Gas phase beckmann rearrangement of cyclohexanone oxime over zirconia-supported boria catalyst

Gas phase beckmann rearrangement of cyclohexanone oxime over zirconia-supported boria catalyst

Applied Catalysis A: General 188 (1999) 361–368 Gas phase beckmann rearrangement of cyclohexanone oxime over zirconia-supported boria catalyst Bo-Qin...

221KB Sizes 0 Downloads 2 Views

Applied Catalysis A: General 188 (1999) 361–368

Gas phase beckmann rearrangement of cyclohexanone oxime over zirconia-supported boria catalyst Bo-Qing Xu a,∗ , Shi-Biao Cheng b , Shan Jiang b , Qi-Ming Zhu a a

State Key Lab of C1 Chemistry and Department of Chemistry, Tsinghua University, Beijing 100084, China b School of Chemical Engineering, Dalian University of Technology, Dalian 116012, China Received 22 February 1999; accepted 22 June 1999

Abstract Gas phase Beckmann rearrangement of cyclohexanone oxime is studied with B2 O3 /ZrO2 as the catalyst at 250–350◦ C. It is shown that B2 O3 /ZrO2 is highly active and selective for the synthesis of ε-caprolactam at 300–320◦ C. Activity and selectivity of the B2 O3 /ZrO2 catalyst are compared with other boria catalysts supported on Al2 O3 , TiO2 , SiO2 , MgO, and HZSM-5 (Si/Al = 25) on an 10% B2 O3 basis at 300◦ C. At this boria load level, B2 O3 /ZrO2 produces the best catalysis for the lactam synthesis. The load level of boria affects the behavior of B2 O3 /ZrO2 catalyst: increasing the boria load (≤20%) results in an increase of the lactam selectivity, but the lactam yield follows variation of the oxime conversion which shows a maximum at 10% boria load. Acidity and boria dispersion structure of the B2 O3 /ZrO2 catalysts are measured, respectively, by NH3 -TPD and IR spectroscopy. It is shown that the lactam yield is correlated with the number of intermediate strong acid sites that are responsible for ammonia desorption at 200–400◦ C. The optimum boria load (10%) of the B2 O3 /ZrO2 catalyst corresponds to two overlayers of boria on the support. The first layer of boria interacts directly with the support surface with tetrahedral BO4 as the structure unit; the other layers of boria are constructed by trigonal BO3 units. ©1999 Elsevier Science B.V. All rights reserved. Keywords: Beckmann rearrangement; Caprolactam; Cyclohexanone oxime; Boria; Zirconia; B2 O3 /ZrO2

1. Introduction The Beckmann rearrangement of cyclohexanone oxime is an important industrial reaction for the production of ε-caprolactam. The conventional technology makes use of fuming sulfuric acid as the catalyst in liquid phase. This technique is not environmentally friendly and is considered to be one of the most inefficient chemical processes that produces 2–5 times of invaluable by-product (ammonium sulfate) for every ∗ Corresponding author. Fax: +86-10-6277-0304 E-mail address: [email protected] (B.-Q. Xu)

unit of the desired lactam product. It has been a long desire that this process be replaced by one that bases on solid acid catalyst [1,2]. There are mainly two types of solid acid catalysts that have been explored in the literature for the Beckmann reaction. One type of the catalysts is based on zeolites or zeolitic materials [2–6], the other is based on supported boria on oxide supports [7–10]. Isolated neutral or weakly acidic silanol groups seem to be the catalytic sites on the zeolite catalysts [3]. However, intermediate to strong acid sites are required for the catalysis on supported boria catalysts [7–10,14]. Both types of catalysts suffer with rapid deactivation during the reaction.

