Kinetics of n-Butylbenzene Dehydrocyclization over Chromia−Alumina

Markku Laatikainen, Kari Vahteristo*, Susanna Saukkonen, and Matti Lindström. Department of Chemical Technology, Lappeenranta University of Technolog...
0 downloads 0 Views 241KB Size
Ind. Eng. Chem. Res. 1996, 35, 2103-2109

2103

Kinetics of n-Butylbenzene Dehydrocyclization over Chromia-Alumina Markku Laatikainen, Kari Vahteristo,* Susanna Saukkonen, and Matti Lindstro1 m Department of Chemical Technology, Lappeenranta University of Technology, P.O. Box 20, FIN-53851 Lappeenranta, Finland

The kinetics of dehydrogenation, cyclization, and isomerization steps in n-butylbenzene dehydrocyclization over precoked chromia-alumina has been investigated at 683-783 K. Apparent kinetic parameters of a pseudohomogeneous model were calculated from the initialrate and integral-rate data measured with n-butylbenzene and 4-phenyl-1-butene in an isothermal fixed-bed reactor. Dehydrocyclization was shown to proceed by dehydrogenation of n-butylbenzene to n-butenylbenzenes, by their cyclization, and by dehydrogenation of the cyclic products to methylindenes and naphthalene. Direct cyclization of n-butylbenzene was of little importance over the precoked catalyst. According to the initial-rate data, dehydrogenation to n-butenylbenzenes followed an apparent order of 0.5 and the kinetic parameters seemed to be applicable also at higher conversions. The primary dehydrogenation products were 1-phenyl1-butenes, and a surface model based on hydrogen abstraction from the R-carbon has been discussed. The initial cyclization rates measured with 4-phenyl-1-butene were appreciably lower than those observed in integral-rate experiments. The discrepancy was tentatively explained by the higher reactivity of 1-phenyl-1-butenes, which are the main constituents of the n-butenylbenzene fraction at longer contact times. High conversion runs with n-butylbenzene also showed that methylindenes and -indans were partly dealkylated and probably polymerized. The apparent orders ranged from 1 to 1.5 for 1,6-cyclization and from 0.5 to 0.8 for 1,5-cyclization. Skeletal isomerization of n-butylbenzene to sec-butyl- and sec-butenylbenzenes was rapidly subsided during coking, and on the coked catalyst at 783 K the reaction followed an apparent order of -0.5. Introduction Dehydrocyclization of suitable C12-alkylbenzenes constitutes an interesting synthetic route to 2,6-dimethylnaphthalene which is a potential monomer precursor for poly(ethylene-2,6-naphthalenedicarboxylate) (PEN). Several patents have been granted for the preparation of the C12-alkylbenzenes (Taniguchi and Matsuoka, 1975; Abe et al., 1990), and typically 1-(ptolyl)-2-methylbutane or -butene is used as the feed for the dehydrocyclization step (Yamagishi et al., 1991). Because of the complexity of their product mixture, these compounds are not, however, amenable for studies of the reaction mechanism and kinetics. In fact, practically no data are available about the numerous side reactions such as 1,5-cyclization to the substituted indans and indenes. It seems therefore advantageous to study the elementary steps of the dehydrocyclization process, i.e., dehydrogenation, 1,6- and 1,5-cyclizations, isomerization, and fragmentation, with simple alkylbenzenes such as n-butylbenzene. The reactions of n-butylbenzene and n-butenylbenzenes have been studied by several authors using bifunctional dehydrocyclization catalysts such as platinum on alumina and chromia-alumina (Csicsery, 1967, 1968; Pines and Goetschel, 1968) or acidic silicaaluminas (Arnaudov and Dimitrov, 1969). Among other things, these studies have shown the benefits of chromia-alumina as a selective 1,6-cyclization catalyst, although it is much less active than the metal catalysts. n-Butylbenzene has been suggested to dehydrocyclize over chromia-alumina both via cationic mechanisms catalyzed by the acidic sites on alumina and via reac* To whom correspondence should be addressed. Fax: +358 53 6212199. E-mail: [email protected].

