42
Ind. Eng. Chem. Res. 1991,30,42-50
ition of Silicon Methoxide. J . Phys. Chem. 1986,90,6233-6237. Niwa, M.; Yamazaki, K.; Murakami, Y. Separation of Oxygen and Nitrogen due to the Controlled Pore-openingSize of CVD Zeolite A. Chem. Lett. 1989,441-442. Scofield, J. H.Hartree-Slater Subshell Photoionization Cross-Sections at 1254 and 1487 eV. J . Electron Spectrosc. Relat. Phenom. 1976,8, 129-137. Teraoka, Y.; Kunitake, K.; Kagawa, S.; Iwamoto, M. Control of Gas Adsorbability of Zeolites by Chemical Vapor Deposition of Si-
lanes. Nippon-Kagakukai-shi 1989,424-428. Yan, Y.; Verbiest,J.; Hulsters, P. D.; Vansant, E. F. Modification of Mordenite Zeolites by Chemisorption of Disilane and its Influence on the Adsorption Properties. J. Chem. SOC.,Faraday Trans. 1 1989,85, 3095-3105.
Received for reuiew March 16, 1990 Revised manuscript received July 24, 1990 Accepted August 9,1990
Catalytic Hydrogenation and Hydrocracking of Fluorene: Reaction Pathways, Kinetics, and Mechanisms Arunas T. Lapinas,?Michael T. Klein,* and Bruce C. Gates Center for Catalytic Science and Technology, Department of Chemical Engineering, University of Delaware, Newark, Delaware 19716
Aris Macris and James E. Lyons Research and Development, Sun Refining and Marketing Company, P.O. Box 1135, Marcus Hook, Pennsylvania 19061
Catalytic hydrogenation and hydrocracking reaction pathways were determined for fluorene a t 335-380 "C and 153-atm total pressure. A presulfided NiW/A1203catalyst was active for isomerization and hydrogenation, and reaction of fluorene gave 1,2,3,4,4a,9a-hexahydrofluoreneand ultimately perhydrofluorene. The much more acidic (presulfided) NiMo/zeolite Y catalyst was active for hydrogenation and isomerization and also for hydrocracking through the central five-carbon-membered ring, which gave equimolar yields of saturated and aromatic single-ring products. The fluorene disappearance rate constants k [(cm3of solution/(g of catalyst s)] were 0.13 and 0.41 at 380 "C for reaction catalyzed by NiW/A1203 and by NiMo/zeolite, respectively. Arrhenius parameters were [log A [(cm3of solution/(g of catalyst s)], Eect (kcal/mol)] = [5.7, 19.81 and [9.7,30.3], respectively. Hydrocracking of the five-carbon-membered-ring-containing (5-CMRC) fluorene occurred via cleavage pathways distinct from those for 6-CMRC fused-ring polynuclear aromatic compounds. The major fluorene hydrocracking pathway, giving 2 mol of single-ring products, leads to less hydrogen consumption than the hydrocracking of a comparable 6-CMRC compound such as phenanthrene, which gives only 1 mol of single-ring products and lower value light-gas products. These results provide guidelines for modeling the hydrocracking of feeds containing polynuclear aromatic hydrocarbons.
