3822
Ind. Eng. Chem. Res. 1999, 38, 3822-3829
Alkylation of Isobutane with 2-Butene over a HFAU Zeolite. Composition of Coke and Deactivating Effect J. Pater,† F. Cardona,† C. Canaff,† N. S. Gnep,† G. Szabo,‡ and M. Guisnet*,† Laboratoire de Catalyse en Chimie Organique, UMR CNRS 6503, Universite´ de Poitiers, 40, avenue du Recteur Pineau, 86022 Poitiers Cedex, France, and CERT TOTAL, BP 27, 76700 Harfleur, France
The transformation of an isobutane -2-butene mixture with a 40 weight ratio was carried out in liquid phase at 50 °C on an USHY zeolite with a framework Si/Al ratio of 4.5. The reaction products can be classified into seven families: (a) trimethylpentanes resulting from isobutane alkylation with 2-butene or from isobutane self-alkylation, (b) dimethylhexanes, (c) C8 alkenes, (d) heavy products (mainly C9-C12), (e) light products C5-C7 alkanes, (f) n-butane, and (g) nondesorbed products (coke). These latter products, which are responsible for deactivation, were removed from samples coked for different times, after dissolution of the zeolite in a hydrofluoric acid solution, and analyzed by GC, IR, HNMR, MS, and GC/MS coupling. The coke components were identified as alkyl bicyclic olefinic compounds with 12-28 carbon atoms. IR spectroscopy analysis of the coked zeolite samples shows that coke molecules interact with the protonic sites, this interaction being mainly responsible for deactivation. Large differences can be observed in deactivation of the various reaction steps. The steps which are the more demanding in number (e.g., hydride transfer) or in strength (e.g., isomerization) of the acid sites are preferentially deactivated. Introduction Alkylation of isobutane with n-butene is an essential process for production of the so-called reformulated gasoline. Commercially, only hydrofluoric acid (HF) and concentrated sulfuric acid (H2SO4) are used as catalysts.1-5 Alkylates have a high research octane number (≈95) and an only slightly lower motor octane number (≈93).5 Other advantages are their low vapor pressure, their high heat of combustion, the absence of toxicity, and clean combustion.6 However, production of alkylates suffers from a number of drawbacks, in particular, safety problems with HF processes and high catalyst consumption with H2SO4 processes. Therefore, the substitution of these catalysts by nontoxic, noncorrosive, nonpolluting solid catalysts constitutes a very important objective for refiners. A large variety of solid acids but also of supported protonic (H2SO4,7 triflic acid,8 etc.) and Lewis (BF3,9 etc.) acids were shown to be active for the production of alkylates having compositions close to those of commercial alkylates. This is in particular the case for large pore zeolites such as HFAU and HBEA zeolites which, unfortunately, deactivate rapidly owing to the formation of carbonaceous compounds (coke). Despite that, there are very few studies of coking and deactivation of zeolites during isobutane alkylation.10-15 Weitkamp and Maixner10 characterized coke deposited during alkylation of isobutane with butene on a LaY zeolite catalyst by 13C CP/MAS NMR, concluding that coke was essentially paraffinic, perhaps with some multiring naphthenes. More recently, Flego et al.11,12 have examined the carbonaceous deposits formed on a LaHY zeolite by in situ UV-vis, FT-IR, and 13C CP/ MAS NMR. They concluded that deactivation was due to acid site poisoning by high molecular weight oligo† ‡
Universite´ de Poitiers. CERT TOTAL.
