B. Imelik et al, (Editors), Catalysis by Zeolites
© 1980 Elsevier Scientific Publishing Company, Amsterdam - Printed in The Netherlands
ISOBUTANE/BUTENE ALKYLATION ON CERIUM EXCHANGED X AND Y ZEOLITES JENS WEITKAMP Engler-Bunte-Institute, University of Karlsruhe, Richard-Willstatter-Allee 5, D-7500 Karlsruhe 1, Federal Republic of Germany INTRODUCTION While zeolite catalysts have gained tremendous importance in some petroleum refining processes, viz. catalytic cracking, hydrocracking, and isomerization, they failed to receive acceptance in the alkylation of isobutane with light olefins. In this process which yields highly branched paraffins required as gasoline components conventional liquid phase catalysts, i.e. concentrated sulfuric acid or anhydrous hydrogen fluoride are still applied. Chemically, alkylation, isomerization, and cracking have many features in common. In particular they are catalyzed by acids and proceed via carbenium ions. Both the mechanisms and process engineering aspects of hydrocarbon conversion on acid zeolites have been treated comprehensively in recent textbooks (ref. 1-3). In a strongly simplified manner the mechanism of isobutane/butene alkylation catalyzed by, e.g., H2S0 4 or HF has often been interpreted in terms of the following chain reaction initiated by protonation of the alkene and hydride transfer between the resulting sec. butyl cation and isobutane: .~
C 'iii /
c-cf- C -c-c
C-/@ ,'c i-OCTANES
I t :"-----------_.. _--- --.._----.... ------._-----.----_.------- ...t .~----
Double bond shift in the alkene is considered to be rapid and hence the tertiary butyl cation can be alkylated either by 1-butene or 2-butene regardless of which butene is used in the feed. The primary products of the alkylation steps can rearran~e to some extent whereby different dimethylhexyl and trimethylpentyl cations are formed. Hydride transfer between i-octyl cations and isobutane fina~ly yie~ds a mixture of isooctanes and tert. butyl cation which propagates the chain. A 1:1 overall isobutane : butene stoichiometry is predicted according to this scheme.
The true mechanism of reaction is considerably more complicated and many side reactions of alkyl carbeniumions have been specified (ref. 4-7) to account for the observed products. The principal feasibility of isobutane/alkene alkylation on zeolites was demonstrated as early as 1968 (ref. 8-10). Since then surprisingly few groups have published results of research work on this reaction, namely Kirsch, Potts et al. (ref. 8,9,11-13), Minachev et al. (ref. 14-16), and Schollner et al. (ref. 17-19). From their work it is evident that zeolites although highly active initially undergo rapid deactivation due to coke forming side reactions. It was mainly this deactivation which rendered zeolites economically unattractive for isobutane/alkene alkylation processes. Furthermore, the occurrence of deactivation made proper experimental handling of the reaction difficult, the situation being severely aggravated by complex product distributions. For the most part these difficulties were circumvented by integral sampling procedures over relatively long periods which, however, can only furnish a time-averaged picture of the catalytic events. Moreover, unduly sim~ analytical methods were employed in some cases. In an attempt to arrive at a better understanding of isoalkane/alkene alkylation catalyzed by zeolites we developed experimental techniques which we consider to be appropriate to the complexity of the system. They combine instantaneous sampling with high resolution product analysis. The present paper intends to give a detailed description of these methods along with some pertinent results obtained with three cerium exchanged zeolites of the faujasite family. 2 EXPERIMENTAL The pressure apparatus for conversion of liquid hydrocarbon mixtures on solid catalysts is depicted in Fig. 1. Essentially it consists of a vessel equipped with a magnetically driven stirrer for preparation and storage of the liquid feed mixture, a piston-type pump for pulsation-free operation at very small throughputs, and the fixed bed reactor made from stainless steel. During a run the reactor is immerged into a circulating oil bath which allows a thorough control of temperature and an effective removal of the heat of the exothermic reaction. For in-situ pretreatment of the catalyst at higher temperatures the reactor can alternatively be housed in an electric furnace. A six-port and a three-port valve are arranged in such a way that prior to an experiment the desired amount of each feed component can be pumped into the storage vessel. An automatically actuated four-way valve controls the functions of the two cylinders of the pump. By means of another four-way valve the reactor can be by-passed, e.g. for analysis of the feed mixture. In order to define the beginning of a run as accurately as possible the following procedure was applied: The feed mixture was pumped while the reactor filled with nitrogen under reaction pressure was by-passed. When the fourway valve was switched the liquid hydrocarbons replaced the nitrogen cushion. Downstream movement of the phase boundary towards the needle valve was v~ble
through pressure resistant teflon tubing.
