Kinetic study on the effect of solvent in ‘vapor-phase’ Beckmann rearrangement of cyclohexanone oxime on silicalite-1

Kinetic study on the effect of solvent in ‘vapor-phase’ Beckmann rearrangement of cyclohexanone oxime on silicalite-1

Microporous and Mesoporous Materials 35–36 (2000) 173–180 www.elsevier.nl/locate/micromeso Kinetic study on the effect of solvent in ‘vapor-phase’ Be...

162KB Sizes 2 Downloads 20 Views

Microporous and Mesoporous Materials 35–36 (2000) 173–180 www.elsevier.nl/locate/micromeso

Kinetic study on the effect of solvent in ‘vapor-phase’ Beckmann rearrangement of cyclohexanone oxime on silicalite-1 Takayuki Komatsu *, Takeya Maeda, Tatsuaki Yashima Department of Chemistry, Tokyo Institute of Technology, 2-12-1 Ookayama, Meguro-ku, Tokyo 152-8550, Japan Received 6 April 1999; received in revised form 4 June 1999; accepted for publication 6 June 1999 Dedicated to the late Werner O. Haag in appreciation of his outstanding contributions to heterogeneous catalysis and zeolite science

Abstract The effect of solvents in the ‘vapor-phase’ Beckmann rearrangement of cyclohexanone oxime was studied on silicalite-1 catalyst through kinetic measurements. Solvents with various polarities were examined for the reaction. It was found that ethanol, which has medium polarity, was effective in the formation of e-caprolactam. The rate of e-caprolactam formation fitted the Langmuir-type rate equation for the various solvents. This experimental rate equation was explained by the reaction scheme where the desorption of e-caprolactam caused by the attack of a solvent molecule was the rate-determining step. IR measurements confirmed the desorption of e-caprolactam induced by the attack of the ethanol molecule. Ethanol and methanol gave specific improvements in the yield of e-caprolactam probably due to the effect of OH groups. However, alcohols with higher carbon numbers lowered the yield owing to their hydrophobic nature. Therefore, ethanol and methanol are the most effective solvents for the ‘vapor-phase’ Beckmann rearrangement of cyclohexanone oxime on silicalite-1. © 2000 Elsevier Science B.V. All rights reserved. Keywords: Beckmann rearrangement; e-Caprolactam; Cyclohexanone oxime; Kinetic; Silicalite; Solvent; Zeolite

1. Introduction Beckmann rearrangement of cyclohexanone oxime into e-caprolactam, which is the intermediate for Nylon 6, has been carried out in industry using sulfuric acid as a catalyst. However, this process yields undesirable ammonium sulfate in a greater amount than the desired e-caprolactam. Moreover, the use of corrosive sulfuric acid should be avoided for environmental and economic reasons. Since ‘vapor-phase’ Beckmann rearrange* Corresponding author. Fax: +81-3-5734-2758. E-mail address: [email protected] ( T. Komatsu)

ment on solid catalysts would solve these problems, solid acid catalysts, especially zeolites and zeolitelike microporous materials, have been studied for an alternative clean process. ZSM-5 with an extremely low aluminum content, silicalite-1, was first reported to be highly active and selective for the production of e-caprolactam through the ‘vapor-phase’ Beckmann rearrangement [1]. More recently, various zeolites and metallosilicates have been reported to be effective, e.g. MFI titanosilicate [2], MFI borosilicate [3], MEL [4], FER and LTA [5], BEA [6 ], LTL and OFF-ERI [7]. In our case [5], Ca2+-exchanged LTA and H-MFI borosilicate showed higher selectivity to e-caprolactam than H-MOR and H-ZSM-5. Among zeolites with

1387-1811/00/$ - see front matter © 2000 Elsevier Science B.V. All rights reserved. PII: S1 3 8 7 -1 8 1 1 ( 9 9 ) 0 0 21 8 - 8

