Comparative study of TAME synthesis on ion-exchange resin beads and a fibrous ion-exchange catalyst

Transcription

1 Reactive & Functional Polymers 55 (2003) locate/ react Comparative study of TAME synthesis on ion-exchange resin beads and a fibrous ion-exchange catalyst Paivi K. Paakkonen *, A. Outi I. Krause Laboratory of Industrial Chemistry, Helsinki University of Technology, PO Box 600, Helsinki, 0205 HUT, Finland Received August 2002; received in revised form November 2002; accepted November 2002 Abstract The reaction rates to tert.-amyl methyl ether (TAME) with the ion-exchange resin bead catalysts (A6, A35 and XE586) and a fibrous catalyst (SMOPEX-0) were measured as a function of temperature ( K) with stoichiometric amount of reagents fed to a continuous stirred tank reactor. When the reaction rates were assessed against the weight or acid capacities of the catalysts, the activity order was A35. A6. SMOPEX-0. XE586. When the rates were calculated versus the square of the acid capacity, the activities of the catalysts were similar. This indicated a dual-site mechanism. The rates of TAME formation and the isomerisation of isoamylenes as a function of temperature ( K) and the feed MeOH/ isoamylene molar ratio ( ), as well as the decomposition of TAME, were then measured in a batch reactor with the ion-exchange fibre (SMOPEX-0) as catalyst. Kinetic modelling results favoured a single-site mechanism for the isomerisation and a dual-site mechanism for the etherification. The activation energy was determined to be 6.7 kj mol for the isomerisation of 2MB to 2M2B, 92.7 kj mol for the etherification of 2MB to TAME and 93.0 kj mol for the etherification of 2M2B to TAME Elsevier Science B.V. All rights reserved. Keywords: Cationic ion-exchange resins; Isoamylenes; Etherification; tert.-amyl methyl ether (TAME); Reaction kinetics. Introduction copolymers of divinylbenzene (DVB) and styrene, where sulfonic acid is the active site The reformulated gasoline component TAME (Brondstedt acidity). The amount of divinylben- (tert.-amyl methyl ether, 2-methoxy-2- zene varies from 2 to 20 wt%, and determines methylbutane) is commercially manufactured in directly the degree of crosslinking and rigidity liquid phase with use of strongly acidic macro- of the structure. That is, the more DVB in the porous ion-exchange resin beads as catalyst. catalyst the more rigid is the resin and its The conventional ion-exchange catalysts are macroporous structure. The bead catalysts have a bidisperse structure, however, with microporous gel phase present as well. The beads swell *Corresponding author. Tel.: ; fax: in polar solvents, those with less DVB swelling address: paakkonen@polte.hut.fi (P.K. Paakkonen). more than those with more DVB. The catalytic /02/$ see front matter 2003 Elsevier Science B.V. All rights reserved. doi:0.06/s38-548(02)

