Hydrotalcite-type materials as catalysts for the synthesis of dimethyl carbonate from ethylene carbonate and methanol1

Microporous and Mesoporous Materials 22 (1998) 399–407

Hydrotalcite-type materials as catalysts for the synthesis of dimethyl carbonate from ethylene carbonate and methanol1 Yoshiaki Watanabe, Takashi Tatsumi * Engineering Research Institute, School of Engineering, The University of Tokyo, Yayoi, Tokyo 113, Japan Received 5 January 1998; accepted 27 February 1998

Abstract Dimethyl carbonate can be synthesized from ethylene carbonate and methanol by using hydrotalcite-type materials as base catalysts. Especially, Mg–Al hydrotalcite-type materials with low Al concentration in brucite-like layers and high OH− proportion in the intercalated anions give high yield of dimethyl carbonate. © 1998 Elsevier Science B.V. All rights reserved. Keywords: Hydrotalcite-type materials; Base catalysis; Dimethyl carbonate; Ester exchange reaction

1. Introduction Hydrotalcite-type materials are double-layered hydroxides expressed by the following formula: [M2+M3+(OH ) ]+(Am− ) · nH O, where x 2(x+1) 1/m 2 M2+=Mg2+, Zn2+ or Ni2+, and M3+=Al3+ or Fe3+ [1]. These hydrotalcite-like materials are referred to as M N–A in this paper. The crystal x structure consists of positively charged brucite-like octahedral hydroxide layers, which are neutralized by the interlayer anions. The compensation anions, Am−, are organic anions, such as terephthalate anion and sebacate anion, or inorganic anions, such as Cl−, CO2− and polyoxometallate 3 anions [2,3]. These anions in the interlayer can be * Corresponding author. Fax: +81 3 5800 6825; E-mail: [email protected] 1Dedicated to Professor Lovat V.C. Rees in recognition and appreciation of his lifelong devotion to zeolite science and his outstanding achievements in this field. 1387-1811/98/$19.00 © 1998 Elsevier Science B.V. All rights reserved. PII: S1 3 8 7 -1 8 1 1 ( 9 8 ) 0 0 09 9 - 7

exchanged with other anions and the distance between the basal planes is determined by the ionic radius of the anion [4]. Water molecules are also present between the hydroxide layers. Thermal decomposition of these materials has been studied; for Mg Al–CO hydrotalcite, the water between x 3 layers is lost below 473 K, and carbon dioxide and further water produced from the dehydroxylation of octahedral hydroxide layers are lost between 523 and 723 K. The thermally-decomposed material is a mixture of Mg–Al mixed oxide. This oxide recovers a layered structure by treatment with hot water [5]. These hydrotalcite-like materials are used not only as ion-exchangers [6 ], but also as solid base catalysts [7]. Dimethyl carbonate (DMC ) is an important precursor of polycarbonate resins as well as a useful carbonylation and methylation agent. Because of the negligible toxicity of DMC, it is promising as a substitute for phosgene, dimethyl sulfate, or methyl iodide [8]. It also has potential

400

Y. Watanabe, T. Tatsumi / Microporous and Mesoporous Materials 22 (1998) 399–407

as a petrol octane enhancer [9]. DMC has long been synthesized by reaction of methanol with phosgene [10]. A tremendous amount of research has been pursued to establish environmentally compatible non-phosgene routes for the production of DMC. Recently, a number of nonphosgene processes for preparing DMC have been developed. For example, DMC production from methanol, carbon monoxide and oxygen has been industrialized by using Cu-based catalyst [11]. Very recently, a new process has been developed which involves the reaction of methyl nitrite and carbon monoxide catalyzed by Pd catalyst [12]. It is known that DMC can be synthesized concomitantly with ethylene glycol by the following ester exchange reaction between ethylene carbonate ( EC ) and methanol:

For this reaction, various base catalysts and Group IV homogeneous catalysts are reported to be effective [13]. Here we report the applicability of hydrotalcite-like materials to solid base catalysts for DMC synthesis by the ester exchange reaction between EC and methanol.

