Catalyst deactivation in the Beckmann rearrangement of cyclohexanone oxime over HSZM-5 zeolite and silica-alumina catalysts

Catalyst deactivation in the Beckmann rearrangement of cyclohexanone oxime over HSZM-5 zeolite and silica-alumina catalysts

Applied Catalysis A: General 262 (2004) 137–142 Catalyst deactivation in the Beckmann rearrangement of cyclohexanone oxime over HSZM-5 zeolite and si...

267KB Sizes 0 Downloads 5 Views

Applied Catalysis A: General 262 (2004) 137–142

Catalyst deactivation in the Beckmann rearrangement of cyclohexanone oxime over HSZM-5 zeolite and silica-alumina catalysts Takeshige Takahashi∗ , Takami Kai, Eiko Nakao Department of Applied Chemistry and Chemical Engineering, Faculty of Engineering, Kagoshima University, 1-21-40 Kohrimoto, Kagoshima 890-0065, Japan Received in revised form 19 November 2003; accepted 19 November 2003

Abstract The vapor–solid phase Beckmann rearrangement over HZSM-5 zeolite and silica-alumina was carried out to elucidate the effects of reaction temperature and diluent solvent on the selectivity to ε-caprolactam (CL) and the catalyst deactivation. When the HZSM-5 was regenerated with oxygen atmosphere after the rearrangement using methanol as the solvent, the cyclohexanone oxime (CHO) conversion decreased with the regeneration number, whereas the caprolactam selectivity was almost constant. The catalyst deactivation at the fresh zeolite was largest among the regenerated zeolites. Although the crystalline structure did not change with the regeneration, the acid strength of the zeolite decreased with the regeneration number. These results suggest that the acid sites in the extra-lattice of the zeolite were responsible for the catalyst deactivation. It was found that the yield of the oligomers of ε-caprolactam, such as dimers, trimers and tetramers, changed with the catalyst deactivation constant. The rearrangement was carried out over silica alumina to obtain a significant amount of oligomers. As a result, the catalyst deactivation was related to the selectivity of the tetramer. The tetramer yield significantly changed with the solvent. These results indicate that the oligomer as a coke precursor was easily removed by vapor of a polar solvent, such as alcohol or acetonitrile. © 2004 Elsevier B.V. All rights reserved. Keywords: e (epsilon)-Caprolactam; Acid catalyst; Beckmann rearrangement; Catalyst deactivation

1. Introduction The Beckmann rearrangement of cyclohexanone oxime (CHO) is an important process to synthesize ε-caprolactam (CL) as a starting compound for nylon-6. Since the current CL manufacturing process is carried out using concentrated sulfuric acid as the catalyst, a large amount of ammonium sulfate was produced as a by-product. During the past decade, many investigators carried out the rearrangement over solid catalyst with high CL selectivity and with low catalyst deactivation rate. High-silica ZSM-5 zeolites [1,2], high-silica metallosilicates with ZSM-5 crystalline structure [3], beta-type zeolite [4], FSM-16 [5] and tantalum pillared-ilerite [6] were found to be effective for Beckmann rearrangement. Ichihashi [7] summarized the reaction mechanism for the rearrangement and the role of high-silica ZSM-5 zeolite as an effective catalyst. They concluded that the reaction mechanism of the vapor-phase re∗ Corresponding author. Tel.: +81-99-285-8360, fax: +81-99-257-5895. E-mail address: [email protected] (T. Takahashi).

0926-860X/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.apcata.2003.11.040

arrangement was the same as that of the liquid-phase rearrangement. They confirmed that the silica nests were active sites for the rearrangement proposed by Hoeldrich et al. [8] Recently, Sumitomo Chemical Co. Ltd. constructed a new commercial plant to manufacture the CL using a solid catalyst composed of high-silica ZSM-5 zeolite. The reactor in the new plant is a fluidized bed [7], because the catalyst life is not sufficient to use in a fixed bed type reactor. Although the catalyst deactivation was responsible for the deposition coke on the catalyst surface, the mechanism of coke formation on the catalyst with weak acid sites has not yet been clarified. The authors found that the catalyst deactivation rate was strongly dependent on the diluent solvent [9]. When methanol was used as the solvent for the rearrangement over high-silica HZSM-5 (Si/Al = 500), the selectivity of CL increased up to 96% and the catalyst life defined by the half-life period was 2200 h. Furthermore, it was found that the selectivity of oligomers, such as dimers, trimers and tetramers of CL, increased with the catalyst deactivation rate. The oligomers could be the precursor of the coke deposited on the catalyst surface.