174

T. Komatsu et al. / Microporous and Mesoporous Materials 35–36 (2000) 173–180

the MEL structure [4], titanosilicate and silicalite gave higher selectivities than aluminosilicate (H-ZSM-11). In the case of H-ZSM-5 [1], the selectivity increased with increasing Si/Al ratio of the zeolite. These observations suggest that the strong acid sites are not the active site for the selective formation of e-caprolactam. In order to feed cyclohexanone oxime, a solid at ambient temperature, as a gaseous reactant, it is necessary to dissolve it in a solvent. Benzene has often been used for this purpose [1,2,4]. Results with various solvents revealed that the solvent strongly affected the activity, selectivity and stability of zeolite catalysts. In the case of H-BEA [6 ], alcohols gave higher selectivity to e-caprolactam than benzene; 1-hexanol exhibited the highest selectivity of 96 mol.%. On H-ZSM-5 [8], both conversion and selectivity showed the highest values (82% and 88 mol.%, respectively) with ethanol among various solvents. In the case of MFI borosilicate [3], ethanol provided a rather stable activity, while the conversion decreased markedly with time on stream when other solvents, such as benzene and toluene, were used. In spite of these investigations on the influence of solvents, there is no definite explanation of the reason for the enhancement in catalytic properties by the specific solvents. In this study, kinetic studies for the ‘vapor-phase’ Beckmann rearrangement of cyclohexanone oxime are carried out on a silicalite-1 catalyst. The purpose of this study is to clarify the effect of solvents on the catalytic properties of silicalite-1 for the formation of e-caprolactam.

0.98Na O : 41.1H O. The zeolite was ion 2 2 exchanged with an aqueous solution of ammonium nitrate at 343 K for 3 days with decantations twice a day, dried and calcined in air at 773 K for 5 h to form H-ZSM-5. The Si/Al atomic ratio and the specific surface area measured by N adsorption 2 were 20 and 430 m2 g−1, respectively. Silicalite with MFI structure (silicalite-1) was synthesized hydrothermally at 433 K for 72 h in a similar manner to that of ZSM-5 without using aluminum sulfate. The gel composition was SiO : 0.25TPABr 2 : 0.44Na O : 28.4H O. The zeolite was treated 2 2 with hydrochloric acid at 298 K for 1 day to eliminate aluminum impurities, dried and calcined in air at 773 K for 5 h. The Si/Al ratio and specific surface area were ca. 1300 and 460 m2 g−1, respectively. 2.2. Reaction The Beckmann rearrangement of cyclohexanone oxime was carried out in a continuous flow reactor under atmospheric pressure. A known amount of catalyst placed in the reactor (i.d.= 17 mm) was dehydrated at 723 K in flowing helium for 1 h. Then the reaction was started at 593 K by feeding cyclohexanone oxime diluted with a solvent ( Table 1) using helium as a carrier. Products and unreacted cyclohexanone oxime collected in a trap at 273 K were analyzed by gas chromatography with an interval of 30 min on stream. The standard conditions are W/F of 1.2 g h mol−1, and partial pressures of cyclohexanone oxime and the solvent were 0.44 and 23 kPa, respectively. The selectivity

2. Experimental

Table 1 Polarity of solvents

2.1. Catalyst preparation

Solvent

Dipole moment (D)

ZSM-5 was synthesized hydrothermally at 423 K for 20 h. Sodium silicate solution ( Water glass No. 3, Nippon Chemical Ind.), aluminum sulfate ( Wako Pure Chemical Ind.) and tetrapropylammonium bromide ( TPABr, Wako Pure Chemical Ind.) were used as silicon and aluminum sources and as template, respectively. Gel composition was SiO : 0.02Al O : 0.09TPABr : 2 2 3

Benzene Cyclohexane Toluene Diethyl ether Ethanol Methanol Pyridine Acetone Acetonitrile

