The Chemxal Engrneenng Journal, 44 (1990) 97 - 106 97

Methyl Tertiary Butyl Ether Formation in a Catalytic Bed Reactor - Kinetic and Modelling Study

A AL1 and S BHATIA*

Department of Chemzcal Engmeenng, Indzan Instztute of Technology, P 0 I.I T, Kanpur 206 016 (UP), (Indza)

(Received 8 July, 1988, m final form 18 October, 1989)

ABSTRACT

Synthesis of methyl tertiary butyl ether (MTBE) usmg methanol and wobutene was s tudzed usrng macroporous cation exchange resm, Amberlyst 15 m the hydrogen form, as a catalyst m the temperature range 328 - 348 K. The reaction was carried an a fixed bed catalytic reactor opem ted at a tmosphenc pressure with liquad methanol and gaseous wobutene. The effects of weight hourly space velocity (WHSV), catalyst particle size and reactron temperature were studied. The reac- taon mtes increased with an mcrease in recaprocal space velocity and reaction tem- perature. Interparticle diffusional effects were absent under experimental condations. The reaction rate data were m terpreted usmg a heterogeneous model based on Langmuv- Hmshelwood kmetics. The reactaon enthalpy was also calculated from kmetic analysis and was found to be compatible with the thermo- dynamic value A LangmuirHmshelwood- Hougen- Wa tson (LHHW) rate model fatted kmetrc data well.

presence of a cation exchange resin. The reac- tion IS reversible and exothermrc (A& = -8.8 kcal mol-‘) and carned out in the temperature range 313 - 363 K. The acidic cation exchange resin offers many advantages over the acid catalysts [4]. These resins are very sensitive to temperature rise and start losmg then activrty above 393 K. Amberlyst 15, a macroporous cation exchange resm, 1s the most supenor catalyst, especially m non-aqueous media.

1 INTRODUCTION

The phase down of lead m automobile gasolme smce the late 1970s has created a need for high octane unleaded gasohne components. In response to this need, pro- cesses have been developed for alkyl ethers that are formed by the addition of iso-olefins (more correctly, tertrary olefins), such as lsobutylene and rsoamylene, and its produc- tion IS simple and mexpensive [ 1 - 31. Methyl tertiary butyl ether (MTBE) IS made by reac- tion of methanol with lsobutene in the

*To whom correspondence should be addressed

The drrect addition of olefins to alcohols catalysed by ion exchange resm to grve ethers was first investrgated by Anclllottl and Percarollo [5,6] They confirmed that ion exchange resin displayed an activity higher than soluble anhydrous p-toluene sulphonic acid Kmetrc orders were deter- mined with respect to the concentrations of reactants and -S03H groups. A zero order dependence of mitral rate on methanol con- centratlons greater than 4 mol l-l, with negative orders at lower concentrations were reported. Glcquel and Torck [4] examined the influence of methanol con- centration on the activity of Amberlyst 15 resm to form MTBE, z.e on the reaction rates, together with, the polarity of the medium on the reaction thermodynamics. Joon Suk Song et al. [7] studied gas phase synthesis of MTBE from Methanol and C4 reffmate, and its decomposltlon over hetero- poly compounds at atmosphenc pressure. Catalytic activity for both synthesis and decompositron of MTBE depends on the acid strength and acid amount of catalysts. Liquid phase synthesis of MTBE has also been reported [ 8). MTBE IS generally made through smgle- or two-stage processes Al- though MTBE process has been developed by several companies, much of the mformatron and process detarls have been patented.

0300-9467/90/$3 50 @ Elsevier Sequoia/Printed m The Netherlands

98

Decomposition of MTBE kmetics has been studied by Cumll et al. [9]. They analysed their kmetic data using Langmuh-Hmshel- wood-Hougen-Watson (LHHW) model. The validity of these models in explaining the gas phase kmetics of reactions catalysed by ion exchange resms for a number of reactions has been demonstrated [lo, 111. The aim of the present work was to study the kinetics of MTBE synthesis usmg Amberlyst 15 as a catalyst in a fixed bed reactor. The important variables such as reaction temperature and space velocity (WHSV) affectmg the yields of MTBE were studied. The kmetic data were analysed using a heterogeneous model based on the LHHW approach.