2. Experimental 2.1. Preparation of hydrotalcite-type compounds 2.1.1. Preparation of Mg Al–NO and Ni Al–NO x 3 2 3 (Method I) Mg Al–NO was prepared by adoption of the x 3 induced hydrolysis method developed by Taylor [14]. Dissolved CO was removed by bubbling 2 N through deionized water, which was used 2 for the preparation of all solutions. In the case of Mg Al–NO (Sample A), Mg(NO ) · 6H O 2 3 32 2 (80 mmol ) and Al(NO ) · 9H O (40 mmol ) were 33 2 dissolved in 65 ml deionized water under N atmo2 sphere. This solution was added dropwise to an aqueous solution of NaOH (375 mmol in 100 ml ) in 90 min, with vigorous stirring. The pH of the mixture was 13. Following the addition, the mix-

ture was kept at 346 K for 18 h with stirring. The resultant slurry was cooled to room temperature, and then the solid product was separated by filtration, washed thoroughly with deionized water and dried at 393 K overnight. Mg Al–NO was synthesized at various pH by 2 3 changing the concentration of aqueous solution of NaOH. The pH of the mixture using aqueous NaOH solutions (1.5 M, 2.5 M and 4.7 M ) were 5.7, 9.2 and >14, respectively. The samples thus prepared are referred to as Samples B, C and D. For preparation of Ni Al–NO , Ni(NO ) · 2 3 32 6H O was used in place of Mg(NO ) · 6H O and 2 32 2 the pH of the mixture of was 11. 2.1.2. Preparation of Mg–Al mixed oxide by thermal decomposition of Mg Al–NO and 2 3 restoration of layered structure Mg–Al mixed oxide (Sample A∞) was prepared by calcining Sample A in air at 723 K for 18 h. Restoration of a layered structure was performed as follows: 1 g of the calcined material was stirred in 100 ml distilled water at 373 K for 40 min. The material thus obtained was referred to as Sample A◊. 2.1.3. Preparation of Mg Al–NO (Method II) x 3 Aqueous solution of NaOH (3.75 M ) was added dropwise to the mixture solution of Mg(NO ) · 6H O and Al(NO ) · 9H O. The 32 2 33 2 following procedures were the same as Method I. Hydrotalcite-like materials having various Mg/Al ratios were prepared at a pH of 13. Samples E, F and G were obtained from the synthesis mixture with an Mg/Al ratio of 2, 3 and 4, respectively. 2.1.4. Preparation of Mg Al–A (A=OH−, Cl−, x SO2−) 4 Mg Al–OH, Mg Al–Cl and Mg Al–SO 2 2 2 4 (Samples H, I and J, respectively) were prepared by ion-exchange of Mg Al–NO prepared by the 2 3 same method as Sample A. Mg Al–NO (1.0 g) 2 3 was added to a solution of sodium compounds containing the corresponding anion dissolved in 100 ml deionized water. The anion was in 50– 100% stoichiometric excess of NO− present in 3 Mg Al–NO . The mixture was stirred for 2 h at 2 3 343 K. After two ion-exchange treatments, the product was separated by filtration, washed and dried at 393 K overnight.

Y. Watanabe, T. Tatsumi / Microporous and Mesoporous Materials 22 (1998) 399–407

2.1.5. Preparation of Zn Al–NO x 3 In the case of Zn Al–NO , Zn(NO ) · 6H O 2 3 32 2 (80 mmol ) and Al(NO ) · 9H O (40 mmol ) were 33 2 dissolved in 65 ml deionized water under N 2 atmosphere. For Zn Al–NO , Zn(NO ) · 6H O 3 3 32 2 (90 mmol ) and Al(NO ) · 9H O (30 mmol ) were 33 2 dissolved in deionized water. A solution of 4 M NaOH was added dropwise to the solution of Zn2+ and Al3+ in 90 min with vigorous stirring, the pH of the reaction mixture being adjusted to 7.0. Following the addition, the mixture was kept at 346 K for 18 h with good mixing. The product was separated by filtration, washed and dried at 393 K overnight.

401

2.3. Catalytic reactions Ester exchange reactions were carried out in a flask containing catalyst (89 mg) without any particular treatment, ethylene carbonate (2.72 g) and methanol (3.96 g). The mixture was refluxed for 3 h. Products were analyzed on a Shimadzu GC-14A gas chromatograph with a flame ionization detector using a DB-5 (30 m) capillary column.