S0888-5885(96)00027-9 CCC: $12.00

tions taking place on chromia (Pines and Goetschel, 1966). The former route leads mainly to methylindans and -indenes, while the latter gives the products containing a six-membered ring. Isomerization of the substituted five-membered rings to the six-membered rings seems to be possible only on strongly acidic catalysts (Arnaudov and Dimitrov, 1969). Data obtained with 4-phenyl-1-butene, which is one of the dehydrogenation products of n-butylbenzene, indicate that the side-chain unsaturation enhances cyclization, especially in the acid-catalyzed mechanism (Csicsery, 1968). The formation of alkenes has been considered as an indispensable step also in alkane aromatization (Rozengart and Kazanskii, 1971). In spite of the large number of studies on dehydrocyclization over chromia-alumina, the kinetics of the reactions involved is poorly understood. Therefore, the individual steps of n-butylbenzene dehydrocyclization were investigated by means of initial- and integral-rate measurements at 683-783 K. Because of rapid initial deactivation and low selectivity of the fresh catalyst, only the precoked catalyst was used. Experimental Section Materials. The purity of n-butylbenzene (Fluka Chemie) and 4-phenyl-1-butene (Aldrich Chemical Co.) was 99 wt %. Some experiments were also carried out with a methylindene mixture which was prepared by cyclodehydration of benzylacetone over HZSM-5 at 300 °C (Pagnotta et al., 1986). After distillation the mixture contained 65% of 2-methylindene and 7% of 1-methylindene. Nitrogen (99.9%) and hydrogen (99.98%) were supplied by AGA, Finland, and they were used without further purification. Chromia-alumina catalyst was prepared by impregnating calcined alumina (Alcoa F-200) with an aqueous © 1996 American Chemical Society

2104

Ind. Eng. Chem. Res., Vol. 35, No. 7, 1996

Scheme 1

solution of chromium trioxide (Merck, 99%). After impregnation the catalyst was dried at ambient temperature for 1 day and at 100 °C for 3 days. Finally the catalyst was calcined in air at 600 °C. The chromia content was 15 wt %, and the final surface area of the catalyst was 110 m2/g. Massive chromia having a BET surface area of 20 m2/g was prepared by thermal decomposition of CrO3 at 500 °C. Apparatus. Kinetic data were obtained by means of a standard isothermal fixed-bed reactor. The length of the 12 mm quartz tube was 300 mm, and the catalyst bed was packed in the middle of the tube between layers of silicon carbide. The amount of catalyst was 0.2 g in the initial-rate experiments and 1.8 g in the integralrate experiments. The temperature of the reactor was controlled within (1 K. The liquid reactant was fed into the reactor with a piston pump (Millipore Waters 590), and the gas flows were regulated by means of massflow controllers (Brooks 5850E). The product stream was condensed, and samples were taken from the liquid phase for analysis. Gaseous products were not analyzed. The amount of coke was determined thermogravimetrically. Product Analysis. The liquid samples were analyzed using a HP 5890 Series II Plus gas chromatograph equipped with a phenyl methyl silicone capillary column (HP-5, length 30 m, i.d. 0.32 mm) and a flame ionization detector. The column temperature was raised at a rate of 3 °C/min after 3 min at 100 °C. The detector and injector temperatures were 300 and 280 °C. The split ratio was 59. 2-Methylnaphthalene was used as an internal standard, and response factors were determined by means of pure components. The reaction products analyzed are depicted in Scheme 1. Procedure. Because the chromia-alumina catalyst showed a rapid decline in activity when contacted with the hydrocarbon feed, a stabilization period was found

necessary in order to obtain repeatable results. After regeneration overnight with air and activation for 1 h with H2 at 783 K, the catalyst was kept on-stream at WHSV ) 4.5 h-1 for 150 min before starting the kinetic experiments. During this period the conversion of n-butylbenzene dropped from about 60 to 33%, and the final coke content was 8.5 wt %. The initial-rate data were measured at a constant WHSV which was adjusted to result in conversions of less than 5%. The partial pressure of the hydrocarbon feed was varied in the range of 15-85 kPa, while the total pressure was 101 kPa. After each pressure adjustment the system was stabilized about 15 min before sampling. The stability of the catalyst activity was checked by duplicate runs in the beginning and end of a cycle. After completion of the cycle, the catalyst was regenerated, activated, and stabilized. The integral reactor data were obtained at a constant feed partial pressure, and the contact time was varied by changing the WHSV. The absence of internal mass-transfer limitations was verified by runs with two particle size fractions (mean particle diameters of 0.05 and 1 mm) of the same catalyst. Results Catalyst Deactivation. In order to study the effect of coking on each reaction step, the product composition was monitored during the stabilization period. The results obtained at 783 K with n-butylbenzene and 4-phenyl-1-butene are depicted in Figure 1 as relative activities. The values were calculated from the liquid product compositions, and the activities equal to 1 were assigned to the first sample taken. A marked change in the catalyst properties during the first 20 min on-stream was observed with both fresh and regenerated catalysts. The total conversion decreased monotonously, which seems to rule out the

Ind. Eng. Chem. Res., Vol. 35, No. 7, 1996 2105

Figure 1. Effect of coking on fragmentation (b), skeletal isomerization (9), dehydrogenation (2), double-bond isomerization (O), 1,6-cyclization (1), 1,5-cyclization ([), and hydrogenation (0) of n-butylbenzene (a) and 4-phenyl-1-butene (b) at 783 K over chromia-alumina.