138

T. Takahashi et al. / Applied Catalysis A: General 262 (2004) 137–142

In the present study, the Beckmann rearrangement over HZSM-5 or amorphous silica-alumina catalyst was carried out to clarify the effect of oligomer production on the deactivation rate. Because the selectivity of oligomers over high-silica HZSM-5 zeolite was very low, low-silica HSZM-5 zeolite (Si/Al = 90) and amorphous silica alumina were used for the rearrangement. As a result, the selectivity of CL and the catalyst life were lower than those of high-silica HSZM-5. Furthermore, the rearrangement was repeated over the same catalyst after regeneration with air atmosphere. The coke or coke precursor was burned out by the regeneration.

To examine the effect of the diluent solvent on the catalyst deactivation, acetonitrile, methanol, benzene, 2-propanol and 1-hexanol were used. A reaction temperature was selected between 573 and 673 K. The acidity and acid strength of the catalysts were determined using an ammonia temperature desorption method (ammonia TPD). The procedure was described in a previous paper [11]. The concentration of the strong acid sites was defined as the amount of ammonia retained after the presaturated catalyst was swept with a nitrogen stream at 570 K. The surface area of the catalyst was measured by the nitrogen adsorption method, in which the Langmuir equation (zeolite) or BET equation (silica-alumina catalyst) was used for the calculation.

2. Experimental The HZSM-5 zeolite (Si/Al = 90) was synthesized in our laboratory according to a previous report [10]. Amorphous silica-alumina was supplied by the Japan Catalysis Society as a reference catalyst named JRC-SA-1. After the powder catalyst was compressed into a small pellet, it was crushed and sieved through a 32–40 mesh. The vapor-phase Beckmann rearrangement was carried out at atmospheric pressure in a fixed bed type reactor made of stainless steel (inner diameter = 4 mm). After being loaded into the reactor, the catalyst was heated to desorb water at a reaction temperature for 1 h in carbon dioxide stream. Cyclohexanone oxime (diluted 50 mass% by a diluent solvent) as the starting compound was supplied from a microfeeder at a constant flow rate to the evaporator. After the vaporized mixture was diluted with carbon dioxide (CHO concentration is 20 mol.%), it was fed into the reactor for a prescribed period to determine the deactivation constant. The reactor effluent was collected in chilled toluene at prescribed time intervals. The products were analyzed with a gas chromatograph equipped with a capillary glass column (60 m) and FID detector.

3. Results and discussion Fig. 1 shows the relationship between the cyclohexanone oxime conversion and the ε-caprolactam selectivity and process time over HZSM-5 zeolite (Si/Al = 90) at 673 and 723 K for 60 min of process time. After the rearrangement, the catalyst was regenerated in air atmosphere for 5 h to burn out the coke deposited on the catalyst. The rearrangement–regeneration sequence was repeated 10 times at 723 K and 8 times at 673 K. Although CL selectivity was almost constant for the process times and reaction temperatures, CHO conversion decreased with process time. Since the rearrangement rate of CHO was represented by a first-order equation [11], the mass balance equation for the fixed bed catalytic reactor at a constant reaction temperature can be represented by Eq. (1):    W 1 1 (1) = ln FA0 k(t)CA0 1 − xA where W is the mass of the catalyst (kg), FA0 the molar flow rate of CHO (mol s−1 ), k(t) the rate constant at any pro-

Fig. 1. Relationships among CHO conversion, CL selectivity and repetition number. Catalyst: HZSM-5 (Si/Al = 90).

T. Takahashi et al. / Applied Catalysis A: General 262 (2004) 137–142

139

Fig. 2. Effect of the repetition number on the deactivation constant and the initial rate constant: (a) deactivation constant (left); (b) initial rate constant (right).

cess time t (m3 kg−1 s−1 ), CA0 the inlet CHO concentration at reaction temperature (mol m−3 ), xA the CHO conversion (−) and t is the process time (s). When the Beckmann rearrangement was carried out over an acid catalyst, the rate constant exponentially decreased due to the coke formation. The relationship between the rate constant and process time is represented by Eq. (2): k(t) = k0 exp(−b · t)

(2) (m3 kg−1 s−1 )

and b where k0 is the initial rate constant is the deactivation constant (s−1 ). The initial rate constant and deactivation constant were calculated from the relationship between the CHO conversion and the process time, as shown in Fig. 1. Fig. 2 demonstrates the relationship between the deactivation constant and the initial rate constant and process time. The deactivation constant at 573 K is larger than that at 623 K, as shown in Fig. 2. On the other hand, the initial rate constant at 573 K is smaller than that at 623 K. These relationships were observed over a high-silica HZSM-5 [10]. Since the coke precursor was preferentially adsorbed at lower temperature, the active sites for the rearrangement were covered by the coke precursor and the deactivation constant becomes large at 673 K. The initial rate constant suddenly decreases by the first regeneration, and the constants at both temperatures gradually decrease with the regeneration number. When the acid concentration of the fresh catalyst and first regenerated catalyst was measured by an ammonia-TPD method, the concentration of the first regenerated catalyst was two-thirds that of the fresh one. However, it was found that the crystalline structure of the zeolites did not change with the regeneration. Furthermore, the surface area did not decrease in the first regeneration. These results suggest that the decrease in the initial rate constant was responsible for the disappearance of the acid sites on the extra-lattice of the zeolite. Since the concentration of the active sites gradually decreases with the regeneration,