3. Results and discussion

2.2. Physical methods

3.1. Effect of layered structure of hydrotalcite-type material

XRD patterns were recorded on a Rigaku Denki RINT 2400 diffractometer. Chemical composition was determined with a Kevex mX7000 energy dispersive spectrometer ( EDS) and a Yanaco MT2 CHN recorder. The absolute contents of intercalated NO− and CO2− in samples were measured 3 3 by CHN analysis. The amount of OH− was estimated as a balance on the assumption that NO− , CO2− , and OH− neutralize the electric 3 3 charge on brucite-like layers. FE-SEM micrographs were obtained with a Hitachi S-900 scanning electron microscope. The surface area of the catalyst was determined by N -BET measurement 2 on a BEL Japan Belsorp 28SA analyzer. Before the N -BET measurement, samples were kept in 2 vacuum at 403 K for 5 h. The XRD patterns indicated that no decomposition of layered structure occurred during this pretreatment. CO -TPD 2 was measured on a Bell Japan Multitask TPD instrument equipped with an ANELVA Q-Mass spectrometer by the following procedure: sample pretreated in vacuum at 403 K for 3 h was kept under CO (20 torr) at 323 K for 15 min. After 2 evacuation for 1 h, TPD measurement was performed at a heating rate of 5 K/min under He flow (50 ml/min). In order to distinguish adsorbed CO from CO generated from intercalated 2 2 CO2− during decomposition of hydrotalcite-type 3 materials, blank TPD experiments were performed without the pre-exposure to CO (20 torr) and the 2 difference spectra were obtained to evaluate the amounts and strength of the basic site.

X-ray diffraction patterns of Samples A, A∞ and A◊ are shown in Fig. 1. Sample A exhibited a characteristic pattern of well crystallized product with a bidimensional structure. The d spacing calculated from the reflection at around 2h=ca. ˚ . The height of the interlayer was 10° is 7.74 A ˚ because the thickness of estimated to be ca. 3.0 A the brucite-like octahedral hydroxide layer com˚ [15]. The XRD posed of Mg and Al was 4.77 A pattern of Sample A∞ prepared by calcining Sample A at 723 K showed transformation of the hydrotalcite into a mixed oxide with the MgO structure. Although Sample A◊ prepared by hydrating Sample A∞ showed the same XRD pattern as Sample A, the peak intensities of Sample A◊ were weaker than those of Sample A. It is considered that Sample A◊ partially recovered the hydrotalcite-type structure. As shown in Fig. 2, Sample A consisted of very small hexagonal particles with a size of ca. 75 nm, and these small particles were coagulated to particles larger than 200 nm by calcination. The BET surface area of Sample A measured by molecular N was 52 m2/g. It was 2 reported that the distance of two neighboring ˚ metallic cations in brucite-like layer was 3.1 A [16 ]. The outer surface of 75 nm hexagonal particles of Sample A was estimated to be 112 m2/g from SEM micrographs by using the d spacing ˚ ) and the above length (3.1 A ˚ ) of two (7.74 A neighboring metallic cations. For this calculation, the thickness of 75 nm hexagonal particles was assumed to be 15 nm because it was known that

402

Y. Watanabe, T. Tatsumi / Microporous and Mesoporous Materials 22 (1998) 399–407

Fig. 1. XRD patterns of Samples A, A∞ and A◊.

such hydrotalcite-type materials as Sample A had a breadth to thickness ratio between 5 and 10 [1,17]. The observed BET surface area could be smaller than the calculated one because some of the particles are aggregated. Thus, the BET surface area might be due to only the outer surface because the gallery height of Sample A was too small for molecular N to enter. The particle shape of 2 Sample A◊ was quite different from that of Sample A and its size was much larger than the size of Sample A. The surface area of Sample A◊ estimated from SEM micrographs was <5 m2/g and the BET surface area of Sample A◊ was below the accuracy of measurement. Table 1 shows the results of DMC synthesis catalyzed by Samples A, A∞ and A◊. Sample A showed the highest activity. Although Sample A∞ exhibited higher BET (192 m2/g) than Sample A, Sample A∞ showed much lower activity than Sample A. Thus the double-layered hydroxide structure seems to be responsible for the activity of the ester exchange reaction. It is considered that this reaction proceeds on the outer surface of hydrotalcite-like materials because the size of EC is much larger than the gallery height between the