2 40 P.K. Paakkonen, A.O.I. Krause / Reactive & Functional Polymers 55 (2003) activity of the resins is determined by the the feed composition and also were resin dependent. amount of sulfonic acid; typically this varies The kinetic measurements with partially from 4.8 (e.g. conventionally sulfonated Am- neutralised resins showed that with Amberlyst berlyst 5) to 5.2 mmol g (e.g. hyperconstant 5 for example, the dependence of the rate sulfonated Amberlyst 35) []. on the concentration of sulfonic acid The novel ion-exchange fibre catalyst groups was almost third order, in agreement SMOPEX-0, manufactured by Smoptech, is with the results of Ancillotti et al. [5]. prepared by supporting sulfonic acids on polycrocalorimetric Rhodes et al. [6] applied a flow mi- (ethylene-graft-polystyrene). The catalyst has technique to characterise the been studied previously in our laboratory in the acidities of macroporous and gel-type resin etherification of C -alkenes [2] and shown to catalysts by ammonia adsorption from the gas 8 have favourable properties in the synthesis of phase. One of the test reactions was TAME ethers bulkier than TAME. SMOPEX-0 was synthesis. The results showed that catalytic superior to Amberlyst 35 in the formation of activities were not related in a simple way to the 2-methoxy-2,4,4-trimethylpentane owing to its concentrations of acid sites measured by amopen structure and thus negligible mass-transfer monia adsorption, but there was a noticeable limiting effect. correlation between the strengths of the acid Several groups have studied the ion-exchange sites measured in this way and catalytic acresins used as catalysts for etherification resite strength can be a useful indicator of activity tivities of the resins. They concluded that acid actions. Parra et al. [3] compared the catalytic activity of 2 different ion-exchange resins as for this type of acid catalyst. catalysts for MTBE synthesis in liquid phase. Previously, we studied the kinetics of TAME Experimental data showed that the ion-exformation with Amberlyst 6 (2% DVB, acid capacity 5.0 mmol g ) and introduced a kinetic changers with higher acidic capacity were the model of the Eley-Rideal type [7] and a revised most active; thus, the rate of MTBE formation model of the Langmuir-Hinshelwood type [8] was 29% greater with Amberlyst 35 than with for the synthesis of TAME. In the present study Amberlyst 5. Different polynomials containing we compare the activities of different commerthree variables (acid capacity, BET surface area cial cationic ion-exchange resin beads from and average pore diameter) were fitted to the Rohm&Haas and a fibre catalyst from Smopdata and the best one statistically showed that tech. Kinetic modelling for the synthesis of the acid capacity had a stronger influence on the TAME with SMOPEX-0 was carried out to catalytic activity than the other two variables. compare the values of the kinetic parameters The authors concluded that the resins with with those obtained with Amberlyst 6. Mechgreater density of the sulfonic groups were more anistic conclusions are drawn on the basis of the active because the reaction of MTBE synthesis experimental steady state reaction rates obtained involves a concerted mechanism in which adjawith the different catalysts and the kinetic cent sulfonic groups participate. Pannemann and modelling results. Beenackers [4] have reported initial reaction rate constants for MTBE synthesis with several macroporous resins and demonstrated significant differences between the resins. In their 2. Experimental view, the catalytic activity depends on the three- 2.. Part I. Experiments with different dimensional structure of the resins (level of catalysts crosslinking, porosity, surface area) and on the sulfonic acid content and its distribution within The reaction rates to TAME with bead the particles. The activation energies varied with catalysts (A6, A35 and XE586) and fibre

3 P.K. Paakkonen, A.O.I. Krause / Reactive & Functional Polymers 55 (2003) catalyst (SMOPEX-0) were measured at dif- mixture of isoamylenes (Fluka Chemika, 2M2B, ferent temperatures ( K) with reagents technical grade) with a composition of 2M2B 93 fed in stoichiometric amount to a continuous wt% and 2MB 7 wt%, p.a. grade MeOH stirred tank reactor (CSTR). (Riedel-de Haen, 99.8 wt%), and p.a. grade isopentane (Fluka Chemika). The ether, TAME, 2.2. Part II. Batch reactor experiments with was supplied by Fortum Gas&Oil and the purity SMOPEX-0 was 98 wt%. The rates of TAME formation and the iso Catalysts merisation of isoamylenes with SMOPEX-0 as catalyst were measured as a function of Commercial macroporous cation ion-extemperature ( K) and the feed MeOH/ change beads in hydrogen form (Amberlyst 6, isoamylene molar ratio (0.5,.0, 2.0) in a batch Amberlyst 35, XE586 from Rohm&Haas) and reactor. The decomposition of TAME to MeOH ion-exchange fibre (SMOPEX-0 from Smopand isoamylenes was also measured. Isopentane tech) were used as catalysts in the experiments. was used as solvent. The experiments are SMOPEX-0 is prepared by grafting a polypresented in Table. ethylene fibre with styrene and sulfonating the grafted fibre with chlorosulfonic acid. The 2.3. Chemicals SMOPEX-0 was ground to a very thin powder and 0.7 g of the powder was placed in the The following reagents were used in the reactor as slurry ( g in the TAME experiments: p.a. grade 2-methyl--butene decomposition experiments). Before the experi- (2MB, Aldrich, 99.8 wt%), redistilled 2- ments, the catalysts were washed with methanol methyl-2-butene (2M2B, Aldrich, 99 wt%), (bead catalysts) or ethanol (fibre catalyst). The Table Description of the batch reactor experiments carried out with SMOPEX-0 No. T (8C) Reactant MeOH/ alkene Molar fraction in the feed molar ratio MeOH Alkene Isopentane Ether 60 IA IA IA IA IA IA IA IA IA IA IA TAME TAME TAME MB MB MB MB M2B M2B M2B