c, He C3He:
--- ---- _...,
LIQUID FEED COMPONENTS
Scheme of the apparatus.
As shown in Fig. 2 the liquid effluent from the reactor is depressurized in a needle valve, vaporized, and diluted with nitrogen. The subsequent sampling system embraces a device for instantaneous or differential sampling with glass ampules, a cooling trap for collecting a liquid sample, and the system for on-line GC analyses.
' :' J'
GASEOUS IN GLASS
ON - LINE
Scheme of the sampling system.
The sampling system with glass ampules is an adaptation of the method first applied by Pichler and G~rtner in a study on catalytic cracking (ref. 20). The top of a glass capillary connected to an evacuated and heated ampule is immerged into the gaseous product stream. The sealing
system resembles a GC injection port. At a predetermined time the top of the capillary can be broken, e.g. by rotating, so that the ampule fills up with gaseous product. Immediately thereafter the ampule is withdrawn and sealed. In this state the sample inside the ampule can be stored without loss of any components until analysis. The whole sampling procedure can be repeated in very short intervals of one minute or less. The liquid sample is normally collected at ca. 10 °c during the period of one or several hours. Although this sample is not needed for quantitative evaluation of a run it is useful for qualitative assignment of peaks in the chromatograms. Moreover, collecting hydrocarbons at subambient temperatures avoids any condensation in the downstream system for on-line sampling. The latter contains a six-port valve with the loop and a precolumn in which remaining Cs+ hydrocarbons are removed and which can be backflushed via an eight-port valve. The main purpose of the on-line GCanalyses is to monitor the degree of butene conversion during a run. The final evaluation is based on the analyses of the samples in the ampules. For analysis an ampule is crushed in a specially designed apparatus which is heated and connected to the injection port of a capillary GLC unit with a flame ionization detector. Stainless steel columns of 100 m length and 0.25 rom internal diameter with polypropylene glycol, squalane, and OV-l0l as stationary phases are used. For satisfactory resolution a temperature program starting at subambient temperature, e.g. -15 °c along with a low heating rate of 1.0 PC/min or even less must be applied. Typically, the time required for one analysis amounts to 4 h. The products contain isobutane in a very large excess. Since the size of our samples was chosen such as to obtain maximum sensitivities for the rest of the hydrocarbons, the isobutane peak inevitably was beyond the linear range of the detector. Hence, isobutane could not be included in the material balances. Assignment of the paraffinic peaks up to C9 is based on commercially available reference SUbstances, retention index data compiled in the literature (ref. 21), and our earlier work on hydrocracking (ref. 22). No attempt was made to identify CIO or higher hydrocarbons. In order to recognize olefinic peaks additional analyses were carried out with the liquid sample in the same GLC equipment with and without application of precolumns. Both an olefin subtracting precolumn filled with a H2S04/H3P04 coated carrier and a hydrogenation precolumn filled with a Pd/A1203 catalyst were - utilized. These methods revealed that even with the high resolution capillary GLC procedures some overlap between certain octane and octene peaks occurred. Starting from NaX and NaY from Union Carbide the catalysts were prepared by conventional ion exchange cycles with an aqueous solution of 0.03 mol-% Ce( N03)3 at 80 °C. In the preparation of the highly exchanged CeY sample two intermediate calcination steps were involved as described elsewhere (ref. 23). The molar ratios Al:Ce:Na in the final catalysts were
