Você está na página 1de 7

Ind. Eng. Chem. Res.

1998, 37, 3575-3581 3575

Catalytic Activity and Deactivation of Acidic Ion-Exchange Resins


in Methyl tert-Butyl Ether Liquid-Phase Synthesis
D. Parra,† J. F. Izquierdo,*,† F. Cunill,† J. Tejero,† C. Fité,† M. Iborra,† and M. Vila‡
Chemical Engineering Department, Faculty of Chemistry, University of Barcelona, C/Martı́ i Franqués 1,
08028 Barcelona, Spain, and Escombreras Research Center, Repsol Petróleo S.A., 30350 Valle de Escombreras,
Cartagena (Murcia), Spain

This paper presents a comparative study of catalytic activity of 12 different styrene-


divinylbenzene ion-exchange resins as catalysts for MTBE (methyl tert-butyl ether) liquid-phase
synthesis. Experiments were carried out at 1.6 MPa and 323 K using a stationary fixed-bed
fed with methanol and C4 olefinic cut as the source of isobutylene. The isobutylene/methanol
molar ratio studied was 0.91 and the LHSV employed in the continuous operation was 85 h-1.
Experimental data show that ion exchangers with higher acidic capacity were the most active
ones. It seems that this parameter plays the greatest influence on the catalytic activity for
MTBE synthesis whereas other properties such as surface area and average pore diameter hardly
influence it.

Introduction There are several published approaches to explain the


behavior of an ion exchanger as a catalyst. From a
It is hard to explain the kinetic mechanism of the
morphological point of view, it is to be noted that in the
etherification reaction of isobutylene with methanol in
model proposed by Guyot (1988), who claims that
liquid phase. At low concentrations of methanol, the
macroporous resins are composed of three families of
catalysis is by the polymer-bound -SO3H groups, which
pores, the structure contains large agglomerates of
are very active, but if the concentration of methanol is
microspheres (100-200 nm), which look like cauliflow-
increased, the alcohol dissociates the acid groups and
ers, and each microsphere shows smaller nuclei (10-
solvated protons become now the catalytic agents, which
30 nm) more or less fused together. In betweeen the
are catalytically less active (Gates and Rodriguez, 1973;
nuclei, there is a family of very small pores (5-15 nm),
Chakrabarti and Sharma, 1993). A proof of the dif-
which are mainly responsible for the high surface area
ficulty to explain methyl tert-butyl ether (MTBE) mech-
(> 500 m2/g) of these materials. Between the micro-
anism synthesis is found in the literature itself because
spheres, a second family of intermediate pores is
researchers have used both pseudohomogeneous and
observed (20-50 nm), which may account for moderate
heterogeneous kinetic models to describe the kinetics
surface areas (up to 100 m2/g). A third family of large
of this reaction (Voloch et al., 1986; Subramanian and
pores (50-100 nm) is located between the agglomerates;
Bathia, 1987; Colombo and Dalloro, 1983; Rehfinger and
this kind of pores yields very low surface area but large
Hoffman, 1990; Gicquel and Tork, 1983; Parra et al.,
pore volume up to 3 mL/g. Resins are able to swell to
1994; Tejero et al., 1996; Panneman and Beenackers,
varying degrees in different solvent media. So, the
1995a). Information in the available literature about
catalytic behavior of a resin may be analyzed in terms
the influence of catalyst morphology, acid concentration,
of accessibility of its functional groups in different media
and solvent effect on catalytic activity of an ion ex-
taking into account their situation inside the resin.
changer in MTBE synthesis is rather scarce (Colombo
and Dalloro, 1983; Jerábek et al., 1993; Rehfinger and Other theories emphasize the dependence of the
Hoffmann, 1990; Panneman and Beenackers, 1995b). catalytic activity on the acidity and on the accessibility
In general, the reaction rate of an ion-exchanger (Widdecke 1988; Buttersack, 1988). The acidity de-
catalyzed reaction in a nonaqueous medium depends pends on the type and number of acid groups within
both on changes of the polymer structure and on the resin and is affected by the degree of cross-linking.
changes in the concentration of accessible acid groups The accessibility is related to the permeability of the
(Rodriguez and Setinek, 1975; Jerábek and Setinek, resin and it can be influenced by the interaction of the
1987). In nonaqueous medium, the polymer matrix solvent and adsorbed molecules with the functionalized
structure influences directly the reaction course since polymer.
the possibility of forming undissociated sulfonic groups So, it can be drawn from the general theories that
depends necessarily on the conformation of polymer the catalytic activity of the resins depends mainly on
chains. Depending on the polarity of the medium and its initial morphology and on its interaction with the
on the size of the reactant and product molecules, steric reacting medium including the solvent and the com-
restrictions may also influence the catalytic activity pounds that can be found in the reacting system.
(Jerábek and Setinek, 1987) because some of the inner The properties of the MTBE reacting system change
active centers could not participate in the reaction. with the reaction progress. In this context, contradic-
tory papers can be found, probably due to the different
* To whom correspondence should be addressed. Fax: +34- initial mixtures and kinds of resins used to study this
3-4021291. E-mail: felipe@angel.qui.ub.es. reaction. Rehfinger and Hoffman (1990) and Jerábek
† University of Barcelona. et al. (1993) found no influence of the degree of cross-
‡ Escombreras Research Center, Repsol Petróleo S.A. linking, the internal surface area, and the exchange
S0888-5885(98)00007-4 CCC: $15.00 © 1998 American Chemical Society
Published on Web 08/14/1998
3576 Ind. Eng. Chem. Res., Vol. 37, No. 9, 1998