2 EXPERIMENTAL DETAILS

2.1. Chemzcal reagents and catalyst High analytical grade methanol (purity,

99%) was supplied by Ranbaxy Chemicals, Punlab, tertiary butanol (LR grade) by S.D. Fme Chemicals Ltd., Bombay, and concentrated sulphunc acid (AR grade) by Glaxo Laboratories (India) Ltd., Bombay. High purity nitrogen gas was used as a carrier gas. Hydrogen gas used m the flame iomza- tion detector was obtamed from Indian Oxygen Ltd., Kanpur.

Amberlyst 15, manufactured by Rohm and Haas Company, U.S.A., was sieve analysed to different fractions. About 15 - 20 particles were taken at random from each fraction and their sizes were measured using an optical microscope fitted with a calibrated eye-piece. The average particle size was calculated from these measurements for each fraction.

2.2. Apparatus The present kinetic study of MTBE

synthesis was carried out in a fixed bed reactor using methanol (AR) and isobutene (2-methyl propene) as reactants. Amberlyst 15, a strongly acidic macroporous cation exchange resin, was used as a catalyst.

A schematic diagram of the set up is shown m Fig. 1. Isobutene was generated m the high pressure Parr reactor. The isobutene gas flow rate was controlled by a needle valve fitted on the reactor. The methanol flow rate was controlled by a metermg pump (FMI). Soap bubble meters were used to measure isobutene inlet and outlet flow rates. The reactor is made of glass and has an annulus which is filled with glycerm, to act as a heating Jacket. The outer surface of the reactor is wrapped by heatmg tape which heats the Jacket and thus the catalyst bed.

1 Ice water tub

2 FMI lab pump

3 Burette

4 Centrifugal pump

5 Llquld sample

6 Unraacted lsobutone

7 Condenser

0 Heating tape

9 Glycerlno jacket

10 Catalyst bed

11 Glass wool

12 Soap bubble meter

13 Parr reactor

14 GLycerme bath

TC Temperature controller

TCR Temperature controller

and recorder

r

-_ & 27 -__

1 X -. --

Fig 1 Expenmental set up

99

There is a provision of a thermowell m the reactor so that the temperature of the catalyst bed can be controlled at the desired tempera- ture usmg an ‘Aplab’ digital temperature controller withm an accuracy of kO.1 K. The Parr reactor temperature was also controlled by another temperature controller withm a temperature accuracy of +l K.

2.3 Procedure Isobutene was generated by heatmg tertiary

butanol using concentrated sulphunc acid as catalyst at a temperature of 333 - 343 K m the glass lmed Parr reactor. The isobutene flow was controlled by a needle valve on the Parr reactor, before bemg sent m the fixed bed reactor. Controlled flow of methanol mto the fixed bed reactor was achieved with the help of a calibrated metering pump.

The fixed bed reactor contams Amberlyst 15 catalyst m hydrogen form which is packed inside the core of reactor (diameter 18 mm). The glass wool was placed below and above the resin catalyst bed for uniform distribution of reactants. The catalyst was conditioned and converted mto the hydrogen form before use.

In a typical experimental run isobutene was generated m the Parr reactor by reacting 250 ml of tertiary butanol and 20 ml of concentrated HzS04 at a temperature of 333 - 343 K. 30 ml mm-i of isobutene was passed continuously to the bed through a soap bubble meter. A known flow rate of methanol was passed to the reactor bed usmg a cahbrated metering pump. The desired reactor temperature was attamed by clrculatmg hot ethylene glycol from a constant temperature bath controlled withm a temperature of +O.l K. Higher temperatures of the bed was attamed by usmg the heating tape surroundmg the bed. The steady state temperature was attamed after 1 h. The products and unreacted reactants were collected once the steady state was achieved and analysed by gas hquid chromatography (GLC). The unreacted isobutene gas flow rate at the reactor outlet was measured by another soap bubble meter. The mass balance of isobutene consumption was checked by the inflow and outflow of isobutene gas m the reactor.