3.30 : 1.00 : 0.13, 6.85: 1.00 : 3.55, and 2.20 : 1.00 : 0.078 for CeX-96, CeY-46, and CeY-98, respectively. The figures denote the formal degree of cerium exchange as calculated simply from the cerium and sodium contents. Whereas the aluminum/cation balance is fulfilled within c~ 5 % for CeX-96 and CeY-46 the CeY-98 sample shows an overall cation excess which, according to literature data (ref. 24), is due to the calcination steps. Each catalyst was pressed binder-free and ground to 0.25 - 0.50 rom. 2.00 cm3 of this particle size fraction were employed in each run. After in-situ pretreatment at 350 °c in a purge of dried nitrogen the mass of CeX-96, CeY-46, and CeY-98 were 1.52, 1.09, and 1.42 g, respectively. The molar ratios of the components in the feed mixtures were as follows: i-butane: 1-butene : propane 11.1 : 1.00 : 0.42 and 11.6 : 1.00 : 0.37 for CeX-96 and Ce-Y-46, respectively, and i-butane: cis-2-butene : propane 11.0 : 1.00 : 0.38 for CeY-98. Propane served as an internal standard for GC analyses. In preliminary experiments it was ascertained that propane was neither consumed nor formed during the reaction. The liquid feed rate at ambient temperature, the pressure, and reaction temperature amounted to 7.5 cm3/h, 3.1 MPa, and 80 °c, respectively. 3
Qualitative description of isobutane/butene conversion on faujasites When a zeolite of sufficiently high acidity and sufficiently large pore size, e.g. CeX or CeY, is exposed to an isobutane/butene mixture the following results can be revealed by instantaneous sampling: During an initial stage lasting typically 20-40 min under our conditions the butene is completely removed. Immediately after start-up of the reaction a complex mixture of Cs+ isoalkanes, i.e. alkylate, can be detected. Moreover, some small amounts of n-butane are formed (ref. 23). It is particularly noteworthy that the Cs+ product is entirely free from olefins or any type of cyclic hydrocarbons during this initial alkylation stage. After a certain time on stream butenes appear as a mixture of 1-butene, cis- and trans-2-butene. Almost simultaneously olefin~ are detected among the Cs+ hydrocarbons, especially isooctenes. The amount of butenes and the content of olefins in the Cs+ product increase rapidly with time on stream. In the following sections appropriate ways are outlined to describe such a highly non-stationary behaviour in a quantitative manner. 3.2
Differential and integrated yields For each-ampule drawn the yield of individual products or groups of products on butene charged can be evaluated due to the use of an internal standard. These yields in terms of gig butene charged will be referred to as differential yields. In Fig. 3 the differential yields of CS-C12 hydrocarbons and butenes as well as their sum are plotted versus time on stream for one experiment. From the liquid feed rate, the composition of the feed, the densities of the components (assuming ideal behaviour of the liquid
mixture), and the mass of catalyst the cumulative butene charge per unit mass of catalyst which has been referred to as catalyst age (ref. 11) can be calculated. It is given as a second abscissa in Fig. 3. TIME ON STREAM 1 min
100 200 300 to O;:-_ _.,..-_ _.:.;-:-_ _-e-r-_ _--=T=---_----._ _----'-'r-__..--..,
ou, ~.. :;
~ CD UJ '" >......
.,vCs - C,2 0
CATALYST AGE Ilg Butene Charged 1 g Catalyst)
Fig. 3. Conversion of isobutane/cis-2-bu~ene on CeY-98. Differential yields versus time on stream and catalyst age. Integrating the differential yields, e.g. graphically, up to a certain catalyst age gives the integrated yields in terms of gig catalyst. In the above example integrated yields of the CS-C12 hydrocarbons and of the total hydrocarbons at a catalyst age of 1.4 g butene charged/g catalyst are 0.36 and 1.175 gig catalyst, respectively. The deficiency of hydrocarbons at the selected catalyst age is then given by 1.4 - 1.175 = 0.225 gig catalyst. To account for this deficiency the following effects have to be considered: 1. Coke was formed from butene and accumulated on the zeolite. 2. Butene was converted into isobutane which could not be determined quantitatively. 3. Quantitative application of our sampling technique with glass ampules is limited to hydrocarbons up to ca. C12 • Any higher products would not be detected quantitatively even if they were desorbed from the catalyst. However, in the case of CeY-98 C13+ products were found to be absent in the liquid sample where they would have been collected if they had formed. Hydrocarbon deficiency is then due to the formation of coke and isobutane. Although for the moment we are unable to separate both effects we believe that most of the deficiency can be attributed to coke formation. 3.3
Alkylation versus alkene oligomerization. Integrated yields of alkylate In Fig. 4 the composition of the main carbon number fraction, i.e. Ca, is given in terms of the mol-fraction of octanes and octenes for the run with CeY-98. Due to peak overlap the values are less accurate at lower octane mol-fractions. The dotted part of the curve should be regarded as an upper limit of the octane mol-fraction.