0926-860X/99/$ – see front matter ©1999 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 6 - 8 6 0 X ( 9 9 ) 0 0 2 5 5 - 0

362

B.-Q. Xu et al. / Applied Catalysis A: General 188 (1999) 361–368

The performance of supported boria catalysts can be improved by proper control of variable factors such as nature of the support, load and dispersion of boria [7–10]. For many catalysts and catalytic reactions, the use of zirconia as catalyst support has often proved to produce better catalysis than other conventional oxide supports [11–13]. Recently, we found that boria supported on zirconia (B2 O3 /ZrO2 ) shows, among many modified zirconia catalysts such as zirconias doped with alkali metal, halogen, sulfate, or phosphate ions, the best catalysis for the synthesis of ε-caprolactam; the average lactam yield during 6 h time on stream (TOS) was 95% [14]. This finding has stimulated a systematic research on the preparation and characterization of B2 O3 /ZrO2 catalyst for the Beckmann reaction. In the present study, we show that the catalytic performance of B2 O3 /ZrO2 is superior to boria supported on Al2 O3 , TiO2 , SiO2 , MgO, and HZSM-5. IR and NH3 -TPD measurements are performed to characterize the structure and acidity of the catalysts with different boria load. Response of catalyst structure, surface acidity and catalytic performance to the change of boria load in B2 O3 /ZrO2 catalyst are reported for a clarification of the relationship between boria load, surface acidity and the catalytic performance for the Beckmann reaction.

2. Experimental Samples of the catalyst were prepared by incipient wetness impregnation with aqueous solution of boric acid. ZrO2 , TiO2 , SiO2 , Al2 O3 , MgO and HZSM-5 (Si/Al = 25) were used, respectively, as the support. These supports were calcined at 500◦ C for 8 h before the loading of boria. The ZrO2 support was prepared by calcination in air at 500◦ C of a zirconium hydroxide which was obtained by hydrolysis of zirconylchloride (ZrOCl2 ·8H2 O, analytical grade) with concentrated ammonia water by control of the final pH = 10. This ZrO2 has a surafce area of 60 m2 /g and an average pore diameter of 10 nm. TiO2 and SiO2 were obtained from Beijing chemical factory, and Al2 O3 was prepared by calcination of Al(OH)3 (Beijing chemical factory) in air at 500◦ C. MgO was purchased form Merck. Unless otherwise specified, the boria load was about 10 wt.% in the catalysts. Calcination of

the supported catalyst was done in air at 350◦ C for 12 h. The Beckmann rearrangement of cyclohexanone oxime was conducted at 250–350◦ C under atmospheric pressure in a flow pyrex reactor (i.d. 8 mm) with 1.0 g catalyst. A mixture of the oxime, benzene (solvent) and nitrogen (carrier gas) with molar ratio of 1 : 13 : 28 was passed into the reactor at WHSV = 0.32 h−1 in terms of the oxime. The oxime was dissolved in the solvent benzene (10 wt.%) and was fed into the reactor with a micro-pump. Before run of the reaction the catalyst was pretreated at 350◦ C for 1.5 h in flow of dry nitrogen (60 ml/min). Usually, the initial 30 min of the reaction was deducted from the reaction time on stream (TOS) since material balance in this initial period was poor. After this 30 min of reaction products from the reactor were collected each hour in an ice/water trap and were analyzed by GC (GC-1102 of Shanghai Analytic Instrument Factory) equipped with a 20% PEG 20M (2 m) column and a FID detector. IR spectra of the catalyst were recorded on a Nicolet 200 FTIR spectrometer with the KBr pellet method. NH3 -TPD experiments were conducted with 150 mg sample in a conventional flow type TPD system equiped with a mass spectrometer as the detector. The catalysts were dehydrated at 350◦ C with ultra high pure helium before the adsorption of NH3 at r.t. The temperature ramp of the TPD measurement was 10 K/min.

3. Results 3.1. Effect of reaction temperature In the previous study, gas phase Beckmann rearrangement of cyclohexanone oxime over zirconia based catalysts was carried out at 300◦ C and 350◦ C [14]. A 10% B2 O3 /ZrO2 catalyst showed the highest lactam yield (>94%) at either temperature for 6 h of reaction. In the present work, the rearrangement is carried out with temperatures in the range of 250–350◦ C over this catalyst to optimize the reaction temperature. A big effect of the reaction temperature on the catalyst activity (conversion of the oxime) and stability is seen in Fig. 1, but selectivity to the desired

B.-Q. Xu et al. / Applied Catalysis A: General 188 (1999) 361–368

363

Fig. 1. Conversion of cyclohexanone oxime over 10% B2 O3 /ZrO2 catalyst at different temperature.

