb0.6 Z to Eg ) 3.80 eV for W4.8 Z. The results are shown in Figure 5b. The edge energy for Nb0.6 Z was nearly equal to that reported by Gao et al. for isolated NbO4 species in Nb-MCM4126 and close to that of niobium oxalate. With increasing Nb loading, the absorption edge energy decreases toward a value typical of polymeric Nb structures (for example, Nb2O5,nH2O).26 This suggests that for the NbOx/ZrO2 system, the size of the Nb domains increases with Nb loading. However, these values remain higher than that measured for niobium oxide. 6. Catalytic Activity Measurements. At 473 K, the zirconia support exhibited a very weak propene formation rate (6.8 × 10-5 mol h-1 g-1). The activity of the NbOx/ZrO2 catalysts was, typically, more important. Propene only was detected. Acetone was not observed confirming the acidic character of these solids. Figure 6 shows the propene formation rate as a function of Nb loading. Catalysts with Nb loadings below 1.2 at/nm2 exhibited
14592 J. Phys. Chem. B, Vol. 109, No. 30, 2005
Figure 6. Propene formation rate as a function of Nb surface density in NbOx/ZrO2 catalysts. Reaction temperature, 473 K; molar space velocity, 43.4 mmol propanol h-1 g-1.
an activity comparable to that of the zirconia support. Above this threshold, propene formation rate increased rapidly with Nb content. Discussion Influence of Nb Addition on the Texture and Composition of the Support. The results shown in Table 1 clearly indicate that the BET surface area of the zirconia support and its composition (monoclinic/tetragonal ratio) were little affected by Nb addition. This is consistent with a previous study by Jehng et al. of the same system obtained in a similar fashion.13 Nature of Deposited Nb Species. Analysis of the structure of the supported Nb phase by XRD and Raman spectroscopy shows no evidence of Nb2O5 or niobic acid formation. This suggests that the Nb phase is present essentially as a surface species. The formation of such species is supported by Raman and infrared data, which shows the presence of a band at 969996 cm-1 attributed to surface monooxoniobate species.15,21 It is also consistent with the Raman and IR results reported by Wachs and co-workers for the same system.8,13-15 Additional evidence for the presence of dimeric and polymeric niobate species, for Nb loadings around 1.2 at Nb/nm2, may be inferred from Raman results, which show a band at 820 cm-1 characteristic of asymmetric Nb-O-Nb.21 These results are consistent with those reported by Jehng et al.14 Analysis of Raman (Figure 1a) and infrared (Figure 3a) spectra in the region characteristic of monooxoniobate species indicates the presence of several bands at ca. 996, 989, 980, 969, and 960 cm-1. Similar findings were reported for the WOx/ ZrO2 system and were attributed to different degrees of polymerization of W species.2,29,31 The high wavenumber band (996 cm-1) increases with increasing Nb content (Figure 2) indicating the development of the more polymerized species. Note that the proposed increase in the degree of polymerization with Nb loading is consistent with UV-visible spectroscopy results that show a progressive decrease in the energy of the Nb absorption edge with increasing loading. Acidity of Surface Species. Lewis Acidity. Lewis acid sites were detected by lutidine adsorption. The results indicated a decrease in the abundance of these sites with increasing Nb loading up to 3.6 Nb atom/nm2 (Figure 4a). Similar results were reported by Datka et al.9 using pyridine as a probe molecule. For the zirconia support, Lewis acid sites are associated with coordinatively unsaturated Zr4+ sites. One can thus attribute the