Figure 2. Initial dehydrogenation rates of n-butylbenzene at 683 (1), 723 (2), 753 (9, 0), and 783 K (b). Filled symbols: feed diluted with N2. Open symbols: feed diluted with N2/H2 having a mole ratio of 2.

possibility that the change was due to incomplete reduction of the catalyst. The apparent increase in n-butylbenzene dehydrogenation activity was partly due to elimination of the reverse hydrogenation reaction at the initial stages of coking (see Figure 1b). Moreover, the steep change at about 20 min was related to the increase of the trans-1-phenyl-1-butene content in the n-butenylbenzenes fraction (see Scheme 1 for the structures of the isomers) from about 35 mol % to the apparent equilibrium value of 51 mol %. At the same time, the initial content of 4-phenyl-1-butene was appreciably higher than expected. In the run with 4-phenyl-1-butene, on the other hand, the disappearance of hydrogenation, skeletal isomerization, and fragmentation activities was balanced by isomerization to the other n-butenylbenzenes and by cyclization reactions. In Figure 1b the total conversion remained at about 97% during the first 40 min. In analogy with the n-butylbenzene run, the initial trans1-phenyl-1-butene content was appreciably less than the apparent equilibrium value. Initial-Rate Experiments. Dehydrogenation of nbutylbenzene yielded the five n-butenylbenzenes isomers shown in Scheme 1. Further abstraction of hydrogen can lead to the corresponding phenylbutadienes, but neither cis- nor trans-isomer could be identified in the reaction products. The initial-rate data for dehydrogenation at temperatures 683, 723, 753, and 783 K are given in Figure 2. The initial rate was calculated from the outlet molar flow rates of the five

isomers. The filled symbols refer to experiments where the feed was diluted with nitrogen, while the data shown as open symbols were obtained in a N2/H2 mixture with a molar ratio of 2. The rate constants and orders corresponding to an apparent rate equation (eq 1) were obtained by linear regression, and the values are listed in Table 1. The Arrhenius plot shown in Figure 3 yielded an apparent activation energy of 115 kJ/mol which agrees reasonably well with the values of 101 kJ/mol reported for n-butane dehydrogenation over chromia-alumina (Carra and Forni, 1971) and 113 kJ/ mol for ethane dehydrogenation over chromia-silica (Lugo and Lundsford, 1985).

dn˘ j rj )

dmcat

) kj

∏i (pi/pe)R(j,i)

(1)

Here rj is the production rate of component j, nj is its molar flow rate, and mcat is the mass of the catalyst. The rate constant and the order with respect to reactant i are represented by kj and R(j,i), while pi and pQ stand for the partial pressure of reactant i and the standard pressure (101.3 kPa), respectively. At 783 K the dehydrogenation product contained about 85% of 1-phenyl-1-butenes with a trans/cis ratio of about 4.3. As shown in Figure 4, this initial isomer content markedly exceeds the apparent equilibrium value of 61% observed at longer contact times. Dehydrogenation of n-butylbenzene therefore seems to produce primarily 1-phenyl-1-butenes which subsequently experience double-bond isomerization. The trans/cis ratio was independent of the contact time, implying a rapid interconversion between the geometrical isomers. At 753 K the situation was similar, except that the equilibrium content of 1-phenyl-1-butenes and their trans/cis ratio were slightly higher, namely, 66% and 4.9. These figures agree fairly well with 66% and 4.3 obtained earlier at 753 K (Csicsery, 1968). The increase of both values with decreasing temperature is in accordance with thermodynamic values calculated with the Benson group contribution method (Reid et al., 1987). Experiments with 4-phenyl-1-butene showed that isomerization to the thermodynamically more favorable olefins took place readily. The primary isomerization products were not, however, the most abundant 1-phenyl-1-butenes, but the 1-phenyl-2-butenes were. In the initial-rate experiments at 783 K the ratio of the latter to the former was about 5, while the apparent equilibrium ratio was 0.54. The trans/cis ratio of the 1-phenyl2-butenes was 1.7 and did not depend on the extent of the reaction. Because of a substantial contribution of thermal isomerization, no kinetic analysis was attempted. For any practical purposes, however, doublebond isomerization can be assumed fast enough to maintain the equilibrium distribution. Hydrogenation was the main reaction experienced by 4-phenyl-1-butene over the clean catalyst, but as shown in Figure 1b the activity of the coked catalyst was negligibly low. Only approximate values for the kinetic parameters were therefore obtained, and they are listed in Table 1. The order with respect to hydrogen, 0.1, was estimated from the data measured at feed H2 pressures of 10 and 40 kPa, respectively. Because no data were available at 783 K, this value was used also in order to calculate the hydrogenation rate constant kh,783. The activation energy estimated from the kh values was 35 kJ/mol. This value seems surprisingly small, even