z '" ~
TINE ON STREANI min
Fig. 4. Conversion of isobutane/cis-2-butene on CeY-98. Composition of the Ce-fraction. It is evident from Fig. 4 that initially alkylate is formed with an utmost degree of selectivity while late in the run the reaction is better described in terms of butene oligomerization. In order to evaluate the integrated yield of alkylate for a given experiment one must define a criterion which ranks a product as alkylate or otherwise. We consider a product to be an alkylate if its content of alkanes in the Ca-fraction is 90 % or higher. In the case of CeY-98 this requirement is fulfilled up to a time on stream of 30 min. CeX- 96
CeY - 98
CeY- 46 ISOBUTANE I I-BUTENE
ISOBUTANE I '-BUTENE
ISOBUTANE / CIS-2 - BUTENE
ON STREAM I min
m." « Ol U oll: ,J:"
a:: .. ~
> .. :E:
o C 5 -C 12
a:: ... 0
AGE I (g Butene Charged I g Catalyst I
Fig. 5. Integrated yields of alkylate for three cerium exchanged faujasites. In Fig. 5 the differential yields obtained with the three cerium exchanged zeolites are presented. In each case the end point of the curve has been
fixed according to the above criterion. Note that the catalyst age scales are somewhat different in relation to the time on stream scales reflecting mainly small differences in the mass of catalysts. Integration up to the end points of the CS-C12 curves gives integrated yields of alkylate amounting to 9, 70, and 125 mglg catalyst for CeY-46, CeY-98, and CeX-96, respectively. From these results and Fig. 5 some interesting conclusions can be drawn. Whereas according to the idealized mechanism the differential yield of alkylate should be slightly above 2 gig butene charged, the experimental values even do not reach 1 gig butene charged. Moreover, our results based on differential sampling reveal that the differential yields of alkylate pass through a maximum. Although it is too early to develop a detailed interpretation, the strong adsorption of the olefin and the accumulation of coke are considered to be important factors which influence the yield of alkylate. The integrated yields for CeY-46 and CeY-98 clearly show that a high degree of cerium exchange is desirable for isobutanelbutene alkylation. Moreover, it is found that CeX-96 is superior to CeY-98 which at least qualitatively corresponds to a higher aluminum and cation content and hence a higher number of acid sites per unit mass in the X zeolite. 3.4
Carbon number distributions Table 1 gives carbon number distributions of the Cs + products obtained with the three catalysts during their alkylation stage. It follows from these data that although Cs is the main carbon number fraction alkylation on the zeolites just as with H2S0 4 or HF as catalysts (ref. 4-7) always yields products with other carbon numbers. TABLE 1. Carbon number distributions (wt.-%) of alkylates Catalyst Time on stream (min)
Cs Cs C7 Cs Cg CIO-C12
5.7 5.9 5.9 65.9 6.1 10.5
30 22.6 11.9 9.0 47.0 6.4 3.1
6.1 5.0 5.3 53.8 8.1 21.7
26.0 14.4 10.4 45.8 2.3 1.1
8.9 7.6 7.9 49.8 7.0 18.8
The data for CeY-98 and CeX-96 reveal a distinct influence of time on stream on the carbon number distributions. In the fresh state both catalysts yield large amounts of Cs - C7 products while towards the end of the alkylation stage considerably more heavy hydrocarbons are formed. The occurrence of C~ - C7 and C9-Cll paraffins is generally explained by the so-called destructive alkylation, i.e. the intermediate formation of higher alkyl carbenium Uns with 12 or 16 carbon atoms followed by one or more steps of cracking. Obviously, the zeolites loose a great deal of their cracking
activity during on stream use. 3.5
Distributions of individual isomers Alkylation on the zeolites at 80 °c selectively leads to i-alkanes with at least one tertiary carbon atom. The n-alkanes as well as i-alkanes lacking a tertiary carbon atom (e.g. 2,2-dimethylhexane, 2,2,3,3-tetramethylbutane, etc.) were found to be absent. The only exceptions of this rule, which was found to be valid for all carbon numbers, were n-pentane and noctane which occasionally occurred in traces. In Table 2 the distributions of the octane isomers are listed for the samples already selected in Table 1. TABLE 2.