Table 1. Physical Properties of Different Ion Exchangers, Methanol Conversions, and Reaction Rates Obtained at 323
K
trade name C (mequiv/g) A (m2/g) A′ (m2/g) V (mL/g) D (Å) dp (mm) XM (%) r (mol/h‚mequiv)
Bayer K2631 4.83 41.5 ---- 0.67 650 0.63 6.9 0.0143
Bayer OC-1501 5.47 25.0 163.8 0.52 832 0.66 11.0 0.020
Amberlyst 15 4.75 42.0 156.9 0.36 343 0.74 7.2 0.0151
Amberlyst 35 5.32 34.0 165.7 0.28 329 0.51 10.4 0.0195
Dowex M32 4.78 29.0 0.33 455 0.63 6.8 0.0143
Purolite CT151 5.40 25.0 151.2 0.30 252 0.43 8.1 0.0150
Purolite CT165 5.00 6.2 0.16 1148 0.43 4.1 0.0081
Purolite CT169 4.90 48.1 0.38 342 0.43 7.7 0.0158
Purolite CT171 4.94 31.0 0.47 597 0.40 7.8 0.0158
Purolite CT175 4.98 29.0 0.48 662 0.40 8.2 0.0164
Purolite CT175/2824 5.30 24.1 157.4 0.44 745 0.43 10.4 0.0196
Purolite CT179 5.25 35.0 220.1 0.33 386 0.43 9.6 0.0182

