of Molecular Catalysis, 72 (1992)
Acidic properties and stability of sulfate-promoted metal oxides Mohamed Waqif, Jean Bachelier, Odette Saur and Jean-Claude Lavalley* Laboratoire Catalyse et S’ctrochimie, VRA.CNZ3S.0414ZSMRA, 6 Boulevard du Mar&2ul Juin, 14050 Caen C&ex (I%-ante) (Received July 18, 1991; accepted October 23, 1991)
Abstract Methanol dehydration, used as a test reaction, and CO and pyridine adsorption studied by infrared spectroscopy, allow comparison of the acidity of pure and sulfated metal oxides such as ZrCe, TiO,-anatase, -rutile and AleOe. Sulfation enhances the strength of the weakest Lewis acid sites but poisons the strongest. It also creates Brdnsted acidity in the case of highly loaded samples. This latter effect explains the high activity of persulfated TiOz and ZrOp samples due to the formation of sulfate species different from those previously characterized on low loaded samples. They have been tentatively described as polymeric forms or SO,-like species. In contrast, sulfation tends to decrease the very high activity of AleO,. The active species are stable at 300 “C on alumina and zirconia, but are quickly decomposed on the TlOe samples. This low stability, compared to that observed under vacuum, is due to interaction wlth the methanol and/or water formed, increasing the ionic character of the sulfate species, and making them less stable.
Introduction In recent years, sulfated metal oxides have been the subject of an increasing interest for several reasons: (i) sulfation of Claus catalystspoisoning the active sites for both the conversion of H,S + SOa, and hydrolysis of COS or CS2 [ 1 ]; (ii) dry removal of SOa from flue gas generally involving sulfation of the adsorbents [2,3]; (iii) doping of metal oxides by sulfate ions, enhancing their acidity. In particular, FeaOa, TiOa and ZrOa doped with SOd2- have been described as superacidic solids [4, 51. We have previously published the infrared study of sulfate species formed on A1203,Ti02  or Zr02 [ 71 by SO2 oxidation. On these oxides, after evacuation at 450 “C, a species with only one S- 0 group has been characterized by an IR absorption band near 1380 cm-‘. Other species may appear when increasing the sulfate amount, species such as S20r2- [ 71, SO,-like [S] and subsurface sulfates  being postulated. We have also compared the thermal stability under vacuum and the H2 reducibility of these sulfate species on different oxides [lo]. However the stability of the sulfates may be different under vacuum than in the presence of reagents. For instance, in their studies *Author to whom correspondence should be addressed.
Q 1992 - Elsevler Sequoia. All rights reserved
related to sulfated Fe203, Lee and Park showed an interaction between the S04’- species, and pyridine, NHa [ 111 or 1-butene [ 12] which favours the sulfate decomposition. Certainly the presence of water also modifies their stability because, on alumina, their structure was found to be different on highly dehydroxylated samples than on non-dehydroxylated samples [ 61. The aim of this paper is to compare the acidity and the stability of different oxides doped with various amounts of sulfates, particularly in the presence of reagents. Dehydration of methanol to dimethyl ether has been chosen as a probe reaction. This test is sensitive to sample acidity [13-151. Moreover, it generates water and could therefore provide information on the stability of the sulfate species in a wet atmosphere, according to the metal oxide. The results are completed by IR spectroscopic measurements in order to characterize the acidity and the sulfate amount of the different catalysts, before and after the reaction test. Experimental Metal,oxides, ZrOa (130 m2 g-r), Ti02-anatase (120 m2 g-r), TiO,-rutile (70 m2 g-.‘) and A1203(100 m2 g-‘) were kindly donated by Rhone-Poulenc. Sulfate ions were introduced by impregnation of pure oxides using titrated aqueous solutions of (NHJ2S04. The samples were then dried at 120 “C and calcined at 450 “C. The surface areas were somewhat decreased (cu. 20%) after this treatment. The sulfate amount on the samples after calcination at 450 “C was determined by sulfur analysis. This showed a sulfate uptake similar to the amount added when this was less than N 300 pmol g-‘, but much less than expected for higher sulfate additions. For example, addition of 1000 pm01 g- ’ of sulfate resulted in an impregnation of only about 600 prnol g- ‘. The notation used and the sulfate amounts for the different samples are reported in Table 1. For IR studies, self-supported discs of about 10 mg crne2 were used. They were activated by evacuation at 450 “C for 2 h. All the spectra were recorded at room temperature using a Nicolet MX-1 FT-IR spectrometer. Different CO pressures of up to 30 torr were introduced at room temperature. Pyridine (P= 2 torr) was also introduced at r.t. and then evacuated at 150 “C. Spectra of adsorbed species given by CO and CSH6Nwere obtained by subtracting that of the activated catalyst, and when applicable, that of the gas phase. Dehydration of methanol to dimethyl ether and water was carried out in a flow reactor. Dry methanol was carried from a thermostatted saturator (0 “C) in a flow of helium gas. The calibrated flow (1.2 1 h- ‘), containing 4% CHaOH, was passed through a thermoregulated Pyrex glass reactor containing a fixed bed of catalyst (0.160 g with particle size in the 0.4-l mm range). These particles were obtained by crushing a pressed wafer of
the material. All samples were tested in the same conditions, which allows direct comparison of their conversion. In a typical run, the catalyst was heated overnight at 410 “C under helium flow. After cooling, the reaction was performed at different temperatures in the 100-300 “C range, and then maintained at 300 “C for a few hours to study the deactivation rate. An online gas chromatography system equipped with a TCD detector and a 5 m Porapak Q column heated at 85 “C, was used to analyse the gas leaving the reactor.