the deactivation constant also decreased with the regeneration number, as shown in Fig. 2. Fig. 3 shows the gas chromatogram of the product over HZSM-5 (Si/Al = 90). It was found that the yield of high-boiling products was related to the catalyst deactivation The molecular weight of high-boiling products A, B and C measured on a gas chromatograph–mass spectrometer coincided with the dimers, trimers and tetramers of CL. The yield of these high-boiling products increased over a catalyst with rapid deactivation. However, because the yields of these high-boiling products over HZSM-5 (Si/Al = 90) were too low to determine the exactly, a silica-alumina catalyst was used to obtain a large amount of the high-boiling products. When a silica-alumina catalyst was used for the rearrangement at 573 K, the catalyst deactivation pattern was the same as that over the HZSM-5 zeolite. Furthermore, the shape of the gas chromatogram obtained over silica-alumina was almost the same as that in Fig. 3. These results indicate that the catalytic activity and the behavior of the catalyst deactivation of the silica-alumina were similar to those of the HZSM-5 zeolite, despite the fact that the micropore sizes of the silica-alumina and HZSM-5 zeolite were different. Fig. 4 shows the relationship between the CHO conversion, CL selectivity and deactivation constant, and the reaction temperature using methanol as the solvent. The conversion was controlled below 100% to obtain the rate constant and the deactivation constant by the change in catalyst mass, that is, 0.075 g of the catalyst used at 573 K was reduced to 0.030 g at 673 K. The CL selectivity was almost constant over the range of reaction temperature as shown in Fig. 4. The deactivation constant decreased with the reaction temperature up to 648 K, but the constant increased above 648 K. The reason why the deactivation constant decreased with reaction temperature has already been described in this paper. When the reaction temperature exceeded 648 K, the

140

T. Takahashi et al. / Applied Catalysis A: General 262 (2004) 137–142

Fig. 3. Gas chromatogram of product.

coke precursor changed to coke with very low vapor pressure. When 2-propanol was used as the solvent, the lowest deactivation was observed at 673 K. These results indicate that the deactivation rate is decided from the balance of the oligomerization rate and its removal rate. As a result, all investigations of the CHO Beckmann rearrangement over a solid catalyst were carried out from 723 to 748 K [1,3–6,12]. Fig. 5 shows the effect of the diluent solvent on CL selectivity and the deactivation constant over the silica-alumina catalyst at 723 K. The deactivation constant obtained from 1-hexanol and benzene was larger than that from methanol, acetonitrile and 2-propanol. When high-silica HSZM-5 was used for the catalyst, the deactivation constant obtained from methanol was the lowest among the solvents. However, the deactivation constant obtained from 2-propanol was lowest among the solvents over the silica-alumina catalyst. The deactivation constant obtained from 2-propanol over the

silica-alumina catalyst was still 100 times greater than that over high-silica HZSM-5 [10]. The deactivation constants obtained from a polar solvent were smaller than those from a non-polar solvent. These results suggest that the solvent vapors remove the oligomers from the catalyst surface. Since the polar solvent can remove much of the oligomers, the deactivation constant from the polar solvent is smaller than that from the non-polar solvent. Figs. 6–8 show the relationship between the deactivation constant and the average yields of the high-boiling products, A, B and C, obtained over the silica-alumina catalyst at 723 K. The average yield of the high-boiling products was calculated from 10 to 60 min of process time. The deactivation constant obtained from several solvents was plotted on the figures. The relationship between the high-boiling product C and the deactivation constant appeared to be linear. To obtain the quantitative relationship between high-boiling

Fig. 4. Effect of the reaction temperature on CHO conversion, CL selectivity and deactivation constant.

T. Takahashi et al. / Applied Catalysis A: General 262 (2004) 137–142

141

Fig. 5. Effect of the solvent on CL selectivity and deactivation constant.

products and the deactivation constant, the correlation coefficients were obtained from Figs. 6 to 8. The coefficients represented in Figs. 6–8 show that the high-boiling products C better related to the deactivation constant than other high-boiling products. These results indicate that the production of a tetramer is responsible for the catalyst deactivation. Fig. 9 shows the relationship between the deactivation constant and the average yield of the high-boiling product C using 2-propanol as the solvent at 573 to 673 K. A linear relationship between the deactivation constant and the yield of product C is obtained. Many mechanisms for the production of CL from CHO over an acid catalyst in a vapor-solid system have been proposed. Recently, Ichihashi [7] proposed the mechanism for the Beckmann rearrangement over high-silica HSZM-5 zeolite. They concluded that each silanol nest is an active site for the rearrangement, and that the reaction mechanism was almost the same as the rearrangement in the liquid phase catalyzed by sulfuric acid. When the CL produced on the

Fig. 6. Relationship between the deactivation constant and high-boiling A.