Fig. 3. Oxime conversion (a) and lactam selectivity (b) of the Beckmann rearrangement over supported boria (10 wt.%) catalysts at 300◦ C. Fig. 2. Effect of reaction temperature on the activity (-䉬-) and selectivity (-4-) of 10% B2 O3 /ZrO2 catalyst: data are averaged for TOS ≤ 6 h.

lactam, which decreases by <7% in 9 h from ca. 97% at TOS = 1 h, is not so much affected. By-products of the rearrangement are dominated by cyclohexanone (<7% of all products) with temperature not higher than 320◦ C. At higher temperature, formations of cyclohexenones (ca. 3%) and 5-cynopent-1-ene (ca. 1–2%) also become detectable. Fig. 2 shows the average oxime conversion and the average lactam selectivity for an 6 h period of TOS. The average oxime conversion increases with the temperature upto 320◦ C at which the conversion basically reaches 100%. The average lactam selectivity seems little affected by the temperature below 320◦ C, it decreases, however, at higher reaction temperature. Apparently, the optimum temperature for the lactam synthesis is 300–320◦ C. In other experiments of this work, the reaction temperature is set to 300◦ C as a standard temperature for the reaction test.

3.2. Comparison with other supports A total of six oxides, ZrO2 , TiO2 , SiO2 , Al2 O3 , MgO and HZSM-5 (Si/Al = 25), are used, respectively, to support 10% B2 O3 for the rearrangement at 300◦ C for a period of TOS ≥6 h. Fig. 3 shows the time course of these catalysts. While slow deactivation is observed with the B2 O3 catalysts supported on ZrO2 , SiO2 , and Al2 O3 , the catalysts supported on TiO2 and HZSM-5 deactivate rapidly as is shown by the decline with TOS of the oxime conversion (Fig. 3a). In contrast to the oxime conversion, the lactam selectivity does not change significantly during the reaction (Fig. 3b). The catalysts supported on ZrO2 , TiO2 , and HZSM-5 are highly selective for the lactam synthesis (ca. 95%). Medium lactam selectivity (ca. 70%) is obtained over SiO2 and Al2 O3 supported catalysts (Fig. 3b), which is consistent with earlier studies in the literature [7–10]. B2 O3 supported by the basic MgO [11] exhibits little activity for the oxime conversion (<6%; not shown in the figure) and produces no lactam but undesired cyclohexanone (50–23%) and aniline (50–77%) products.

364

B.-Q. Xu et al. / Applied Catalysis A: General 188 (1999) 361–368

Fig. 4. Effect of boria load on the activity and selectivity of B2 O3 /ZrO2 catalyst at 300◦ C: data are averaged for TOS ≤ 6 h.

The average lactam yield during 6 h TOS varies with the support in the order of ZrO2 (lactam yield: (75.2%) > SiO2 (67.9%) = TiO2 94.8%) > Al2 O3 (67.4%) = HZSM-5 (67.0%) > MgO (0%). Evidently, B2 O3 /ZrO2 shows the best performance for the lactam synthesis at the 10% level of boria load. Although the lactam selectivity is comparable over ZrO2 (96%), TiO2 (94%) and HZSM-5 (95%) supported catalysts (Fig. 3b), the low yields of the lactam over these latter two catalysts are a consequence of their rapid deactivation (Fig. 3a). These results indicate that the B2 O3 /ZrO2 catalyst is highly active as well as selective and relatively stable for the lactam synthesis. 3.3. Effect of the boria load Improvement of the B2 O3 /ZrO2 catalyst is made with optimizing the boria load. The catalysts with different boria load are compared for the Beckmann reaction at 300◦ C for a period of TOS ≥6 h. Fig. 4 shows the average oxime conversion as well as the lactam selectivity for TOS = 6 h. The change of boria load (0–20 wt.%) has different effects on the activity and selectivity of the rearrangement. A maximum oxime conversion appears on the catalyst with ca. 10% B2 O3 , and surprisingly, the lactam selectivity also increases with increasing the load up to this same load level for the maximum catalyst activity. Beyond this load the lactam selectivity does not change. It is clear that all of the B2 O3 /ZrO2 catalysts are far superior to the support alone (with zero boria load, Fig. 4) for the Beckmann reaction. The support without boria is initially selective for the formation of cyclo-