Reactivity Under the reaction conditions used, dimethyl ether and water are the only products formed. Variation of the conversion level as a function of the reaction temperature (Z’,,+) are reported in Pig. 1. It appears that among the four pure oxides, only alumina is very active, the conversion reaching that expected in equilibrium conditions [ 151 when Tread>250 “C. On the other hand, at 250 “C, methanol conversion is hardly noticeable on r-utile and zirconia, and is only 5% on anatase. L-02
PC TiO2 Rutllo
TlO2 Anatau x 80.
04 -+.--:+ 100 160
Fig. 1. Variation of the conversion level as a function of the reaction temperature for (a) zirconia; (b) aIumiI&q(c) anatase; (d) rutiie (samples I, II, III, IV).
The effect of the sulfate depends upon the nature of the metal oxide and the sulfate amount. On ZrOz, and both samples of TiOa, addition of SOd2- increases the conversion of methanol. However it is still low for Zr02II, Ti02-AJII and TiO,-RIII. It does not strictly correlate with the amount of sulfate, since the activity markedly increases from Zr02-II or Ti02-III. In contrast, sulfation of ahnnina slightly decreases its conversion: when equilibrium is not reached, for example at 220 “C, the lower the amount of sulfate, the higher the conversion. Figure 2 shows the variation of conversion with time on stream at 300 “C for highly sulfated samples. The activity stays almost constant for A1203IV, and Zr02-IV, whereas it sharply decreases after a short time on stream for Ti02-IV samples. In this case the conversion tends to be similar to that of the corresponding pure oxides, anatase being a little more active than x-utile. Infrared results Spectra of adsorbed [email protected]
species Figure 3 shows the IR spectra of the different oxides with relatively low (samples Il) or high (samples IV) sulfate amounts, and those of the latter after reaction with methanol at 300 “C for a relatively long time (cu. 20 h). All the samples were evacuated at 450 “C. Spectra of low-loaded sulfated samples have already been discussed in previous studies [6, 71, except in the case of rutile. It is in fact much more complex than that of anatase, due to the presence of extra bands in the 1300-1320 cm-’ range indicating the presence of different types of sulfate species. The increase in the SOd2- amount enhances the intensity of the different bands, and shii the characteristic v(S -0) vibration near 1380 cm-’ towards higher wavenumbers. This shift has already been discussed in the case of A120s [ 161 and Zr02 [ 7 ] and explained by the formation of new species.
Fig. 2. Variation of the conversion level vs. time on stream at 300 “C for the highly sulfated metal oxides (samples IV).
Fig. 3. IR spectra of the sulfate species after evacuation of the samples at 450 “C on: A (a) ZrO& (b) ZrO&II; ZrO&V (c) before; (d) after methanol reaction test; B - (a) A120& A1208-IV(b) before; (c) after methanol reaction test; C - anatase; (a) TiO,-AX; Ti02-A.IV (b) before; (c) after methanol reaction test; D - rutile; (a) Ti02-R.11;Ti02-R.IV (b) before; (c) after methanol reaction test.