0 0.33 0.38 1.25 1.44 1.66 2.15 2.90 3.92

T. Komatsu et al. / Microporous and Mesoporous Materials 35–36 (2000) 173–180

175

to e-caprolactam was calculated based on the amount of reacted cyclohexanone oxime. 2.3. IR measurements IR spectra were obtained at room temperature on a Shimadzu FTIR-8100 spectrometer with a self-supporting wafer (25 mg cm−2) of the catalyst. The catalyst set in an IR cell was evacuated at 723 K for 1 h. Then a piece of e-caprolactam (ca. 50 mg) was placed in the cell at 298 K followed by a brief evacuation. The cell was closed and heated at 723 K for 20 min to adsorb e-caprolactam on the catalyst. After that, remaining e-caprolactam was eliminated by evacuation at 473 K for 30 min, and a spectrum of adsorbed e-caprolactam was obtained. The vapor of solvent (1.3 kPa) was then introduced to the catalyst at 473 K for 10 min. After evacuation at 473 K for 30 min, a spectrum was again recorded to determine the influence of the solvent.

Fig. 1. Change in cyclohexanone oxime conversion (#) and e-caprolactam selectivity ($) with time on stream on silicalite-1 with benzene as solvent. Reaction conditions: T=593 K, W/F=1.2 g h mol−1, P(cyclohexanone oxime)=0.44 kPa, and P(benzene)=23.1 kPa in He carrier.

3. Results and discussion Silicalite-1 with MFI structure has been reported to be active and selective for the ‘vaporphase’ Beckmann rearrangement of cyclohexanone oxime [1]. In order to clarify the effect of solvents by kinetic measurements, it would be better to use a highly selective catalyst for the formation of e-caprolactam since this would eliminate the influence of side reactions. Fig. 1 shows the conversion of cyclohexanone oxime and the selectivity to e-caprolactam on silicalite-1 with benzene as the solvent. Although the conversion decreased slightly with time on stream, the selectivity did not change significantly, exhibiting very high values (>98 mol.%). The deactivation was reported to result from the adsorption of volatile material on the acid site of highly siliceous H-ZSM-5 [9] or the pore blocking effect by nitrogen-containing species inside the pores of MFI borosilicate [10]. By-products detected were cyclohexanone, cyclohexenone, 1-cyanopentane and 5-cyanopent-1-ene. We used various solvents having dipole moments of 0–3.92 D ( Table 1) in order to investigate the effect of the polarity of the solvents. Fig. 2

Fig. 2. Effect of the polarity of solvents (P=23.1 kPa) on the yield of e-caprolactam at 165 min on stream. Other reaction conditions were the same as those in Fig. 1.

shows the yield of e-caprolactam through the reaction with these solvents at 165 min on stream. A high solvent/reactant molar ratio of 53/1 led to the complete solvation of cyclohexanone oxime in every solvent at room temperature. The mass balance before and after the catalyst bed was always higher than 95 C-%. Ro¨seler et al. [3] reported the deactivation of MFI borosilicate in the same reaction using several solvents. They found that the degree of deactivation depended on the type of solvent. For example, the conversion of cyclohexanone oxime, which was initially close to 100%, decreased markedly with toluene and benzene, while it was almost constant with ethanol within 8 h on stream. In our case, however, the initial conversion was much lower (<30%) for each solvent and the deactivation was not so severe

176

T. Komatsu et al. / Microporous and Mesoporous Materials 35–36 (2000) 173–180

at 165 min on stream. As shown in Fig. 2, using benzene, the yield was 15 mol.% as calculated from Fig. 1 as well. The yield increased with increasing dipole moment and reached a maximum value (28 mol.%) with ethanol, which has a medium dipole moment of 1.44 D. Further increase in the dipole moment lowered the yield: 16 mol.% with acetonitrile, which has the highest dipole moment of 3.92 D. Although there could be another interpretation for the data of Fig. 2, which is that the yield does not depend on the polarity except for ethanol and methanol, we will discuss a possible reason for the high yields with these alcohols in relation to their medium polarity. The selectivity to e-caprolactam was always high (>95 mol.%) with each solvent. Significant amounts of by-products formed through the reaction of solvents by themselves were never detected, probably because silicalite-1 has a much lower acidity than H-ZSM-5, which will catalyze the dehydration of ethanol, etc. under the reaction conditions applied. It should be noted that basic solvents such as pyridine gave comparable yields, although those will neutralize the acid sites if present. This is again explained by the extremely low acidity of silicalite-1. Fig. 2 clearly shows that ethanol provided the highest rate for the formation of e-caprolactam. We propose that the solvent with medium polarity is preferable for the Beckmann rearrangement on silicalite-1. The reason for this solvent effect was kinetically studied. The conversion was controlled below 10% by changing W/F for the differential reactor conditions. Fig. 3 shows the change in the rate of e-caprolactam formation, R, and selectivity with time on stream when ethanol was used as the solvent. It is clear that both the formation rate and the selectivity did not change significantly with time on stream, indicating that silicalite-1 has a good stability under the reaction conditions for the kinetic study. We used the data obtained at 165 min on stream, hereafter. Fig. 4 shows the change in the rate of e-caprolactam formation with the partial pressure of cyclohexanone oxime, P . Ox The formation rate increased with increasing partial pressure. It is revealed that the reaction order for the Beckmann rearrangement is positive but not first-order with respect to the partial pressure