brucite-like layers. Sample A◊ with the layered structure recovered showed higher activity than Sample A∞; however, Sample A◊ gave lower DMC yield than Sample A. Although the lower activity of Sample A◊ would be partly due to its smaller surface area, interestingly the difference in the activity between these samples was not so great compared to the difference in the surface area. This might be due to the difference in morphology between the two samples. It is speculated that the reaction occurs at the edge part of the layered structure. 3.2. Effect of pH at synthesis of hydrotalcite-type material Table 2 shows the results of reactions catalyzed by hydrotalcite-type materials consisting of various pairs of metals. Mg Al–NO (Sample A) exhibited 2 3 higher activity than Zn Al–NO or Ni Al–NO . It 2 3 2 3 occurred to us that the difference in activities might be related to the difference in pH during the synthesis of samples; Mg Al–NO , Ni Al–NO 2 3 2 3 and Zn Al–NO were synthesized at pH 13, 11 2 3 and 7.0, respectively.

Y. Watanabe, T. Tatsumi / Microporous and Mesoporous Materials 22 (1998) 399–407

Sample A

Sample A'

Sample A''

Sample A''

Fig. FE-SEM micrographs of Samples A, A∞ and A◊.

403

404

Y. Watanabe, T. Tatsumi / Microporous and Mesoporous Materials 22 (1998) 399–407

Table 1 Effect of layered structure of hydrotalcite-type material on synthesis of DMC Sample

Main content

EC conversion (%)

DMC yield (%)

A A∞ A◊

Mg Al–NO 2 3 MgO+Al O 2 3 Mg Al–OH 2

54 36 46

40 9.9 28

Reaction conditions: MeOH/EC=4 N with MeOH being refluxed, 3 h. 2

(mol/mol );

(MeOH+EC )/cat.=75

(w/w).

Catalyst

weight

89 mg,

under

Table 2 Synthesis of DMC catalyzed by various hydrotalcite-type materials Catalyst

pHa

EC conversion (%)

DMC yield (%)

NO /Mg Al (Sample A) 3 2 NO /Ni Al 3 2 NO /Zn Al 3 2 NO /Zn Al 3 3

13 11 7.0 7.0

54 22 11 20

40 18 0.37 4.4

Reaction conditions: MeOH/EC=4 (mol/mol ); (MeOH+EC )/cat.=75 (w/w). Catalyst weight 89 mg, under N with MeOH being 2 refluxed, 3 h. a pH of the mother liquor for the synthesis of catalyst.

To investigate the effect of pH during the synthesis of hydrotalcite-type materials, Mg Al–NO 2 3 (Samples B, C and D) as well as Sample A were synthesized at various pH values (5.7, 9.2 and >14) by Method I. The Mg/Al ratios of Samples B, C, A and D were 0.5, 1.5, 1.8 and 1.8, respectively. It was observed by XRD that all the samples had layered structure patterns and that both the pattern of the layered structure and that of the

boehmite was observed for Sample B with high Al concentration. The EC conversion and DMC yield on these samples are shown in Fig. 3. Mg Al–NO synthesized at higher pH gave rise to 2 3 higher activity and selectivity for DMC. This may be due to the high OH− content of Mg Al–NO 2 3 synthesized at high pH, as will be described in Section 3.4. Samples A∞, B∞, C∞ and D∞ were prepared by calcination of Samples A, B, C and D at

Fig. 3. Synthesis of DMC on Mg Al-NO prepared under various pH values. Reaction conditions: MeOH/EC=4 (mol/mol ); 2 3 (MeOH+EC )/cat.=75 (w/w). Catalyst weight 89 mg, under N with MeOH being refluxed, 3 h. 2

405

Y. Watanabe, T. Tatsumi / Microporous and Mesoporous Materials 22 (1998) 399–407 Table 3 Synthesis of DMC catalyzed by Mg Al–NO x 3 Sample

Mg/Al

EC conversion (%)

DMC yield (%)

A E F G

1.8 1.8 2.1 2.5

54 60 66 70

40 45 54 58

Reaction conditions: MeOH/EC=4 (mol/mol ); (MeOH+ EC )/cat.=75 (w/w). Catalyst weight 89 mg, under N with 2 MeOH being refluxed, 3 h.