capacity of the resin on the rate of MTBE synthesis, but of resins that differed markedly in their properties were
Panneman and Beenackers (1995b) showed that the compared at the same experimental conditions.
catalytic activity of an ion-exchange resin on the same The second goal of this paper is to know the deacti-
reaction depends on its three-dimensional structure, vating effect of ammonium and Fe(III) on the etherifi-
which is influenced by the level of cross-linking, the cation reaction. To test catalytic activity of partially
porosity, the internal surface area, the sulfonic acid sulfonated ion exchangers, reaction rates were compared
content, and its distribution within the particles. at the same experimental conditions.
On the other hand, another important factor in
commercial operation that influences the catalytic activ- Experimental Section
ity is the deactivation of the catalyst. The acid sites of
the cation-exchange resin catalyst may easily be poi- (i) Continuous Operation. Twelve commercial
soned by different ions that can be present in the feeds macroporous styrene-divinylbenzene ion exchangers
due to corrosion in pipelines or can be carried to the were used as catalysts in MTBE synthesis. Their
reactor through misoperations in preceding units. morphological properties are summarized in Table 1.
Brockwell et al. (1991) make a distinction between Prior to use, all ion exchangers were washed with
two different modes of catalyst deactivation. The first distilled water and dried at 110 °C for more than 14 h
mode is plug flow neutralization and the second is at atmospheric pressure. Afterward they were stored
diffused flow neutralization. With plug flow neutraliza- in a desiccator over sulfuric acid (98%). The water
tion, the poison accumulates on the catalyst in a plug- content in all resins, measured by Karl Fischer titration,
flow manner as it impinges on the catalyst bed. The was less than 3%. Before the experiment, the catalyst
uptake of deactivator is rapid and stoichiometric. As bed was preheated in the reactor at operating temper-
deactivation progresses, plug flow neutralization results ature in the presence of the methanol stream to ensure
in a foreshortening of the active catalyst bed length. that an amount of methanol greater than 10 times the
Examples of plug-flow deactivators are all salts and reactor volume was passed through it. As a result, the
other Lewis and Brönsted bases. Diffused flow neu- residual water content in the resin was reduced to less
tralization is a mode of deactivation in which catalytic than 1.6%. tert-Butyl alcohol produced from this amount
sites are neutralized uniformly throughout the bed of water was not detected in the experiments. Other
rather than in progression from the point of impinge- byproducts (dimethyl ether and diisobutylene) were not
ment. Diffusional deactivators are generally very weak found in the runs.
bases that either weakly bind the catalytic sites or are In all experiments, carried out in the steady state,
species that generate more strongly binding species in commercial particle sizes of different resins have been
situ. Examples of diffusional deactivators are propi- used to study their behavior in operating conditions of
onitrile, dimethylformamide, dimethyl sulfide, hydrogen industrial reactors. Fresh dry catalyst sample (0.874
sulfide, and amine formers such as acetonitrile, which g) was always placed in a continuous packed-bed mi-
is responsible for most of the catalyst deactivation croreactor (15 cm length; 4.4 mm i.d.) and the small
experienced in MTBE reactors due to formation of remaining reactor volume was filled with quartz of 0.63
ammonium ion (Marston, 1994). < dp < 0.8 mm to avoid bed expansion.
It has not been possible to find experimental studies The experiments in the continuous regime were
about the catalytic behavior of an ion exchanger partly carried out in a setup made of stainless steel at 1.6 MPa
deactivated by ammonium or Fe(III) in MTBE synthe- and 323 K by feeding an appropriate methanol-C4
sis. Krause and Hammarström (1987) propose that the olefinic cut mixture. This pressure was enough to
loss of the activity of the resin partly deactivated by maintain the reacting system in liquid phase and this
sodium or ammonium ions is the same for tert-amyl temperature was chosen to avoid temperature axial
methyl ether (TAME) synthesis in liquid phase, and gradients. The methanol molar flow rate was always
Kmosták and Setinek (1981) found that sodium ions 0.874 mol/h, the molar ratio of isobutylene/methanol
deactivate the catalyst more than Fe(III) ions for employed was 0.91, and the liquid hourly space velocity
isomerization of cyclohexene and dehydration of 1-pro- (LHSV) used was 85 h-1. C4 cut from a steam cracking
panol in the gas phase. (SC) plant containing 45-55% isobutylene was supplied
So, the first aim of this work is to contribute further by Repsol Petróleo S.A. Methanol for high-performance
to the elucidation of the effect of different morphologic liquid chromatography (HPLC) (<0.01% water, 99%
properties of an ion exchanger (surface area, average pure) was supplied by Romil Chemicals Ltd.
pore diameter, and acidic capacity) on its catalytic Two sampling valves injected 0.2 µL into an HP 5890
activity in MTBE synthesis. For this purpose, a number gas-liquid chromatography (GLC) apparatus equipped
Ind. Eng. Chem. Res., Vol. 37, No. 9, 1998 3577