Onfroy et al. observed decrease in the abundance of these sites to an increased coverage of the support by surface Nb species. Pyridine is a more appropriate probe molecule than lutidine for monitoring the strength of Lewis acid sites because of the more pronounced shift of the ν8a band. An increased strength of Lewis acidity was observed with increasing Nb content.11 A similar behavior was reported for the WOx/ZrO2 system.2,32,33 The increased strength of Lewis acidity can be attributed to the electron withdrawing of an NbOx cluster in proximity of Lewis acid sites from the support sites.7 Brønsted Acidity. Lutidine thermodesorption results (Figure 4b) indicate that Brønsted acid sites appear, only, after a threshold of Nb loading (1.2 at Nb/nm2) is reached. The need for a minimum loading of a supported phase for the formation of Brønsted acid sites has been reported by Wachs34 for aluminasupported NbOx, MoOx, ReOx, and VOx systems. No such behavior was reported for the corresponding ZrO2-supported systems. Although the solids were prepared in a fashion similar to the one adopted in the present study, pyridine adsorption results for the NbOx/ZrO2 system by Datka et al.9 did not evidence the formation of Brønsted acid sites. This is not in contradiction with lutidine adsorption data. Lutidine is a more basic probe molecule as compared to pyridine and as such can be more readily protonated. Thus, weaker Brønsted acid sites can be detected. Furthermore, the higher integrated molar absorption coefficient of protonated lutidine as compared to protonated pyridine ( ) 6.8 cm µmol-1 18 vs 1.8 cm µmol-1 35,36) makes lutidine inherently a more sensitive probe for Brønsted acid sites. Catalytic Activity. Catalytic activity measurements of NbOx/ ZrO2 (Figure 6) clearly show that catalysts containing ca. 1.2 Nb atom/nm2 exhibit little activity. For higher loadings, a drastic increase in activity is observed. Alcohol dehydration activity of Nb-based catalysts supported on SiO2, Al2O3, TiO2, or ZrO2 has been observed.6,8 However, to the best of our knowledge, the need to exceed a threshold of Nb loading for the activity to develop has not been reported for the NbOx/ZrO2 system. Correlation between Acidity and Catalytic Performance. Figure 7a shows the evolution of the rate for propene formation as a function of Nb content. Also reported in Figure 7a is the abundance of Brønsted acid sites estimated from lutidine desorption data at 523 K. Clearly, a good correlation is found between the catalytic activity and the abundance of Brønsted acid sites. Catalysts containing Nb loadings e1.2 at/nm2 exhibit little propene formation activity (0.14-0.38 mmol h-1 g-1) and no Brønsted acid sites. The drastic increase in the catalytic activity for the catalyst containing 2.5 Nb atom/nm2 is concomitant with significant formation of Brønsted acid sites. With further increases in Nb content, a similar evolution of the catalytic activity and abundance of Brønsted acid sites is observed. This is better evidenced in Figure 7b where the rate of propene formation is plotted vs the density of Brønsted acid sites. It should be noted that no direct correlation was observed between the concentration of Lewis acid sites and the catalytic activity. It is of interest to compare these results to those reported for the corresponding WOx/ZrO2 system.2 Both systems exhibit a similar threshold of metal loading (W or Nb) for the appearance of Brønsted acid sites and catalytic activity. They also show a direct correlation between the abundance of Brønsted acid sites and the catalytic activity. However when compared under the same conditions, the WOx/ZrO2 system is up to 2 orders of magnitude more active than the corresponding Nb catalysts. This can be in part related to the observed difference in the strength
Niobium Oxide Supported on Zirconia
Figure 7. (a) Propene formation rate (0) and number of Brønsted acid sites (9) vs Nb density; (b) correlation between the number of Brønsted acid sites determined by lutidine desorption at 523 K and the propene formation rate.