Introduction Hydrocracking is a flexible refining process that allows production of gasoline and light oils from less valuable petroleum stocks such as residua, cycle oils, gas oils, and cracked and straight-run naphthas (Sullivan and Scott, 1983). Fuels derived from blends from coal, shale, and tar sands also provide potential feeds for hydrocracking processes (Guin et al., 1976; Petrakis et al., 1983a-c; Allen et al., 1984). Hydrocracking of the polycyclic aromatic compounds present in these stocks to give lower molecular weight species with increased H:C ratios requires bifunctional catalysts having hydrogenation and cracking activity. In general, amorphous catalysts are used for heteroatom removal and aromatic hydrogenation, as in the production of lube oils, whereas zeolite-containing catalysts are used for hydrocracking heavy feeds to produce kerosene, jet and diesel fuels, gasoline, and petrochemical feedstocks. A ring within the framework of a polycyclic molecule in a hydrocarbon feed comprises either five or six carbon atoms. One important class of these hydrocarbons includes compounds containing only six-carbon-membered rings,
* Author to whom correspondence should be addressed. 'Present address: Corporate Research Science Laboratories, Exxon Research and Engineering Company, Clinton Township, Route 22 East, Annandale, N J 08801. 0888-5885/91/2630-0042$02.50/0
and another includes compounds containing both sixcarbon-membered rings and five-carbon-memberedrings. Aromatic compounds in the former class, such as naphthalene, anthracene, phenanthrene, and pyrene, hydrocrack through pathways including terminal ring hydrogenation followed by possible isomerization, ring-opening, and dealkylation. Hydrogen consumption and the yield of light gases are relatively high (Qader and Hill, 1972; Qader et al., 1973; Qader, 1973; Wu and Haynes, 1975; Huang et al., 1977; Veluswamy, 1977; Shabtai et al., 1978; Wuu, 1983). Much less attention has been paid to five-carbon-membered-ring-containing (5-CMRC) compounds, such as fluorene and fluoranthene, which are, however, present in many crude oils and coal liquids in roughly as high a mass fraction as the 6-CMRC fused-ring aromatics (Guin et al., 1976; Petrakis et al., 1983a-c). Fluoranthene hydrogenation catalyzed by a presulfided NiW/A1203 and hydrocracking catalyzed by a NiMo/zeolite (Lapinas et al., 1987) proceed via kinetically significant cleavage pathways distinct from those observed for the 6-CMRC compound class. Hydrocracking through the five-carbon-memberedring is a major pathway and leads t o lower gaseous and higher liquid p r o d u c t y i e l d s . a higher degree of product aromaticity, and a lower hydrogen consumption than the hydrocracking pathway outlined above for 6-CMRC fused-ring aromatic hydrocarbons. These differences have 0 1991 American Chemical Society
Ind. Eng. Chem. Res., Vol. 30, No. 1, 1991 43 Table I. Summary of Catalytic Batch Reactor Experiments NiW/AI2O3 NiMo/zeolite Y catalyst solvent n-hexadecane cyclopentane 153 total pressure, atm 153 335,380 350,380 temperature, O C to 300 to 300 reaction time. min
major processing implications, which provide the motivation for the research presented here. The reaction of fluorene with hydrogen in the presence of solid catalysts has not been studied. The reaction of fluorene catalyzed by Lewis acids (Salim and Bell, 1984) and in donor solvents (Constant et al., 1986) can proceed through central-ring cracking without prior hydrogenation. Cleavage of both the phenyl-phenyl (Salim and Bell, 1984) and phenyl-methylene (Salim and Bell, 1984; Constant et al., 1986) bonds leads to diphenylmethane and 2methylbiphenyl,respectively, as primary reaction products. Here we describe reactions of fluorene in the presence of hydrogen and NiW/A1203and NiMo/zeolite catalysts. Experiments were designed to characterize the performance of these catalysts at temperatures and pressures of industrial processes (Sullivan and Scott, 1983; Unzelman and Gerber, 1965; Ward, 1975),and a specific goal was the development of a quantitative model for fluorene hydrogenation and hydrocracking. Comparison of this model to a model presented earlier (Lapinas et al., 1987) for fluoranthene hydroprocessing allowed development of a preliminary mechanistic interpretation of the conversion of 5-CMRC compounds. This is expected to contribute to the basis for process modeling. Experimental Section Hydroprocessing of fluorene was investigated in isothermal batch (reaction) experiments at temperatures ranging from 335 to 380 "C and a total pressure of 153 atm (2250 psig). Table I is a summary of the conditions of the experiments, including the catalysts, solvents, temperatures, pressures, and holding times. Materials. Fluorene (Aldrich, 98%), high-pressure hydrogen (Linde, 3500 psi), n-hexadecane solvent (Humphrey Chemical Co., specially distilled), cyclopentane solvent (Aldrich, 95+ % ), gas chromatography standards, and sulfiding materials, such as carbon disulfide (Fisher Scientific) and hydrogen sulfide (Matheson Gas Products, 90% Hz/lO% H2S), were all obtained from commercial sources and used as received. The commercial catalysts and their properties are summarized in Table 11. Both the Shell 454 NiW/A1203hydrogenation catalyst and the NiMo/zeolite Y hydrocracking catalyst were equilibrated over a few months' stream time in hydroprocessing applications at Sun Company prior to the present investigation. This procedure minimized the influence of catalyst deactivation in the experiments reported here. The catalysts were sulfided before use. The extrudates were first ground to 80-100 mesh, and then a 30 cm3/min flow at 1 atm of an H2S/Hz gas mixture (10% HzS) was passed over the resulting particles as they were heated to and held for 2 h at 400 "C. The sulfided catalyst particles were then introduced into the reactors after cooling for 1 h (Bhinde, 1979). Reactions. Reaction kinetics were measured with a 300-cm3stirred batch reactor (Autoclave Engineers) that allowed for reactant injection at time zero, periodic liquid and gas sampling, efficient liquid- and gas-phase agitation, and precise temperature and pressure measurement and control. The details of the procedures and the solvent were chosen to minimize the complications arising from cracking of the solvent.
