Applied Catalysis A: General 197 (2000) 303309

Intrinsic kinetics of MTBE synthesis from methanol and isobutene using a synthesised MFI type zeolite
M.A. Ali, B.J. Brisdon, W.J. Thomas
Department of Chemical Engineering, University of Bath, Claverton Down, Bath BA2 7AY, UK Received 25 June 1999; received in revised form 29 October 1999; accepted 29 October 1999

Abstract The intrinsic kinetics of the reaction of methanol with isobutene, catalysed by a synthesised MFI type zeolite ZCIC-10, was studied in a batch reactor operating in the range 1520 bar and 70100 C. Intraparticle and interparticle mass transfer effects were eliminated by judicious choice of catalyst particle size and conditions of uid agitation. Orders of reaction with respect to the reactants and principal product methyl tertiary butyl ether (MTBE) were determined for the range of temperature and pressure studied and curve tting techniques were applied to the experimental concentrationtime data obtained. LangmuirHinshelwood kinetics in which the surface reaction is rate determining is the most satisfactory t to the data. The activation energy for the reaction was found to be 140.8 kJ mol1 . 2000 Elsevier Science B.V. All rights reserved.
Keywords: MTBE; Synthesis; Kinetics; MFI zeolite

1. Introduction Concomitant with the legislation and consequent demand for unleaded motor fuels, there has been a resurgence in the production of suitable additives for blending with gasoline to form high-octane petroleum spirits for use in automobiles. Additives used include methyl tertiary butyl ether (MTBE), ethyl tertiary butyl ether (ETBE), tertiary amyl methyl ether (TAME), tertiary amyl ethyl ether (TAEE) and di-isopropyl ether (DIPE). The most widely used of these additives is MTBE on account of performance and cost [1]. Commercial production of MTBE by the liquid phase re Corresponding author. Tel.: +44-1225-826-575; fax: +44-1225-826-894. E-mail address: [Link]@[Link] (W.J. Thomas)

action of methanol with butene in the presence of a sulphonated ion-exchange resin catalyst occurs below 100 C. Byproducts are di-isobutene and tertiary butyl alcohol, the amounts of which increase with increase in the reaction temperature. Disadvantages in using such an ion-exchange resin is its thermal instability and tendency to leach acid from the resin surface during reaction. The batch reaction of methanol and butylene on a particular ion-exchange resin (Amberlyst 15) was studied by Al-Jarallah et al. [2] who showed the reaction to be rst-order in methanol, half-order in isobutene and retarded by product. It was concluded that the reaction kinetics could be represented by an EleyRideal mechanism in which methanol, adsorbed on the resin, reacted with butylene absorbed by the liquid methanol. A similar mechanism was reported

0926-860X/00/$ see front matter 2000 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 6 - 8 6 0 X ( 9 9 ) 0 0 4 9 3 - 7

304

M.A. Ali et al. / Applied Catalysis A: General 197 (2000) 303309

by Caetano et al. [3]. On the other hand, studies with a continuous differential reactor using a Bayer resin catalyst K2631 in the liquid phase showed the reaction to be rst-order in isobutene [4]. The vapour phase synthesis of MTBE using the Amberlyst 15 resin indicated that the catalyst is sensitive to methanol concentration [5]. Analysis of reported researches of the vapour phase synthesis of MTBE over resin catalysts showed that LangmuirHinshelwood kinetics tted the experimental results [611]. Notwithstanding the apparent discrepancies in mechanisms reported for the MTBE synthesis on resin catalysts in liquid and vapour phases, several attempts have been made to seek more stable and active catalysts using zeolitic and other microporous materials. Chu and Kuhl [12] investigated the synthesis of MTBE using a variety of zeolite catalysts, including ZMS-5, and Hoelderich [13] published a critical review of the use of zeolites for the formation and cleavage of CO bonds in ne chemicals in general. More recently, Chang et al. [14] re-emphasized the potential of these materials for catalytic MTBE formation. The study reported in this paper relates to the use of an MFI type synthesised zeolite (ZCIC 10) with a silicon to aluminium molar ratio of 10. This ratio of Si/Al was shown [15] to give the highest yield of MTBE for a number of zeolite preparations with Si/Al ratio ranging from 10 to 100.

ratio of this catalyst was 10 : 1 and it had a surface area of 321 m2 g1 and a median pore diameter of 5 105 m. Following the introduction of a measured volume of methanol and mass of catalyst into the autoclave, initial temperature was set and the vessel pressurised to a pressure corresponding to the selected reaction temperature and at which the isobutylene would be in the liquid phase (1520 bar). A measured volume of liquid isobutene was then fed by nitrogen displacement to the vessel contents via a preheated inlet tube and valve. Progress of the reaction was monitored by collecting, at predetermined time intervals, samples of the reaction mixture in a stainless steel sampling tube. These samples were weighed and chemically analysed using a high-pressure gas chromatography column. The HP chromatographic column operated at 30 C was a 50 m 0.32 mm glass capillary with a 0.5 m lm thickness of polydimethylsiloxane stationary phase.