of Brønsted acid sites. Lutidine thermodesorption experiments at 573 K indeed show significant abundance of Brønsted acid sites for active WOx/ZrO2 catalysts vs negligible amounts for the corresponding NbOx/ZrO2 catalysts. Correlation between the Nature of Surface Species and Acidity. Figure 8a compares, as a function of Nb loading, the abundance of Brønsted acid sites estimated from lutidine desorption at 523 K, with the intensity of the band at 996 cm-1 attributed to polymeric NbOx species. The results show a similar evolution for both parameters. Catalysts that are essentially inactive for propene formation show no indication of the presence of polymeric NbOx species and exhibit little or no Brønsted acidity. The onset and development of the catalytic activity correlate with the evolution of polymeric NbOx species and the abundance of Brønsted acid sites. This suggests that Brønsted acid sites are associated with polymeric NbOx species (Figure 8b). Note that niobic acid was shown to be active for 2-propanol dehydration;37 however, no direct evidence of its presence in our catalysts was found. It is worth noting the similarity in the development of acid sites and surface species of WOx/ZrO2 and NbOx/ZrO2 catalysts. Both systems exhibited a similar threshold for the appearance of Brønsted acid sites and comparable evolution with the loading of the supported phase. In both instances, Brønsted acid sites were associated with high wavenumber ν(WdO) or ν(NbdO) band attributed to extensively polymerized species.2 This suggests that for these two zirconia-supported systems the evolution of the surface structure is determined by the zirconia support whereas the acidity appears to depend on the nature of the supported phase.
J. Phys. Chem. B, Vol. 109, No. 30, 2005 14593
Figure 8. (a) Raman band intensity (0) and number of Brønsted acid sites (9) vs Nb density; (b) correlation between the number of Brønsted acid sites determined by lutidine desorption at 523 K and the intensity of the ν(NbdO) Raman band located at 996 cm-1.
Conclusions A series of NbOx/ZrO2 catalysts containing between 0.6 and 4.8 Nb at/nm2 has been prepared by incipient wetness impregnation. XRD and Raman results indicate that the supported Nb phase is essentially present as a surface species. In situ characterization by Raman, infrared, and UV-visible spectroscopy indicated an increased degree of polymerization of NbOx species with increasing Nb content. The adsorption of 2,6-dimethylpyridine followed by infrared spectroscopy indicated that Brønsted acid sites were only detected for catalysts containing Nb loading higher than 1.2 at/ nm2. The abundance of Brønsted acid sites increased with further increases of Nb density. The threshold observed for the formation of Brønsted acidity correlated with the appearance of catalytic activity for 2-propanol dehydration. A direct relationship was observed between the rate of propene formation and the abundance of Brønsted acid sites. Analysis of the structure of Nb phase indicated that the catalytic activity and the Brønsted acidity are associated with polymeric NbOx species. Acknowledgment. The Van Gogh exchange Program is gratefully acknowledged for a grant. We thank Prof. B. M. Weckhuysen and Dr. T. Visser (Department of Inorganic Chemistry and Catalysis, University of Utrecht, The Netherlands) for making available the Raman and UV spectrometers and for their advice. Thanks are also due to M. N. Metzner (Lab. SIFCOM, UMR CNRS 6176, ENSICAEN-Universite´ de Caen) for performing the XRD measurements.
