Bisphenol A synthesis modeling of industrial reactor

and catalyst deactivation

Zdenek Prokop, Libuse Hankova, Karel Jerabek
*
Institute of Chemical Process Fundamentals, Academy of Sciences of the Czech Republic, Rozvojova 135, Praha 6 165 02, Czech Republic
Received 10 July 2003; received in revised form 3 December 2003; accepted 16 February 2004
Available online 4 June 2004
Abstract
Bisphenol A is an important starting material for production of epoxy resins and polycarbonates, which is manu-
factured by ion exchanger resin catalyzed condensation of acetone and phenol. Much higher reaction rate is achieved
when in the reaction environment are present meracpto-group containing compounds. Alternatively, the promoter can
be either dissolved in the reaction mixture or immobilized in the resin catalyst. Important problem of the process using
the promoted catalyst is relatively rapid catalyst deactivation. Analysis of the reactor operation and mathematical
modeling of its behavior indicate that the cause of the deactivation is a poison in the reaction mixture inhibiting the
promoter. Proper design of laboratory experiments allows investigation of the deactivation process within acceptable
time period and testing of possible preventive actions.
2004 Elsevier B.V. All rights reserved.
Keywords: Bisphenol A; Catalyst deactivation; Ion exchanger catalyst
1. Introduction
Bisphenol A (BPA), an important starting ma-
terial for production of epoxy resins and polycar-
bonates, is manufactured by acid catalyzed
condensation of acetone and phenol.
OH HO
CH
3
3
CH
C
OH
O
CH
3
C
3
CH
+
BPA
H
+
+ H
2
O
(-SH)
OH HO
CH
3
3
CH
C
OH
O
CH
3
C
3
CH
+
BPA
H
+
+
OH HO
CH
3
3
CH
C
OH
O
CH
3
C
3
CH
+
2
BPA
H
+
+
(-SH)
Despite the signicance of the reaction, in lit-
erature is possible to nd surprisingly small num-
ber of papers devoted to technological aspects of
BPA synthesis [19]. Industrial processes for BPA
production usually use ow-trough reactors lled
with strong acid ion exchange resins as catalysts.
Substantial performance improvement is achieved
if in the reaction environment are present merca-
pto-group containing compounds. Presence of the
promoter increases the reaction rate by factor 710
and signicantly improves the selectivity also. The
promoter can be dissolved in the reaction mixture
(e.g. 0.5 wt% of ethylmercaptan) or also immobi-
lized in the resin catalyst. The immobilization is
possible to achieve by neutralization of a part
*
Corresponding author. Tel.: +420-2-203-903-32; fax: +420-
2-209-206-61-332.
E-mail address: kjer@icpf.cas.cz (K. Jerabek).
1381-5148/$ - see front matter 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.reactfunctpolym.2004.02.013
Reactive & Functional Polymers 60 (2004) 7783
www.elsevier.com/locate/react
REACTIVE
&
FUNCTIONAL
POLYMERS
(optimally about 15%) of the acidic groups with 2-
mercaptoethyl amine (cysteamine), either directly
by this compound [10] or after reaction of the
acidic resin with thiazolidine [11], which is in the
presence of the acid catalyst hydrolyzed to cyste-
amine and ketone. The promoter immobilization
eliminates possible problems with smelly emissions
from the process equipments, simplies the process
layout and eliminates possibility of contamination
of the product with sulfur-containing products.
Life of unpromoted ion exchanger catalyst in in-
dustrial condition may be as long as a few years
and the main mechanism of its deactivation is
building-in of polyphenol-type high molecular
weight compounds formed inside the catalyst
grains into the catalyst polymeric matrix. How-
ever, the heterogeneous catalyst containing both
the acid groups and the promoter is more prone to
a deactivation and need to be replaced much
sooner. We have studied character and possible
prevention of this phenomenon on the base of
analysis of actual process data from an industrial
reactor and in laboratory model experiments.
2. Experimental
2.1. Chemicals
As the catalyst was used strongly acidic resin
Ostion KSC-1 (Spolchemie,