3. Results and discussion Intrinsic kinetics are often obscured or completely masked by intra and interparticle mass transport effects. It was therefore considered necessary to search for experimental conditions which would at least minimise or, at best, eliminate transport resistances. Suitable conditions which normally minimise interparticle transport effects in a batch reactor are vigorous uid agitation. Intraparticle diffusion effects may be avoided by judicious choice of catalyst particle size. Small sized particles are preferred, but too small a size can lead to loss of particulates during sampling. Once appropriate experimental conditions have been chosen, intrinsic kinetics would be observed. 3.1. Elimination of transport resistances Interparticle transport effects were examined by measuring the initial reaction rate at three different speeds of rotation of the uid impeller. The data represented in Fig. 1 was obtained by rst observing the increase in MTBE concentration as a function of time for each of the stirring speeds adopted. Each one of the three curves were then tted to a polynomial and

2. Experimental The vessel in which the reaction between liquid methanol and butylene was effected consisted of a closed batch autoclave equipped with an impeller and internal cooling coil. The reactor head assembly was provided with a gas inlet valve, a tube for taking liquid samples from the reaction mixture, a gas sampling valve and a thermowell. The complete autoclave was electrically heated and the internal pressure registered with a pressure gauge. Temperature of the reacting mixture was controlled by circulating oil through the internal cooling coil to remove heat produced by the exothermic reaction. The MFI type zeolite catalyst was prepared by rapid crystallisation, and calcined at 873 K in air, prior to ion-exchange to remove sodium ions followed by further calcination. The Si/Al

M.A. Ali et al. / Applied Catalysis A: General 197 (2000) 303309

305

the criterion for minimal intraparticle diffusion resistance. It was therefore concluded that the prepared zeolite of particle size 5 105 m was appropriate for determining the intrinsic kinetics of MTBE synthesis. 3.2. Order of reaction The reaction orders with respect to each of the components methanol and isobutene were determined by observing the initial rate of MTBE formation with, in turn, one of reactants in excess. Representing the rate of reaction by a power law, the net rate of formation of MTBE would be dcC a b c = k cA cB k cC dt where C represents MTBE, A methanol and B isobutylene. The indices a, b, and c are the reaction orders with respect to each component. Initial rates were evaluated so that the inuence of the product C was negligible. A plot of log (initial rate) versus the concentration of the component not in excess therefore yields the indices a and b. Initial rates were found by computing gradients, determined at zero time, to polynomial representations of MTBE concentration as a function of time. Figs. 2 and 3 yield average slopes of 0.94 for methanol and 0.95 for isobutylene for the four

Fig. 1. Initial reaction rate as a function of impeller speed.

differentiated with respect to time to obtain an initial reaction rate for each speed of agitation. It is clear from Fig. 1 that there is a decrease in initial reaction rate below stirring speeds of approximately 950 rpm, indicating that transport of reactants from uid to solid particulates has an effect on the intrinsic kinetics of reaction. Above 1000 rpm, however, the effects become negligible. The impeller rotational speed was thus chosen as 1000 rpm in order to measure kinetics unimpeded by uid to solid transport resistances. Resistance to intraparticle diffusion, on the other hand, may be reduced to negligible proportions by using sufciently small catalyst particles. The criteria for observing chemically controlled reaction rates is that the Thiele modulus , in terms of measurable parameters, should be [16] = dp 2 R cDe
1/2

< 0.2

where dp is the catalyst particle size (m), R the reaction rate (measured on the basis of unit mass of catalyst) and c the concentration expressed as moles per unit catalyst mass (mol g1 ). Utilising conservative values from the preliminary experiments [15] with dp = 104 m, De (for Knudsen diffusion in a microporous zeolite [17]) = 107 m2 s1 , R = 15 103 mol g1 s1 and c = 102 mol g1 ; the value of the Thiele modulus is 0.003, well within

Fig. 2. Loglog plot of initial rate of MTBE formation as a function of methanol concentration.