14594 J. Phys. Chem. B, Vol. 109, No. 30, 2005 References and Notes (1) Tanabe, K.; Ho¨lderich, W. F. Appl. Catal. A 1999, 181, 399. (2) Onfroy, T.; Clet, G.; Houalla, M. J. Phys. Chem. B 2005, 109, 3345. (3) Nowak, I.; Ziolek, M. Chem. ReV. 1999, 99, 3603. (4) Tanabe, K.; Okazaki, S. Appl. Catal. A 1995, 133, 191. (5) Viparelli, P.; Ciambelli, P.; Lisi, L.; Ruoppolo, G.; Russo, G.; Volta, J. C. Appl. Catal. A 1999, 184, 291. (6) Ichikuni, N.; Shirai, M.; Iwasawa, Y. Catal. Today 1996, 28, 49. (7) Tavares Da Silva, C. L.; Loyola Camorin, V. L.; Zotin, J. L.; Rocco Duarte Pereira, M. L.; da Costa Faro, A. C., Jr. Catal. Today 2000, 57, 209. (8) Jehng, J.-M.; Wachs, I. E. Catal. Today 1990, 8, 37. (9) Datka, J.; Turek, A. M.; Jehng, J. M.; Wachs, I. E. J. Catal. 1992, 135, 186. (10) Pittman, R. M.; Bell, A. T. Catal. Lett. 1994, 24, 1. (11) Onfroy, T.; Clet, G.; Bukallah, S. B.; Hercules, D. M.; Houalla, M. Catal. Lett. 2003, 89, 15. (12) Pittman, R. M.; Bell, A. T. J. Phys. Chem. 1993, 97, 12178. (13) Jehng, J.-M.; Wachs, I. E. J. Mol. Catal. 1991, 67, 369. (14) Jehng, J.-M.; Wachs, I. E. J. Phys. Chem. 1991, 95, 7373. (15) Burcham, L. J.; Datka, J.; Wachs, I. E. J. Phys. Chem. B 1999, 103, 6015. (16) Toraya, H.; Yoshimura, M.; Somiya, S. J. Am. Ceram. Soc. 1984, 67, C119. (17) Kim, B. K.; Hahn, J. W.; Han, K. R. J. Mater. Sci. Lett. 1997, 16, 669. (18) Onfroy, T.; Clet, G.; Houalla, M. Microporous Mesoporous Mater. 2005, 82, 99. (19) Zhao, B.; Xu, X.; Gao, J.; Fu, Q.; Tang, Y. J. Raman Spectrosc. 1996, 27, 549. (20) Brayner, R.; Bozon-Verduraz, F. Phys. Chem. Chem. Phys. 2003, 5, 1457.
Onfroy et al. (21) Weckhuysen, B. M.; Jehng, J.-M.; Wachs, I. E. J. Phys. Chem. B 2000, 104, 7382. (22) Lahousse, C.; Aboulayt, A.; Mauge´, F.; Bachelier, J.; Lavalley, J. C. J. Mol. Catal. 1993, 84, 283. (23) Travert, A.; Manoilova, O. V.; Tsyganenko, A. A.; Mauge´, F.; Lavalley, J. C. J. Phys. Chem. B, 2002, 106, 1350. (24) Jacobs, P. A.; Heylen, C. F. J. Catal. 1974, 34, 267. (25) Fournier, M.; Louis, C.; Che, M.; Chaquin, P.; Masure, D. J. Catal. 1989, 119, 400. (26) Gao, X.; Wachs, I. E.; Wong, M. S.; Ying, J. Y. J. Catal. 2001, 203, 18. (27) Mendes, F. M. T.; Perez, C. A.; Soares, R. R.; Noronha, F. B.; Schmal, M. Catal. Today 2003, 78, 449. (28) Tanaka, T.; Nojima, H.; Yoshida, H.; Nakagawa, H.; Funabiki, T.; Yoshida, S. Catal. Today 1993, 16, 297. (29) Gutie´rrez-Alejandre, A.; Castillo, P.; Ramirez, J.; Ramis, G.; Busca, G. Appl. Catal. A 2001, 216, 181. (30) Barton, D. G.; Shtein, M.; Wilson, R. D.; Soled, S. L.; Iglesia, E. J. Phys. Chem. B 1999, 103, 630. (31) Scheithauer, M.; Grasselli, R. K.; Kno¨zinger, H. Langmuir 1998, 14, 3019. (32) Santiesteban, J. G.; Vartuli, J. C.; Han, S.; Bastian, R. D.; Chang, C. D. J. Catal. 1997, 168, 431. (33) Baertsch, C. D.; Soled, S. L.; Iglesia, E. J. Phys. Chem. B 2001, 105, 1320. (34) Wachs, I. E. Catal. Today 1996, 27, 437. (35) Emeis, C. A. J. Catal. 1993, 141, 347. (36) Khabtou, S.; Chevreau, T.; Lavalley, J. C. Microporous Mater. 1994, 3, 133. (37) Ve´drine, J. C.; Coudurier, G.; Ouqour, A.; Pries de Oliveira, P. G.; Volta, J. C. Catal. Today 1996, 28, 3.