with flame ionization detector (FID). One valve took free of external mass transport processes. Runs carried
samples from the reactor inlet and the other one from out by maintaining feed composition and catalyst mass,
the reactor outlet. A 50 m × 0.2 mm × 0.5 µm methyl but using different molar flow rates, demonstrated that
silicone capillary column was used to separate MTBE, for LHSV > 60 h-1 reaction rate does not depend on
methanol, and C4 cut compounds (isobutylene and flow rate. As a result, a LHSV ) 85 h-1 was used.
1-butene could not be separated; as a consequence, only Fourteen different replicated experiments were made
methanol and MTBE were used as the reference com- to calculate the experimental error of methanol conver-
pounds to measure the reaction progress). More de- sion ((0.2% with a probability level of 95%).
tailed information about setup, procedure, and sample Table 1 shows the morphological properties of ion
analysis can be found elsewhere (Parra et al., 1994). exchangers used along with the results of methanol
Catalyst activity was measured as methanol conver- conversion and reaction rates obtained after the experi-
sion and also as a reaction rate at steady-state condi- ments carried out in steady-state mode. The reaction
tions to know the efficiency of the resins from an rates were calculated by taking into account that for
industrial point of view. To reduce the effect of reacting the methanol conversion values obtained, the reactor
medium and the development of temperature gradients, was differential (Parra et al., 1994). So, the reaction
the methanol conversion was lower than 11%. Reaction rates were determined by
rates were also calculated to look into, from a scientific
point of view, the influence of different properties of the XMF
resins. In each run the flow rates were maintained (3-4 r) (1)
h) until chromatographic analysis was the same within 100CW
the limits of the experimental error to ensure that
steady state was reached. Some experiments were At first sight, the following differences about the
repeated to determine the experimental error. properties of the resins must be pointed out: Purolite
(ii) Batch Operation. The experiments in batch CT165 has a particular morphology, its Brunauer-
mode were carried out in a stainless steel jacketed 250 Emmett-Teller (BET) surface area (A) and pore volume
mL autoclave. A calculated amount of methanol and are the smallest in the group but its average pore
about 2.9 g of dry resin (water content less than 3%) diameter is the highest one. It seems to be a mi-
were first charged into the reactor and, after checking croporous resin. Despite the fact that Purolite CT151
for leakages, heated to the desired temperature (always has a high acidic capacity, it has a surprisingly small
333 K) and then an amount of C4 liquid, given by the average pore diameter in comparison with the other
desired isobutylene/methanol ratio, was introduced resins studied. Bayer OC-1501, Amberlyst 35, and
directly from a cylinder. At this time, the magnetic- Purolite CT175/2824 have the highest acidic capacity
drive turbine was switched on at 500 rpm and the of ion exchangers selected. Finally, Purolite CT 169 has
reactor pressure was set at 1.6 MPa throughout the the largest value of surface area (A), but its other
experiment in order to maintain the liquid phase. The properties are similar to the other resins.
time of introduction of the olefinic fraction into the As can be seen in Table 1, the catalytic behavior of
reactor was considered as the starting point for the the resins differs greatly. Methanol conversions vary
reaction. Further information on this apparatus is from 4.1% (Purolite CT165) to 11.0% (Bayer OC-1501).
described elsewhere (Izquierdo et al., 1992). Equal conversion values for Amberlyst 35 and Purolite
The reactor was connected directly to the sampling CT175/2824 are found, close to that for Bayer OC-1501.
valve (Valco 4-CL4WE) which injects a small pressur- Traditional catalysts in MTBE synthesis such as
ized liquid volume (0.2 µL) to the gas chromatograph Bayer K2631 or Amberlyst 15 show lesser catalytic
(HP5890A). The same capillary column described above activities than the new generation of commercial resins
was used for sample analysis. No byproducts (tert-butyl from the same manufacturers, Bayer OC-1501 and
alcohol, dimethyl ether, and diisobutylene) were de- Amberlyst 35. This fact could be explained because of
tected in the experiments. the increase of density of sulfonic groups produced after
Catalytic activity was measured as initial reaction a special sulfonation procedure which added a second
rate calculated from the slope of methanol conversion sulfonic group to the aromatic rings already sulfonated
versus time data. Some experiments were repeated to rather than sulfonate the unsulfonated aromatic rings.
evaluate the experimental error on reaction rate. So, this sulfonation procedure makes it possible to
(iii) Partial Neutralization of Ion Exchangers. achieve a higher number of acidic groups per unit of
Four resins were chosen to be partially neutralized by mass of these new resins.
Fe(III) and ammonium ions. Two standards of iron and To assess the quantitative influence of morphological
one of ammonium were used. A known amount (always properties in dried state of the different resins on
60 g) of a given dried ion exchanger was placed into a methanol conversion, different polynomials were fitted
beaker and 400 mL of solution, either FeCl3‚6H2O or to the data. From a statistical point of view, the best
ammonium hydroxide, was added. After standing for significant polynomial fitted to the data was
24 h with stirring, the catalyst was washed with water
in a column until there was negative reaction on the XM ) (2.924 × 10-3)C3A - (1.751 × 10-4)C2A2 +
corresponding anion. After this, the resin was redried
in a oven at 105 °C for 14 h. The concentrations of (1.598 × 10-5)C3D (2)
exchangeable acid sites left were measured by the same
method as for the fresh resin (Bayer, 1990). This fourth-order polynomial only contains three vari-
ables, namely, acid capacity (C), BET surface area (A),
Results and Discussion and average pore diameter (D). We have not considered
in this mathematic analysis pore volume (V), because
(i) Fresh Catalysts. Preliminary experiments were it depends on the surface area and average pore
made at 323 K to ensure that rate data were generated diameter, or approximate average particle diameter (dp),
3578 Ind. Eng. Chem. Res., Vol. 37, No. 9, 1998