306

M.A. Ali et al. / Applied Catalysis A: General 197 (2000) 303309

Fig. 3. Loglog plot of initial rate of MTBE formation as a function of isobutylene concentration.

kinetics in the temperature range studied. It is therefore assumed, purely for pragmatic purposes, that a LangmuirHinshelwood or EleyRideal kinetic model is a reasonable starting point for matching experimental results with a mathematical model. Bearing in mind that experiments to determine reaction orders from measured initial reaction rates showed that the order with respect to each reactant is 1 (i.e. initially a second-order reaction rate), ve rate models of the LangmuirHinshelwood and EleyRideal type which would conform to second-order initial kinetics were examined. Parameters in each of ve models examined [15] were (i) the surface reaction rate constant KS , (ii) the adsorption equilibrium constants KA , KB and KC , and (iii) the thermodynamic equilibrium constant K and (iv) the reaction orders a, b and c. Values of a, b and c were chosen over the range 0.53.0 while the thermodynamic equilibrium constant was calculated from the equation lnK = 4.8 + (7300/T ) + 1.17 102 T 4.8 106 T 2 + 2.5 109 T 3 previously utilised by others [10,11,12]. A computer software package Sigma Plot was used for model parameter estimation. The software

temperatures of operation. The reaction order with respect to each of the reactants is therefore virtually unity. 3.3. The reaction prole For each of the four reaction temperatures studied, the variation of concentration as a function of time was observed for both the two reactants and the product. Fig. 4 is a typical prole obtained at 100 C and initial concentrations of 1.0 mol g1 methanol and 10 mol g1 isobutene. Similar proles were obtained for 70, 80 and 90 C. In each case, the reaction was allowed to proceed until the MTBE concentration was asymptotic to its equilibrium value. All of the proles were curve tted by a cubic polynomial of the form cC = ( + t + t 2 + t 3 ) mol g1 Values of the constants , , and which were found are given in Table 1. The polynomials were subsequently differentiated to give reaction rates as a function of time to produce observed reaction rates for matching against those predicted by a suitable model. The object of the rate measurements conducted was not to establish a molecular mechanism but to search for a suitable expression which will represent the

Fig. 4. Prole of reaction showing component concentrations as a function of time.

M.A. Ali et al. / Applied Catalysis A: General 197 (2000) 303309 Table 1 Constants in polynomial tting MTBE concentration vs. timea T (K) 343 353 363 373
a

307

104 mol g1 4.94 4.15 15.0 22.0

103 mol g1 h1 4.10 6.80 12.4 19.3

104 mol g1 h2 1.63 7.93 35.1 79.0

10 mol g1 h3 2.19 3.68 30.4 80.4

Initial concentrations: 1 mol g1 methanol, 10 mol g1 isobutene; temperature: 100 C.

nds the coefcients of the independent variables which give the best t between the chosen model and the experimental data expressed in terms of rate-concentrations data. The software programme is interactive and it was run until convergence was obtained. A non-linear least squares regression algorithm contained within the software enabled residuals , which are the difference between estimated and actual values of parameters, to be calculated and the square root of the absolute sum of squares of residuals minimised. Of the ve models tested, a LangmuirHinshelwood model with rate determining surface reaction between adsorbed methanol and adsorbed isobutene gives the most satisfactory agree 2 ranging from ment with the observed results, 3 to 8.9 103 , thus representing a very 1.63 10 good t. Adsorption of each reactant on adjacent surface sites might possibly occur by virtue of bonding between the oxygen of methanol and the hydrogen of isobutene to hydrogen and oxygen sites, respectively, of the acidic silica. The rate expression for this model is
a b rA = KS KA KB a b c (cA cB cC )/K (1 + KA cA + KB cB + KC cC )a+b

1.5 found by Al-Jararallah et al. [2] for an acid resin catalyst. This result, although inconsistent with the principle of microscopic reversibility, merely represents an artefact of the computational modelling rather than any mechanistic revelation. Fig. 5 demonstrates that the surface reaction rate constant KS increases with temperature as required, the constant obeying the equation KS = 3.92 1019 exp 14935 T mol g1 s1

The activation energy for reaction extracted from the above equation is 140.8 kJ mol1 . This value may be compared with an activation energy of 130.2 kJ mol1 found for the batch synthesis of MTBE from methanol

Criteria for model discrimination included (i) the rate constant KS and the adsorption equilibrium constants KA , KB and KC should be positive; (ii) the constant KS should increase with increase in temperature; (iii) the adsorption equilibrium constants KA , KB and KC should decrease with increase in temperature in accordance with the thermodynamic requirement of the vant Hoff equation. The best t to the experimental data was obtained for a = b = 1, as indeed, the experiments designed to ascertain reaction orders revealed. The best value for c was 2.5 in contrast to the value of

Fig. 5. Variation of rate constant KS with absolute temperature T.