2106

Ind. Eng. Chem. Res., Vol. 35, No. 7, 1996

Table 1. Kinetic Parameters Calculated from the Initial-Rate Data by Means of the Apparent Rate Equation (Equation 1) and the Langmuir-Hinshelwood Rate Equation (Equation 2) 783 K kdh (10-6 mol/s gcat) R(dh,BB) kdh′ (10-6 mol/s gcat) KBB (10-3 kPa-1)

2.2 (0.14) 0.53 (0.05) 9.0 (1.1) 5.8 (1.5)

kh (10-8 mol/s gcat) R(h,4PB) R(h,H2)

1.4 (0.60) 0.80 (0.26)

753 K

723 K

Dehydrogenation of BB 1.2 (0.25) 0.67 (0.05) 0.51 (0.06) 0.42 (0.03) 4.8 (0.4) 2.4 (0.2) 7.0 (1.2) 9.1 (2.0)

Ea (kJ/mol)

0.15 (0.02) 0.38 (0.10)

115

170

105

75

Hydrogenation of 4PB 0.89 (0.08) 1.0 (0.03) 0.1

k15,BB (10-7 mol/s gcat) R(15,BB) k15,4PB (10-7 mol/s gcat) R(15,4PB)

0.76 (0.03) 0.18 (0.08) 10 (6.1) 0.84 (0.10)

0.31 (0.01) 0.05 (0.06) 6.8 (2.2) 0.74 (0.21)

1,5-Cyclization 0.05b -0.3b 1.6 (0.03) 0.40 (0.04)

k16,BB (10-7 mol/s gcat) R(16,BB) k16,4PB (10-7 mol/s gcat) R(16,4PB)

1.7 (0.09) 0.75 (0.12) 9.1 (6.0) 1.5 (0.13)

0.53 (0.01) 0.87 (0.09) 4.9 (0.91) 1.4 (0.22)

1,6-Cyclization 0.24 (0.005) 0.81 (0.04) 1.9 (0.08) 1.2 (0.06)

a

683 K

35b

0.77 (0.04) 0.52 (0.09)

0.42 (0.02) 0.85 (0.12)

A (mol/s gcat)

3 × 10-4 b

145b

450b

130

380

155

2550

140

1850

The values in parentheses indicate the standard errors in the parameter values. b Unreliable values.

Figure 5. Initial rates for 1,5 cyclization (a) and 1,6-cyclization (b) of 4-phenyl-1-butene at 683 (1), 723 (2, 4), 753 (9), and 783 K (b). Filled symbols: pH2 ) 10 kPa. Open symbols: pH2 ) 40 kPa. Figure 3. Arrhenius plots for dehydrogenation of n-butylbenzene (b), 1,5-cyclization of 4-phenyl-1-butene (2), and 1,6-cyclization of 4-phenyl-1-butene (1) and n-butylbenzene ([). The length of the error bars is twice the standard error given in Table 1.