723 K for 18 h, respectively. As shown in Fig. 3, the calcination resulted in a decrease in the activity. Samples A∞ and D∞, with the same Al concentration, showed similar EC conversion. Sample B∞ showed slightly lower activity than the others. This might be due to its high Al concentration. This is in agreement with the observation by Miyata et al. [18], who reported that the basicity of hydrotalcitederived mixed oxide increased with increasing Mg/Al ratio up to 3. However, the difference in catalytic activities among the calcined samples was less remarkable than that among uncalcined samples, probably because the calcined samples have lost the intercalated anions upon the decomposition of the layered structure. 3.3. Effect of Mg/Al ratio of Mg Al–NO x 3 Samples A and E were synthesized from a solution with Mg/Al ratio of 2.0. In the case of Samples F and G, the Mg/Al ratio of the solution was set at 3.0 and 4.0, respectively. The XRD

pattern for every sample consists of sharp and symmetrical peaks, being characteristic of a layered structure and similar to the pattern of Sample A. The BET surface areas of Samples A, E, F, and G were 52, 56, 48, and 43 m2/g, respectively. Results of DMC synthesis catalyzed by these samples are shown in Table 3. Although there was no significant difference in Al concentration between Samples A and E, Sample E prepared by Method II showed slightly higher activity than Sample A. It was observed that EC conversion and DMC yield increased with decreasing Al content for Samples E, F and G in spite of the decreased BET surface area. It was reported that the amount of basicity on hydrotalcite-type materials increased with decreasing concentration of Al in the brucite-like layer [19]. Kaneda et al. [20], investigating hydrotalcite-type materials as basic catalysts for Baeyer–Villiger oxidation, showed that the activity increased with increasing Mg/Al ratio up to 5. For the ester exchange reaction, the best results were obtained with Sample G (Mg/Al=2.5); EC conversion and DMC yield were 70% and 58%, respectively. It is noteworthy that this ester exchange reaction is an equilibrium reaction and the EC conversion (%) and DMC yield (%) at equilibrium at 338 K are estimated to be 81% and 64%, respectively. 3.4. Effect of intercalated anions of hydrotalcitetype materials Table 4 shows anion content existing between hydroxide layers of Samples A, H, I and J and the

Table 4 Effect of intercalated anion of hydrotalcite-type material on DMC synthesis Sample

A H I J D G

Mg/Al

1.8 1.7 2.1 1.9 1.8 2.5

Anion ratio OH−

CO2− 3

NO− 3

Cl− or SO2− 4

0.27 0.29 0.21 0 0.33 0.28

0.29 0.32 0.12 0.10 0.25 0.20

0.15 0.07 tr. 0 0.17 0.32

0 0 0.55 (Cl−) 0.45 (SO2−) 4 0 0

OH−content (10 mmol/g catalyst)

EC conversion (%)

DMC yield (%)

1.2 1.3 0.8 0.0 1.5 1.0

54 63 32 31 64 70

40 54 6.1 8.9 54 58

Reaction conditions: MeOH/EC=4 (mol/mol ); (MeOH+EC )/cat.=75 (w/w). Catalyst weight 89 mg, under N with MeOH being 2 refluxed, 3 h.

406

Y. Watanabe, T. Tatsumi / Microporous and Mesoporous Materials 22 (1998) 399–407

Samples D and G with varying anion proportions were also examined. Sample D having the same Mg/Al ratio as Sample A showed higher activity than Sample A, probably because of the high OH− proportion of Sample D, which was synthesized at higher pH than Sample A. Sample G with low Al content showed higher EC conversion and DMC yield than Sample D regardless of its low OH− content compared to Sample D. From these results, it was confirmed that the activities of hydrotalcite-type materials increased with decreasing Al content in brucite-type layers as described in the previous section. Although Sample H contained the largest amount of OH− on a weight basis, its activity was lower than Sample G. As shown in Fig. 4, particles were aggregated during ion-exchange reaction and the BET surface area of Sample H was too small to be measured, whereas the surface area of Sample G was 43 m2/g.

Fig. 4. FE-SEM micrograph of Sample H.

activity of these samples. It was observed that the proportion of NO− in intercalated anions easily 3 decreased during ion-exchange reaction. These results were in agreement with the order of stability for anions intercalated between the hydrotalcite layers, OH−>SO2− >F−>Cl−>Br−>NO− 4 3 >I − [21]. Sample H with low NO− and high 3 OH− proportions showed higher activity than Sample A. The EC conversion and DMC yield on Samples I and J were much lower than Sample A. This might be due to the lower OH− content. Thus, it is obvious that the intercalated anions affect the amount and strength of basic sites on hydrotalcite-type materials as will be described below.