4 42 P.K. Paakkonen, A.O.I. Krause / Reactive & Functional Polymers 55 (2003) of the catalysts was checked by repeating the experiment at the first temperature after comple- tion of the experiments at the other tempera- tures. The catalysts showed no deactivation. Depending on the temperature intervals, it took 2 4 h to obtain a steady state in CSTR. bead catalysts were stored in methanol and the fibre catalyst was dried in an oven (00 8C) to remove moisture and other impurities and stored dry in a desiccator. The properties of the catalysts are summarised in Table Apparatus and procedures Part I. CSTR Part II. Batch reactor The reaction rates to TAME with the bead The kinetic experiments with SMOPEX-0 3 catalysts (A6, A35, XE586) and the fibre were carried out in a batch reactor (80 cm, catalyst (SMOPEX-0) were measured at 323, stainless steel). The reaction mixture ( 50 g 343 and 353 K. Reagents were fed in stoichioperature was controlled within K by initially) was stirred magnetically, and the temmetric amount to a continuous stirred tank 3 reactor (CSTR, 55.6 cm, stainless steel) where immersing the reactor in a thermostated water the reaction mixture was magnetically stirred. bath. To guarantee liquid-phase operation at The stirrer speed was set to 950 rpm to elimievery temperature, the pressure was set at 0.7 nate the influence of external diffusion control MPa. Vertical mixing baffles on the reactor walls on the reaction rates [9]. The A6 (0.295 g ensured complete mixing. The samples ( dry) and XE586 ( g dry) were placed in a g) were removed manually via a sample valve metal gauze basket (60 mesh), and the fibre from the top of the reactor. The valve and the catalyst ( g dry) and A35 (0.393 g dry) sample were cooled with ice during the sam- were introduced to the reactor as slurry. The pling in order to avoid evaporation. The samples temperature ( K) was controlled within were taken as a function of contact time 60.2 K by immersing the reactor in a thermotook from 3 to 6 h depending on the feed (amount of catalyst 3 time), and experiments stated water bath. The pressure was kept constant at 0.7 MPa to ensure liquid-phase operation (shorter times for TAME decomposition). at all temperatures. The pulse-free flow rate ( g h ) of the feed was controlled by 2.6. Analysis a liquid mass flow controller. A Mettler PM 6000 balance was used to measure the actual The products were analysed with a Hewlettflow at the outlet of the reactor system. The Packard gas chromatograph 5890 Series II, compositions of the feed and the reactor effluent equipped with a flame ionisation detector and were analysed on-line with a gas chromatograph using a HP 3396A integrator. The compounds equipped with an automated liquid sampling were separated in a glass capillary column DB- valve. During the experiments the effect of (length 60 m, film thickness.0 mm, column temperature was varied randomly. The stability diameter mm; J&W Scientific). During Table 2 Properties of the catalysts A6 A35 XE586 SMOPEX-0 Crosslink level Medium High Medium Exchange capacity (mmol g ) Surface area (m g ) Average pore diameter (A) Porosity Particle diameter (mm) Mean: 0.7 Powder