Figure 4. Effect of contact time on the distribution of the dehydrogenation products from n-butylbenzene at 783 K: (b) 4-phenyl-1-butene, (2) cis-1-phenyl-2-butene, (9) trans-1-phenyl2-butene, (1) cis-1-phenyl-1-butene, ([) trans-1-phenyl-1-butene.

though a value of 27 kJ/mol for hydrogenation of n-hexenes over chromia-alumina has been reported (Saffert et al., 1986). Cyclization of n-butylbenzene and 4-phenyl-1-butene yielded several products containing five- and six-

membered rings, and the structural formulas are shown in Scheme 1. The initial 1,5- and 1,6-cyclization rates from 4-phenyl-1-butene are depicted in Figure 5, and the kinetic parameters corresponding to the apparent rate equation (1) are listed in Table 1. The data obtained with n-butylbenzene were analyzed in a similar way. The cyclization rates were, however, an order of magnitude smaller, and the rapid cyclization of the phenylbutenes had to be taken into account even at the initial-rate conditions. The parameters estimated from the corrected data are given in Table 1. The Arrhenius plots shown in Figure 3 gave rather similar activation energies in all cases. The values ranging from 130 to 150 kJ/mol are, however, substantially smaller than the value of 214 kJ/mol reported for aromatization of n-hexenes (Saffert et al., 1986). The 1,5-cyclization of 4-phenyl-1-butene yielded mainly 1-methylindene, and only traces of the thermodynamically more stable 2-methylindene were detected. The equilibrium isomer ratio ranges from 0.11 at 683 K to 0.15 at 783 K (Daubert and Danner, 1989). The methylindan content in the 1,5-cyclization products was 5% at 783 K and about 30% at 683 K. These values are higher than expected on the basis of thermodynamic data of indan and indene (Daubert and Danner, 1989). Even at 683 K the equilibrium mixture contains only 6% of indan and at 783 K none. When n-butylbenzene was used, the methylindene isomers were produced in a ratio ranging from about 4 at 783 K to nearly 1 at

Ind. Eng. Chem. Res., Vol. 35, No. 7, 1996 2107

Figure 6. Formation of tetralin (b), 1,2-dihydronaphthalene (9), and naphthalene (2) from n-butylbenzene at 783 K. pBB0 ) 50 kPa. The partial pressures of tetralin and dihydronaphthalene have been multiplied by 50 and 5, respectively.

683 K. The methylindan analysis in the n-butylbenzene runs was interferred with by overlapping impurity peaks, and no accurate results were obtained. 1,6-Cyclization seemed to proceed by a consecutive mechanism via tetralin and dihydronaphthalene to naphthalene. Thermodynamic calculations showed that in the temperature range studied naphthalene is the only product at equilibrium. At low conversions nbutylbenzene yielded approximately equal amounts of tetralin and dihydronaphthalene, and only at the highest temperature small amounts of naphthalene were observed. With longer contact times, however, naphthalene became the major product as shown in Figure 6. When starting from 4-phenyl-1-butene, on the other hand, only about 10% of the 1,6-cyclization products in the initial-rate experiments was tetralin, while the dihydronaphthalene content was as high as 65%. At higher conversions naphthalene became the major component in analogy with the n-butylbenzene runs. Integral-Rate Experiments. The experimental data obtained with n-butylbenzene at 783 K and with different contact times are depicted in Figure 7. The liquid product compositions were converted to partial pressures by taking into account stoichiometric amounts of gaseous products. Besides the reactions considered earlier, the data of skeletal isomerization and fragmentation were also taken into account. Skeletal isomerization yielding sec-butylbenzene and sec-butenylbenzenes (see Scheme 1) proceeded readily over the clean catalyst, but after the rapid deactivation the rate became too low to be measured at the initial-rate conditions. Fragmentation reactions, on the other hand, were predominantly noncatalytic, and no detailed kinetic analysis was attempted. According to the initialrate data, 4-phenyl-1-butene was fragmented about 7 times faster than n-butylbenzene and the reaction followed an apparent order of 1.5. Figure 7 also shows the predictions of a pseudohomogeneous model based on Scheme 1. Each reaction was described by the apparent rate equation (eq 1), and the rate constants were estimated by means of a leastsquares minimization method combined with the fifthorder Runge-Kutta integration routine (Press et al., 1986). In order to simplify the procedure, the reaction orders calculated from the initial-rate data (Table 1) were used. The best-fit value for the dehydrogenation rate constant, 2.6 × 10-6 mol/s gcat, agreed reasonably well with

Figure 7. Comparison of experimental and calculated outlet pressures of n-butylbenzene (b), n-butenylbenzenes (9), fragmentation products (2), skeletal isomerization products (0), 1,5cyclization products (O), and 1,6-cyclization products (1) in n-butylbenzene dehydrocyclization at 783 K over chromiaalumina. pBB0 ) 50 kPa. The solid lines were calculated according to Scheme 1 by using eq 1 and the parameter values given in the text.