3.5. CO -TPD 2 Table 5 compares the amount of adsorbed CO 2 and the DMC yield on the hydrotalcite-type catalysts. Obviously, the adsorbed CO depended on 2 the OH− content, and the samples adsorbing larger amounts of CO gave a higher yield of DMC, 2 indicating that the amount of basic sites on hydrotalcite-type materials is adjustable by changing intercalated anions and that hydrotalcite-type materials containing a large number of base sites effectively catalyze the ester exchange reaction. The amount of CO adsorbed on Sample G was 2 about 1.2 times more than on Sample A. This might be due to the low Al concentration in brucite-like layers. However, as shown in Fig. 5, the amount of relatively strong basic sites for Sample G was less than for Sample A. This might

Table 5 Relationship between adsorbed CO amount and catalytic activity 2 Sample

Mg/Al

OH− content (mmol/g catalyst)

Adsorbed C (mmol/g catalyst)

DMC yield (%)

A G J

1.8 2.5 1.8

1.2 1.0 0.0

0.086 0.100 0.057

40 58 8.9

Y. Watanabe, T. Tatsumi / Microporous and Mesoporous Materials 22 (1998) 399–407

407

Fig. 5. CO -TPD of Samples A, G and I. 2

be due to the low concentration of OH− as intercalated anions. However, Sample G gave the highest DMC yield, suggesting that weak basic sites are sufficiently active in this ester exchange reaction. Samples were used for reaction without any treatment and must have adsorbed CO . 2 Moreover, samples could adsorb CO dissolving 2 in the substrate mixture of methanol and EC. The adsorbed CO might inhibit the inherent activities 2 of samples because the ester exchange reaction was performed at low temperature (ca. 338 K ). Especially, strong basic sites might be preferentially blocked. This might be the cause for the apparent high activity of the sample with high amount of relatively weak basic sites. 4. Conclusions For the synthesis of DMC by ester exchange reaction between EC and methanol, hydrotalcitetype materials were found effective as catalysts. Especially, the materials containing hydroxide layers with low Al content or a large amount of OH− as intercalated anion were suitable catalysts for this reaction. References [1] W.T. Reichle, Chemtech. (1986) 58. [2] T. Kwon, T.J. Pinnavaia, J. Mol. Catal. 74 (1992) 23.

[3] R.S. Weber, P. Gallezot, F. Lefebvre, S.L. Suib, Microporous Mater. 1 (1993) 223. [4] M.K. Drezdzon, Inorg. Chem. 27 (1998) 4628. [5] C. Kennedy, J. William, J. Chem. Soc., Chem. Commun. (1989) 926. [6 ] E. Suzuki, M. Okamoto, Y. Ono, J. Mol. Catal. 61 (1990) 283. [7] S. Ueno, K. Ebitani, A. Ookubo, K. Kaneda, Appl. Surf. Sci. 121 (1997) 366. [8] A.-A.G. Shaikh, S. Sivaram, Ind. Eng. Chem. Res. 31 (1992) 1167. [9] A.L. Bhattacharya, Prepr. Am. Chem. Soc., Div. Fuel Chem. 40 (1995) 119. [10] E. Abrams, in: Kirk-Othmer (Eds.), Encyclopedia of Chemical Technology, Wiley, New York, 3rd edn., 1979, p. 758. [11] U. Romano, C. Tesei, M. Massi Mauri, P. Rebora, Ind. Eng. Chem., Prod. Res. Dev. 19 (1980) 396. [12] K. Nishihira, T. Matsuzaki, S. Tanaka, Shokubai (Catalyst) 37 (1995) 68. [13] J.F. Knifton, R.G. Duranleau, J. Mol. Catal. 67 (1991) 389. [14] R.M. Taylor, Clay Miner. 19 (1984) 591. [15] S. Miyata, Clays Clay Miner. 23 (1975) 369. [16 ] C. Fromdel, Am. Miner. 26 (1941) 295. [17] C.P. Kelkar, A.A. Schutz, Microporous Mater. 10 (1997) 163. [18] S. Miyata, T. Kumura, H. Hattori, K. Tanabe, Nippon Kagaku Zasshi 92 (1971) 514. [19] F. Cavani, F. Trifiro, A. Voccad, Catal. Today 11 (1991) 173. [20] K. Kaneda, S. Ueno, T. Imaoka, J. Chem. Soc., Chem. Commun. (1994) 797. [21] S. Miyata, Clays Clay Miner. 31 (1983) 305.