308

M.A. Ali et al. / Applied Catalysis A: General 197 (2000) 303309

Fig. 6. Variation of adsorption equilibrium constant KA with absolute temperature T.

Fig. 8. Variation of adsorption equilibrium constant KC with absolute temperature T.

KB = 5.14 1023 exp KC = 1.74 1019 exp

19170 T 14522 T

g mol1 g mol1

4. Conclusions The experimental work reported demonstrates that an MFI synthesised zeolite (ZCIC 10) with a silicon to aluminium molar ratio of 10 is a good catalyst for the synthesis of MTBE from isobutylene and methanol. In a batch reactor operating in the temperature range 70100 C, and pressures between 15 and 20 bar, equilibrium concentrations of the reaction component are established within 50 h. Analysis of the time dependent concentration of each reaction component in the approach to equilibrium indicates that the synthesis reaction can be described by a LangmuirHinshelwood model, the parameters of which are comparable with previously reported work [2,3]. A recently published paper [18] on the other hand reports apparent activation energies of only 32 kJ mol1 for the synthesis reaction and 51 kJ mol1 for the decomposition of MTBE. These values were

Fig. 7. Variation of adsorption equilibrium constant KB with absolute temperature T.

and isobutylene [3]. Figs. 68 show that KA , KB and KC all decrease with temperature and are represented by KA = 1.36 1019 exp 16492 T g mol1

M.A. Ali et al. / Applied Catalysis A: General 197 (2000) 303309

309

obtained using a temperature relaxation technique using solid NMR spectroscopy in which conditions were neither isothermal nor isobaric, and hence, do not provide a fair comparison. For batch liquid phase experiments, values of 130.2 kJ mol1 obtained for an Amberlyst resin catalyst [3] and 92.4 kJ mol1 for a different Amberlyst resin catalyst [19] compare more favorably with the results quoted in this paper. References
[1] R. Csikos, I. Pallay, B. Torck, in: Proceedings of 10th World Petroleum Congress, Bucharest, Vol. 10, No. 5, 1980, p. 107. [2] A.M. Al-Jarallah, M.A.B. Siddiqui, A.K. Lee, Can. J. Chem. Eng. 66 (1988) 802. [3] N.S. Caetano, J.M. Loureiro, A.E. Rodriguez, Chem. Eng. Sci. 49 (1994) 4589. [4] D. Parra, J. Tejero, F. Cunill, M. Iborra, J.E. Izquierdo, Chem. Eng. Sci. 49 (1994) 4563.

[5] A. Giquel, B. Torck, J. Catal. 83 (1983) 9. [6] S. Subramanian, S. Bhatia, Can. J. Chem. Eng. 45 (1987) 613. [7] A. Ali, S. Bhatia, Chem. Eng. J. 44 (1990) 97. [8] T. Zhang, R. Dutta, Ind. Eng. Chem. Res. 34 (1995) 730. [9] J. Tejero, F. Cunill, M. Iborn, J. Mol. Catal. 42 (1987) 257. [10] J. Tejero, F. Cunill, J.F. Izquierdo, Ind. Eng. Chem. Res. 27 (1988) 338. [11] J. Tejero, F. Cunill, J.F. Izquierdo, Ind. Eng. Chem. Res. 28 (1989) 1269. [12] P. Chu, G.H. Kuhl, Ind. Eng. Chem. Res. 26 (1987) 365. [13] W.F. Hoelderich, Catal. Sci. Technol. 1 (1991) 31. [14] K.-H. Chang, G.T. Kim, W.S. Ahn, Ind. Eng. Chem. Res. 31 (1992) 125. [15] M.A. Ali, Ph.D. Thesis, University of Bath, Bath, 1998. [16] J.J. Carberry, Chemical and Catalytic Reaction Engineering, McGraw-Hill, New York, 1976. [17] C.N. Sattereld, Mass Transfer in Heterogeneous Catalysis, MIT Press, Cambridge, 1970. [18] T. Mildner, H. Ernst, D. Freude, W.F. Hlderich, J.A.C.S. 119 (1997) 4258. [19] A. Rehnger, U. Hoffman, Chem. Eng. Sci. 45 (1990) 1619.