5 P.K. Paakkonen, A.O.I. Krause / Reactive & Functional Polymers 55 (2003) the experiments, the response factors were (FT,out 2 F T,in) regularly checked using calibration solutions. r 5]]]]] obs f W H g (w T,out 2 w T,in)m~ tot 5 ]]]]]] 3. Results and discussion MW T catfh g (2) 3.. Part I. Comparison of catalysts (FT,out 2 F T,in) robs 5]]]]] 2 WcatfH g In a search for the kinetic mechanism of (w T,out 2 w T,in)m~ TAME formation, we carried out experiments tot 5 ]]]]]] 2 (3) with three types of beads and also with a MW T catfh g powdered fibre catalyst. Steady-state measurements were carried out with Amberlyst 6, These equations take into account the number Amberlyst 35, XE586 and SMOPEX-0. The of sulfonic acid groups in the different catalysts Amberlyst 6 was sieved and the Amberlyst 35 according to a first order (2) and second order was crushed and sieved to very small particles (3) dependence. The activities of the catalysts (diameter mm). The XE586 was used are compared in Fig. a c. unsieved, because in this catalyst the acid sites Fig. a and b shows that A35 is the most are located solely in permanent macropores, active catalyst, especially at elevated tempera- whereas A35 and A6 have an active gel phase tures ($70 8C), when the reaction rates are with micropores present too. calculated per catalyst mass or sulfonic acid The catalyst weight based reaction rate r concentration. This must be due to hyperobs (mol kg s ) to the product TAME was sulfonation []. The next most active catalyst is cat calculated from the measured amount of TAME A6, and the fibre catalyst shows moderate (g) in the product stream (weight fraction of activity. Activity was surprisingly low for the TAME in the outlet 3 total mass of the stream surface sulfonated XE586. Beforehand we had (g s ), because W is zero since TAME was thought that XE586 might be very active since T,in not fed to the reactor) divided by the molecular the active sites are located in easily accessible mass of TAME and the weight of the dried macropores and there should be no intraparticle catalyst according to Eq. (): diffusion limitations. Since it turned out that this was not the case, and since the fibre catalyst showed only moderate activity, we were forced (FT,out 2 F T,in) r 5]]]]] to conclude that the density of the active sites is obs Wcat the key parameter in rendering a catalyst suit- (w 2 w )m~ able for etherification, or for other acid-cata- T,out T,in tot 5 ]]]]]] () lysed reactions in organic phase, such as dimeri- MW T cat sation. The most interesting observation is that, when the rates are calculated according to Eq. To compare the influence of the sulfonic acid (3), i.e. when the rate is expressed in inverse concentration of the different catalysts on the proportion to the square of the sulfonic acid observed reaction rates, the reaction rate to concentration, all catalysts showed similar ac- TAME (s ) was also calculated by further tivity, except for XE586, which was slightly dividing Eq. () with the exchange capacity more active (Fig. c). This result suggests that (mol kg ) of the studied catalysts according to the etherification reaction proceeds via a dual- Eq. (2) or with the square of the exchange site mechanism, at least in stoichiometric con- 2 capacity (mol kg ) according to Eq. (3): ditions. This is a finding that we surely did not cat

6 44 P.K. Paakkonen, A.O.I. Krause / Reactive & Functional Polymers 55 (2003) Fig.. Comparison of the catalysts: (a) weight-based reaction rate, (b) reaction rates against acid capacity, and (c) reaction rates against square of the acid capacity.

7 P.K. Paakkonen, A.O.I. Krause / Reactive & Functional Polymers 55 (2003) expect beforehand, and obviously it would be interesting to compare the catalysts also in other than stoichiometric conditions. In this study only the effect of temperature was studied, but temperature is kinetically were strong variable, so that these results in our opinion are very intriguing. This surely merits further study Part II. Batch reactor experiments with SMOPEX-0 The calculation of the initial rates was made by regression from the slopes of the initial General observations straight lines of the experimental ether and Two types of main reactions of isoamylenes 2MiB (i5,2) concentrations (mmol) as a took place simultaneously in the experiments: function of contact time (h g dry cat). The calcu- the etherification with the alcohol and the lated initial rates of the reactions at 333, 338, isomerisation of the double bond between the a- 343 and 353 K are presented in Table 3. At the and the b-positions (feed: 2MB) or between temperatures studied, the initial rate of ether the b- and the a-positions (feed: 2M2B). The formation was about twice that of isomerisation, same was also observed earlier with other when an equimolar ratio of methanol and 2MB catalysts [7,0]: was used as a feed, whereas the initial rate of etherification was 0 times that of isomerisation when an equimolar ratio of methanol and 2M2B was used. The etherification of 2MB was two to three times as fast as the etherifica- (4a) tion of 2M2B. The decomposition of TAME (4b) (5) Table 3 Initial rates for the formation of products in main and side reactions and conversions of alkenes after 6 h No. Temp., TAME, 2M2B, 2MB, TAOH, DIA, Conv. (%) K mmol h g mmol h g mmol h g mmol h g mmol h g after 6 h cat cat cat cat cat