Figure 1. Reaction rates of MTBE synthesis at 323 K versus


average pore diameter.
Figure 3. Reaction rates of MTBE synthesis at 323 K as a
function of acidic capacity.
surface area for the swollen state of some of the resins
studied (Jerábek, 1997) by inverse steric exclusion
chromatography (ISEC) corroborate this affirmation (see
Table 1). As can be seen, there is not a relationship
between the BET surface area, determined in dry state,
and the surface area in the swollen state.
After this, we studied in detail the influence of acidic
capacity on reaction rate for all resins and we found out
some behavior deviations. Figure 3 shows the reaction
rate against the acidic capacities of the resins studied.
As a general rule, it can be seen that the higher the
acidic capacity, the higher the reaction rate, but there
are two resins that do not fulfill this trend, namely,
Purolite CT 151 and especially Purolite CT 165. Despite
Figure 2. Reaction rates of MTBE synthesis at 323 K versus BET the fact that Purolite CT 151 has a high acidic capacity,
surface area. it has the smallest average pore diameter in the dry
state. It seems that this resin could not suffer a
because the values obtained from manufacturers are not significant swelling in contact with the MTBE reaction
accurate enough. mixture, and so the accessibility of reactants to the
It can be seen from this polynomial that the positive ensembles of sulfonic groups placed inside the polymer
influence of acidic capacity on methanol conversion is mass is not satisfactory. Purolite CT 165 has a high
stronger than that of the other two variables. It seems acidic capacity and the smallest BET surface area,
that the accessibility of methanol and isobutylene to the though its average pore diameter in the dry state is the
greater part of the acidic active centers is quite good largest one. It seems that this resin, which is mi-
for the majority of the resins chosen. croporous (Jerábek, 1997), could have a small density
Several authors assure that kinetics of MTBE syn- of accessible sulfonic groups and so concerted mecha-
thesis catalyzed by strong acidic ion exchangers pro- nisms are difficult to form.
ceeds with the participation of ensembles of sulfonic (ii) Deactivated Catalysts. Four representative
groups (Parra et al., 1994; Tejero et al., 1996; Jerábek resins (Bayer K 2631, Bayer OC-1501, Amberlyst 35,
et al., 1993). Most of these ensembles are placed within and Purolite CT 175) were chosen to be partially
the polymeric matrix and their accessibility to reactants neutralized by Fe(III) and ammonium ions in batch
may be sometimes limited. operation. These cations were selected because they can
On the whole, the strong influence of acidic capacity frequently be found in industrial reactors for MTBE
on catalytic activity of the resins studied for MTBE synthesis, so it is expedient to measure the loss of
synthesis can be explained by supposing that the activity of these resins due to partial neutralization by
polymeric mass of ion exchanger, which suffers a both cations. Two neutralization standards for Fe(III)
moderate swelling due to the presence of significant and only one for ammonium were used. To measure
quantities of methanol, allows reactants to permeate the deactivation produced, initial reaction rates of
with relative freedom and to react faster as the density MTBE formation were determined from the slope of the
of active centers increases. Because of that, the weak curve at zero time of experimental values of methanol
influence of average pore diameter and BET surface conversion versus time. Figure 4 shows the plots of
area on catalytic activity, whose values have been methanol conversion against time for fresh and partly
obtained from dry state, could be explained by supposing deactivated resin K2631. It can be seen that the initial
that the initial structure of dry ion exchanger would slope of the curve for fresh resin is higher than those
have changed due to the reaction microenvironment, for resins neutralized by, respectively, low standard of
and so pore sizes and surface area would have changed Fe(III), high standard of Fe(III), and ammonium. Some
after the moderate swelling. Figures 1 and 2 demon- experiments were repeated to evaluate the experimental
strate this affirmation because there are not significant error on initial reaction rate, which was (0.0011 mol
and defined relationships either between reaction rate h-1 mequiv-1.
and surface area or between reaction rate and average The concentrations of pollutants in the contaminated
pore volume for the resins studied. The results of resins, the remaining acidic capacity after deactivation,
Ind. Eng. Chem. Res., Vol. 37, No. 9, 1998 3579