Distributions of individual Ca isomers (mol-i)
Catalyst Time on stream (min) 2-M-Heptane 3-M-Heptane + 3-E-Hexane 4-M-Heptane 2,3-DM-Hexane 2,4-DM-Hexane 2,5-DM-Hexane 3,4-DM-Hexane a 3-E-2-M-Pentane 2, 2, 3-TM-Pentane 2,2,4-TM-Pentane 2,3,3-TM-Pentane 2,3,4-TM-Pentane
1.7 0.2 9.4 3.2 0.3 45.8 2.2 3.3 3.0 16.5 14.3
0.4 0.1 4.8 6.2 3.0 6.9 0.6 4.4 22.3 27.9 23.2
0.8 0.1 11. 7 4.9 2.3 24.3 1.8 2.8 12.2 19.6 19.3
1.1 0.2 12.9 7.0 4.2 13.4 1.8 2.9 18.5 20.8 16.8
4.7 5.8 1.6 6.9 0.3 4.4 23.3 30.7 22.0
aBo t h diastereomers These data show that monobranched isooctanes are formed in almost negligible amounts. Moreover, they reveal that of the two structures predicted by the idealized mechanism as primary products of isobutane/butene alkylation, i.e. 2,2-dimethylhexane and 2,2,3-trimethylpentane~ one is absent whereas the other occurs in relatively small concentrations. While these discrepancies are usually accounted for by postulating rearrangement steps of i-octyl cations (ref. 4,5) the pronounced dependencies of the experimental isomer distributions upon both the degree of cerium exchange in the Y zeolites and time on stream are not well understood at present. We believe that beside the idealized mechanism there are principally different routes leading to i-octanes. One of these can be seen in the so-called selfalkylation of the olefin, particularly in the alkylation of sec. butyl cation with 2-butene which leads to the carbon skeleton of 3,4-dimethylhexane and to a much lesser extent in the alkylation of sec. butyl cation with l-butene which results in the formation of the skeleton of 3-methylheptane. The contribu~ion of self-alkylation is obviously more pronounced when the zeolite
is less acid, as in the case of CeY-46, and towards the end of the alkylation stage. A third route to isooctanes is the breakdown of polymeric material. Such a pathway has been claimed by Weeks and Bolton (ref. 25) to produce mainly 2,3- and 2,5-dimethylhexane as well as 2,2,4-trimethylpentane. 4
CONCLUSIONS Isobutane/alkene alkylation, which is perhaps one of the most complex catalytic reactions, requires special experimental techniques to cope with the simultaneous occurrence of non-stationary behaviour and complex product distributions. Differential sampling in intervals, which are short compared to the characteristic time of non-stationary behaviour, appears to be a powerful tool especially in combination with high resolution analyses. A systematic application of the methods outlined in this paper is under way in our laboratory and will, hopefully, lead to a more detailed picture of the low temperature chemistry of alkyl carbenium ions, including rearrangements and low temperature cleavage. Furthermore, the results can be expected to shed light on special properties of zeolite catalysts, e.g. hydride transfer power, and their response to coke deposition, which are