Fig. 5. Product selectivity over the zirconia support without boria at 300◦ C.

hexanone (60–70%); considerable selectivity for the lactam (<50%) develops at longer TOS (Fig. 5). The loading of boria upto ca. 10% leads not only to increase in the oxime conversion and the lactam selectivity, but also to improved catalyst stability as indicated in Fig. 6, which shows the detailed time course behavior of the B2 O3 /ZrO2 catalysts. While significant deactivation of the catalyst cannot be avoided at longer TOS (Fig. 6a), decrease of the lactam selectivity never exceeds 7% in 10 h (Fig. 6b). Thus, the curves of the lactam yield (Fig. 6c) depend almost exclusively on the activity of the supported catalysts. Again, the decrease in the lactam yield is the lowest when the catalyst is optimized for the boria load (10% B2 O3 ). Except over the 5% B2 O3 /ZrO2 catalyst, cyclohexanone is the only significant by-product (2–10%) of the reaction. The decrease with TOS in the oxime conversion (deactivation) induces a gradual increase in formation of this by-product (Fig. 6d), which is responded by a decrease of the desired lactam. In addition to cyclohexanone, small amounts (ca. 2%) of cyclohexenone, 5-cyanopent-1-ene and aniline are also formed over the 5% B2 O3 /ZrO2 catalyst. All of these by-products are much less than those over the zirconia support without loading of the boria (Fig. 5). 3.4. IR and NH3 -TPD measurements Fig. 7 shows the infrared spectra of B2 O3 /ZrO2 catalysts. The sample with 5% B2 O3 shows two broad absorption bands at 1380–1430 and 1000–1150 cm−1 . The band at 1380–1430 cm−1 shifts upto 1460 cm−1 while the band at 100–1150 cm−1 disappears with

B.-Q. Xu et al. / Applied Catalysis A: General 188 (1999) 361–368

365

Fig. 6. Time course behavior of B2 O3 /ZrO2 catalyst at 300◦ C with varing level of boria load. (a) oxime conversion; (b) lactam selectivity; (c) lactam yield; (d) selectivity to cyclohexanone.

Fig. 7. IR spectra of boric acid, borax and B2 O3 /ZrO2 catalysts with 5% (a), 10% (b), 14% (c) and 20% B2 O3 (d).

increasing the boria load. Simultaneously, new bands at 1190 and 885 cm−1 grow in proportion with the boria load. It is well known that there exist two structure

units (i.e. tetrahedral BO4 and trigonal BO3 ) of boron in borates [15–18]. With reference to the spectra of a pure borax, boric acid, and boria (the spectrum for boria is not included in the figure since it resembles that of boric acid), the absorptions near 1460, 1190 and 885 cm−1 represent formation of BO3 units [15,17], like those in boria or boric acid, in the samples with more than 5% B2 O3 . Clearly, these BO3 units increase with the boria load. Since the absorptions at 1190 and 885 cm−1 are absent in the 5% B2 O3 /ZrO2 sample, and the broad band at 1380–1430 cm−1 for this sample is close to the bands of borax (1350–1420 cm−1 ) characterizing BO4 units, the coordination of boron atoms in the 5% B2 O3 /ZrO2 sample is considered mainly in BO4 units. NH3 -TPD is performed to compare the surface acidity of the catalysts. The resultant TPD profiles are shown in Fig. 8. Desorption of adsorbed NH3 from the catalyst covers a broad range of temperature (50–400◦ C) and features three peaks centered at about 100◦ C, 200◦ C, and 300◦ C, respectively. The total desorbed ammonia or total area of the TPD peaks decreases with the boria load, indicating a reduction