Figure 4. Methanol conversion values versus time at 333 K for


fresh and deactivated K2631.
Figure 5. Initial reaction rates of MTBE synthesis at 333 K
Table 2. Remainig Acidic Capacities and Initial versus acidic capacities of the resins before and after being polluted
Reaction Rates at 333 K for the Four Resins Used in by iron and ammonium.
MTBE Synthesis as a Function of Fe(III) and Ammonium
Contents Table 3. Loss of Activity for Different Resins Partially
Neutralized by Iron and the Corresponding Values of
acidic
capacity r0 (mol/ Effective Exchange Capacity for Fe(III)
resin pollutant cFe(III) cam (mequiv/g) h‚mequiv) resin ppm Fe(III) L (mequiv/g) nFe(III)
Bayer K-2631 none 4.83 0.037
Bayer K-2631 3613 0.11 1.70
low Fe(III) 3613 4.72 0.036
15904 0.67 2.35
high Fe(III) 15904 4.16 0.033
Bayer OC-1501 3152 0.14 2.48
ammonium 8433 4.38 0.026 14368 0.55 2.13
Bayer OC-1501 none 5.47 0.041 Purolite CT 175 2915 0.13 2.48
low Fe(III) 3152 5.33 0.041 13630 0.57 2.34
high Fe(III) 14368 4.92 0.040 Amberlyst 35 3386 0.14 2.31
ammonium 9653 4.59 0.027 13890 0.36 1.45
Purolite CT175 none 4.98 0.037
low Fe(III) 2915 4.85 0.036 significantly greater than that of Fe(III). An explana-
high Fe(III) 13630 4.41 0.034
tion for this fact is that the presence of ammonium in
ammonium 9696 4.57 0.025 the catalyst besides neutralization could lead to the
Amberlyst 35 none 5.32 0.042
low Fe(III) 3386 5.18 0.041 formation of hydrogen bonds with some sulfonic groups.
high Fe(III) 13890 4.96 0.035 Clearly Fe(III) is incapable of doing so. Furthermore,
ammonium 9822 4.90 0.031 it is known that the ionic radius of Fe(III) is 0.064 nm
and the ionic radius of ammonium is 0.143 nm. So it is
and the initial reaction rate values can be seen in Table likely that a part of the so-called active sites, situated
2. From these data it can be pointed up that fresh inside the polymer mass, were difficult to reach for the
Bayer OC-1501 and Amberlyst 35 are more active reactants due to steric hindrance produced by am-
(higher initial reaction rate) than Purolite CT175 and monium ion placed at external layers of the polymeric
Bayer K2631. This fact was already found above in mass.
continuous operation. While the deactivation by low Figure 5 shows that there exists a linear relationship
concentrations of Fe(III) did not affect significantly the between the initial rates and the acidic capacities of the
reaction rate, high concentration values of Fe(III) led fresh resins, but it can be also seen that the catalytic
to a different decrease in both exchange capacity and activity of the resins polluted by iron follows the same
initial reaction rate. For this high iron contamination, trend as the fresh resins. Table 3 shows the effective
the resin Bayer OC-1501 was the strongest one; its loss exchange capacity values for Fe(III) obtained after
of activity was only 2%, whereas Amberlyst 35 suffered analyzing the relationship between the loss (L) of acidic
a loss of 17%. The loss of acidic capacity varied from capacity (difference between the acidic capacity of fresh
7% to 14%. resin and the remaining acidic capacity determined
Surprisingly, the behavior of the resins was different after the neutralization by iron) of the resins and their
when they were deactivated by ammonium ion. In Fe(III) content. To calculate the effective exchange
general, the loss of activity was more pronounced, capacity of Fe(III), the following expression has been
especially for Bayer OC-1501. The decrease on initial used:
reaction rates (catalytic activities) for the four resins
was now from 26% to 34%, whereas the loss of exchange (1000)(55.85)L
capacity ranged from 8% to 16%. It must be considered nFe(III) ) (3)
that the standard of ammonium employed in deactivat- cFe(III)
ing the resins was less than the high iron standard (see
Table 2) used and also that the exchange number for From Table 3 we can see that values of nFe(III) varies
Fe(III) is 3 times higher than that of ammonium. from an unexpected value of 1.45 for Amberlyst 35 at
When deactivation by ammonium is compared to that high standard of Fe(III) to 2.48 for Bayer OC1501 and
produced by Fe(III) high standard, it could be said that Purolite CT 175 at low standard of Fe(III), but for the
ammonium ion leads to a deactivation of the catalyst most part values are close to 2. So, an average value of
3580 Ind. Eng. Chem. Res., Vol. 37, No. 9, 1998