8 46 P.K. Paakkonen, A.O.I. Krause / Reactive & Functional Polymers 55 (2003) with dilution of 50 mol% of isopentane was about four times as fast as the formation of TAME at stoichiometric conditions with dilution of 0 mol% of isopentane. These general trends are in line with earlier observations by (8) Rihko et al. [7] made with Amberlyst 6 as catalyst in TAME synthesis. The tert.-etherification rate of the isoamylenes was times that of the dimerisation Side reactions of the isoamylenes to form DIA. The Formation of tert.-amyl alcohol (TAOH) dimerisation has been observed by Rihko et al. from the isoamylenes and water was detected in [2] with Amberlyst 6 only at low methanol/ the experiments. Because the catalyst was dry alkene ratios. The rates are collected in Table 3. when placed in the reactor, the water needed for the formation of the tert.-amyl alcohol must Conversions have come from the condensation reaction of Conversion of isoamylenes to TAME in methanol: experiments (Table 3) was calculated according to Eq. (9): (6) ntame Conversion to TAME after 6 h 5 ]] (9) n IA,feed Thus, consequently: (7a) Comparison of experiments 4 7 and 8 shows that the conversion to TAME is increased with excess methanol in the feed (Table 3). However, the initial etherification rate to TAME is decreased with excess methanol. In experiments 5 conversions were also calculated by Eq. (0): Overall conversion after 6 h ntame n2mjb 5 ]]]]] (i 5,2 and j 5 2,) n 2MiB,feed (0) (7b) The calculated initial rates are presented in Table 3. The tert.-etherification rate of isoamylenes was times that of the hydration rate of the isoamylenes to form TAOH. With Amberlyst 6 the rate of tert.-etherification was times that of the hydration in the study by Kiviranta-Paakkonen et al. []. Dimerisation of isoamylenes was detected only at higher temperatures ($343 K) and in the experiments where alkenes were fed in excess (MeOH/ IA50.5): Fractional conversion to TAME was greater from 2MB than from 2M2B. Overall conversion of 2MB was greater than the overall conversion of 2M2B (Table 3). These trends agree with the earlier results obtained with A6 as catalyst in the study of Rihko et al. [7] and A5 as catalyst in the study of Krause and Hammarstrom [0] Kinetic modelling Two kinetic models were tested against the data obtained with SMOPEX-0 in the batch reactor. The Eley-Rideal-type model in Eqs.

9 P.K. Paakkonen, A.O.I. Krause / Reactive & Functional Polymers 55 (2003) () and (2) assumes that only the alcohol and at the outlet of the reactor. Altogether 4 the ether but not the alkene are adsorbed on the samples (points of observation) were taken single acid site of the catalyst [7]: during the experiments described in Table. Weight factor (w i) was used for all the rtame 5 compounds (Eq. 8). at at ka MabS2 Dka 3 Ma2bS2 D 2 Ka Mab Ka 2 Ma2b WSRS 5O(n exp 2 n calc) w i (8) ]]]]]]]]]]]]] K S ] T a a K T M M D The activity coefficients were calculated by () the Wilson method; the binary interaction parameters were taken from Rihko-Struckmann et a2b al. [5]. The estimated parameters for the two r 5 kas2 ]] D (2) 3 b mechanisms were the rate coefficients k, k and 3 k5 at 333 K (mol kgcat s ) and their corre- The Langmuir-Hinshelwood-type model in sponding Arrhenius-type activation energies E, Eqs. (3) and (4) assumes that the alcohol and E2 and E 3 (J mol ). In addition, the ratio of the alkene are adsorbed on two adjacent acid adsorption equilibrium constants K T/KM at 333 sites, but the adsorption of alcohol is dominant K was estimated for the Eley-Rideal-type [8,3,4]. The simple model derived from the model. dual-site expression states that the rate of etherification is proportional to the activity ratio of Modelling results and comparison with isoamylenes and methanol: results obtained with Amberlyst 6 9 ka a The estimated parameters for the two kinetic b T r 5]] S 2]]] D a a a K models are presented in Table 4. M b M 9 The smaller residual sum of squares (RSS) ka 3 2b at ]] 2 ]]] (3) for the dual-site mechanism than for the singlea S a a KD M 2b M 2 site mechanism shows that the dual-site mechanism is more appropriate. Also, for three of the 9 ka 5 b a2b r 5]] S 2 ]] D (4) parameter values the standard error (SE %) is ISOM am abk3 one or two percentage points smaller with the The equilibrium compositions have been rethe Langmuir-Hinshelwood mechanism than with ported by Rihko-Struckmann et al. [5]. The Eley-Rideal mechanism. This is in line with temperature dependencies of the equilibrium the earlier results we obtained with Amberlyst constants are presented in Eqs. (5) (7): 6 [8] and with the results presented in Part I of this study. The ratio of the adsorption equilib- K 5 exp( /T ) (5) rium constants (K T/K M) for the ER mechanism has a value much smaller than one, which K2 5 exp( /T ) (6) means that the alcohol is the dominant molecule to adsorb on the acid site of the catalyst. This is K 5 K /K (7) in contrast to the finding of Karinen and Krause 3 2 [6] for C -alkene etherification with the fibre 8 The estimation was carried out with the catalyst. In C8 etherification, the ratio obtained KINFIT estimation program by minimising with was the value of 2.5 in the parameter estimation procedure, which indicated that the ether ad- sorbs more readily than the alcohol on the fibre catalyst site. the Levenberg-Marquardt method the weighted sum of residual squares (WSRS) between the experimental and calculated compositions (mol)