This also occurs in the case of TiOa samples. For such high sulfate amounts, the bands appearing near 1100 cm- ’ are not well resolved, due to overlapping. In any case, strong bands between 1200 and 1100 cm-’ do not appear, ahowing us to exclude bulk-like sulfate species [ 171. Methanol dehydration at 300 “C does not modify the spectrum of sulfate species on the AlaO&V sample evacuated at 450 “C (Fig. 3Bc). On ZrOzIV, only a few differences are observed. The band observed at 1410 cm-’ on the fresh catalyst is shifted to 1400 cm-’ and its intensity has slightly decreased (Fig. 3Ad). On the other hand, the spectra of Ti02-A.IV and TiOaRN, before and after reaction, are very different. The intensity of the bands has very much decreased (Fig. 3Cc, 3Dc), indicating a large diminution in the amount of sulfate species. Elementary analysis of the suIfur content in the samples after reaction is in agreement with such a result: S% in weight varies from 1.7 to 0.13 (Ti02-AN) and from 2.1 to 0.8 for Ti02-R.N.
Acidityrneasurem.ents CO adsorption at r.t. was used to study the effect of sulfation on the strength of the Lewis acid sites. Thii strength can be evaluated from the wavenumber of the V&O) band, which is always at a higher frequency than that of gaseous CO (2143 cm-‘), this being typical of CO molecules linked to coordinatively unsaturated surface cations through a c-dative bond [ 18 1. It is found at 2210 cm-’ on alumina, at 2192 cm-’ on ZrOz-I and Ti02-R.1 (Table 2) while two y(C0) bands are observed on TiOa-A.I, as already reported [ 19-211. Sulfation increases the wavenumber of CO species coordinated on ZrOa, Ti02-R and AlaO (Table 2) showing an increase in the strength of the Lewis acid sites, as already reported on ZrOa [ 221. However it leads to a decrease in the intensity of the z&O) band (F’ig. 4A). In the case of anatase, only one band at 2194 cm-’ is observed on TiOa-A.IV (Fig. 4B, Table 2), suggesting that sulfate species specilklly poison the strongest Lewis acidic sites, giving rise on the pure sample to the 2207 cm-’ band. Pyridine allows the detection of both Lewis and Briinsted acid sites. On the pure metal oxides, or on those with a small amount of sulfate species (samples II), no pyridinium species (bands at 1540 cm- ‘) have been detected, at least on the samples evacuated at 450 “C. Only Lewis acidic sites are in evidence by the ~a, band observed at a wavenumber higher than 1600 cm-‘, As generally reported, AlaO possesses the strongest Lewis acid sites, while those on ZrOz,TiOa-Aand TiOa-Rhave similar strengths (Table 2). Impregnation with a high amount of sulfate creates Brijnsted acidity whatever the oxide (Fig. 5). Results observed on zirconia (Fig. 5Bc, d) suggest that its appearance accompanies a decrease in the number of Lewis acid sites. Discussion Results relative to pure metal oxides Use of probes like CO and pyridine shows that the pure metal oxides possess Lewis acid sites whose strength decreases from AlaOa to TiOa and ZrOa. Two types of sites are clearly observed on the anatase sample, as already reported [ 19-2 11. Khadzhiivanov et al. [ 2 1 ] explained the difference between the sites, taking into account the coordination number (V or IV) of the Ti4+ ions and the coordination of the ligands, which led to the exclusion of CO adsorption on per&a-coordinatedTi4+ lying on the (001) face. A recent coupled HRTEM-FI’IR study of TiOa anatase, showed that the weakest Lewis sites are associated with Ti4+ on the exposed faces, while the strongest are localized on edges and/or corners [ 231. Bolis’et al.  recently reported the existence of two types of Lewis acid site on ZrOa. The more abundant, but less energetic, of these belongs to flat faces, whereas the other is situated at defective sites. Rutile has been studied much less; Zaki and KnGnger [ 251 reported a z&O) band at 2182 cm-’ in the presence of gas at r.t. Pyridine adsorption on AlaO conlkns the heterogeneity of the Lewis acid sites [ 261, and their comparative high strengths (Table 2).
134 Q cq
4. IR spectra of CO adsorbed species in presence of CO gas (P=30 torr)A-OnZrOz: (a) sample I, @) sample II, (c) sample IV; B - On anatase: (a) sample I, (b) sample IV.
Fig. 5. IR spectra of pyridine adsorbed species. A - On alumina: (a) sample I, (b) sample II, (c) sample IV (after pyridine evacuation at 100 “C); B - (a) On Ti02-RN, (b) on Ti02-A.Iv, (c) On ZrO&V, (d) on Zr02-I (after pyridine evacuation at 160 “C).