the value given in Table 1, but the hydrogenation reaction seemed to be far more important than suggested by the initial-rate data. The estimate for kh was 4.1 × 10-6 mol/s gcat, which is about 300 times higher than the value in Table 1. The initial-rate data implied that the main 1,6cyclization route goes via n-butenylbenzenes, and therefore only the rate constant k16,PB was estimated, while the rate constant of direct cyclization, k16,BB, was taken from Table 1. The value obtained, 12 × 10-6 mol/s gcat, was much larger than the value derived from the initialrate data, and the discrepancy may be due to different reactivities of the olefin isomers. In the initial-rate experiments 4-phenyl-1-butene was the only isomer available, while at higher conversions the main isomer was trans-1-phenyl-1-butene which was shown to be closely related with the increase in the cyclization activity during coking (see Figure 1). The hypothesis was tested by data from a 4-phenyl-1-butene run carried out at 783 K and 1/WHSV ) 0.26 h. At these conditions the n-butenylbenzene fraction contained 55% of 1-phenyl-1-butenes. The outlet pressure of the 1,6-cyclization products was 5.7 kPa which corresponds to a cyclization rate constant of 10 × 10-6 mol/s gcat. Although based only on one experimental point, this value seems to support the idea of the abnormally low reactivity of the 4-phenyl-1-butene isomer. The experimental data shown in Figure 7 implied that part of the 1,5-cyclization products react further, and a consecutive step for the methylindans and -indenes was required in the kinetic model to obtain any acceptable fit in the data. The curve shown in Figure 8 was calculated with a cyclization rate constant k15,PB ) 1.1 × 10-6 mol/s gcat and by assuming a first-order consecutive step with a rate constant of 16 × 10-6 mol/s gcat. The direct cyclization rate was calculated with the kinetic parameters given in Table 1. The nature of the consecutive reaction was studied in experiments with the methylindene mixture as the feed. Substantial amounts of indene and indan were initially produced over the precoked catalyst at 783 K, but the dealkylation activity vanished over a few minutes. These runs also showed that no expansion of the five-membered ring to

2108

Ind. Eng. Chem. Res., Vol. 35, No. 7, 1996

Figure 8. Initial dehydrogenation rates of n-butylbenzene at 723 (2), 753 (9, 0), and 783 K (b) and the best-fit plots of eq 2. The solid lines were calculated with the parameter values given in Table 1 and with KH2 ) 0.014 kPa-1. Filled symbols: feed diluted with N2. Open symbols: feed diluted with N2/H2 having a mole ratio of 2.

a six-membered ring takes place under these conditions. Moreover, practically no sec-butylbenzene was observed, indicating that the consecutive step cannot be explained by ring opening. Skeletal isomerization produced at short contact times mainly sec-butylbenzene, which subsequently dehydrogenated to the two sec-butenylbenzenes shown in Scheme 1. No isobutylbenzene was observed. In analogy with the n-butenylbenzenes, the trans/cis ratio equal to 1.6 was independent of the contact time. The best-fit values for the isomerization rate constant and order with respect to n-butylbenzene were 1.6 × 10-8 mol/s gcat and -0.5, respectively. In order to elucidate the skeletal isomerization mechanism, unimpregnated alumina and massive chromia were also tested at 783 K with nbutylbenzene as the feed. No sec-butylbenzene was detected over alumina, while small amounts were produced over chromia during the first minutes of the experiment. However, even on the uncoked chromia the ratio of sec-butylbenzene to the 1,6-cyclization products was only about 0.02, while on uncoked chromia-alumina the two reactions were of comparable importance. Discussion The catalytic activity of supported or unsupported chromias is generally ascribed to the coordinatively unsaturated metal and oxide ions formed during the reduction treatment and to the acidic sites of the support (Poole and MacIver, 1967; Burwell et al., 1969). The activity and selectivity of the chromia catalysts are, however, markedly altered by coke deposition. The reactions considered here seem to fall in two categories of distinctly different sensitivity to coking: skeletal isomerization, hydrogenation, and fragmentation deactivated much more readily than dehydrogenation, doublebond isomerization, and cyclization reactions. This is also of practical importance, because all rapidly poisoned reactions are undesirable in the cyclization process. The changes of the product composition during coking and the kinetic data indicated that cyclization of nbutylbenzene proceeds predominantly via the olefinic intermediates. As implied by the variations of the cyclization activity during coking, a direct cyclization is also possible. This mechanism is probably operative only on the most energetic sites available on the clean