Table 11. Properties of NiW/A1201and NiMo/Zeolite Catalysts catalyst NiW/A1203 NiMo/zeolite Y pellet diameter, cm 0.159 0.318 bulk density, g/cm3 1.03 0.61 surface area, m2/g 133 625 (zeolite) 0.34 pore volume, cm3/g 24.45 unit cell size of zeolite, A composition, w t % Ni 4.36 7 W 27.2 Mo 13 SiOz 0.11 65.6 A1203 58.2 33.6
When the NiW/AlZO3catalyst was used, the solvent cracking was minimal, and the autoclave was typically loaded with 150 cm3 of high-purity n-hexadecane solvent, 0.3 g of sulfided catalyst, and 0.07 g of carbon disulfide. After the system had been purged with hydrogen, it was heated to a temperature 5 "C higher than the reaction temperature to allow for cooling upon injection of reactant. The hot reactor was then pressurized to 102 atm with hydrogen. Concurrently, a loading device containing 35 cm3of solvent and 1.2 g of fluorene was heated to 180-190 "C to ensure that the contents were in the liquid phase. Samples were removed from the reactor before the arbitrary time zero to allow measurement of the degree of the background thermal and catalytic cracking of the n-hexadecane. A 0.5-pm pore size filter prevented removal of catalyst during sampling. A 51-atm excess driving force in the loader allowed rapid injection of the contents into the reactor and simultaneous pressurization of the reactor with 153 atm of hydrogen. Stabilization of the reactor temperature occurred in less than 2 min, whereafter liquid samples were withdrawn periodically for analysis. In experiments characterizing the reactions of fluorene in the presence of the NiMo/zeolite catalyst, when cracking of both the reactant and n-hexadecane would occur, the reactor was loaded instead with 80 cm3 of cyclopentane, 0.045 g of catalyst, and 0.05 g of carbon disulfide. The severe hydrocracking of n-hexadecane catalyzed by the zeolite (Lapinas et al., 1987) would have obscured observation by gas chromatography of important lower molecular weight products. Any cracking of cyclopentane solvent would not interfere with product identification. After loading, the reactor was purged of air with hydrogen and heated to a temperature 5 "C higher than the desired reaction temperature to allow for cooling upon injection of reactant. An initial reactor loading of 80 cm3 of cyclopentane and hydrogen at 20 atm resulted in total pressures of 110 and 121 atm for temperatures of 350 and 380 "C, respectively. The loading device containing 20 cm3 of cyclopentane and 0.17 g of fluorene was pressurized to provide a driving force to assure rapid injection of the contents into the reactor and simultaneous pressurization of the reactor to 153 atm. Analytical Chemistry. Reaction products were identified by GC/MS (Hewlett-Packard 5970 mass-selective detector at the University of Delaware and the MS analytical laboratory at Sun Company). Products were analyzed by GC. Condensed-phase products were separated in a 50-m SE-54 fused-silica capillary column in both the GC/MS and GC and detected by flame ionization in a Hewlett-Packard 5880 gas chromatograph. Separately determined GC response factors, relative to the external standard, dibenzyl ether, allowed determination of product yields. These analyses permitted calculation of the GCobservable products index (OPI), defined as the mass of
44 Ind. Eng. Chem. Res., Vol. 30, No. 1, 1991 Table 111. Catalytic Reaction of Fluorene with Hydrogen in the Batch Autoclave catalyst NiMo/zeolreaction conditions NiW/A1203 ite Y temperature, OC time, min conversion of fluorene
380 60 0.56
380 380 380 60 180 180 0.52 0.87 0.96 product molar yields"