10 48 P.K. Paakkonen, A.O.I. Krause / Reactive & Functional Polymers 55 (2003) Table 4 Modelling results with SMOPEX-0 as catalyst Parameter Eley-Rideal model, Langmuir-Hinshelwood model, 95% confidence limits 95% confidence limits Value SE % Lower Upper Value SE % Lower Upper k (mol kg s ) k 3 (mol kg s ) E 2 (J mol ) k 5 (mol kg s ) K T/KM E (J mol ) E (J mol ) RSS Since the etherification is a dual-site reaction Table 5 and the isomerisation a single-site reaction, Activation energies obtained with A6 [8] and SMOPEX-0 comparison of the dual-site results obtained Amberlyst 6 SMOPEX-0 with SMOPEX-0 as catalyst with the results E (kj mol ) E 2 (kj mol ) obtained with Amberlyst 6 [8], requires the E 3 (kj mol ) rate constants to be divided by the square of the acid capacity (k, k 3) and the simple acid the highest temperature of 353 K. These results capacity (k 5) of the catalysts. The acid capacity indicate that the diffusion of isoamylenes to the of Amberlyst 6 was 5.0 mmol g and the acid active site is less hindered with the fibre catalyst capacity of the fibre was 3.4 mmol g. The than within the bead catalyst. results are presented in Fig. 2. The figure shows The activation energies obtained earlier with that the values of the etherification rate parame- A6 [8] and here with SMOPEX-0 are comters (k and k 3) are approximately equal at pared in Table 5 (LH model). lower temperatures of 333 K and 343 K. At 353 The activation energies obtained with the K the values with of the etherification rate bead catalyst (A6) are marginally lower than parameters (k and k 3) A6 are slightly smaller the activation energies obtained with the fibre than the values with SMOPEX-0. The iso- (SMOPEX) catalyst. This result further indimerisation rate parameter (k 5) is noticeably cates that there were slight intraparticle massgreater with the fibre than the bead catalyst at transfer limitations within the bead catalyst, because unsieved catalyst was used in the previous experiments [7,8]. It has been shown that no mass-transfer limitations are associated with the fibre catalyst [2,7]. 4. Conclusions Steady-state reaction rates for TAME synthesis were measured with three bead catalysts Fig. 2. Comparison of the etherification rate parameters (k and (A6, A35, XE586) and a fibre catalyst k 3; H ) s and the isomerisation rate parameters (k 5; s ) with Amberlyst 6 [8] and SMOPEX-0 as catalyst. Temperatures (SMOPEX-0) in a continuous stirred tank 333, 343 and 353 K. reactor. A35 was the most active catalyst,