Fig. 7. Relationship between the deactivation constant and high-boiling B.

active site was easily desorbed from the sites, CL selectivity increased and the catalyst deactivation constant decreased. On the other hand, when CL was strongly adsorbed on the active site, CL consecutively oligomerized from the dimer

Fig. 8. Relationship between the deactivation constant and high-boiling C.

142

T. Takahashi et al. / Applied Catalysis A: General 262 (2004) 137–142

Fig. 9. Relationship between the deactivation constant and the average yield of C.

to the tetramer on the acid site. The oligomer adsorbed on the acid sites changed to coke by a further reaction. The catalyst deactivation enhanced on the strong acid site was responsible for the strong adsorption of the oligomer or coke precursor. Since a part of the oligomers deposited on the catalyst was removed by the vapor of the diluent solvent, the concentration of oligomers detected by gas chromatography in the vapor phase was related to the concentration of the oligomers deposited on the catalyst, that is, the concentration of the oligomers in the vapor phase increased with the concentration on the catalyst. Since the removal ability from the catalyst surface was different for each solvent, the catalyst deactivation constant was also dependent on the solvent, as shown in Fig. 5. It is considered that the active sites for the Beckmann rearrangement and for coke formation are the same, that is, the silanol nest with acidity is playing an important role in both reactions. As a result, it is concluded that the CL selectivity increase and the catalyst deactivation decrease were dependent on the removal of CL from the active sites, and the polar solvent vapor should accelerate the removal of CL.

4. Conclusions The vapor phase Beckmann rearrangement of cyclohexanone oxime was carried out over HSZM-5 zeolite and

silica-alumina catalyst to clarify the effects of the solvent and reaction temperature on the CL selectivity and the catalyst deactivation rate. When the HZSM-5 was regenerated with oxygen atmosphere after the rearrangement on using methanol as the solvent, CHO conversion decreased with the regeneration number, whereas the caprolactam selectivity was almost constant. The catalyst deactivation at the fresh zeolite was the largest among the regenerated zeolites. The characterization of the HZSM-5 revealed that the acid sites in the extra-lattice of the zeolite were responsible for the catalyst deactivation. Since the yields of the oligomers of CL, such as dimers, trimers and tetramers, changed with the catalyst deactivation constant, the concentration of the oligomers was exactly determined over the silica-alumina catalyst. It was observed that the yield of the tetramer significantly changed with the solvent. These results indicate that the oligomer as a coke precursor was easily removed by polar solvent vapor, such as alcohol or acetonitrile.

Acknowledgements This work was financially supported in part by a Grant-in-Aid for Scientific Research (Scientific Research for Priority Areas: Catalytic Molecular Reaction Engineering; No. 13126218) from the Ministry of Education, Science, Sports and Culture of Japan.

References [1] H. Sato, K. Hirose, N. Ishii, Y. Umeda, US Patent 4,709,024 (1987) to Sumitomo Chemical Co. Ltd. [2] H. Ichihashi, M. Kitamura, Catal. Today 73 (2002) 23. [3] M.N.A. Nasution, T. Takahashi, T. Kai, Stud. Surf. Sci. Catal. 105 (1997) 1189. [4] L.X. Dai, R. Hayasaka, Y. Iwaki, K. Koyama, T. Tatsumi, Chem. Commun. (1996) 1071. [5] D. Shouro, Y. Moriya, T. Nakajima, S. Mishima, Appl. Catal. A Gen. 198 (2000) 275. [6] Y. Ko, M.H. Kim, Y.S. Uh, Chem. Commun. (2000) 829. [7] H. Ichihashi, in: Proceedings of the Fourth Tokyo Conference on Advanced Catalytic Science and Technology, 2003, p. 99. [8] G.P. Heitmann, G. Dahlhoff, W.F. Hoeldrich, J. Catal. 90 (1999) 12. [9] T. Takahashi, T. Kai, Stud. Surf. Sci. Catal. 135 (2001) 27-P-16 in CD Rom. [10] T. Takahashi, M.N.A. Narution, T. Kai, Shokubai 37 (1995) 512. [11] T. Takahashi, K. Ueno, T. Kai, Can. J. Chem. Eng. 69 (1991) 1096. [12] T. Yashima, N. Oka, T. Komatsu, Catal. Today 38 (1997) 249.