366

B.-Q. Xu et al. / Applied Catalysis A: General 188 (1999) 361–368

Fig. 8. TPD profiles of ammonia adsorbed on B2 O3 /ZrO2 catalysts with 5% (a), 10% (b), and 20% B2 O3 (c): the shaded area in profile (c) shows an example for the deconvolution for measuring the acidity of intermediate strong acid sites.

of the total number of acid sites. The area percentage of the low temperature peak at 100◦ C decreases, while the percentage of the high temperature peak near 300◦ C increases with increasing the boria load. This latter high temperature peak becomes dominant when the boria load is higher than 14% (Fig. 8c).

4. Discussion The present data suggest that B2 O3 /ZrO2 is highly active and selective for the Beckmann rearrangement of cyclohexanone oxime. At 10% boria load level the catalyst supported on zirconia shows much higher yield for the lactam syntheis than the ones on the other supports used in the present study. This 10% boria load happens to match the optimum boria load over the zirconia support (Figs. 4 and 6). While it seems difficult to extend this observation, based on the samples with 10% boria load, to the cases when boria catalysts supported on the other supports are optimized with boria loaded, the very high selectiv-

ity for the desired lactam over this 10% B2 O3 /ZrO2 catalyst (90–97%) exceeds over those (<90%) documented for impregnation B2 O3 /SiO2 and B2 O3 /Al2 O3 catalysts [7–10]. The optimum boria load was shown to be 20% on the silica [7] and 20–25% on alumina [8], respectively. The result that zirconia support alone shows little selectivity for the desired lactam suggests that the active sites of B2 O3 /ZrO2 catalysts are connected with the surface of dispersed boria. This behavior of catalysis is quite similar to that reported for the catalyst systems of B2 O3 /SiO2 and B2 O3 /Al2 O3 [7–10]. It is common for these supported boria catalysts that the lactam selectivity increases with increasing the boria load. Judgement of the optimum boria load of these catalysts can only be made from the maximum activity or conversion of the oxime (Figs. 4 and 6). The performance of the B2 O3 /MgO catalyst, which produces no lactam but cyclohexanone and aniline and is similar to that of Na-ZrO2 and Na-Al2 O3 in previous reports [9,14], confirms that the formation of ε-caprolactam does not proceed over a surface that possesses strong basic sites. On the other hand, strong acid sites, like those required for the isomerization of xylene [19] and those on sulfated or halogenated zirconia [14], are also not effective for the lactam synthesis. Surface sites of intermediate acid strength were shown to be responsible for the selective formation of the lactam [7,9,10,14]. These acid sites were characterized by heat of ammonia adsorption of greater than 80 kJ/mol over B2 O3 /SiO2 catalysts [7], and by desorption of adsorbed ammonia between 200◦ C and 350◦ C over B2 O3 /Al2 O3 catalysts [9,10]. In the present work the TPD profiles over B2 O3 /ZrO2 seem best deconvoluted to three peaks with maximum desorption rate near 100◦ C, 200◦ C and 300◦ C. There are no stronger acid sites which are capable to adsorb ammonia molecules at above 400◦ C (Fig. 8). Since the desorption of ammonia at 200–350◦ C over these B2 O3 /ZrO2 catalysts are contributed by the two peaks at around 200◦ C and 300◦ C, we extend the upper temperature to 400◦ C to cover the whole peak centered at 300◦ C for counting the intermediate strong acid sites on B2 O3 /ZrO2 catalysts. In Fig. 9 the relative area of ammonia desorption between 200◦ C and 400◦ C (amount of intermediate strong acid sites) and their percentage to the total desorption area are plot-

B.-Q. Xu et al. / Applied Catalysis A: General 188 (1999) 361–368

Fig. 9. Influence of boria load on the acidity (area of ammonia desorption at 200–400◦ C; -䉬-) and on the desorption percentage (-4-) over B2 O3 /ZrO2 catalyst.

