Hydroisomerization of n-C5 and n-C6 mixtures on zeolite catalysts

Jul 14, 1981 - Shlnnar, R.; Shapira, D.; Zakal, S. Ind. Eng. Chem. Process Des. Dev. 1981, 20, 581. Shlnnar, R.; Feng, C. "Thermodynamic Constraints o...
0 downloads 0 Views 1MB Size
Ind. Eng. Chem. Process Des. Dev. 1982, 21, 750-760

750

Krikosian. 0. M. “Thermodynamics Propetties of Wyoming Coals”; Technical Note 72-18, UCID-16587, Livermore, CA, Lawrence Livermore Laboratory, 1972. May, W. G.; Mueiler, R. H.; Sweetser, S. B. Ind. Eng. Chem. 1058, 5 0 , 1289. Rostrup-Nlelsen, J. R. J . Catal. 1972, 27, 343. Othmer, H. G. (2”.Eng. Scl. 1078, 37,993. Shlnnar, R. CHEMTECH 1978. 8, S86. Shinnar, R.; Kuo, J. C. W. “Gasifier Study for Mobil Coal to Gasoline Processes”; FE-2766-13, 1978. Shinnar, R.; Shapka, D.; Zakai, S. Ind. Eng. Chem. Process Des. D e v . 1081, 20, 581. Shinnar, R.; Feng, C. “Thermodynamic Constraints of Catalytic Processes”; 1981, to be published.

Squires, A. M. Trans. Inst. Chem. Eng. 1061, 39,3. Stom, and Webster Engr. Co. “Comparative Evaluation of I#&I and Low T e n peratwe Gas Cleaning for Coal Gasiflcation-Combined Cycle Power Systems”; EPRI-AF-416, 1977. Wei, J.; Prater, C. D. A&. &tal. 1082, 13. 204. Wen, C. Y.; Huebler. J. I d . Eng. Chem. Process Des. D e v . 1985, 4 , 147. Wen, C. Y.; Tone, S. ACS Symp. Ssr. 1078, 72, 56. Yoon, H.;Wei. J.; Denn. M. M. “Modelling and Analysis of Moving Bed Coal Gasifiers”; Final Report, EPRI AF-590, 1977. Zaharadnik, R. L.; Grace, R. J. A&. Chem. Ser. 1974, No. 131, 127.

Received for review July 14, 1981 Accepted May 11, 1982

Hydroisomerization of n -C5 and n -C6 Mixtures on Zeolite Catalysts James J. Spivey’ Research Triangle Institute, Research Triangle Park, North Carolha 27709

Phllllp A. Bryant’ Chemical Engineering Department, Louisiana State University, Baton Rouge, Louisiana 70803

Hydroisomerization of n C , and n-C, mixtures was carried out on a Pt-H-mordenlte and a Pd-H-faujasite catalyst. A different type of anomalous behavior was observed on each catalyst in the conversion of mixtures relative to the conversion of pure components. Statistical analysis of various Langmuir-Hinshelwood type models for the hydroisomerization rate constant show that this behavior can be related to the adsorption parameters of each system.

Introduction Hydroisomerization of low molecular weight paraffins has become an increasingly important conversion process in recent years. This is due to the phase-down of antiknock concentrations in motor gasoline and the resulting need for higher octane, Le., more branched hydrocarbon constituents. The research octane number of streams containing C5 and (& mostly straight-chain hydrocarbons can be boosted from about 70 to over 80 with the extent of improvement depending on the isomerization temperature and C 5 / C 6ratio (Kouwenhoven and Langhout, 1971). Zeolite catalysts have been employed almost exclusively in new efforts in this area. Paraffin isomerization requires a stable catalyst with high activity to take advantage of the higher equilibrium conversions to branched products at lower temperatures (Chick et al., 1977). Zeolites have been shown to be highly active for this type of reaction and have additional advantages such as relative insensitivity to moisture, sulfur, and nitrogen (Minachev et al., 1972). Little has been reported in the literature on systematic studies of mixed n-Cs and n-C6 feeds although it is this type of feed that is typical of hydroisomerization feed stocks for octane enhancement. Any anomalies observed in hydroisomerization of such mixed feeds are important in the design of commercial systems and in the investigation of the mechanism and dynamics on a molecular level. Exxon Research and Development Labs, Baton Rouge, LA 70821. 0196-4305/82/1121-0750$01.25/0