2.4. Analysas of products Liquid samples were analysed by GLC

(manufactured by Chromatography and Instruments Co., Baroda). A 1 m long, 6 mm outside diameter column contammg 10% SE30 on Chromosorb W was used for the analysis of methanol, MTBE and tertiary butanol. A flame ionization detector (FID) module was employed for the detection of peaks. The flow rates of nitrogen and hydro- gen gases were 15 ml mm-l each. The column and detector temperature was maintamed at 371+ 1 K. The column resolved the peaks of methanol, MTBE and tertiary butanol success- fully.

3 RESULTS AND DISCUSSION

Synthesis of MTBE was studied m a fixed bed reactor m the temperature range 328 - 348 K m the presence of Amberlyst 15 resin m the hydrogen form as a catalyst. Selectivity studies have been omitted because this reac- tion happens to be highly selective, z.e. it gives selectivity above 97% Besides this, such a study is of not much practical use because the byproducts, i.e. tertiary butyl alcohol (TBA) and dnsobutyene (DIB), also possess good blending values. Expenmental runs were performed with an average methanol flow rate of 0.53 gmol h-l and an isobutene flow rate of 0.0803 gmol h-’ respectively. Since the methanol molar flow rate is higher than the isobutene molar flow rate, fractional conversion has been based on isobutene. It can conveniently be defined as the ratio of isobutene reacted or MTBE formed to isobutene fed. The reciprocal space velocities ranged from 5.6 to 32.8 g h gmol-‘. Experi- mental runs were also performed to check the mtraparticle and interphase mass transfer effects.

The fractional conversion of isobutene is plotted against reciprocal space velocity for various reaction temperatures. The apparent reaction rates were determmed from the slopes at various W/F values. These X us. W/F curves were fitted by the lmear regres- sion method [ 131 m order to mmimlze errors.

3.1 Effect of reciprocal space velocity W/F Figure 2 shows the effect of reciprocal

space velocity on the fractional conversion.

100

hypothesized the reaction between methanol and acid group as

SOsH + ROH + ROH,+ + SO, (1)

He argued that the S03H group was more acidic than the solvated proton. Thus the rate of MTBE formation will increase with the mcrease m SOsH group concentration at a methanol concentration greater than 4 mol l-i, where methanol is no longer able to effect the rate as a solvent. Smce m the present study the methanol concentration was 24.68 mol l-‘, catalyst concentration seems to be a maJor factor m determming the reaction rate and conversion.

It can also be observed m Fig. 2 that frac- tional conversion keeps on increasing with an mcrease m space tnne (W/F), though it seems to be on the way to equihbnum, z.e when the net rate of reaction equals zero. By the trend of the curve m Fig. 3, it seems that eqmlib- num may not be achieved before attammg a space time of 60 - 70 g h gmol-‘. One can expect high conversions at these values of space time.

348K

3I.3 K

338 K

333 K

328 K

Flow rats of MeOH = 036 cc/mm

Flow rate ot lsobutene i 30cdmm

PartIck s,*e = 0 42mm

L I ‘0

I I I I I I I

S 10 15 20 25 30 35 LO

W/F (gmcat-hrlg mole1

Fig 2 Effect of reciprocal space velocity and tem- perature on converslon

It can be mferred from Fig. 2 that conversion mcreases with an increase in reciprocal space velocity. Secondly, withm the range of space times studied for a given temperature, the reaction did not reach its equihbnum (as the reaction 1s reversible). The effect of temperature on fractional conversion for a given space velocity will be discussed later.