Methanol dehydration has already been used as a test reaction to compare the acidity of various pure alumina samples 1131.It is a sensitive reaction, since the conversion depends on the crystallinity of the samples, which can be related to the relative number of tetrahedral and octahedral A13+ sites. This sensitivity towards the strength of Lewis acid sites explains well the higher conversion observed on AlaOa. Of the other three metal oxides, anatase seems to be slightly more active (Fig. 1). Sulfation eflect It has been reported that addition of phosphate ions to amorphous alumina increased its Lewis acidity, created Briinsted acidity and enhanced its catalytic activity to methanol dehydration [ 141. Use of CO and pyridine as probe molecules shows that sulfate addition increases the strength of the Lewis sites. In the case of Ti02-A, for which two types of Lewis acid sites are in evidence, a speciilc effect is shown, sulfates poisoning the strongest acid sites but enhancing the acidity of the weakest. Moreover, all high-loaded sulfate samples possess Briinsted acidity (Fig. 5), whereas the number of Lewis acid sites seems less. This Brijnsted acidity is certainly generated by surface hydrogenosulfate ions, although such species are diflicult to detect. A previous study has shown that they are not associated with a well-defined v(OH) band, due to strong hydrogen bond interactions [ 271.Since the strength of the Lewis acid sites of ZrOz and TiOz increases by sulfation, it must increase their activity in methanol dehydration.The observed effect is, however, much higher than expected on persulfated samples, and it does not correlate to the amount of sulfate species in TiOa and z1-0~.Moreover, it seems difficult to explain the decrease in alumina activity by sulfation, taking account of only the Lewis acidity. The mechanism of methanol dehydration to dimethyl ether has been studied less than that of other alcohols such as isopropanol [ZS]. However Bandiera and Naccache [ 151 suggested a mechanism involving both acidic and basic active sites on a deal~inated mordenite. We have previously shown that sulfate adsorption involved basic oxygen atoms, which could explain the weaker activity of sulfated alumina samples. An important experimental result reported in this study is the leap in activity observed by introducing a high amount of sulfates on zirconia and titania samples. Pyridine adsorption shows that such a high amount induces Briinsted acidity, suggesting an important role played by Brijnsted acid sites. Indeed, this type of site has been considered to be active in the case of Hmordenite and sulfonated polystyrene catalysts . We may therefore consider that methanol dehydration could occur by two mechanisms,involving either strong Lewis acid sites, as in the case of alma, or Briinsted acid sites, such as on highly sulfated zirconia and titania samples. On these highloaded samples (samples IV), IR spectra show that supplementary sulfate species are formed. They are characterized by an absorption near 1400 cm-‘, attributed either to S20r2- [ 71 or SOS-likespecies . Their presence creates Briinsted acidity and explains the high activity of these samples Iv. Komarov
and Sinilo [ 321 also reported that the total acidity of zirconium oxide samples depended on the content of sulfate ions. Another possibility is the creation of Briinstedacid sites by the coordination of methanol on the strongest Lewis acid sites on AlaO [30, 311. In such a case, the mechanism would also involve Briinsted acidity but the sites would be created by the coordination of the first methanol molecules. Stability of sulfate species The reactivity results provide data on the stability of sulfated samples vs. temperature in the presence of methanol and water formed. Figure 2 shows that persulfated alumina and zirconia samples retain a high activity at 300 “C, in contrast to titanium oxide samples. Comparison of IR spectra before and after reaction shows that deactivation of the latter is due to a partial elimination of sulfate species during the reaction. The stability of the covalent sulfate species varies in the following order TiOa < ZrO,
t, M-O_zS=O M-O
KS/ + Hz0
‘$ SC’ H
.a ‘h 0
Many papers have reported that sulfated titania or zirconia samples present high acidity, called superacidity [4,5,33,34]. These sulfated samples are generally prepared by impregnation of the corresponding metal hydroxide with sulfuric acid or ammonium sulfate, followed by calcination. Although the reported spectra appear quite similar to those reported here, it is not fully understood whether the state of the support (oxide or hydroxide) during introduction of the sulfates influences the activity and the stability of the catalysts, especially in the case of high loaded samples. It would be of interest to clarify this point. References 1 2 3 4 5 6
M. Graulier and D. Papee, Energy Process., (1974) 2. D. P. McArthur, H. D. Simpson and K. Baron, oil Go-s J., (1983) 70. J. W. Byrne, B. K. Speronello and E. L. Imenberger, oil Gas J., (1984) 101. K. Tanabe, M. Misono, Y. Ono and H. Hattori, Stud. Sue Sci. CataL, 51 (1989) 199. T. Yamaguchi, AppL C&aL, 62 (1990) 1. 0. Saur, M. Beusitel, A. B. Mohammed Saad, J. C. Lavalley, C. P. Tripp and B. A. Morrow, J. Cat&., 99 (1986) 104.