Table 4. Catalytic Exchange Capacity and Fictitious the reactants with Fe(III) and especially with nitrogen
Loss of Activity Values for Different Resins Partially organic compounds should be avoided in industrial
Neutralized by Ammonium and the Corresponding
operation.
Values of Effective Exchange Capacity for This Ion
ppm CEC L′
resin ammonium (mequiv/g) (mequiv/g) nam
Conclusions
Bayer K-2631 8433 3.40 1.42 3.04 After testing of 12 macroporous resins as catalysts
Bayer OC-1501 9653 3.54 1.93 3.60 on MTBE synthesis, it has been found that the acidic
Purolite CT 175 9696 3.27 1.71 3.17 capacity is the most significant variable for this reaction
Amberlyst 35 9822 4.06 1.26 2.31 This conclusion agrees with other similar studies (Krause
and Hammarström, 1987; Panneman and Beenackers,
2 can be considered as the effective exchange capacity 1995b). It seems that the resins which have greater
of Fe(III) for these resins. density of sulfonic groups were the most active because
The results of initial rates for ammonium are signifi- the reaction of MTBE synthesis involves a concerted
cantly lower than for iron and it was found that there mechanism where some adjacent sulfonic groups par-
is a certain linear relationship between initial rates and ticipate.
the remaining acidic capacities of the resins polluted Four resins were partially deactivated using Fe(III)
by ammonium (Figure 5). We can see in Table 2 that and ammonium ions to evaluate the influence of these
resins polluted by ammonium really have lesser effec- ions on the activity of the resins. The results showed
tive exchange capacities, from a kinetic standpoint, than that for both ions there is a linear relation between the
the ones obtained by titration. It can be inferred again initial reaction rates and the remaining acidic capacities
that ammonium not only neutralizes sulfonic groups but of the resins, but while iron fulfils the same correlation
also obstructs the access of the reactants into the as the fresh resins, ammonium produces standards of
polymer mass where there is a high density of active activity much smaller than those produced by Fe(III).
sites. This conclusion can be confirmed after calculating It seems that ammonium could provoke steric hindrance
the acidic capacities that resins would have, taking into due to the formation of hydrogen bonds with sulfonic
account the initial rates obtained (catalytic exchange groups located in outer layers of the polymeric mass of
capacity) and analyzing the relationship between the the resins studied.
fictitious loss of activity and the quantities of am-
monium ion found in them. Acknowledgment
The fictitious loss of activity (L′) of a resin can be
determined as a difference between the acidic capacity We thank Repsol Petróleo S.A. for its financial sup-
of the fresh resin and the catalytic exchange capacity port and Mr. A. Gonzalez, M. Somoza, and J. A.
calculated by using the significant linear correlation Gonzalez for their technical support.
between initial rates and acidic capacities of fresh resins
and polluted by Fe(III). The appropriate expression to Nomenclature
calculate the catalytic exchange capacity (CEC) values
A ) BET surface area, m2/g
is
A′ ) ISEC surface area, m2/g
C ) acidic capacity, mequiv/g
r0 - 0.00012 cam ) concentration of ammonium in the resin, ppm
CEC ) (4)
0.0076 cFe(III) ) concentration of Fe(III) in the resin, ppm
CEC ) catalytic exchange capacity, mequiv.g-1
To calculate the effective exchange capacity of am- D ) average pore diameter, Å
monium we used the expression: dp ) approximate value of average particle diameter, mm
F ) methanol flow rate, mol.h-1
L ) loss of acidic capacity by Fe(III), meq.g-1
(1000)(18)L′
nam ) (5) L′ ) loss of acidic capacity by ammonium, L′ ) C - CEC,
cam mequiv‚g-1
nam ) effective exchange capacity for ammonium, eq/at-g
Results obtained for CEC, L′ and nam are presented in ammonium
Table 4. It can be seen that the values of effective nFe(III) ) effective exchange capacity for Fe(III), eq/at-g iron
exchange capacity for ammonium range from 2.31 for r ) reaction rate, mol h-1 mequiv-1
Amberlyst 35 to 3.60 for Bayer OC-1501 but an average r0 ) initial reaction rate, mol h-1 mequiv-1
value of 3 could be considered as the representative one. V ) pore volume, mL/g
If we compare both average values of effective ex- W ) mass of dry resin, g
change capacity for Fe(III) and ammonium, it can be XM ) methanol conversion, %
inferred that deactivation by ammonium is surprisingly
much higher than by Fe(III). The standard exchange Literature Cited
capacity of Fe(III) is 3 and the effective value obtained Bayer. Product and Operation Information, 1990.
was only 2. So it can be thought that an important Brockwell, H. L.; Sarathy, P. R.; Trotta, R. Synthesize Ethers.
amount of Fe(III) deactivates the resin by ion exchang- Hydrocarbon Process. 1991, 70, 133.
ing but there is a small part of it placed into the resin Buttersack, C.; Accessibility and Catalytic Activity of Sulfonic Acid
that does not affect the active centers of the catalyst. Ion-Exchange Resins in Different Solvents. React. Polym. 1988,
10, 143.
On the other hand, the standard exchange capacity Chakrabarti, A.; Sharma M. M. Cationic Ion Exchange Resins as
of ammonium is 1 and the effective value found was 3. Catalyst. React. Polym. 1993, 20, 1.
So this ion deactivates the resin not only by neutraliza- Colombo, F.; Dalloro, L. MTBE Synthesis: Activity of Different
tion but also by blocking sulfonic groups. So contact of Catalysts and Kinetics. Geterog. Katal. 1983, 2, 407.
Ind. Eng. Chem. Res., Vol. 37, No. 9, 1998 3581