at present very little understood. ACKNOWLEDGEMENTS I thank Mr. W. Stober for assistance with the experiments. Financial support by the German Science Foundation (Deutsche Forschungsgemeinschaft) is gratefully acknowledged. REFERENCES P.A. Jacobs, Carboniogenic Activity of Zeolites, Elsevier Scientific Publishing Co., Amsterdam, Oxford, New York, 1977, 253 pp. 2 M.L. Poutsma, in J.A. Rabo (Ed.), Zeolite Chemistry and Catalysis, Am. Chern. Soc. Monograph, Vol. 171, Am. Chern. Soc., Washington, D.C., 1976, pp. 437-528. 3 A.P. Bolton, ibid., pp. 714-779. 4 L. Schmerling, in G.A. Olah (Ed.), Friedel-Crafts and Related Reactions, Vol. II, Part 2, Interscience Publ., New York, London, Sidney, 1964, pp , 1075-1131. 5 R.M. Kennedy, in P.H. Emmett (Ed.), Catalysis, Vol. 6, Reinhold Publ. Corp., New York, 1958, pp. 1-41. 6 T. Hutson, Jr. and G.E. Hays, Preprints, Div. Petro Chern., Am. Chem. soc , , 22 (1977) 325-342. 7 L.F. Albright, ibid., 22 (1977) 391-398. 8 F.W. Kirsch, J.D. Potts and D.S. Barmby, Oil Gas J., 66, No. 29, July 15 (1968) 120-127. 9 F.W. Kirsch, J.D. Potts and D.S. Barmby, Preprints, Div. Petro Chem., Am. Chern. Soc., 13, No.1 (1968) 153-164. 10 W.E. Garwood and P.B. Venuto, J. Catal., 11 (1968) 175-177. 11 F.W. Kirsch and J.D. Potts, Preprints, Div. Petro Chern., Am. Chern. Soc., 15, No.3 (1970) A-l09 - A-121. 12 F.W. Kirsch, J.L. Lauer and J.D. Potts, ibid., 16, No.2 (1971) B-24 B-39. 13 F.W. Kirsch, J.D. Potts and D.S. Barmby, J. Catal., 27 (1972) 142-150. 14 E.S. Mortikov, S.M. Zen'kovskii, N.V. Mostovoi, N.F. Kononov, L.I. Golomshtok and Kh. M. Minachev, Izv. Akad. Nauk SSSR, Ser. Khim., 7 (1974) 1551-1554.
E.S. Mortikov, S.M. Zen'kovskii, N.V. Mostovoi, N.F. Kononovand L.I. Golomshtok, ibid., 10 (1974) 2237-2239. 16 Kh. M. Minachev, E.S. Mortikov, S.M. Zen'kovskii, N.V. Mostovoi and N.F. Kononov, Preprints, Div. Petro Chern., Am. Chern. Soc., 22 (1977) 1020-1024. 17 R. Schollner, H. Holzel and M. Partisch, Wiss. Zeitschr. Karl-MarxUniv. Leipzig, Math.-Naturwiss. R., 23 (1974) 631-641. 18 R. Schollner and H. Holzel, Journal f. Prakt. Chemie, 317 (1975) 694-704. 19 R. Schollner and H. Holzel, Zeitschr. Chern., 15 (1975) 469-475. 20. H. Pichler and R. Gartner, Brennstoff-Chem., 43 (1962) 336-340. 21 A. Matukuma, in C.L.A. Harbourn (Ed.), Gas Chromatography 1968, The Institute of Petroleum, London, 1969, pp. 55-75. 22 J. Weitkamp, in J.W. Ward and S.A. Quader (Eds.), Hydrocracking and Hydrotreating, Am. Chern. Soc. Symp. Series, Vol. 20, Am. Chern. Soc., Washington, D.C., (1975) pp. 1-27. 23 J. Weitkamp, Proc. 5th Intern. Conf. Zeolites, Naples, June 2-6, 1980, in press. 24 A.P. Bolton, in R.B. Anderson and P.T. Dawson (Eds.), Experimental Methods in Catalytic Research, Vol. 2, Academic Press, New York, San Francisco, London (1976) pp. 1-42. 25 T.J. Weeks, Jr. and A.P. Bolton, J. Chern. Soc., Farad. Trans. I, 70 (1974) 1676-1684.