product hexahydrofluorene (HHF) isohexahydrofluorene (a) isohexahydrofluorene (b) perhydrofluorene (PHF) (methylcyclopenty1)methylenebenzene (MCP:M:B)c (methylcyclopenty1)toluene(MPC:TOL) cyclohexylmethylenebenzene (CH:M:B)d cyclohexyltoluene (CH:TOL) 2-methylbiphenyl naphthalene tetralin cis- and trans-decalin methylindan indan ethylbenzene toluene methylcyclohexane (MCH) dimethylcyclopentane (DMCP) cyclohexane benzene methylcyclopentane (MCP)
0.33 trb tr 0.09
0.31 tr tr 0.41
0.04 0.016 0.029 tr 0.009
0.03 0.023 0.036 tr 0.016
0.002 0.009 0.001 tr 0.013 0.031 tr 0.008 0.027 0.023 0.026 0.068
0.001 0.008 0.004 tr 0.006 0.079 tr 0.041 0.107 0.054 0.109 0.208 0.291 0.047 0.327 0.291
o
20
60
40
a0
io0
120
140
160
i s 0 200
t/min
0.073
0.016 0.099 0.094
Figure 2. Product molar yields for reaction of fluorene catalyzed by NiW/A1203 at 380 OC and 153 atm.
Fluorene
tl
k1
HHF
iso-HHF isomers
I, mol/mol reactant fed. tr = trace amount. (Methylcyclopenty1)phenylmethane. dCyclohexylphenylmethane. A. Reaction with NiW/A120, catalyst
PHF
Figure 3. Fluorene hydrogenation network. Fluorene
isomers of hexahydrofluorene (ioo.HHF isomers)
Perhydrofluorone (PHF)
1.2,3,4,49,9a.he~ahydr0flu0rene
B. Additional Products from reaction w i t h NiMoizeolile catalyst
o-Cyclohrx yltoluene (CH:TOL)
Methylcyclopentylmethylenebenlene (MCP:M:B)
Cyclohexylmethylenebenzene (CH:M:B)
o.Methylcyclopentyltoluene (MCP:TOL)
Figure 1. Structures of some products of fluorene conversion.
all observed products, including fluorene, divided by the initial mass of fluorene, and also the determination of product molar yields y = moles of product/initial moles of fluorene (0 I y I 1)and their variation with time in the reactor. Results and Discussion Fluorene hydrogenation in the absence of measurable hydrocracking was catalyzed by NiW/A1203,whereas both hydrogenation and hydrocracking were catalyzed by
NiMo/zeolite. Representative yields of reaction products are summarized in Table 111, and their chemical structures are illustrated in Figure 1. NiW/A1203Catalyst. The major products (see Figure 1A for structures) of the reaction of fluorene in the presence of the NiW/A1203 catalyst at 335 and 380 O C were 1,2,3,4,4a,9a-hexahydrofluorene (HHF), perhydrofluorene (PHF), and isomers of HHF (iso-HHF) in trace amounts. The OPI was 1.0 f 0.2. Fluorene hydrogenation a t 380 " C is summarized in Figure 2,which shows that the major products were again HHF and PHF. Fluorene underwent complete conversion after 220 min reaction time, and its initial rate of disappearance essentially matched the initial rate of formation of HHF; the initial rate of formation of PHF was apparently zero. Moreover, the rate of PHF formation WPS maximal at the point of the maximum in the yield of HHF, which occurred in the range of 140 min. Thus we conclude that HHF was a primary hydrogenation product and PHF likely formed from HHF. Trace amounts of iso-HHF(a) and iso-HHF(b) were observed. Reaction a t 335 " C was qualitatively similar. These data are the basis for the fluorene hydrogenation network proposed in Figure 3. Fluorene undergoes hydrogenation to HHF, the lone primary hydrogenation product, and its subsequent reaction is further hydrogenation to PHF, the final stable product. The identities of the trace products, iso-HHF(a) and iso-HHF(b), suggest that they are formed by isomerization of HHF. The rate of fluorene disappearance was essentially first order in fluorene, as shown by the linearity of plots of In
Ind. Eng. Chem. Res., Vol. 30, No. 1, 1991 45 1.o
Table IV, Fluorene Hydrogenation and Hydrocracking Rate Constants'
k, log A, E,&, reaction cm3 g-' cm3 g-l s-l kcal mol-' Hydrogenation with NiW/A1203 (T = 380 " C ) 1 0.13 5.8 19.9 2 0.15 1.7 7.5
0.8
Hydrocracking Pathways Observed with NiMo/Zeolite Y (2' = 350 " C ) 0.13 9.7 30 0.90 9.1 26 0.16 8.9 28 0.04 21 6.2 0.002 25 6.2 0.03 27 8.0 0.03 23 6.7 0.13 17 5.1 21 0.03 5.7
0.4
4Reactions are defined in Figures 3 and 6.