Fig. 3. Change in selectivity ($) and the rate of e-caprolactam formation (6) with time on stream on silicalite-1 with ethanol as solvent. P(cyclohexanone oxime)=3.5 kPa. Other conditions were the same as those in Fig. 1.

Fig. 4. Relation between the rate of e-caprolactam formation and the partial pressure of cyclohexanone oxime for ethanol as solvent. Reaction conditions: T=593 K, W/F=1.2 g h mol−1, P(ethanol )=23.1 kPa, and time on stream of 165 min.

of cyclohexanone oxime. We obtained a similar relation when other solvents were used. The data for each solvent were tried to fit the Langmuirtype rate equation: kKP Ox . (1) 1+KP Ox Fig. 5 shows the relation between P /R and Ox P for the kinetic data obtained for each solvent. Ox A linear relationship was clearly observed for each solvent, indicating that the formation rate follows the Langmuir-type rate Eq. (1). From the slope and intercept of each line, values of k and K, which correspond to the rate constant and equilibrium constant of adsorption, respectively, were obtained. The values are shown in Table 2 with the dipole moments of the solvents. We considered the reaction scheme, which would obey the rate Eq. (1). The Langmuir– R=

T. Komatsu et al. / Microporous and Mesoporous Materials 35–36 (2000) 173–180

177

molecule, Sv, to form the product e-caprolactam, Cp, and an adsorbed solvent, Sv(a), Eq. (4). Sv(a) is also formed through the adsorption of solvent on the vacant active site, Eq. (5). The active site, A, was proposed to be the neutral SiMOH or an extremely weak acid site of highly siliceous H-ZSM-5 [11]. We considered three conditions, where step (2), (3) or (4) is the rate-determining step to obtain the Langmuir-type rate Eq. (1). When the step (2) is the slowest and the others are in equilibrium, the formation rate of e-caprolactam is expressed as follows: R=k P [A], 2 Ox

Fig. 5. Plot of the Langmuir-type rate Eq. (1) for the solvents of (a) benzene, (b) toluene, (c) ethanol, (d) acetone, and (e) acetonitrile. P(solvent)=23.1 kPa. Other reaction conditions were the same as those in Fig. 4. Table 2 Kinetic parameters for the Beckmann rearrangement of cyclohexanone oxime with various solvents Solvent

Dipole moment (D)

k (104 mol h−1 g−1)

K (10−2 Pa−1)

Benzene Toluene Ethanol Acetone Acetonitrile

0 0.38 1.44 2.90 3.92

6.4 8.7 13 8.7 10

3.2 2.9 1.2 0.83 0.30

Hinshelwood mechanism was taken as follows: Ox+AOx(a),

(2)

Ox(a)Cp(a),

(3)

Cp(a)+SvCp+Sv(a),

(4)

Sv+ASv(a).