This study examines the kinetics of n-C5/n-C,hydroisomerization on a mordenite and a faujasite typical of developmental catalysts. The objectives are to describe the effect of reactant partial pressures on a simplified rate constant, explain any anomalies in terms of fundamental kinetic quantities, and compare the performance of the two catalysts in terms of kinetic parameters. Zeolite Catalysts Zeolites are a special class of crystalline aluminosilicate compounds which have specific pore dimensions and pore structure. These pores may be tailored to exclude certain reactants and thus obtain high selectivity. In addition, the special acidic and adsorptive properties of the zeolite surface promote specific conversion reactions such as hydroisomerization. By introducing both noble metal and acid sites into a zeolite using special techniques, a dual function catalyst is formed. Such a catalyst is active for isomerization. It is thought that hydroisomerization of a normal paraffin on such a catalyst proceeds through a mechanism whereby olefiis are formed at the metallic site by dehydrogenation of the normal paraffin feed and are then adsorbed at an acidic site or the catalyst surface. A t this acidic site, a carbonium ion is formed which undergoes skeletal rearrangement. The reeulting isocarbonium ion is converted to an isoolefin which is then hydrogenated at a metallic site and desorbed (Kouwenhoven, 1973). Mordenite. Mordenite is unique among zeolites in that the structure consists of parallel elliptical channels which do not intersect. Mordenite is the most active catalyst 0 1982 American Chemical Society

Ind. Eng. Chem. Process Des. Dev., Vol. 21. No. 4. 1982

751

k could be expressed as

strong acid sites. Garanin et al. (1972) found that the hydroisomerization of n-C5over a 0.5 wt % Pd-Ca-Y was consistent with the generally accepted dual-function mechanism. The authors state that their results are consistent with the rate law r = kPc,/PH,.However, Bolton and Lanewala (1970) question the existence of an olefinic intermediate in the isomerization of n-C6 over a O.f, wt % Pd-Y catalyst. They propose a cyclic intermediate which may be cracked into various hexenes which are then hydrogenated at a metallic site to complete the reaction. Kouwenhoven (1973) says that this mechanism does not give a more straightforward explanation than the classical dual-function mechanism. Saito and Iwasaki (1976) studied the isomerization of n-C5on a Pt-H-Y of varying metal content. Below 0.3 wt % Pt, hydrogenation-dehydrogenation was rate-limiting while above 0.3 wt 9% Pt isomerization was rate-limiting. Comparison of Mordenite and Faujasite. Several studies comparing the activity and properties of mordenite and faujasite catalysts have been carried out. Gray and Cobb (1975) state that the active sites of mordenite are similar to their faujasite counterparts. Despite similarities, each catalyst has its own unique properties. For example, steric effects are more pronounced on mordenite since it has a slightly smaller pore size than type Y faujasite and since its structure does not contain the large macropores of faujasites (Yashima and Hara, 1972). Addition of a noble metal does not increase n-C5 isomerization activity on H-M as it does on H-Y (Voorhies and Bryant, 1968). Braun et al. (1977) examined the influence of various noble metals incorporated into H-M and Ca-Y on the activity and selectivity in n-C5and n-C6 hydroisomerization. They found that at a 0.5 w t % loading, Pt-Ca-Y was less active than Pd-Ca-Y but that Pt-H-M was more active than Pd-H-M. The selectivity of the mordenites was superior to that of the faujasites as measured by the fraction of isomers in the product. The selectivity of the two Pd zeolites was slightly higher than that of the two Pt zeolites. They also show that the method of incorporating the noble metal into the crystalline structure as well as the nature of the metallic particles on the zeolite surface can significantly affect the catalyst activity.