(1- X)vs time (Lapinas, 1989) up to high conversions (0.99 after 220 min). Estimation of the pseudo-first-order rate constants of Figure 3 was by the optimization algorithm COMPLEX (Beveridge and Schechter, 1970). The temperature dependence of these estimates yielded Arrhenius parameters of [log A [cm3of solution/(g of catalyst s)],Ea& (kcal/mol)] = [4.8, 171 for the disappearance of fluorene. For comparison, reaction of fluoranthene catalyzed by NiW/A1203yielded [log A, Ea&]= [6.4,19] (Lapinas et al., 1987). The hydrogenation of fluoranthene is therefore approximately an order of magnitude faster than that of fluorene. The activation energies for fluorene disappearance are within the range indicated in the literature, typically 15-40 kcal/mol. The rate constants of Figure 3 are summarized in Table IV. The curves in Figure 2 represent the molar yields of fluorene, HHF, and PHF calculated by using these rate constants and the network of Figure 3. Agreement between the data and the fit is satisfactory. NiMo/Zeolite Catalyst. Reaction of fluorene at 350 and 380 "C in the presence of the NiMo/zeolite catalyst yielded the numerous hydrogenation, isomerization, and cracking products listed in Table 111. The OPI was 1.0 f 0.1. The major multiring products were HHF, iso-HHF(a), iso-HHF(b), cyclohexylmethylenebenzene (CH:M:B), (methylcyclopenty1)methylenebenzene (MCP:M:B), cyclohexyltoluene (CH:TOL), (methylcyclopenty1)toluene (MCP:TOL), tetralin, and indan. The colons are meant to delineate the moieties within the hydrocarbons of molecular weight 174 (HC174 = CHM:B, MCPM:B, CH:TOL, MCPTOL) illustrated in Figure 1B. The major single-ring products were methylcyclohexane (MCH), dimethylcyclopentane (DMCP), toluene, cyclohexane, benzene, and methylcyclopentane (MCP). Several minor products were also observed, as shown in Table 111. The discussion that follows first focuses on fluorene and its primary reaction products and then on the formation and disappearance of secondary products in the order of multiring to single-ring species. The goal of this discussion is a statement of the reaction network. Figure 4 illustrates the time dependence of the yields of the hydrogenation and isomerization products at 350 "C and 153 atm. Fluorene was about 60% converted after 200 min. The yield of HHF passed through a maximum after about 80 min, when the rate of its secondary reaction to iso-HHF(b) was high. The rate of formation of isoHHF(a) also appeared to be maximal at the point of the maximum in HHF yield. Figure 5 allows closer scrutiny of the yields of HHF and its products, namely, the sum of the iso-HHF isomers, tetralin, indan, and the HC174 lump. The latter are
0.6
0.2
0.0
timin
Figure 4. Initial hydrogenation and isomerization product yields for reaction of fluorene with hydrogen catalyzed by NiMo/zeolite at 350 " C and 153 atm.