mers adsorbed as alkenyl carbenium ions. However, as was previously underlined,16 the composition of coke, i.e., the distribution of their components as a function of their nature and of their size, cannot be obtained by these in situ techniques. The only way to obtain this composition is to remove coke from the zeolites and to analyze it by adequate techniques. With zeolites, most of the coke molecules are formed and trapped inside the micropores; hence, dissolution of the zeolite, for instance in hydrofluoric acid, is necessary to liberate them. Fortunately, coke formed in low-temperature reactions, as is the case for alkylation, is completely soluble in organic solvents and therefore can be analyzed easily by classical techniques such as GC, 1H NMR, IR, MS, and GC/MS.16 This procedure was used in this work for determining the composition of coke formed during isobutane alkylation with 2-butene carried out in a fixed bed at 50 °C over a HFAU zeolite. A high value of the isobutane/butene ratio equal to 40 was chosen so as to limit the deactivation and to compare accurately the deactivating effect of coke on the various reactions involved in isobutane-butene transformation. Experimental Section Alkylation Conditions. The alkylation of isobutane with 2-butene was carried out in an up flow fixed bed reactor, as previously described.17 The operating conditions were as follows: 3 g of a USHY zeolite, temperature of 50 °C, pressure of 23 bar applied to keep the hydrocarbons in the liquid state, isobutane/2-butene weight ratio of 40, flow rate of the isobutane-2-butene mixture of 12 mL h-1, i.e., molar flow rate of butenes and isobutane of 2.9 × 10-3 and 112.4 × 10-3 mol h-1, respectively. The effluents were online analyzed by GC chromatography, as previously described.17 The yield in each product was taken as the ratio between the weight percentage of this product in the effluent and the weight percentage of butene in the feed.
10.1021/ie9902232 CCC: $18.00 © 1999 American Chemical Society Published on Web 09/11/1999
Ind. Eng. Chem. Res., Vol. 38, No. 10, 1999 3823 Table 1. Physicochemical Characteristics of the USHY Zeolite total Si/Al ratio framework Si/Al ratio unit cell formula pore volume micropore volume Bro¨nsted acidity Lewis acidity a
3.0 4.5 Na0.4H34.5Al34.9Si157.1O384, 17.8 EFALa 0.360 cm3 g-1 0.256 cm3 g-1 720 µmol g-1 190 µmol g-1
EFAL ) extraframework aluminum atoms.
Catalyst Sample. The USHY zeolite resulted from overnight calcination at 773 K under dry air flow of a CBV 500 sample supplied by PQ zeolites BV. The calcinated sample was characterized by various techniques: elementary analysis, nitrogen adsorption-desorption, IR spectrosocopy (OH and TOT bands and adsorption of pyridine). The main physicochemical characteristics are reported in Table 1. It should be emphasized that the number of protonic sites retaining pyridine adsorbed at 150 °C, 720 µmol g-1 (Table 1), is much lower than the theoretical number of acid sites estimated from the unit cell formula, 2590 µmol g-1, which can be related (i) to the inaccessibility of certain protonic sites (located in sodalite cages) or to their weakness and (ii) to exchange of protonic sites by cationic extraframework species.18 The IR spectrum in the region of OH stretching vibrations is very complex: besides the three characteristic bands corresponding to OH located in the supercages (3630 cm-1) or in the sodalite cages (3565 cm-1) and to silanols (3745 cm-1), various other bands are observed at 3599 and 3525 cm-1, which correspond to the OH groups of the supercages or of the sodalite cages in interaction with extraframework species (these OH groups are strongly acidic), and at 3680 and 3600 cm-1, which correspond to nonacidic OH groups probably due to extraframework AlOH species. Coke Characterization. Coked zeolite samples were recovered at different time-on-stream through the following procedure: the flow rate of reactant was stopped and the pressure decreased to atmospheric pressure, then the reactor was kept under nitrogen flow for 5 min at the reaction temperature and the coked sample recovered for analysis. The coke content was determined with an elementary analysis system (Thermoquest). Two methods were used for coke recovery. In the first one, the catalyst sample was directly treated in a Soxhlet with methylene chloride, whereas in the second one, the zeolite was dissolved in a hydrofluoric acid solution before Soxhlet extraction. Methylene chloride was eliminated, and the recovered compounds were analyzed by various techniques: GC, IR, 1H NMR, and GC/MS coupling. Results Influence of Time-on-Stream on the Rate of Formation of Reaction Products. Whatever the time-on-stream, the reaction products have from 4 to 14 carbon atoms. These products can be classified into six families: (a) trimethylpentanes (TMP), i.e., the normal products of isobutane alkylation with 2-butene, (b) dimethylhexanes (DMH), (c) C8 alkenes, (C8)), (d) heavy products C9+, mainly C9-C12, (e) light products: C5-C7 alkanes, and (f) n-butane. It should be remarked that all the alkanes, except n-butane, are ramified. A
Figure 1. Alkylation of isobutane with 2-butene over HFAU zeolite. Butene conversion and total yield in reaction products versus TOS.