The value of the hydrocarbon adsorption constant, KHC, was much greater than the hydrogen adsorption constant, KH #aujasite. The structure of faujasite differs from that of mordenite. Faujasite is composed of channels in the form of truncated tetrahedra with windows into larger macropores. Faujasite catalysts that are active for isomerization reactions have been made by incorporating hydrogen and noble metals into the crystalline structure. Reactions of single pure paraffins on dual-function faujasites have been studied extensively, though little has been reported on the hydroisomerization of mixtures. Minachev et al. (1972) found that Pt-Ca-Y exhibited an increasing isomerization activity with increasing metal content up to 0.5 wt 9%Pt. At higher Pt contents, isomerization activity did not change, indicating that skeletal rearrangement was rate-limiting. Topchieva and Dorogochinskaya (1973) found that increasing reaction temperature led to cracking of a pure n-C6 feed. It was also found that at a fixed hydrogen to hydrocarbon mole ratio (H,/HC), decreasing total pressure resulted in increased cracking. Yegiazarov et al. (1975) found that highly acidic Pd-H-Y retarded n-C5 hydroisomerization apparently due to retarded desorption of the olefinic intermediate from

Experimental Section The reactions were carried out in a tubular Inconel reactor with an i.d. of 1.6 cm using approximately 10 cm3of a mixture including both inert mullite and catalyst (Figure 1). Catalyst particles were contained in a fixed bed supported on each end by plugs of glass wool and microporous metal frits. The temperature was regulated using a fluidized sand bath into which the reactor was submerged. The reactants flowed into the reactor through a coil submerged in the sand bath in order to bring them to the proper temperature. Both the sand bath and the reactor temperatures were recorded. The pressure of the reactor was maintained using a back-pressure regulator and monitored using a pressure gauge. Figure 2 shows the arrangement of the reactor and support system. Reagent grade hydrocarbons were fed to the reactor using Ruska metering pumps. High-purity electrolytic hydrogen was fed through a flow control valve. The two reagents were mixed and preheated before entering the reactor. The product stream was sampled for chromatographic analysis immediately downstream of the back-pressure regulator. The product stream then passed through a water saturator to a wet test meter where the flow rate was measured. Chromatography was carried out in a Perkin-Elmer Model 990 GC using a 3 m X 0.64 cm column with 20% Ucon HB-280X on 60/80 Chromasorb

developed so far for isomerization of saturated hydrocarbons. Beecher et al. (1968) found that acid leaching produces an alumina-deficient structure which decreases diffusional resistance normally associated with unmodified mordenite. Acid leaching also raises the Si02/A1203ratio and thus decreases catalyst acidity. This implies an optimum Si02/A1203ratio for isomerization. Hopper and Voorhies (1972) found that a ratio of 18/1 resulted in maximum n-C5 isomerization activity for one modified Pd-H-M. Weller and Brauer (1969) observed a maximum in n-C6 cracking activity as a function of Si02/A1203ratio at a value of about 17.5/1. Eberly et al. (1971) correlated catalyst acidity with n-C5 hydroisomerization activity for a series of alumina-deficient mordenites. They found that activity increased with acidity but the relationship was not linear. They also state that for n-C5,decreased diffusional resistance associated with the removal of alumina does not affect the catalytic activity. Satterfield and Chiu (1974) also note decreased diffusional resistance with the removal of alumina. They state, however, that considerable differences in pore size distribution and other properties may exist in alumina-deficient mordenites even with the same Si02/A1203ratio. These differences can cause varying experimental results from catalyst to catalyst in reactions like hydroisomerization. Minachev et al. (1972) postulate that n-C5isomerization on H-mordenite occurs via a three-step hydride abstraction mechanism. Their rate expression shows that the dependence of the rate on P,,.c,/PH takes the form of a Langmuir isotherm. They also founh that on their 0.5 wt % Pd-H-M, the isomerization of the carbonium ion formed by hydride abstraction was the rate-controlling step. Hopper and Voorhies (1972) were able to correlate data from the hydroisomerization of cyclohexane over a series of 0.6 wt % Pd-H-M catalysts of varying Si02/A1203ratios by using a dual site Langmuir-Hinshelwood model. They found that for the simplified reaction cyclohexane

k

methylcyclopentane

752

Ind. Eng. Chem. Process Des. Dev., Vol. 21, No. 4, 1982 ,

REACT ANTS^

t

PRODUCTS

HEATED SAND EA1 VESSEL

INLETFOR FLUIDIZING AIR

-

2

f POROUS FRIT

Figure 1. Reactor and sand bath.