Ust nad Labem,
Czech Republic) with the nominal content of di-
vinylbenzene 4 mol% and exchange capacity 5.35
mmol/g. Before use it was washed on column with
diluted hydrochloric acid (10 wt%) and then with
deionized water till the eluate was free of chloride
ions (negative test with silver chloride solution).
To the weighed amount of the resin with know
acid content was added solution of cysteamine
hydrochloride (purum, Fluka, Switzerland) in the
amount corresponding to 15% of the acidic groups
in the resin. The solution was left in contact with
the resin for 24 h, with occasional stirring. After
that the resin was washed again with deionized
water. For the catalytic experiments was used au-
thentic industrial reaction mixture containing
phenol and acetone in molar ratio 10:1 (Spolche-
mie,

Ust nad Labem, Czech Republic).
2.2. Apparatus
For to keep the reaction mixture liquid, all the
setup, including head of the reaction mixture
pump (Sanwa Tsusho Co., Japan), was enclosed in
a heated compartment. As the reactor was used
glass column 6.6 150 mm (Alltech, USA) sur-
rounded by mantle connected with water thermo-
stat maintaining the reactor temperature with
precision better than 0, 1 C. Acetone conversion
in the reaction mixture was evaluated from the
ratio of phenol and BPA concentrations deter-
mined by liquid chromatography on 25 cm column
lled with Nucleosil C8 7l (MarchereyNagel,
Germany) using mixture of 70% methanol and
30% water as the mobile phase. For spectropho-
tometric detection (Spectromonitor 3200, Thermo
Separation Products, USA) was used the wave-
length 290 nm, at which was achieved advanta-
geous sensitivity ratio for phenol and BPA.
3. Results and discussion
3.1. Evaluation of behavior of industrial reactor
From a Czech factory producing BPA using
promoted ion exchanger catalyst in an adiabatic
reactor we obtained records of process parameters
during the life of the catalyst charge. The evalua-
tion of data on behavior of industrial adiabatic
reactor was based on comparison of the actual
catalyst charge in the reactor with the theoretical
catalyst amount computed using the laboratory
determined kinetics of bisphenol A synthesis (Eq.
(1)) on a promoted ion exchanger catalysts [3]. In
the Eq. (1) for the reaction rate r is the parameter k
the rate constant and K
E
is the thermodynamic
equilibrium constant. Reactant concentrations c
and adsorption equilibrium coecients K are
designated by lower index letters a, p, b, w for
acetone, phenol, BPA and water, respectively
r
kc
a
c
2
p
1=K
E
c
b
c
w

1 K
2
a
c
2
a
K
p
c
p
K
w
C
w

4
: 1
Within the technologically signicant tempera-
ture range (4585 C) it is possible to use constant
78 Z. Prokop et al. / Reactive & Functional Polymers 60 (2004) 7783
values for the adsorption coecients for acetone
and phenol (K
a
0:479 dm
3
mol
1
, K
p
0:110
dm
3
mol
1
). Temperature dependence of the other
equation constants is described by Eqs. (2)(4),
where T is temperature in K.
kdm
3
h kgmol
4
24600 exp20500=8:314T;
2
Kmol dm
3
4:3 10
15
exp9000=8:314T;
3
K
w
dm
3
mol
1
0:00214 exp21200=8:314T:
4
For the analysis of behavior of the industrial
reactor operating with ion exchanger catalyst
promoted by neutralization of 15% of its acidic
groups with cysteamine were available the fol-
lowing operational data: whole reaction mixture
input (kg/h), acetone input (kg/h), temperature at
the reactor input (t
1
, usually 4550 C), tempera-
ture at the reactor output (t
2
, usually 7585 C),
production of BPA (kg/h) and weight of the ion
exchanger catalyst (dry) (kg).
All variables in the kinetic equation, including
the local reaction temperature were expressed as
function of the acetone conversion x. The upper
index 0 denotes the concentrations at the reactor
input
c
a
c
0
a
1 x; 5
c
p
c
0
p
2c
0
a
x; 6
c
b
c
0
a
x; 7
c
w
c
0
w
c
0
a
x; 8
T 273 t
2
t
1
=x
output
x t
1
: 9
In Eq. (9) x
output
is the acetone conversion at the
reactor output computed on the base of reactor
operational data.
The industrial adiabatic reactor is of plug-ow
type and hence, by the integration of the kinetic
equation Eq. (1) in limits of the acetone conversion
from zero to the actually observed value is then
possible to obtain theoretical value of the re-
ciprocal space velocity W =F
theor
(Eq. (10))
W =F
theor:

Z
xx
output
x0
1
rx
dx: 10
The ratio of the theoretical reciprocal space
velocity W =F
theor
(Eq. (10)) and actual reciprocal
space velocity W =F
actual
represents the eciency of
the catalyst charge in the reactor (g) as a fraction
of the activity of fresh catalyst determined during
laboratory tests (Eq. (11))
g W =F
theor
=W =F
actual
: 11
The analyses of changes in catalyst charge e-
ciency were performed using data collected during
a number of operation periods of the catalyst
charges. In Fig. 1 are compared changes of acetone
conversion at the reactor output and correspond-
ing computed eciency of the catalyst charge
during a typical operation period. It shows that the
seemingly not too high drop of the conversion
from about 80% on the fresh catalyst to 55% at the
end of the operation period corresponds to much
more dramatic change of the reactor eciency. At
the beginning of the lifetime of the catalyst charge
the computed eciency of the catalyst was close to
100% and it is demonstration of good accuracy of
the kinetic model. With the increasing time on
stream the eciency drops substantially to values
as low as 20%. At the end of the operation period
the reactor performance was evidently equal to
that achievable with a fresh catalyst in amount of
only a small fraction of the reactor charge.
Adiabatic character of the reactor used in the
process made possible the reactor performance by
measurements of the temperature prole within
the catalyst bed and its changes during the reactor
operation period. Obtained results are shown in
Fig. 2.
Changes of temperature proles with catalyst
lifetime (Fig. 2) clearly show that the activity of the
catalyst does not drop simultaneously in all parts
of the bed. At the beginning of the catalyst life the
adiabatic increase of the temperature starts at the
top of the catalyst bed. With increase of the time
on stream the point of temperature rise is shifted
deeper and deeper. The deactivation evidently oc-
curs rst in the layers of the catalyst close to the
reactor input and with the increased amount of
the reaction mixture passed through the reactor
Z. Prokop et al. / Reactive & Functional Polymers 60 (2004) 7783 79
the deactivated catalyst zone spreads further. This
character of the changes in temperature proles
with the time on stream suggests that a poison
present in the reaction mixture deactivates the
catalyst. The shifts of the starting points of the
temperature proles corresponding to various
Amount of reaction mixture passed through the reactor [metric t]
0 2000 4000 6000 8000
A
c
e
t
o
n
e

c
o
n
v
e
r
s
i
o
n

[
%
]

o
r

E
f
f
i
c
i
e
n
c
y

[
%

o
f

t
h
e

f
r
e
s
h
]
0
20
40
60
80
100
Acetone conversion
Efficiency
Fig. 1. Change of eciency of the catalyst and acetone conversion in the industrial reactor during a typical operation period.
45 65 85
Temperature,
o
C
C
a
t
a
l
y
s
t