0.30
1
+
E
E
-k
0.20
2
0.10
0.00 0
200
100
300
timin
Figure 5. Molar yields of cleaved products and their apparent precursors during reaction of fluorene with hydrogen catalyzed by NiMo/zeolite at 350 O C and 153 atm.
products of the cleavage of one carbon-carbon bond of HHF and of the iso-HHF isomers. The major products in the HC1,., lump, MCP:M:B and CH:M:B, were accompanied by lesser amounts of MCPTOL and CH:TOL. The yields of tetralin and indan increased monotonically to values of 0.072 and 0.057, respectively, a t 310 min. The data of Figure 5 suggest that tetralin and indan may also be hydrocracking products of HHF and/or the iso-HHF isomers. The single-ring products DMCP, MCH, and benzene first appeared after 90 min and reached yields of 0.13,0.23, and 0.21, respectively, at 360 min. Thus cleavage of the alkyl-phenyl bonds of MCP:M:B and CH:M:B evidently led to benzene plus the corresponding saturated-ring compound. Similar behavior was observed for the reaction of MDP:TOL and CH:TOL, to give cyclohexane, MCP, and toluene. The reaction of fluorene with hydrogen a t 380 OC was qualitatively similar. Quantitative differences are reflected in the values of the rate parameters summarized in Table IV. The foregoing results provided a basis for formulation of the fluorene hydrocracking network shown in Figure 6, which comprises a single primary reaction pathway and six secondary pathways, labeled Pl-P6. Fluorene undergoes initial hydrogenation to give HHF, the reactions of which include isomerization to give iso-HHF(a) and iso-HHF(b). HHF and its isomers are of molecular weight 172 and follow the secondary reaction pathways Pl-P6. P1, P2, P4, and P5 depict bond scission to give the four products in the HC174 lump, namely, CH:M:B, CH:TOL,
46 Ind. Eng. Chem. Res., Vol. 30, No. 1, 1991
cm Fluorene
CH:M:B
CH:TU
1
1
MCP:TOL
1
MCP:M:B
1
Figure 6. Fluorene hydrocracking network (pseudo-first-order rate constants a t 350 "C).
MCP:TOL, and MCP:M:B. CH:M:B (formed in P1) undergoes further cracking to give MCH and benzene. CH:TOL (formed in P2) cracks to give cyclohexane and toluene. Akin to P2, pathway P4 involves cracking of MCP:TOL to give MCP and toluene, and MCP:M:B is converted in P5, which is akin to P1, to give DMCP and benzene. Pathways P3 and P6 in Figure 6 represent the cracking of a hydrogenated terminal ring and lead to the formation of tetralin and indan, respectively; the hydrocracking pathways of those compounds to give single-ring cracked products are well-known (Qader and Hill, 1972; Qader, 1973; Haynes et al., 1983). Parameter estimation using the network of Figure 6 and the COMPLEX optimization algorithm yielded the rate constants of Table IV. Best-fit activation energies are within the literature range of 15-40 kcal/mol for the hydrocracking of polynuclear compounds (Qader et al., 1973; Lapinas et al., 1987; Gates et al., 1979; Stephens and Chapman, 1983). Comparison of the solid lines that represent the network correlation and the experimental points in Figures 4 and 5 show satisfactory agreement. Separate experiments with HHF and the HC174compounds as reactants would aid in improvement of the network and kinetics analysis. Reaction Mechanisms. According to the traditional view, the sequence of reactions in the catalytic hydrocracking of fused-ring aromatic compounds is ring hydrogenation at metal sites followed by cracking at acidic sites. Cracking of 6-CMRC compounds frequently involves the opening and dealkylation of saturated terminal rings formed by hydrogenation (Lemberton and Guisnet, 1984). In contrast, the 5-CMRC polycyclic fluoranthene followed a unique pathway (Lapinas et al., 1987), whereby initial hydrogenation to the primary product 1,2,3,3a-tetrahydrofluoranthene (THFL) was followed by cleavage to give 1-phenyltetralin, the further cracking of which gave tetralin and benzene. This sequence represents cracking through a central fused ring, made possible, mechanistically, through a series of intramolecular hydride shifts.