ted against the boria load of B2 O3 /ZrO2 catalysts 1 . On recalling of the results for the Beckmann reaction (Fig. 4) it is easy to realize that the catalytic activity of B2 O3 /ZrO2 catalyst is in proportional to the number of intermediate strong acid sites or area of ammonia desorption at 200–400◦ C. The lactam selectivity, on the other hand, is in parallel with the desorption percentage. These data are at variance of that observed with B2 O3 /Al2 O3 catalysts [9,10]. According to Curtin et al. [9,10] no clear relationship existed between oxime conversion and the surface acidity of B2 O3 /Al2 O3 catalysts; the lactam selectivity over those B2 O3 /Al2 O3 catalysts was found to be related to the number of intermediate strong acid sites from which desorption of ammonia took place at 200–350◦ C [9,10]. These observations are in contrast with those reported for the zeolite catalysts on which nearly neutral silanol groups are responsible for the gas phase Beckmann rearrangement reaction [3,6]. It has appeared that the acidity requirement for the vapor phase reaction cannot be extrapolated when the reaction is performed in liquid phase where strong Brönsted acid sites are responsible for the catalysis [20]. The change of catalysis and acidity of B2 O3 /ZrO2 catalyst with boria load may be related to the dispersion structure of boria. IR measurement of the B2 O3 /ZrO2 catalysts (Fig. 7) reveals that tetrahedral oxygen-coordinated boron (BO4 ) units are the main 1 Qualitatively, Fig. 9 is independent of the selection of the upper temperature for evaluating the acidity: the shape of the curves do not change when the area and percentage are counted as those are responsible for ammonia desorption between 200◦ C and 350◦ C.

367

structure of boria dispersion at low boria load (5% B2 O3 ). Trigonal BO3 units increase with increasing the boria load. Since monolayer dispersion is usually expected with low boria load, this first overlayer of boria on the zirconia surface should be composed of the BO4 units. The development of BO3 units at higher boria load is most probably related with a stack of boria on the interconnected BO4 units of the monolayer. In a separate work the dispersion structure of boria on zirconia surface has been characterized with various physical measurements (XPS, FT-Raman and 11 B MASNMR, etc.). The results, which will be published elsewhere [21], confirm the structure of BO4 units in the first monolayer and indicate that the monolayer threshold is at 0.050 ± 0.001 g B2 O3 /g-ZrO2 or 12 ␮mol B2 O3 /m2 -ZrO2 . Using the density of crystalline boria of 2.44 g/cm3 the thickness of boria layer at the 5% load level is calculated to be 0.34 nm, which is very well consistent with a spacing of only one-layer thickness. It turns out that the optimum boria load (10% B2 O3 ) for the lactam synthesis (Fig. 4) just happens at an average thickness of ca. 2 surface layers. Then, it is conclusive that the monolayer dispersed BO4 state of boria is not an efficient catalyst for the lactam synthesis. The best performance of catalysis of B2 O3 /ZrO2 catalyst is determined by a combination state of dispersed boria species which contains mainly BO3 as the structure units with a BO4 sublayer lays in between the support surface and the BO3 structures. The formation of BO3 structure units by stacking boria onto the first monolayer of their BO4 colleagues on the zirconia surface is crucial for the selective synthesis of the lactam. The decrease in oxime conversion at high stack thickness or high levels of boria load may be explained by reduction of the number of such lactam selective BO3 units. The higher stack thickness buries more BO3 structure and these buried BO3 units become not available for the reaction. As a result, the number of acid sites and the activity of the catalyst decrease at higher level of boria load (>10% B2 O3 ). To a better understanding of the loss of acid sites and catalyst activity at high boria load, we measured the surface area and porosity of the B2 O3 /ZrO2 catalysts by nitrogen adsorption (Fig. 10). As being expected, both the surface area and pore volume of the catalyst decrease with the boria load when it becomes higher than the optimum value (10% B2 O3 ).

368

B.-Q. Xu et al. / Applied Catalysis A: General 188 (1999) 361–368

References

Fig. 10. Influence of boria load on the surface area (-4-) and pore volume (-䉬-) of B2 O3 /ZrO2 catalyst.