b
e
d

h
e
i
g
h
t
Start of
operation
Middle of the
catalyst life
End of operation
Reactor output
Reactor input
Fig. 2. Temperature proles measured along the height of the catalyst bed.
80 Z. Prokop et al. / Reactive & Functional Polymers 60 (2004) 7783
phases of the catalyst life are connected with ad-
justments of process parameters attempting to
keep the process production in spite of failing
catalyst performance.
3.2. Laboratory modeling of catalyst deactivation
Deactivation of one charge of the industrial
reactor takes more than 1000 h. For laboratory
modeling of the catalyst poisoning it is necessary
to shorten this time scale by an order of magni-
tude. In the case of the observed frontal deacti-
vation of the catalyst bed it is possible to achieve
this by designing the laboratory reactor as a model
of the element of the input part of the industrial
catalyst bed where the deactivation starts. We have
used a ow- through glass tubular reactor with the
inner diameter 6.6 mm and the cross-section area
0.37 cm
2
. For to achieve linear ow velocity of the
reactant in this tube similar to that in the industrial
reactor, there was necessary to feed the reaction
mixture at ow rate 40100 cm
3
/h. For to shorten
the time needed for the catalyst deactivation as
much as possible, during the laboratory model
experiments was used the low rate 100 cm
3
/h.
Changes in acetone conversion determined
from liquid chromatography analyses of BPA
content in the reaction mixture were analyzed
similarly as described above for the industrial re-
actor and were expressed as the eciency of the
catalyst charge. Results are shown in Fig. 3.
As the empty symbols are depicted the results of
two independent runs using the same batch of the
reaction mixture. In both runs the eciency of the
catalyst with the increasing of the amount of re-
action mixture passed through the reactor de-
creased to about 2030% of the starting level.
Satisfactory agreement of the results of both runs
shows good reproducibility of the experiments.
The ordinary reasons for ion exchanger catalyst
deactivation are either poisoning of the acidic
centers or diminishing of their accessibility [8].
However, in our case the titration of the deacti-
vated resin with alkali solution shown that the
total number of acidic centers per whole reactor
charge is in the deactivated catalyst practically the
same as in the fresh one. Also, analysis of the
swollen-state morphology of the resin catalysts by
inverse steric exclusion chromatography [12] did
not shown appreciable morphology dierences
between the fresh and deactivated catalyst. It
proved that the observed poisoning does not di-
minish signicantly either the number of the acidic
centers or their accessibility. As the only probable
mechanism of the poisoning eect remains possi-
bility that the poison destroys the promoter
0.0
20.0
40.0
60.0
80.0
100.0
0 5000 10000 15000 20000
Amount of the reaction mixture passed through the
catalyst [g/g]
E
f
f
i
c
i
e
n
c
y