It is evident that the longer the amount of tune the reactants spend m the bed, more the conversion or yield. The reactants, methanol m particular, have more access to the active sites at higher space times W/F. For a given flow rate of methanol F, as the catalyst weight W (meq. H+ ions) is increased z.e. W/F IS mcreased, a higher yield of MTBE 1s obtamed at a given temperature. Thus as the catalyst concentration (acid group) mcreases, the MTBE yield is also mcreased till equmb- num IS reached. It has been reported m the literature [ 51 that the rate of MTBE forma- tion depends on the acid group concentration at about third order. Uematsu et al. [14] reported m 1972 that the concentration of carbonmm ions formed per unit tune m- creased with increase m functional sulfo groups. With utilization of higher concentra- tions of hydrogen ions (H+) a relatively high yield of MTBE may be obtamed. Gates [lo, 111, m his work on alcohol dehydration,

3.2. Effect of tempemture The effect of temperature on the fractional

conversion is also shown m Fig. 2. The reac- tion temperature was varied from 328 to

\

Flow rate 01 MeOH = 036cdmK1

Flow rate of lsobutono i 30cc/mln

Part& see = 0 12mm

3CBK

343 K

338K

333 K

328K

L I ‘0

I I I I I 1 I 5 10 15 20 25 30 35 LO

W/F (pm hr/g mole)

Fig 3 Effect of reciprocal space velocity (W/F) on reaction rate

101

348 K. The data clearly show that fractional conversion mcreases with an mcrease m temperature. It shows that for the same space time the total amount of MTBE produced is higher at higher temperatures; this is due to the effect of temperature on the reaction rates.

3.3. Effect of resm part&? szze In order to check the mtraparticle diffu-

sion, two different resm particle sizes (0.60 and 0.32 mm) were taken and a tune on-stream study was performed at a given temperature and space velocity. The effect of resin particle size 1s shown m Fig. 4, where fractional conversion is plotted agamst time on-stream. It is clear from Fig. 4 that particle size has virtually no effect on the fractional conversion m a time on-stream study. Thus it may be concluded that mtraparticle diffu- sional resistance is negligible and not im- portant. It can also be seen that after about 60 mm the plot of fractional conversion (X) us. time on-stream is Just a honzontal he.

This leads us to the fact that the catalyst does not deactivate for at least 3 h.

0 50 -

* Q

/

O&5- /

I ,

0 LO - I /

035- dI

x I’

i 030- I

,

r I PartICk sne f

I 2 025- , d OSOmm

” / 0 032mm

3 s 020

- t

Q

; 0 /’

I= 015 ,’

- t

I

010 ,’ I

I

Flow rote of MeOH i 036cc/mm

Flow rate of lsobutone : 30 cc/mm

W/F : 17 65 gm hr/g mole

Temperature = 333 K

005

r I

Ok I I I I I I I I I 0 20 LO 60 60 100 120 l&O 160 160

Tlmo on stream (mm)

Fig 4 To check Internal dlffuslon resistance

3.4 Test for mterphase daffuszon resistance A set of experimental runs were performed

to check mterphase diffusional resistance. Reactions were carried out at two different feed rates and amounts of catalyst, fractional

conversion us. tune on-stream for the reaction under study. Smce the data obtamed for two different feed rates fall on the same curve, it can be concluded that external mass transfer hmitation is negligible durmg the experimentation.