Gates, B. C.; Rodriguez, W. General and Specific Acid Catalysis Panneman, H. J.; Beenackers, A. A. C. M. Synthesis of Methyl
in Sulfonic Acid Resin. J. Catal. 1973, 31, 27. tert-Butyl Ether Catalyzed by Acidic Ion-Exchange Resins.
Gicquel, A.; Torck, B. Synthesis of MTBE Catalyzed by Ion- Influence of the Proton Activity. Ind. Eng. Chem. Res. 1995b,
Exchange Resin. Influence of Methanol Concentration and 34, 4318.
Temperature. J. Catal. 1983, 83, 9. Parra, D.; Tejero, J.; Cunill, F.; Iborra, M.; Izquierdo, J.F. Kinetic
Guyot, A. Syntheses and Separations using Functional Polymers; Study of MTBE Liquid-Phase Synthesis using C4 Olefinic Cut.
John Wiley & Sons Ltd.: Chichester, England, 1988; Chapt. 1, Chem. Eng. Sci. 1994, 49, 4563.
p 1. Rehfinger, A.; Hoffmann, U. Kinetics of Methyl tert-Butyl Ether
Izquierdo, J.F.; Cunill, F.; Vila, M.; Tejero, J.; Iborra, M. Equilib- Liquid-Phase Synthesis Catalyzed by Ion Exchange ResinsI.
rium Constants for Methyl tert-Butyl Ether Liquid-Phase Intrinsic Rate Expresion in Liquid-Phase Activities.Chem. Eng.
Synthesis. J. Chem. Eng. Data 1992, 37, 339. Sci. 1990, 45, 1605.
Jerábek, K. Personal communication, 1997. Rodriguez, O.; Setinek, K. Dependence of Etherification Rates on
Jerábek, K.; Setinek, K. Polymer Matrix Influence on Ion Ex- Cross-Linking of Ion Exchange Resins Used as Solid Catalysts.
change Resin-Catalyzed Reactions. J. Mol. Catal. 1987, 39, 161. J. Catal. 1975, 39, 449.
Jerábek, K.; Hochmann, T.; Prokop, Z. Influence of Polymer Subramanian, C.; Bathia, S. Liquid-Phase Synthesis of MTBE
Support Morphology on Ion-Exchanger Catalysts Activity in Catalyzed by Ion Exchange Resin. Can. J. Chem. Eng. 1987,
tert-Alkyl Methyl Ethers Synthesis. New Frontiers in Catalysis 65, 613.
1993; Proceedings of the 10th International Congress on Tejero, J.; Cunill, F.; Izquierdo, J. F.; Iborra, M.; Fité, C.; Parra,
Catalysis, Budapest, Hungary; Elsevier Scientific Publishers: D. Scope and Limitations of Mechanistic Inferences from Kinetic
Amsterdam, 1993. Studies on Acidic Macroporous Resins. The MTBE Liquid-Phase
Kmosták, S.; Setı́nek, K. The Effect of Partial Exchange of Synthesis Case. Appl. Catal. 1996, 134, 21.
Sulphonated Macroporous Styrene-Divinylbenzene Copolymers Voloch, M.; Ladisch, M. R.; Tsao, G. T. Methyl tert-butyl Ether
for Different Ions on their Catalytic Activity.Collect. Czechosl. (MTBE) Process Catalyst Parameters. React. Polym. 1986, 4,
Chem. Commun. 1981, 46, 2354. 91.
Krause, A. O. I.; Hammarström, L. G. Etherification of Isoamylenes Widdecke, H. Syntheses and Separations using Functional Poly-
with Methanol. Appl. Catal. 1987, 30, 313. mers; John Wiley & Sons Ltd.; Chichester, England, 1988;
Marston, C. R. Improve Etherification Plant Efficiency and Safety. Chapt. 4, p 149.
Fuel Reform. 1994, 4, 42. Received for review January 7, 1998
Panneman, H. J.; Beenackers, A. A. C. M. Influence of the Revised manuscript received June 8, 1998
Isobutene/Methanol Ratio and of the Methyl tert-Butyl Ether Accepted June 10, 1998
Content on the Reaction Rate of the Synthesis of Methyl tert-
Butyl Ether. Ind. Eng. Chem. Res. 1995a, 34, 3817. IE980007D