Figure 2. Alkylation of isobutane with 2-butene over HFAU zeolite. Percentages of coke on the USHY zeolite versus TOS.
small amount of 1-butene resulting from double bond shift isomerization (but no isobutene) is also observed. As was already reported,17 the conversion of butene and the yield in the reaction products (taken as the ratio between their weight percentage in the effluent and the weight percentage of butene in the feed) pass through a maximum after 5-10 h reaction (Figure 1). The deactivation of the catalyst which is observed afterward is relatively slow: a decrease of 1.8% and 3.3% per hour of butene conversion and of the total yield, respectively. This decrease in conversion is probably due to the retention on the zeolite of carbonaceous compounds, which will be called “coke” for the sake of simplification. This formation of coke increases rapidly during the first hours of reaction (3 wt % of coke on the zeolite after 2 h) then more slowly (Figure 2). Part of this coke can be removed from the zeolite by direct Soxhlet treatment with methylene chloride and, hence, could correspond to carbonaceous deposits located on the outer surface of the crystallites and in the mesopores (outer coke). The other part which cannot be removed corresponds to coke located inside the zeolite micropores (inner coke). Figure 2 shows that at time-on-stream (TOS) e3 h, only inner coke is formed, whereas at long TOS the amount of outer coke becomes significant. The formation of coke being relatively fast, it is necessary to consider coke as a reaction product. This is done in Figure 3 which gives the yields in the various products as a function of TOS. Whatever the time-onstream, TMP (resulting from alkylation) are the main reaction products. For all the products, except coke and C8), the yield passes through a maximum; the decrease is more pronounced for TMP, C5-C7, and nC4 than for C9+ and DMH. The yield in coke decreases very rapidly with increasing TOS, more rapidly than the yield in the desorbed products. Furthermore, the yield in C8), very low at short TOS, increases for TOS >10 h. Alkenes
3824
Ind. Eng. Chem. Res., Vol. 38, No. 10, 1999
Figure 3. Alkylation of isobutane with 2-butene over HFAU zeolite. Yield in various products versus TOS.
Figure 5. Alkylation of isobutane with 2-butene over HFAU zeolite: distribution of (a) trimethylpentanes (TMP) and (b) dimethylhexanes (DMH) versus TOS. Table 2. Alkylation of Isobutane with 2-Butene over HFAU Zeolite. Trimethylpentanes (TMP) and Dimethylhexanes (DMH) Distributions (%) at Various Time-on-Stream and at Thermodynamic Equilibrium
Figure 4. Alkylation of isobutane with 2-butene over HFAU zeolite. Distribution of C8 hydrocarbons versus TOS.
appear in C9+ at the same TOS as shown from the comparison of the chromatogram of C9+ formed at short and at long TOS values: most of the peaks corresponding to C9+ have longer retention times at long TOS than at short TOS. The distribution of C8 hydrocarbons (TMP, DMH, traces of methylheptanes, and C8)) changes significantly with TOS. At short TOS, TMP are the main C8 products (≈85%) and C8) are formed only in traces. With increasing TOS, the percentage of TMP decreases, that of DMH remains constant, and that of C8) increases (Figure 4). There is also a large change with TOS in the distributions of TMP and DMH (Figure 5). It should be remarked (Table 2) that (i) no 2,2- and 3,3-DMH are observed, (ii) at short TOS the distribution of TMP is not too far from that at thermodynamic equilibrium,19 whereas this is not the case for the distribution of DMH, (iii) at long TOS, the 2,2,3-isomer becomes predominant in TMP and the 3,4-isomer in DMH. Characterization of Coked Zeolite Samples and of Coke. The IR spectra of two zeolite samples used for 2 or 30 h reaction, treated with methylene chloride in order to remove the outer coke, and then in primary vacuum for 2 h at 50 °C (this low temperature was chosen to eliminate the main part of physisorbed water
2,2,3-TMP 2,2,4-TMP 2,3,3-TMP 2,3,4-TMP total 2,2-DMH 2,4-DMH 2,5-DMH 3,4-DMH total
1.5 h
7h
53 h
equilibriuma 19
15.6 50.7 22.0 11.7 100 34.5 35.5 20.0 10.0 100
14.0 46.0 25.5 14.5 100 35.1 33.0 17.2 14.7 100
56 9.5 14 20.5 100 11.5 0 0 88.5 100
12.5 63.5 10.0 9.0 100 4.0 36.2 55.0 4.8 100
a 2,2- and 3,3-DMH were not considered in the equilibrium distribution.