3 5 Krnetac modelhng and analyst Reaction modellmg is essential for mter-

pretation of kinetic data m chemical kmetic studies. Basically two approaches are followed for ion exchange resm catalysed reactions [ 15,161. One such approach is when the resm behaves like a dissolved electrolyte. Under such circumstances it can be treated as a homogeneous reaction m hqmd phase. The other approach is when the resin is treated as a solid porous catalyst, like a conventional catalyst m heterogeneous catalytic reaction

3 5.1 Heterogeneous kmetws The catalytic mechanism occurring in the

presence of a macroporous resm depends on the polarity of the reaction medium, as pointed out by kinetic studies carried out on dehydration reactions of alcohols such as methanol, tertiary butanol and isopropyl alcohol [lo, 111 In macroporous resins, wherein the internal solvent is distnbuted between gel and pores, the Helffench model of homogeneity may not be applicable and, therefore, approaches traditional to hetero- geneous catalysis and based on classical models such as Langmuu-Hmshelwood or Rideal [ 171 can be adopted. In this approach, one assumes that specific competitive adsorp- tion of one or more molecules of reactants or products, with hydrogen counter ions essen- tially fixed at sites near the resin skeleton, caused local concentrations of sorbed species to be different from their values m the pore liquid, even though a lmear distribution law may correctly relate the concentrations m the bulk liquid and m the pore hquid [ 181.

In general, heterogeneous reactions occur either because the reactants are of different phases or the catalyst is of a different phase from the fluid reactants. In the present study the catalyst is m the solid phase and thus different from the phases of the reactants. We are therefore Justified m considermg the reaction as heterogeneous. The different steps involved m heterogeneous reactions are dif-

TABLE 1

Rate expressions for different rate controlling mechamsms (LHHW model)

A+B + C Methanol wobutene Methyl tertiary butyl ether

Reactlon mechamsm Smgle site Dual site

Surface reaction controlling

Adsorption of methanol (A)

r= k,(CACB - CC/K)

(1 +KACA+KCCC)

r = k$CA - CC/KCB)

(CACB - CC/K) r = k&&B

(l+ KACA+ K&B+ G$c)*

r = k, (CA - '%/K~B)

(1 + KBCB + KACCIKCB + K&c)

controlhng

(CACB-Cc/K) Desorptlon of MTBE(C) controlling

r = kd(KK&ACB - CC) r = kdKc (1 + KACA + K&B + KcKCACB)

fusion of reactants and products mto and away from the pores, adsorption of reactants and desorptlon of products on the surface, and reaction on the catalyst surface of adsorbed reactants. It 1s assumed that the rate 1s controlled by a single slow step and all other steps occur at equlllbrmm [ 191. The followmg steps are considered to take place successively at the surface usmg the Langmulr-Hmshelwood model approach. (1) Methanol (A) and/or lsobutene (B) are

adsorbed on the surface of resin (2) Surface reaction between adsorbed reac-

tants. (3) Desorptlon of the product (MTBE) from

the surface The overal reaction 1s represented as

CH3 CH3

CH,OH + & =CH A

cJ!I,

2 & CH3- -0-CH3

AI

(2)

C 3

MeOH + lsobutene MTBE

(A) (B) (C)

Various plausible kmetlc models are derwed dependmg on which of the three steps 1s controllmg. It may thus be adsorption controllmg, surface reaction controlling or desorptlon controlhng.

Two mam reaction mecharusms, based on the above approach, may be speculated. They are (1) a dual site mechanism and (2) a single site mechamsm.

Vanous models based on these mechamsms are presented m Table 1.

Glcquel and Torck [4] proposed a kinetic model which 1s suited for the present reac- tion. It 1s a shght modlflcatlon of single site surface reaction controlhng model and 1s gwen by

k&ACB - k4RCc r=

CA + RCc (3)

where k3 and k4 are the forward and back- ward rate constants at temperature Z’(K), and R 1s the ratio of MTBE/CH30H adsorp- tion coefflclents at T(K)

A non-linear algorithm (B SOLVE ALGORITHM) by Marquardt [ 131, which 1s an extension of the Gauss-Newton method, was used to obtain various constants of the heterogeneous rate expressions as shown m Table 1. This program was supenor to the Gauss-Newton method as it converged with relatively poor startmg guesses for the unknown coefflclents

The obJectwe function was defined as

ObJective function = i (r,.ti - r’exp)2 I=1

(4)

where red 1s the calculated rate from the vmous models, rexp 1s the observed reaction rate and n 1s the number of observations.