11 P.K. Paakkonen, A.O.I. Krause / Reactive & Functional Polymers 55 (2003) especially at elevated temperatures ($70 8C), n i exp, amount of component i measured, when the reaction rates were calculated versus mol catalyst mass or sulfonic acid concentration. r rate of reaction for component i, mol i The next most active catalyst was A6, and the kg s fibre catalyst showed moderate activity. The w weight factor i lowest activity was detected for the surface W catalyst mass, kg cat sulfonated XE586. When the rates were calcu- xi molar fraction of component i lated against the square of the acid capacity, the four catalysts showed similar activity. This Abbreviations suggests that the reaction proceeds with a dual- 2MB 2-methyl--butene site mechanism, at least in stoichiometric con- 2M2B 2-methyl-2-butene ditions. A5 Amberlyst 5 (Rohm&Haas) The effects of temperature and molar ratio of A6 Amberlyst 6 (Rohm&Haas) the reagents on the synthesis and decomposition A35 Amberlyst 35 (Rohm&Haas) rates of TAME were measured in a batch CSTR continuous stirred tank reactor reactor with the novel ion-exchange fibre DIA diisoamylene, 2,3,4,4-tetramethyl-- SMOPEX-0 as the catalyst. Kinetic modelling hexene results suggested a dual-site mechanism for the DME dimethylether TAME synthesis. The rate parameters obtained IA isoamylene mixture (93 wt% 2M2B, 7 in the earlier studies [8] with Amberlyst 6 wt% 2MB) (ion-exchange bead) as catalyst were the same MeOH methanol as those obtained with SMOPEX-0 at 333 K TAME tert.-amyl methyl ether, 2-methoxy-2- and 343 K, but at higher temperature of 353 K methylbutane the rate parameter values for SMOPEX-0 TAOH tert.-amyl alcohol, 2-methyl-2-butanol were higher than those for Amberlyst 6. The RSS residual sum of squares activation energies obtained for the isomerisa- SE standard error tion and etherification reactions of isoamylenes WSRS weighted sum of residual square were 9 30% higher for the fibre catalyst than for the bead catalyst, indicating some diffusion Greek letter limitations associated with the latter catalyst. gi activity coefficient for component i 5. Notation Acknowledgements ai activity of a component i 5 gix i E activation energy, J mol The financial support of the Academy of a F molar flow of component i, mol s Finland through the Graduate School in Chemii [H ] acid capacity of the catalyst, mol kg cal Engineering (GSCE) is highly appreciated. k rate constant, mol kg s Ki adsorption equilibrium constant of component i References K reaction equilibrium constant for a j reaction j, j 5 3 [] B. Corain, M. Zecca, K. Jerabek, J. Mol. Catal. 77 (200) 3. molar mass of the component i [2] R.S. Karinen, A.O.I. Krause, K. Ekman, M. Sundell, R. i Peltonen, Stud. Surf. Sci. Catal. 30 (2000) 34. i calc amount of component i calculated by [3] D. Parra, J.F. Izquierdo, F. Cunill, J. Tejero, C. Fite, M. ~m tot total flow, kg s M n, a model, mol Iborra, M. Vila, Ind. Eng. Chem. Res. 37 (998) 3575.

12 50 P.K. Paakkonen, A.O.I. Krause / Reactive & Functional Polymers 55 (2003) [4] H.-J. Pannemann, A.A.C.M. Beenackers, Ind. Eng. Chem. [2] L.K. Rihko, J.A. Linnekoski, A.O.I. Krause, J. Chem. Eng. Res. 34 (995) 438. Data 39 (994) [5] F. Ancillotti, M.M. Mauri, E. Pescarollo, J. Mol. Catal. 4 [3] I.P. Pavlova, D.N. Caplic, I.V. Isuk, M.E. Basner, CNIIE - (978) 37. Neftechim, Moscow, 986. [6] C.N. Rhodes, D.R. Brown, S. Plant, J.A. Dale, React. Funct. [4] C. Oost, U. Hoffmann, Chem. Eng. Sci. 5 (996) 329. Polym. 40 (999) 87. [5] L.K. Rihko-Struckmann, J.A. Linnekoski, O.S. Pavlov, [7] L.K. Rihko, P.K. Kiviranta-Paakkonen, A.O.I. Krause, Ind. A.O.I. Krause, J. Chem. Eng. Data 45 (2000) 030. Eng. Chem. Res. 36 (997) 64. [6] R.S. Karinen, A.O.I. Krause, Ind. Eng. Chem. Res. 40 [8] P. Kiviranta-Paakkonen, A.O.I. Krause, Chem. Eng. Technol. (200) (2002) (in press). [7] J. Lilja, J. Aumo, T. Salmi, D.Yu. Murzin, P. Maki-Arvela, [9] L.K. Rihko, A.O.I. Krause, Ind. Eng. Chem. Res. 34 (995) M. Sundell, K. Ekman, R. Peltonen, H. Vainio, Appl. Catal. 72. A: General 228 (2002) 253. [0] A.O.I. Krause, L.G. Hammarstrom, Appl. Catal. 30 (987) [] P.K. Kiviranta-Paakkonen, L.K. Struckmann, J.A. Linnekoski, A.O.I. Krause, Ind. Eng. Chem. Res. 37 (998) 8 24.