Figure 6. Alkylation of isobutane with 2-butene over HFAU zeolite. IR spectra of the fresh zeolite and of two samples coked for 2 and 30 h.
without desorbing coke molecules) can be compared in Figure 6 with the one of the fresh zeolite sample. The percentages of coke (inner coke) on the coked samples are approximately 2.5 and 4.5 wt %, respectively. In these samples, the hydroxyl groups appear as a very
Ind. Eng. Chem. Res., Vol. 38, No. 10, 1999 3825
Figure 7. Alkylation of isobutane with 2-butene over HFAU zeolite. 1H NMR spectrum of total coke recovered after 53 h reaction.
large band with a maximum at 3560-3600 cm-1. Part of this band can be related to the displacement toward lower wavenumbers, of bands corresponding most likely to acidic or strongly acidic OH groups,18 owing to their interaction with not very basic coke molecules located near these OH groups. The other part can be related to the bands observed with the fresh zeolite, the corresponding OH groups being likely too far from the coke molecules or too weak for interacting with them. Two significant groups of bands, which were not present in the fresh zeolite, appear between 2850 and 3000 cm-1 and 1300-1500 cm-1 in the coked samples. In the region of the CH stretching modes, the bands at 2980, 2936, and 2877 cm-1 correspond to υasCH(CH3), υasCH(CH2), and υsCH(CH3), respectively. The region between 1500 and 1300 cm-1 exhibits the corresponding CH deformation bands δas CH(CH3) or δasCH(CH2) at 1450 cm-1, δs CH(CH3) at 1393 cm-1, and a band at 1674 cm-1 corresponding to ethylenic double bonds. No band corresponding to aromatics is present. It can therefore be concluded that coke is constituted by aliphatic, highly branched molecules with one or several double bonds. It should be remarked that the intensity of the bands corresponding to the aliphatic groups increases with coke content, whereas that of the CdC band decreases, which indicates a decrease in the olefinicity of the coke. The totality of coke (outer + inner coke) can be recovered in methylene chloride after dissolution in a hydrofluoric acid solution of zeolite samples used for 2, 30, and 53 h, with a coke content of 3, 8, and 10.5 wt %, respectively. This coke, as well as the outer coke recovered from a direct Soxhlet treatment from zeolite samples used for 10 and 30 h, was analyzed by various techniques: GC, IR, 1H NMR, and GC/MS coupling for the total coke and only GC and GC/MS analysis of the outer coke, because of the low amount of recovered material. IR spectroscopy of total coke shows the presence of various bands characteristic of aliphatic compounds with di-, tri-, or tetrasubstituted C-C double bonds. The bands corresponding to CH3 groups are very intense, indicating a high degree of branching of the coke components. In the 1H NMR of total coke (Figure 7), various signals can be observed. The most significant, which appears at 0.84 ppm, corresponds to protons of methyl groups R to saturated C atoms (50-60% of all the protons depending on time-on-stream). The signals at 1.15 and 1.35 ppm (20-25% of the protons) correspond to protons of the CH2 and CH groups R to saturated C atoms. Those at 1.68, 1.72, and 1.79 ppm (10-15% of the protons)