The obJectwe function was mmlmlzed usmg the above method, starting with the mltlal guess values of vanous constants of the model. The final values were obtamed once

050

OL5 F i

0 40 - /

0.35- %

x / 5 030- a / ‘0 ; 025- /

”

z s 020- $ P

s / LL 015- I

:

OlO- 1

:

Oo5 I I

A += 17gm

0 669 mole I hr ~250

D + 6gm 034gmoloIhr

~1765

Part~clo size = 0 42mm

Tomporoture F 333 K

01 ’ I I I I I I I I 0 20 40 60 60 100 120 140 160 l60

Time on stream (mln)

Fig 5 To check external dlffuslon reswtance.

the ObJectwe function attamed a value of 0.00001.

The various models proposed were tested. The models based on adsorption and desorp- tion controllmg did not give a satisfactory fit of the experimental data and were reJected. Models based on surface reaction controlling (dual site and single site) gave positive constants and a satisfactory fit of the data. The surface reaction models which fitted experimental data well are

Smgle site mechanism

ks(cAcB-~C/~) r=

(1 +KAcA +&CC)

Dual site mechanism

(5)

' = ksKAKB (1 + KACA + K&B + K&C)* (6)

Equation (5) has been modified by Gicquel and Torck [4] to mcorporate certain vahd assumptions. These assumptions include that the methanol to isobutene molar ratio is greater than unity and the value of RCLITBE IS high so that the term given m the denomi- nator of equation (5) has been modified

ksC,Cz - k4RCc r=

CA + RCc (7)

Thus we fmd that eqns. (5), (6), and (7) which are based on surface reaction control

103

gave a satisfactory fit. In order to smgle out the correct model, dlscnmmation among rival models is necessary. We have based our discrimination on some a postenon tech- niques [20]. Some of these methods are methods of diagnostic parameters, e.g. hkeh- hood ratio, physico-chemical nature of param- eters, goodness of fit etc.

In our discnmination methods we have discriminated most of the models based on the physico-chemical nature of parameters. Parameters m a mechanistic model have physico-chemical meaning so that they are subJect to constramts: rate coefficients and adsorption equmbnum constants cannot be negative. Also, estimates of activation energies have to be statistically positive and, for adsorption, eqmhbrmm constants enthalpies are negative [20]. Applying this discnmma- tion strategy we found that the three models (eqns. (5), (6) and (7)) satisfied this criteria. An interesting pomt about these three models was that all three of them were basically surface controllmg models. In order to get the best rival model, a method of non- mtrmsic parameters has been adopted. In this method, a non-mtrmsic parameter does not appear m the model, but is mtroduced for the sole purpose of famhtatmg the discnmmation between rival models.

For selection between two rival models, the method developed by Wrlhams and Kloot [21] and brought to the attention agam by Mezalu and Kittrell [ 221 only requires simple lmear regression, an undeniable advantage. The method mvolves definmg a new depen- dent variable z which IS given by

2 = Y - l/2(9, + 92) 93)

where y1 1s the least squares prediction of the dependent variable y under model z. If y1 - jl, is considered as an independent variable, and Model 1 is correct

E(z) = l/2(9, - 92) (9)

If Model 2 is correct, then

E(z) = --l/2(9, - 92) (10)

Both relations (9) and (10) may be combmed into single equation:

E(z) = A(91 - 92) (11)

104

0 0*0r

OOlS- Model 1 (Eq 5)

&

oozo- Model Z(Eq 6)

I I I I I I I

04u 003 0.02 001

x=

z.OL23r

Fig 6 Apphcatlon of method of non-mtrmslc param- eters to dlscrlmmate between rival modes

ratio of MTBE/CHsOH adsorption coef- ficients at T(K) and r is the rate of reaction. The constants obtamed are listed m Table 2. An Arrhemus plot of In ks us. l/T is plotted m Fig. 7.