catalyst, while on the coked surface it is subsided by the less demanding sequential mechanism. The discrepancy between the initial-rate and integral-rate data also suggested that the cyclization rate strongly depends on the structure of the olefin intermediate. Although no direct evidence was obtained in the present study, the role of trans-1-phenyl-1-butene isomer appears especially important. 1-Phenyl-1-butenes were shown to be the primary products of the dehydrogenation step. This result can be explained by a surface process, where the hydrogen abstraction from n-butylbenzene starts from the most acidic R-hydrogen by means of the basic O2- ions. A similar heterolytic dissociative mechanism for hydrocarbon adsorption has been suggested also by earlier investigators (Burwell et al., 1969; Carra and Forni, 1971). In these models the active site is considered as a O2--Cr3+ pair which can be occupied by one hydrocarbon or hydrogen molecule. Assuming that the hydrogen abstraction from the adsorbed n-butylbenzene is the rate-determining step, a Langmuir-Hinshelwood type rate equation given in eq 2 is obtained for the initial dehydrogenation rate.

rdh0 )

kdh′KBBpBB (1 + KBBpBB + KH2pH2)N

(2)

Ki (i ) BB, H2) represents here the adsorption equilibrium constant, and N is the number of sites required by the reaction. A two-site model has been applied to dehydrogenation of C4-hydrocarbons over chromiaalumina (Carra and Forni, 1971). Figure 8 shows the experimental data along with the best-fit curves of eq 2 with N ) 2. The values measured at 753 K in the presence of hydrogen were used to evaluate KH2, and a value of 0.014 kPa-1 was obtained. The other parameters of eq 2 are listed in Table 1. The temperature dependence of kdh′ yielded an activation energy of 105 kJ/mol, in fair agreement with 115 kJ/mol obtained earlier for kdh. Although the simple model seems to work satisfactorily with the dehydrogenation data, extension of the surface-site formalism to the complete reaction system was considered unjustified because of the complexity of the coked catalyst surface. The high reactivity of 1-phenyl-1-butenes in coking and cyclization is probably connected with their dominant role in the dehydrogenation process. The deficiency of trans-1-phenyl-1-butene in the reaction product at initial stages of coking implies that this isomer was largely consumed in the coke formation. It is therefore plausible to think that initially, when a large number of energetic chromia sites are available, dehydrogenation is instantly followed by coking via, e.g., polymerization of the olefins. If this reaction proceeds faster than double-bond isomerization, a deviation from the equilibrium distribution of n-butenylbenzenes becomes possible. Large amounts of polymeric byproducts have been observed during cyclization of 1-phenyl-1-butenes but not of the other isomers over silica-alumina (Arnaudov and Dimitrov, 1969). Over the coked catalyst, on the other hand, the consecutive step becomes cyclization. The relatively large amount of cyclization products observed at the very beginning of the run shown in Figure 1a is probably due to the direct cyclization mechanism from n-butylbenzene to naphthalene. In contrast to the compounds containing six-membered rings, the 1,5-cyclization products seemed not to

Ind. Eng. Chem. Res., Vol. 35, No. 7, 1996 2109

be stable at the conditions studied. The existence of reactions which consume methylindenes and -indans was even more evident in preliminary runs with a commercial chromia-alumina catalyst. The consecutive reaction is highly desirable from a practical point of view because of the problems in separation of the 1,5- and 1,6-cyclization products. Although the final reaction products could not be identified, the only plausible mechanism appears to be polymerization preceded possibly by dealkylation. The dealkylation products indene and indan were regularly detected in the reaction products of the high conversion runs, but their content was always less than 10% of the 1,5-cyclization products. Skeletal isomerization to sec-butylbenzene is considered to proceed on strongly acidic sites of chromiaaluminas, and it offers one possible route to 1,5cyclization (Pines and Goetschel, 1966). The acidity, on the other hand, is conventionally ascribed to the alumina support. It is therefore surprising that no isomerization products were observed over the unimpregnated alumina. Moreover, comparison between chromiaalumina and massive chromia indicated that the skeletal isomerization activity of the unsupported chromia is insignificantly low. The acidic sites must therefore stem from structures formed by the interaction of chromia and alumina, or, alternatively, the reactions do not proceed via the cationic intermediates. In fact, the absence of isobutylbenzene in the reaction products has been considered indicative of a free-radical isomerization mechanism (Pines and Goetschel, 1966). The negative order of the isomerization reaction implies a strong contribution by the adsorption effects. In view of eq 2 the value of N should be appreciably larger than 1, suggesting a mechanism where cooperation of several sites is needed. Acknowledgment TEKES (Technology Developement Centre, Finland) is gratefully acknowledged for financial support. Nomenclature A ) frequency factor (mol/s gcat) Ea ) activation energy (kJ/mol) k ) apparent rate constant (mol/s gcat) k′ ) rate constant in the Langmuir-Hinshelwood equation (mol/s gcat) mcat ) mass of the catalyst (g) n ) molar flow rate (mol/s) N ) number of surface sites p ) pressure (kPa) r ) reaction rate (mol/s gcat) WHSV ) weight hourly space velocity (h-1) Greek Letters R(j,i) ) apparent order with respect to reactant i in reaction j Subscripts BB ) n-butylbenzene