correspond to protons of the CH3 and CH2 R to unsaturated C atoms or to CH2 in cycles. That at 2.2 ppm (810%) corresponds to protons of CH groups R to a double bond. Very small peaks corresponding to olefinic protons (0.5-2% of the protons) can be observed at δ between 4.4 and 5.7 ppm. No peaks corresponding to aromatic protons are observed. GC analysis shows that outer coke as well as total coke are constituted of a very significant number of compounds. The retention times are between 60 and 120 min for the outer coke after 10 and 30 h reaction and between 50 and 110 min and 70 and 130 min for the total coke after 2 and 53 h, respectively. GC/MS coupling shows that most of the components of total and outer coke have a general formula of CnH2n-4 (with n between 12 and 28); i.e., they correspond to hydrocarbons presenting 3 unsaturations and/or cycles. Whatever the TOS, C15-C20 compounds are largely predominant. However, the longer the TOS, the heavier the coke components. To determine the number of double bonds in the coke components, the hydrogenation of total coke formed after 53 h reaction was carried out in a batch reactor at 25 °C, pH2 ) 1 bar (and also under pH2, 50 bar). This hydrogenation leads, as demonstrated by GC/MS coupling, to a mixture of CnH2n-2 compounds, i.e., of compounds resulting from the saturation of one double bond of the coke components. This could mean that either coke molecules have only one double bond (hence two cycles) or that their hydrogenation is uncomplete. However, we have checked on the example of 2,3dimethyl-2-butene wherein, under the operating conditions, tetrasubstituted double bonds underwent hydrogenation. Moreover the quasiabsence of olefinic protons (20 times) than the one found for the best CeFAU catalyst at 80 °C with an isobutane/butene ratio of 11,20 i.e., under conditions for which the deactivation is much faster. Values between 1 and 10 g/g catalyst have been reported for rare earth exchanged HY zeolites by using another type of reactor (continuous stirred-tank reactor) and an isobutane/n-butene ratio between 15 and 30.21 The value found with our catalyst is, however, much lower than that found with commercial HF and H2SO4 catalysts and with a lower isobutane/butene ratio.5 The low value of the yield in alkylate found with zeolites is mainly due to their rapid deactivation owing to a strong retention of reaction products inside the zeolite pores. Indeed, for 2.5 g of alkylate produced per gram of zeolite, 0.3 g of coke is retained on the zeolite. Therefore, the use of zeolites as alkylation catalysts demands a decrease of this deactivation by coke and the development of an economically method of regeneration.22 According to de Jong et al.,23 the key to extend the lifetime of alkylation catalysts is to operate at low olefin concentration throughout the reactor. The best way to limit deactivation is to operate under supercritical conditions. Thus, Clark and Subramanian24 have recently shown that supercritical alkylation using carbon dioxide as a diluent results in a virtually steady alkylate production. Reaction Scheme and Mechanisms. Whatever the TOS, C8 hydrocarbons [only alkanes at short TOS (10 h). The difference found at short TOS can be related to self-alkylation, while that at long TOS is due to the greater participation of n-butene in the formation of C9+ (as furthermore shown by the presence of alkenes in this family), of coke, and of the C5-C7 products. The percentage of isobutane self-alkylation in the formation of TMP was estimated from the value, leading to an equilibrated balance, of the number of butene molecules required for the formation of one molecule of TMP. Figure 9 shows that the percentage of self-alkylation decreases when TOS increases, becoming negligible after 10 h. Formation of Coke and Deactivation. Deactivation is most likely due to the formation inside the pores of heavy reaction products (coke). At very short TOS (10%) remained blocked in the zeolite pores. (2) All the desorbed saturated products (except nbutane) result from a carbenium ion chain mechanism involving alkylation, type A cracking and isomerization, and hydride transfer