Flow rate of MoOH = 036cclmm

Flow rote of Isobutene = 30 cc hln

-06

The dependent variable y m our case has been taken as conversion (X). Calculated conver- sions have been depicted as 9i, jj2, Q3 for models eqns. (5), (6) and (7) respectively.

Figure 6 shows plots of y - l/2(9, + s2) us 9, -jj2 and y -l/2(9, -9s) us. j1 -9s. From this plot we determme the mtrmsic parameters by linear regression The first plot gives the value of X as 0.423 + 0.277 and the second plot gives the value of X as -0.268 f 0 242. This clearly mdlcates Model 3 (eqn 7) is the most suitable followed by Model 2 (eqn. 6). Thus the Glcquel-Torck Model (Model 3) 1s the most suitable model to represent the kmetlcs of this reaction Therefore, the kmetlc model IS

k&CB - k4RCc r=

C, + RCc (12)

where k3 and k4 are the forward and reverse rate constants at temperature Z’(K), R 1s the

-1.91 I I I I 2 65 290 2 95 300 305

(l/T) x ld (I?)

Fig 7 Effect of temperature on forward rate constant (heterogeneous model, eqn (7 ))

The apparent activation enewes for MTBE synthesis and decomposition as determmed from the Arrhenms plots are tabulated m Table 3. The apparent activation energies obtamed m our study agree well with the values reported by other mvestrgators. A comparison of the values reported m the literature have also been hsted m Table 3.

The activation energy values reported from the homogeneous model are higher than those from the heterogeneous model. This trend is consistent with the reported values. In the homogeneous model where dlffuslon is not important, the reaction is assumed to occur m solution, a situation which is not far away from acid catalysis m solution. In the hetero-

TABLE 2

Various rate constants as obtamed from the heterogeneous model (eqn 7)

Temperature (K) k3 x 102 k4 x lo6 [12(gmol g h)-‘1 [l(g h)-‘I

R

1 328 20 0 4 37 1007 2 2 333 35 8 10 26 979 7 3 338 52 1 26 87 648 2 4 343 85 7 145 70 32110

105

TABLE 3

Actlvatlon energies for MTBE synthesu as determmed by previous mvestlgators

E (kJ moT1)

82 0 7115 87 9

912

103 4

74 08

79 0 76 73 68 9

110 4

Catalyst

Amberlyst 15 Amberlyst 15 methyl sulphurlc

acid methyl sulphurlc

acid p-toluene

sulphomc acid Amberlyst 15

Amberlyst 15 Amberlyst 15 Amberlyst 15

Amberlyst 15

System

sohd-hquld sohd-hqmd (mltlal rates) gas( lC&hquld (methanol)

homogeneous

homogeneous

sohd-hquld (fixed bed contmuous flow)

homogeneous heterogeneous heterogeneous

(MTBE synthesis) heterogeneous

(MTBE decomposltlon)

References

Glcquel-Torck [ 41 Anclllottl et al [5, 61 Beauflls and Hellm [ 41

Beaufds and Helhn [4]

Ancdlottl et 01 [6]

Glcquel and Torck [4]

Subramamum and Bhatla [ 8 ] Subramamum and Bhatla [ 81 Present study

Present study

geneous model the activation energy 1s lowered due to the activatron energy of drf- fusion which rs of the order of 25 - 42 kJ mole1 as compared to 12.5 - 25 kJ mol-’ m solution [4]. Furthermore, m the resm structures awulable catalytic sites come across excess methanol which is present. This favours the reaction by lowermg the activa- tion level towards the transition state. This could result in the discrepancy between the activation energy values reported between homogeneous and heterogeneous approaches. The heat of reactron A& obtained from activation energies based on the hetero- geneous model was 41.5 kJ mol-‘. This value is comparable with the thermodynamic reported value of -39.8 f 0.4 kJ mol-’ [ 231. This further confirms that the hetero- geneous model proposed in the present study is a valid representation of the reaction.