PB ) n-butenylbenzenes 4PB ) 4-phenyl-1-butene HC ) hydrocarbon H2 ) hydrogen dh ) dehydrogenation h ) hydrogenation 15 ) 1,5-cyclization 16 ) 1,6-cyclization Superscripts 0 ) feed or initial condition Q ) standard state (101.3 kPa)

Literature Cited Abe, T.; Uchiyama, S.; Ojima, T.; Kida, K. Process for Production of 2,6-Dimethylnaphthalene. Eur. Pat. Appl. EP 362,507, 1990. Arnaudov, S.; Dimitrov, C. Catalytic Conversions of 1-Phenylbutene-3, 1-Phenylbutene-2 and 1-Phenylbutene-1 on SilicaAlumina Catalyst (in Bulgarian). Annu. Univ. Sofia, Fac. Chim. 1969/1970, 64, 343-352. Burwell, R. L.; Haller, G. L.; Taylor, K. C.; Read, J. F. Chemisorptive and Catalytic Behavior of Chromia. Adv. Catal. 1969, 20, 1-96. Carra, S.; Forni, L. Catalytic Dehydrogenation of C4 Hydrocarbons over Chromia-alumina. Catal. Rev. 1971, 5 (1), 159-198. Csicsery, S. Reactions of n-Butylbenzene over Supported Platinum Catalysts. J. Catal. 1967, 9, 336-357. Csicsery, S. Catalytic Reactions of Phenylbutenes. J. Catal. 1968, 12, 183-190. Daubert, T. E.; Danner, R. P. Physical and Thermodynamic Properties of Pure Chemicals; Hemisphere Publ. Co.: New York, 1989. Lugo, H. J.; Lundsford, J. H. The Dehydrogenation of Ethane over Chromia Catalysts, J. Catal. 1985, 91, 155-166. Pagnotta, M.; Cesa, M. C.; Burrington, J. D. Preparation of Indenes. U.S. Patent 4,568,782, 1986. Pines, H.; Goetschel, C. Alumina: Catalyst and Support XXX. Dehydrogenation and Skeletal Isomerization of Butylbenzenes over Chromia-Alumina Catalysts. J. Catal. 1966, 6, 371-379. Poole, C. P., Jr.; MacIver, D. S. The Physical-Chemical Properties of Chromia-Alumina Catalysts. Adv. Catal. 1967, 17, 223-314. Press, W. H.; Flannery, B. P.; Teukolsky, S. A.; Vetterling, V. T. Numerical Recipes; The Art of Scientific Computing; Cambridge University Press: Cambridge, U.K., 1986. Reid, R. C.; Prausnitz, J. M.; Poling, B. E. The Properties of Gases and Liquids, 4th ed.; McGraw-Hill: New York, 1987. Rozengart, M. I.; Kazanskii, B. A. Catalytic C6-Dehydrocyclisation of Aliphatic Hydrocarbons on Oxide Catalysts. Russ. Chem. Rev. 1971, 40 (9), 715-732. Saffert, W.; Gru¨nert, W.; Schellang, W.; Klaus, A.; Nowak, S. Kinetic Study of n-Hexane Dehydrocyclization over Chromium Oxide-Aluminum Oxide Catalyst. Part II. Chem. Tech. 1986, 38, 472-475. Taniguchi, M.; Matsuoka, H. 2-Methyl-1-tolylbutane (in Japanese). Jpn. Kokai Tokkyo Koho JP 50,093,925, 1975. Yamagishi, K.; Yoshihara, J.; Inamasa, K.; Watabe, K. Process for Producing 2,6-Dimethylnaphthalene. Eur. Pat. Appl. EP 430,714, 1991.

Received for review January 17, 1996 Accepted April 4, 1996X IE960027K

X Abstract published in Advance ACS Abstracts, June 1, 1996.