(5)

First, the reactant cyclohexanone oxime, Ox, is adsorbed on an active site, A, of the catalyst to form cyclohexanone oxime adsorbed species, Ox(a), Eq. (2). Ox(a) is converted into e-caprolactam adsorbed species, Cp(a), Eq. (3). This is desorbed by the interaction with a solvent

(6)

where k and [A] are the rate constant of step (2) 2 and the concentration of vacant active sites, respectively. Using the equilibrium constants of steps (3), (4) and (5), K , K and K , the rate Eq. (6) 3 4 5 is transformed into the following equation: R=

k K K [A] P 2 3 4 t Ox , K K +1+K P +(1+K )K P 3 4 5 Sv 3 5 Cp

(7)

where P , P and [A] are the partial pressures Cp Sv t of e-caprolactam and solvent and the total concentration of active sites, respectively. The rate Eq. (7) contains P in its denominator instead of Cp P . It is clear that the Eq. (7) does not coincide Ox with the Langmuir-type rate Eq. (1). In the second case, i.e. step (3) is rate determining, the formation rate is obtained as follows: k K [A] P 3 2 t Ox , R=k [Ox(a)]= 3 1+K P +K P /K +K P 5 Sv 5 Cp 4 2 Ox (8) where k is the rate constant of step (3). As the 3 step (3) is assumed to be the slowest, [Ox(a)] should be much higher than [Cp(a)]. Therefore [Cp(a)] [A]

K P [Ox(a)] = 5 Cp % =K P . 2 Ox K [A] 4

(9)

The experimental data were obtained at a constant value of P . Eq. (8) will then be transformed into Sv

178

T. Komatsu et al. / Microporous and Mesoporous Materials 35–36 (2000) 173–180

the following approximation of Langmuir-type: R#

k∞K∞P Ox , 1+K∞P Ox

(10)

where k∞=k [A] , 3 t

K∞=

K

2 . 1+K P 5 Sv

(11)

Table 2 shows that the values of k (=k∞) for the various solvents differ from each other. As [A] t should be constant, k must be varied with the 3 solvent. However, this is not true because step (3) does not contain the solvent molecule. It is concluded that step (3) is not the rate-determining step. Finally, the third case, which means that step (4) is rate determining, is considered. The rate equation is derived as follows: k K K [A] P P 4 2 3 t Sv Ox R=k [Cp(a)]P = , 4 Sv 1+K P +K (1+K )P 5 Sv 2 3 Ox (12) where k is the rate constant of step (4). When 4 the reaction is carried out with higher W/F, the conversion reaches more than 95% at a reaction temperature of 593 K. Therefore, the equilibrium constant K should be much higher than unity. 3 Then Eq. (12) is transformed into the following approximation of Langmuir type: R#

k◊K◊P

Ox , 1+K◊P Ox

(13)

where k◊=k [A] P , 4 t Sv

K◊=

K K 2 3 . 1+K P 5 Sv

(14)

The rate Eq. (14) clearly coincides with the Langmuir-type Eq. (1). Therefore, it is concluded that the reaction scheme shown below, where the rate-determining step is the desorption of e-caprolactam, Eq. (17), fits the rate Eq. (1)

obtained from the experimental data: Ox+APOx(a),

(15)

Ox(a)PCp(a),

(16)

Cp(a)+SvPCp+Sv(a),

(17)

Sv+APSv(a).

(18)

Sato et al. [11] reported an IR study on the Beckmann rearrangement of cyclohexanone oxime on highly siliceous H-ZSM-5 and reported that e-caprolactam is easily formed but hardly desorbed at lower temperatures and that the low acidity of the catalyst facilitates the desorption of e-caprolactam leading to a high yield. Their results agree well with our scheme above derived from the kinetic measurement. From Table 2, the value of K (=K◊) decreased with increasing polarity of the solvent. This would indicate that the solvent molecule with higher polarity interacts more strongly with the active site to increase K and 5 decrease K , resulting in the lower K◊ value, as 2 can be explained by Eq. (14). On the other hand, the value of k (=k◊) reflects the rate of desorption of e-caprolactam since [A] and P are constant. t Sv Table 2 shows that k (=k◊) increased with increasing polarity of the solvent, except for ethanol. The solvent with higher polarity may accelerate the desorption of e-caprolactam through reaction (17), resulting in the increase in k (=k◊). IR studies were carried out to support the above reaction scheme. Fig. 6 shows IR spectra of e-caprolactam adsorbed on H-ZSM-5 (Si/Al=20). As we reported previously [5], the molecular dimension of e-caprolactam is larger than the pore opening of silicalite-1, resulting in the adsorption only on the external surface of its crystallites. The crystallite size of silicalite-1 is of the order of micrometers and much larger than that of ZSM-5. As a result, the amount of adsorbed e-caprolactam will be much smaller than that on ZSM-5. Moreover, the IR light would be scattered significantly by the large silicalite crystallites, resulting in a quality reduction of the spectra. For those reasons, we could not obtain a good spectrum of e-caprolactam on silicalite-1. As shown in Fig. 6(a), e-caprolactam was adsorbed on H-ZSM-5 at 593 K, the reaction temperature, in a