While deactivation of B2 O3 /ZrO2 catalysts can not be avoided in the present study, it is remarkable that the catalyst optimized for boria load (10% B2 O3 ) shows also the highest catalyst stability for the lactam synthesis (Fig. 6). Our primary work which investigates effects of solvents and carrier gases has revealed that an use of CO2 as the reaction carrier gas has a remarkable effect in preventing deactivation of the 10% B2 O3 /ZrO2 catalyst [22]. Further study is being made to improve the catalyst stability by optimizing other parameters of the catalyst preparation and of the reaction atmosphere. In conclusion, the present work shows that zirconia-supported boria is a highly active as well as selective catalyst for the synthesis of ε-caprolactam by gas phase Beckmann rearrangement of cyclohexanone oxime at 300–320◦ C. The preformance of B2 O3 /ZrO2 catalyst is affected by the boria load. Increasing the boria load (≤20%) results in increase of the lactam selectivity and of the percentage of intermediate strong acid sites. However, the lactam yield coincides with the activity and the number of such acid sites. The optimum boria load (10%) corresponds to two overlayers of boria on the zirconia support.

Acknowledgements Finacial support for this work (grant no. 2890018) from the National Natural Science Foundation of China (NSFC) is greatly acknowledged.

[1] K. Tanabe, Appl. Catal. 113 (1994) 147. [2] P.S. Landis, P.B. Venuto, J. Catal. 6 (1966) 245. [3] H. Sato, K. Hirose, M. Kitamura, Y. Nakamura, Stud. Surf. Sci. Catal. 49 (1989) 1213. [4] T. Takahashi, K. Ueno, T. Kai, Can. J. Chem. Eng. 69 (1991) 1096. [5] A. Thangaraj, S. Sivasanker, P. Ratnasamy, J. Catal. 137 (1992) 252. [6] T. Yashima, K. Miura, T. Komatsu, Stud. Surf. Sci. Catal. 84C (1994) 1897. [7] H. Sato, K. Urabe, Y. Izumi, J. Catal. 102 (1986) 99. [8] H. Sato, S. Hasebe, H. Sakurai, K. Urabe, Y. Izumi, Appl. Catal. 29 (1987) 107. [9] T. Curtin, J.B. McMonagle, B.K. Hodnett, Appl. Catal. 93 (1992) 75. [10] T. Curtin, J.B. McMonagle, B.K. Hodnett, Appl. Catal. 93 (1992) 91. [11] K. Tanabe, M. Misono, Y. Ono, H. Hattori, New Solid Acids and Bases, Kodansha-Elsevier, 1989. [12] T. Yamaguchi, Catal. Today 20 (1994) 199. [13] J.A. Lercher, J.H. Bitter, W. Niessen, K. Seshan, Stud. Surf. Sci. Catal. 101A (1996) 463. [14] S.-B. Cheng, B.-Q. Xu, D.Q. Wang, F. Wang, T. Cai, X. Wang, Chinese J. Catal. (Cuihua Xuebao) 17 (1996) 281. [15] C.E. Weier, J. Res. Natn. Bur. Stand. 70A (1966) 153. [16] J. Haleer, V. Szybalska, Disc. Faraday Soc. 72 (1981) 263. [17] J.L. Parsons, M.E. Milberg, J. Am. Chem. Soc. 43 (1960) 326. [18] S. Arya Devi, D. Philip, G. Aruldhas, J. Solid State Chem. 113 (1994) 157. [19] H. Sakurai, S. Sato, K. Urabe, Y. Izumi, Chem. Lett. (1985) 1783. [20] M.A. Camblor, A. Corma, H. Garcia, V. Semmer-Herledan, S. Valencia, J. Catal. 177 (1998) 267. [21] B.-Q. Xu, S.-B. Cheng, Q.M. Zhu, M.G. White, in preparation. [22] S.-B. Cheng, B.-Q. Xu, T. Cai, X. Wang, Chinese J. Catal. (Cuihua Xuebao) 17 (1996) 330.