138 7 M. Bensitel, 0. Saw-, J. C. Lava&y and B. A. Morrow, Mater. Chem. Phys., 19 (1988) 147. 8 K. Khadzhiivanov and A. Davydov, Kind. Kutd., 29 (1988) 460. 9 Y. Okamoto and T. Imanska, J. Phys. Chem., 92 (1988) 7102. 10 M. Waqif, 0. Saur, J. C. Lavalley, Y. Wang and B. A. Morrow, AppL C&o& 71 (1991) 319. 11 J. S. Lee and D. S. Park, J. CutaL, 120 (1989) 46. 12 J. S. Lee, M. H. Yeont and D. S. Park, J. CataL, 126 (1990) 361. 13 F. Abbattista, S. Delmastro, G. Gozzelino, D. Mazza, M. V&o, G. Busca, V. Lorenzelli and G. Ramis, J. CataL, 117 (1989) 42. 14 F. Abbattista, A. Delmastro, G. Gozzelino, D. Mazza, M. Vallino, G. Busca and V. Lorenzelli, J. Chem. Sot., Faraday !Prans. 1, 86 (1990) 3653. 15 J. Bandiera and C. Naccache, AppL Cutul., 69 (1991) 139. 16 M. Wmif, 0. Saw, J. C. Lavalley, S. Perathoner and G. Centi, J. Phys. Chm, 95 (1991) 4051. 17 M. Ben&e& M. Waqif, 0. Saur and J. C. Lavalley, J. Phys. Chem, 93 (1989) 6581. 18 G. Della Gatta, B. Fubini, G. Ghiotti and C. Morterra, J. CataL, 43 (1976) 90. 19 V. Bolis, B. Fubini, E. Garrone and C. Morterra, J. Chem. Sot., Famday Trans. I, 85 (1989) 1383. 20 G. Busca, H. Saussey, 0. Saur, J. C. Lavalley and V. Lorenzelli, AppL C&B!., 24 (1985) 245. 21 K. I. Khadzhiivanov, A. A. Davydov and D. G. Klisurski, K&et. KataZ., 29 (1988) 161. 22 M. Bensitel, 0. Saur, J. C. Lavalley and G. Mabtion, Mater. Chem. Phys., I7 (1987) 249. 23 C. Morterra, L. Orio, G. Spoto, A. Zecchina and L. Marchese, Cage. Italian Chem. Sot., Per&a, 1989. 24 V. Bolis, C. Morterra, M. Volante, L. Orio and B. Fubini, Lungmuir, 6 (1990) 695. 25 M. Zaki and H. Knozinger, Spectrochim. Actu, 43A (1987) 1455. 26 P. Nortier, P. Fourre, A. B. Mohamed Saad, 0. Saur and J. C. Lavalley, AppL CutaL, 61 (1990) 141. 27 M. Lion, M. Maache, J. C. Lavalley, G. Ramis, G. Busca, P. F. Rossi and V. Lorenzelli, J. Mol. Stmct., 218 (1990) 417. 28 J. B. Nagy, J. P. Lange, A. Gourgue, P. Bodart and Z. Gabelica, Stud. SUV$ Sci CataL, 20 (1985) 127. 29 B. C. Gates and L. N. Johanson, J. CataL, 24 (1969) 69. 30 G. Busca, P. F. Rossi, V. Loreruelli, M. Benaissa, J. Travert and J. C. Lavalley, J. Phys. &em., 89 (1985) 5433. 31 J. P. Gallas and C. Binet, Adv. Mol. Rekzx., Interact. Process., 24 (1982) 191. 32 V. S. Komarov and M. F. Sinilo, Kin&. Ku&L, 29 (1988) 701. 33 M. Hino and K. Arata, J. Chem Sot., C?wm. Commun. (1979) 1148; ibid., (1980) 851. 34 K. Arata and M. Hino, Muter. Chmn. Phys., 26 (1990) 213.