4 CONCLUSIONS

The information gained from the present study may be summanzed as follows.

(1) Amberlyst 15 IS a smtable catalyst for the synthesis of MTBE, which showed the absence of deactivation and dlffusional resis- tance under operatmg conditions.

(2) The reaction rate increased with an increase 111 temperature and reciprocal WHSV.

(3) Kinetic data were correlated usmg heterogeneous models based on the LHHW mecharusm.

REFERENCES

8

9

10

11

12

13

14

15

J. C Davis and P M (1979) 91 S C Stmson, Chem 35

Kohn, Chem Eng , May 21

Eng News, June 25 (1979)

Chem Eng News, January 24 (1983) 27 A Glcquel and B Torck, J Catul, 83 (1983) 9 F Anclllottl, M Mass1 Maun and E Pescarollo, J Catal, 46 (1977) 49 F Ancdlottl, M Mass] Mauri,, E Pescarollo and L Romagnom, J Mol Catal, 4 (1978) 37 Joon Suk Song, Jai Chun Smg, and Wha Yound Lee, Proc 4th APCChE’ 87, Smgapore, May 13 - 15,1987 C Subramamam and S Bhatla, Can J Chem Eng, 65 (1987) 615 F Cumll, J. TeJero and J F Izqulerdo, Appl Catal, 34 (1987) 341 R Thronton and B C Gates, J Catal, 34 (1974) 275 B C. Gates and W L Rodnquez, J Catal, 31 (1973) 27 F Colombo, L. Corl, and P Delogu, Znd Eng Chem Fundam, 22 (1983) 219 J L Kuester and J H Maze, Optzmzzatzon Tech- nzques w&h Fortran, McGraw Hill, New York, 1973 T Uematsu, Bull Chem Sot Jap, 45 (1972) 3329 C N Satterfleld, Heterogeneous Catalysis m Practice, McGraw H111, New York, 1980

106

16 F Helffench, J Am Chem Sot, 76 (1954) 5567

17 B C Gates and L. N Johnson, AZChE J, 17 (1981) 1971

18 M B Bother et al, Znd Engn Chem Fundam, 4 (1965) 315

19 J M. Smith, Chemzcal Engzneermg Kmettcs, McGraw H111, New Delhi, 1981.

20 F G. Froment and L H Hosten, Catalytic Kmetlcs Modellmg, Vol. 2, Catalysis, Sprmger- Verlag, Berlm 1981, pp 132 - 140

21 E J. Wdhams and N H Kloot, Aust J Appl SC1 , 4 (1953) 1

22 R Mezakl and J R. Kltterell, Can J Chem Eng, 44 (1966) 285

23 H Arntz and K Gottheb, J Chem Thermodyn, 17 (1985) 967

APPENDIX A NOMENCLATURE

CA Concentratron of methanol (gmol 1-r)

CB Concentration of lsobutene (gmol 1-l)

Cc Concentration of MTBE (gm011-~) C3 Concentration at outside particle sur-

face (gm011-~) E am Apparent activation energy (kJ mol-‘) F Molar flow rate of MeOH (gmol h-l)

ks

k3

k4

ka

kd K

KA

KB

Kc

Ik s t T W WIF

X

Forward rate constant (L-H Model) Igmol(g W1l Forward rate constant (G-T Model) [l*(gmol g W’l Reverse rate constant (G-T Model) U(g W1l Adsorption rate constant Desorptlon rate constant Equihbrium constant Adsorption equihbnum constant of MeOH (1 mol-‘) Adsorption eqmlibrium constant of iC4 (1 mol-‘) Adsorption equilibnum constant of MTBE (1 mol-I) Apparent rate of reactlon [ gmol(g h)-‘1 Ratio of MTBE/CH,OH adsorption coefficients at T(K) Active &es on the resin particle Time (h) Temperature (K) Weight of the catalyst (g) Space time of reciprocal weight hourly space velocity (g h gmoP) Fractional conversion of iC4