These, in turn, were favored by the presence of a benzylic hydride one bond length's distance from the carbocation formed upon protonation of the phenyl substituent in 1-phenyltetralin. Reaction of fluorene in the presence of each of the catalysts proceeded via primary hydrogenation. Secondary reaction of HHF catalyzed by NiW/A1203gave PHF and trace amounts of the isomerization products iso-HHF(a) and iso-HHF(b), whereas reaction catalyzed by the NiMo/zeolite took place with high selectivity to isomerization and cracking. Consistent mechanistic steps are summarized in Figures 7-10, which provide more detail about the global reaction pathways of Figure 6. P1 and P2, in Figure 7, begin via protonation of the aromatic ring, which affords a carbocation intermediate having as two of its resonance structures A and B that can in turn undergo /%scission to give the secondary carbenium ions A' and, with a subsequent H shift, B'. Their deprotonation, followed by olefin hydrogenation, leads to the formation of CH:M:B and CH:TOL. Both CH:M:B and CH:TOL may then undergo protonation and subsequent @-scission,which recovers the aromaticity of the phenyl substituent and also yields an alkyl carbocation, the deprotonation of which leads to an olefin that readily undergoes hydrogenation. The net productivity of P1 and P2 is equimolar yields of methylcyclohexane and benzene and also cyclohexane and toluene, respectively. The mechanistic steps underlying pathways P4 and P5 are expected to be similar. Formal replacement of the cyclohexyl moieties in Figure 7 with methylcyclopentyl moieties allows generation of intermediate structures, otherwise analogous to A, B, A', and B', that lead to MCP:M:B and MCP:TOL. The ultimately formed single-ring product pairs are benzene and DMCP, and toluene and MCP, from P4 and P5, respectively. The differences between the pair of kinetically similar pathways P1 ( k 3 = 0.16) and P5 (k, = 0.13) and the pair of the kinetically similar paths P2 (k4 = 0.04) and P4 (k,
Ind. Eng. Chem. Res., Vol. 30, No. 1, 1991 47
m m
aJ3 CHMB
Figure 7. Postulated mechanism for hydrocracking HHF along the global reaction pathways P1 and P2 of Figure 6.
an m
P3.
pi*
= 0.03) invite mechanistic spectulation. As noted above, P5 is the formal analogue of P1 and P4 is the analogue of P2. Evidently the @-scissionof intermediate B in P2 and P4, which leads to the formation of an isotoluene-like moiety, has a high energetic requirement. The @-scission of intermediate A in P1 and P5 leads to a fully aromatic ring connected to a cyclohexyl carbenium ion (A'). The net result is that P1 and P5 proceed about four times as fast as P2 and P4, respectively. Figure 8 is a summary of the steps expected to lead to the formation of tetralin from HHF along P3a. Initial protonation affords the carbocation intermediate A, the @-scissionof which gives intermediate A'. These steps are common to P1 in Figure 7. Scission of either of the two cyclohexyl bonds @ to the ion A' affords a relatively unstable primary carbenium ion that would undergo rapid isomerization, via a 1,2-hydride shift, to a more stable secondary carbenium ion (Lemberton and Guisnet, 1984). This would, in turn, undergo further @-scissionto generate propylene and either an allyl or secondary carbocation. Ring closure occurs through addition of the carbocation to the adjacent ring, which yields a stable tertiary carbenium ion, the deprotonation of which yields dialin, which would be expected to be hydrogenated rapidly to give tetralin. Figures 9 and 10 show mechanistic steps expected for the reaction of the iso-HHF isomers to give tetralin, along P3b, and indan, along P6, respectively. Figure 9 shows that protonation followed by @-scissionleads to a secondary carbenium ion, the @-scissionof which leads to the evolution of propylene and the carbenium ion intermediate (D'), the allylic form of which allows ring closure to give tetralin, as described above. Figure 10 shows steps involving a secondary benzylic carbocation that leads to the formation of isobutene and indan. The mechanistic steps proceed through intermediates having relative stabilities that parallel the kinetics deduced from analysis with the network of Figure 6. Recall that, in Figure 8, pathway P3a required the formation of relatively unstable primary carbenium ions, whereas in Figure 9, P3b involved only the more stable secondary carbenium ions. Thus tetralin formation should be largely from P3b. Inspection of Table IV, a summary of the fit of experimental results to the proposed network, shows consistency with this interpretation, where k5/k6