T. Komatsu et al. / Microporous and Mesoporous Materials 35–36 (2000) 173–180

Fig. 6. IR spectra of adsorbed species on H-ZSM-5 (a) after e-caprolactam adsorption and after subsequent exposure to (b) benzene and (d) ethanol, and difference spectra, (c)=(b)−(a) and (e)=(d )−(a).

certain amount. Four distinct absorption bands were observed at 1410, 1440, 1510 and 1640 cm−1. These are attributed to wagging and scissoring of CH , bending of NMH, and stretch2 ing of CNO of e-caprolactam, respectively. Basic e-caprolactam would be adsorbed on the acid sites on the external surface of H-ZSM-5. The major contribution of the external surface as the active sites has also been reported for highly siliceous H-ZSM-5 in the same reaction [12]. After the sample was exposed to vapor of benzene at 473 K followed by the evacuation at 473 K, spectrum (b) was obtained. From the difference spectrum (c) between (a) and (b), it is clear that benzene molecules did not affect the adsorbed e-caprolactam. In the case of ethanol, however, the exposure caused significant changes as shown in the difference spectrum (e), (d )−(a). Negative peaks were observed at 1510 and 1640 cm−1, indicating the desorption of e-caprolactam. A positive peak at 1390 cm−1 is attributed to the symmetric bending of CH . It is indicated that ethanol molecules 3 expelled some of the adsorbed e-caprolactam molecules from the acid sites. This result supports reaction (17), where the solvent accelerates the desorption of e-caprolactam. As shown in Table 2, ethanol gave an especially high value of k (=k◊), suggesting that ethanol has a specific effect on the desorption of e-caprolactam. In Fig. 2, methanol also showed a significantly

179

high yield compared with other non-alcoholic solvents. It is expected that the OH group in the alcohol has the specific effect. Therefore, we tried to clarify the effect of OH by adding water to the reaction mixture. Fig. 7 shows the result obtained on silicalite-1 with ethanol as the solvent. The addition of water appreciably increased the yield of e-caprolactam up to the amount of H O/Ox=1, although this amount corresponded 2 to an H O/C H OH molar ratio of only 1/53. This 2 2 5 is a similar result to that obtained for MFI borosilicate with ethanol as solvent [3]. A further experiment using acetonitrile as solvent revealed that the addition of water (H O/Ox=1.4) also increased 2 the e-caprolactam yield. However, the increase in yield (ca. 5 mol.%) due to the water addition was comparable between ethanol and acetonitrile solvents, although the enhancing effect of water could be stronger in acetonitrile since it has no OH groups. Other alcohols with higher carbon numbers were examined to clarify the effect of OH in the alcohols. Fig. 8 shows the yields of e-caprolactam for various alcohols together with those obtained with water added to the solvent in the amount of H O/Ox=1. Though ethanol and methanol were 2 better solvents as shown in Fig. 2, 1-propanol and 1-butanol showed only comparable yields with those for non-alcoholic solvents. A further increase in carbon number decreased the yield to 3.4 mol.% for 1-hexanol. The reactions were carried out with an excess of these solvents to dissolve cyclohexa-

Fig. 7. Effect of water added to the reactant with ethanol as solvent on the yield of e-caprolactam at 165 min on stream. P(ethanol )=23.1 kPa. Other reaction conditions were the same as those in Fig. 1.