W. Integration of the chromatogram was done using a Varian CDS 111. The catalysts in this study were supplied by Exxon Research and Development Labs. A 0.5 wt 9% platinum H-mordenite (Pt-H-M) with a Si02/A1203ratio of 14/1 and a 0.5 wt 9% palladium H-faujasite type Y (Pd-H-Y) with a Si02/A1203ratio of 6.411, both of 15/35 mesh, were used. Additional properties of the catalyst were not provided. Kinetic Model Formulation. Calculations and comparison with previous studies were made to determine applicable simplifying assumptions leading to the formation of the kinetic model. Pure component studies carried out by Bryant (1966), Hatcher (1968),and Luzarraga (1971) at conditions closely approximating those of this study revealed insignificant gas-to-particle mass transfer resistance. The model formulated herein assumed this to be the case. Reactant fugacities were calculated using the RedlichKwong equation of state a t 514 K, 2.17 MPa, and 10/1 H,/hydrocarbon (H2/HC)ratio, conditions typical of this study. A maximum deviation from ideality of less than 1% was found for the three reactants. Thus the mixture was assumed to be ideal. Calculation of the intrapore mass transfer resistance revealed that significant configurational diffusion resistance, which occurs when the pore diameter approaches the diameter of the diffusing molecule, could be expected in both catalysts (Spry and Sawyer, 1975). In order to study this effect, catalysts of varying pore sizes would be needed. Such catalysts have not been available and any configurational diffusion effects are included in the calculated rate constants. Studies on similar catalysts carried out by Bryant (1966) and Hopper (1969) examined the effect of catalyst particle size on the observed rate constant calcu-

lated in this manner. No variation was observed over a tenfold change in particle size, indicating constant intraparticle diffusion resistance which need not be included in examining the hydroisomerization of mixtures carried out at constant total pressure, temperature, and H2/HC. Hopper and Voorhies (1972) found that intraparticle diffusion was not rate-limiting when a simplified rate constant for cyclohexane hydroisomerization on 0.5 wt % Pd-H-M was used. Plug flow conditions within the reactor were verified using the correlation of Levenspiel and Bischoff (1963) relating the catalyst particle Reynolds number and a "mixing intensity". A thermocouple measuring the temperature at the end of the catalyst bed was monitored routinely during each run. A maximum temperature difference from the sand bath of f1.5 K was observed with a typical temperature difference of f0.5 K indicating that the catalyst bed was isothermal. It can also be shown that the heat of reaction for the isomerization of n-C5and n-C6 results in insignificant temperature effects. Model. The general Langmuir-Hinshelwood model describing reactions of gases on solid catalysts is given by [kinetic term][potential term] rate = (1) [adsorption term] The kinetic term includes the true surface reaction rate constant as well as constants characterizing the surface of the catalyst and in some cases adsorption constants. The potential term for the first-order elementary reaction A e B has the form PA - PB/K. The adsorption term has the form (1 KAPA KflB)", where n is the number of active catalyst sites involved in the reaction, usually 1 or 2. The exact form of the adsorption term may change depending on the rate-limiting step for a given reaction. The material balance equation for an integral reactor in plug flow is given by r = dhTB/dw = k ( c -~CB/K) (2) The rate constant, k, is a lumped term including both the adsorption term and the effectiveness factor. With pure A as feed, and fugacity coefficients of unity, the rate expression may be integrated and rearranged in order to calculate the rate constant from experimental data. The result is

+

+

(3) Assuming that Langmuir adsorption isotherms may be applied to the simple first-order reversible model for hydroisomerization, the first-order rate constant may be expressed as K = R/[1 CKiPi]" (4)

+

where

k

= ko XnKi i

(5)

Kinetic properties of interest may be determined by analyzing the experimental results using these relationships. For example, e q 3 may be rearranged, noting that tH = WpG/p$vRo(l + R ) to yield

A semilog plot of (1- YB/Ye*)vs. 117, should be a straight line with an intercept of 1.0 at any given temperature if

Ind. Eng. Chem. Process Des. Dev., Vol. 21, No. 4, 1982 753 METERING

nPUMP

VENT PRESSURE GAUGE

d

DRIER

s

v

t BACK PRESSURE REGULATOR

PACKED BED WATER SATURATOR

REGULATOR

7

HYDROGEN CYLINDER

Figure 2. Experimental equipment.

the hydroisomerization reaction may be represented by a first-order reversible reaction. It can be shown from eq 4 and 6 that if the adsorption constants for the hydrocarbon species are approximately equal, then

plotting k-l/" w. 1/(1+ 8).If KHc>> KH2,this plot should be a straight line-with a positive slope. If KHc > Kc, and that K H