steps. (3) Deactivation is caused by nondesorbed products (coke). These nondesorbed products were identified as polyalkylbicyclic olefinic compounds with 12-28 carbon atoms. (4) IR spectroscopy analysis shows that coke molecules interact with the protonic sites. This interaction is probably responsible for a large part of deactivation, with pore blockage also affecting the activity at long time-on-stream. (5) The deactivating effect of coke depends on the demanding character of the reaction steps, i.e., on the number and strength of acid sites required for their catalysis. Reactions such as hydride transfer, which require more than one acid site for their catalysis, or isomerization, which requires relatively strong acid sites, are preferentially deactivated. Literature Cited (1) Gary, J. H.; Handwerk, G. E., Petroleum Refining, Technology and Economics, Marcel Dekker: New York, 1994; Chapter II, p 231. (2) Albright, L. F.; Goldsby, A. R. (Eds.) Industrial and Laboratory Alkylations; ACS Symposium Series, Vol. 55, American Chemical Society, Washington, DC, 1977. (3) Torck, B. Ge´ nie et Proce´ de´ s Chimiques, Techniques de l′Inge´ nieur, Paris, 1983; Tome J4, J5680-1. (4) Corma, A.; Martinez, A. Catal. Rev.-Sci. Eng. 1993, 35 (4), 483. (5) Weitkamp, J.; Traa, Y. Handbook of Heterogeneous Catalysis; Ertl, G., Kno¨zinger, H., Weitkamp, J., Eds.; Wiley-VCH: Weinheim, Germany, 1997; Vol. 4, pp 2039-2069. (6) Liolios, G. C. NPRA, Annual Meeting, San Francisco, California, 1989. (7) Joly, J.-F.; Benazzi, E.; Marcilly, C. European Patent Application 539 277, applied for by Institut Franc¸ ais du Pe´trole, April 28, 1993. (8) Hommeltolft, S. I.; Topsoe, H. F. A. U.S. Patent 5 245 100, assigned to Haldor Topsoe A/S, September 14, 1993. (9) Cooper, M. D.; King, D. L.; Sanderson, W. A. WO Patent Application 92/04977, applied for by Catalytica, Inc., April 2, 1992. (10) Weitkamp, J.; Maixner, S. Zeolites 1987, 7, 6-8. (11) Flego, C.; Kiricsi, I.; Parker, W. O., Jr.; Clerici, M. G. Appl. Catal. A 1995, 124, 107-119. (12) Flego, C.; Galasso, L.; Kiricsi, I.; Clerici, M. G. Stud. Surf. Sci. Catal. 1994, 88, 585-590.
Ind. Eng. Chem. Res., Vol. 38, No. 10, 1999 3829 (13) Querini, C. A.; Roa, E. Appl. Catal. A 1997, 163, 199-215. (14) Sto¨cker, M.; Mostad, H.; Rorvik, T. Catal. Lett. 1994, 28, 203-209. (15) Nivarthy, G. S.; He, Y.; Seshan, K.; Lercher, J. A. J. Catal. 1998, 176, 192-203. (16) Guisnet, M.; Magnoux, P. Stud. Surf. Sci. Catal. 1994, 88, 53-68. (17) Cardona, F.; Gnep, N. S.; Guisnet, M.; Szabo, G.; Nascimento, P. Appl. Catal. A 1995, 128, 243-257. (18) Morin, S.; Ayrault, P.; Gnep, N. S.; Guisnet, M. Appl. Catal. A 1998, 166, 281-292. (19) Stull, D. R.; Westrum, E. F., Jr.; Sinke, G. C. The Chemical Thermodynamics of Organic Compounds; J. Wiley & Sons: New York, 1969. (20) Weitkamp, J. Stud. Surf. Sci. Catal. 1980, 5, 65-75. (21) Kirsch, F. W.; Potts, J. D. Preprints, Div. Petr. Chem. Am. Chem. Soc. 1970, 15 (3), A109-A121. (22) Weitkamp, J.; Traa, Y. Catal. Today 1999, 49, 193-199.
(23) de Jong, K. P.; Mesters, C. M. A. M.; Peferoen, D. G. D.; van Brugge, P. T. M.; de Grost, C. Chem. Eng. Sci. 1996, 51, 20532060. (24) Clark, M. C.; Subramaniam, B. Ind. Eng. Chem. Res. 1998, 37, 1243-1250. (25) Deno, N. C.; Peterson, H. J.; Saines, G. S. Chem. Rev. 1960, 60, 7-14. (26) Corma, A.; Fornes, V.; Martinez, A.; Orchilles, A. V. ACS Symp. Ser. 1998, 368, 542. (27) Giannetto, G.; Sansare, S.; Guisnet, M. J. Chem. Soc. Chem. Commun. 1986, 1302. (28) Magnoux, P.; Roger, P.; Canaff, C.; Fouche´, V.; Gnep, N. S.; Guisnet, M. Stud. Surf. Sci. Catal. 1987, 34, 317.
Received for review March 24, 1999 Revised manuscript received July 19, 1999 Accepted July 23, 1999 IE9902232