180

T. Komatsu et al. / Microporous and Mesoporous Materials 35–36 (2000) 173–180

Fig. 8. e-Caprolactam yields obtained on silicalite-1 with various 1-alcohol solvents before (solid bar) and after the addition of water (H O/oxime=1, open bar). Other reaction conditions 2 were the same as those in Fig. 1.

none oxime (oxime/solvent molar ratio of 53/1). Therefore, the external surface of silicalite-1 will be almost covered by the solvent molecules under these reaction conditions. In the case of 1-alcohols, these will be adsorbed with their OH groups on the surface, resulting in the formation of an aliphatic layer composed of hydrocarbon chains. When this chain is long, the aliphatic layer may prevent the cyclohexanone oxime from approaching the active sites on the catalyst surface. This would lower the concentration of adsorbed cyclohexanone oxime, [Ox(a)], and accordingly that of adsorbed e-caprolactam, [Cp(a)], which clearly reduces the reaction rate, as can be seen from Eq. (12). In the case of H-BEA zeolite, Dai et al. [6 ] have found that 1-hexanol gave the best selectivity to e-caprolactam among C –C 1 7 1-alcohol solvents under the conditions where conversion of cyclohexanone oxime was close to 100% for each solvent. The different influence of 1-hexanol on the formation of e-caprolactam may be caused by the difference in the pore dimension of the catalysts and reaction conditions. The addition of water was again carried out for each alcohol as shown by open bars in Fig. 8. The improvement in yield by the water addition was most pronounced for methanol, but was insignificant for 1-pentanol. 1-Hexanol did not dissolve

water completely in a 1-hexanol/H O molar ratio 2 of 53/1, indicating its strong hydrophobicity. The influence of water would again be explained by the hydrophobic aliphatic layer formed on the catalyst surface since water molecules must approach the adsorbed e-caprolactam, Cp(a), on the active site to accelerate the desorption of Cp(a) and increase the reaction rate. It is concluded that alcohols with short hydrocarbon chains, such as ethanol and methanol, are the most effective solvents, which expel e-caprolactam from the active sites in order to accelerate the Beckmann rearrangement without forming a hydrophobic layer disadvantageous to the cyclohexanone oxime approaching the active site.

References [1] H. Sato, N. Ishii, K. Hirose, S. Nakamura, in: Y. Murakami, A. Iijima, J.W. Ward ( Eds.), New Development in Zeolite Science and Technology, Elsevier and Kodansha, Amsterdam and Tokyo, 1986, p. 755. [2] A. Thangaraj, S. Sivasanker, P. Ratnasamy, J. Catal. 137 (1992) 252. [3] J. Ro¨seler, G. Heitmenn, W.F. Ho¨lderich, Appl. Catal. A: General 144 (1996) 319. [4] J.S. Reddy, R. Ravishankar, S. Sivasanker, P. Ratnasamy, Catal. Lett. 17 (1993) 139. [5] T. Yashima, K. Miura, T. Komatsu, in: J. Weitkamp, H.G. Karge, H. Pfeifer, W.F. Holderich (Eds.), Zeolites and Related Microporous Materials: State of the Art 1994, Stud. Surf. Sci. Catal. 84, Elsevier, Amsterdam, 1994, p. 1897. [6 ] L.-X. Dai, R. Hayasaka, Y. Iwaki, K.A. Koyano, T. Tatsumi, J. Chem. Soc., Chem. Commun. (1996) 1071. [7] L.-X. Dai, Y. Iwaki, K. Koyama, T. Tatsumi, Appl. Surf. Sci. 121 (1997) 335. [8] T. Yashima, N. Oka, T. Komatsu, Catal. Today 38 (1997) 249. [9] T. Takahashi, M. Nishi, Y. Tagawa, T. Kai, Microporous Mater. 3 (1995) 467. [10] P. Albers, K. Seibold, T. Haas, G. Prescher, W.F. Ho¨lderich, J. Catal. 176 (1998) 561. [11] H. Sato, K. Hirose, Y. Nakamura, Chem. Lett (1993) 1987. [12] H. Sato, Catal. Rev. – Sci. Eng. 39 (1997) 395.