[
%
]
Fig. 3. Changes of the eciency of the catalyst bed in laboratory modeling of the catalyst deactivation. , results obtained in two
independent runs with 0.4 g catalyst modied with cysteamine, reaction mixture of phenol and acetone in mol. ratio 10, ow rate 100
cm
3
/h, 70 C. d results obtained with the guard bed (8 31 mm) placed before the reactor impute and composed of ion exchange
resin saturated with cysteamine (other conditions identical).
Z. Prokop et al. / Reactive & Functional Polymers 60 (2004) 7783 81
component of the catalyst, that is cysteamine
bound to a part of the acidic centers. With this
explanation is in agreement the shape of the de-
activation curves in Fig. 3, leveling at the eciency
2030%, which is the level of performance close to
that of unpromoted catalyst [3]. In such a case
there should be possible to eliminate the poison
from the reaction mixture by passing the mixture
through a suitable guard bed located before main
body of the catalyst. Such method is described in
our patent from 1993 [13] and practically identical
of the catalyst was later patented also in USA [14].
Eciency of this of this arrangement was tested
in an experiment in which the reaction mixture be-
fore entering the reactor passedthrougha guardbed
of ion exchange resin saturated with cysteamine.
Results of this experiment are in Fig. 3 reported as
full symbols. There is evident that the guard bed
virtually prevented deactivation of the catalyst. The
poison contained in the industrial reaction mixture
was adsorbed in the cysteamine-saturated resin.
This experiment proved feasibility of stabilization
of the performance of promoted catalyst for BPA
synthesis by purication of the reaction mixture.
As a test of possible risks of application of
cysteamine saturated resin as the poison lter we
also checked possible hypothesis that the perfor-
mance of the main catalyst bed could be inuenced
by leakage of cysteamine from the guard bed. In
this test behind the guard bed (6.5 g of the resin
saturated with cysteamine) was connected the re-
actor lled with 0.935 g of the unpromoted cata-
lyst. BPA synthesis is very sensitive to the presence
of a promoter [3] and if there would be a leakage
of cysteamine, the activity of the unpromoted
catalyst should increase. There was observed no
change in the catalyst bed performance even after
long-time exposition (54 h) of the unpromoted
catalyst to the stream passing through the pro-
moter-saturated guard bed. It proves that the
stabilization eect of the guard bed on the reactor
performance cannot be explained by a leakage of
the promoter into the reactor.
Stabilization of the activity of promoted resin
catalyst by ltration of the authentic industrial
reaction mixture through the guard bed experi-
mentally proved that for the deactivation of the
promoted ion exchanger catalysts for synthesis of
BPA is responsible compound(s), which strongly
bonds with the mercapto groups of the promoter.
The American patent [5] mentions as the possible
poison hydroxyacetone. This compound can be
present as impurity in phenol produced by cumene
hydroperoxide method. Our own experiments with
articial additions of hydroxyacetone into the re-
action mixture conrmed eciency of this com-
pound as specic poison diminishing the activity
of promoted ion exchanger catalyst for BPA syn-
thesis. However, there is uncertain that the culprit
in our industrial process is really hydroxyacetone.
Accumulated industrial data shows that the ob-
served decrease of the reactor performance indi-
cates poisoning of about 0.20.6 g of the promoted
catalyst by 1 kg of the reaction mixture. If the
poison would be a simple compound with mol.
weight less than 100, its concentration in the re-
action mixture should be lower than 0.05 wt%.
Analytical determination of so small concentra-
tions of components in the mixture composed
predominantly of phenol is rather dicult. Anal-
yses of the industrial mixture on GCMS HP 6890
(HewletPackard, USA) using 30 m capillary col-
umn DB-5MA with inner diam. 0.25 mm per-
formed in our analytical laboratory shown beside
acetone and phenol the presence of a number of
minor components, however neither of them was
identied as hydroxyacetone. Trial analysis of a
sample to which 0.1 wt% hydroxyacetone was
added shown that such concentration is in the used
arrangement well identiable, but the detection of
components having concentration lover than 0.05
wt% could be uncertain. Positive unequivocal de-
termination of the identity of the poison actually
existing in the industrial reaction mixture is at
present behind our analytical possibilities but the
quest continues.
References
[1] R.A. Reinicker, B.C. Gates, AIChE J. 20 (1974) 993.
[2] K. Jerabek, J. Odnoha, K. Setnek, Appl. Catal. 52 (1987)
1272.
[3] K. Jerabek, Li Gwang Hun, K. Setnek nek, (Eng) Coll.
Czech. Chem. Commun. 54 (1989) 321.
[4] A.P. Singh, Catal. Lett. 16 (1992) 431435.
[5] M. Kiedik, A. Krueger, Przem. Chem. 71 (1992) 335337.
82 Z. Prokop et al. / Reactive & Functional Polymers 60 (2004) 7783
[6] M. Kiedik, A. Krueger, Przem. Chem. 72 (1993) 306307.
[7] K. Nowinska, W. Kaleta, Appl. Catal. A 203 (2000) 91100.
[8] A. Krueger, W. Balcerowiak, E. Grzywa, React. Funct.
Polym. 45 (2000) 1118.
[9] K. Jerabek, L. Hankova, Z. Prokop, E.G. Lundquist, 232
(2002) 181188.
[10] B.W. McNutt, B.B. Gammill, Dow Chemical Co., U. S.
Patent 3,394,089, 1968.
[11] B.B. Gammill, G.R. Ladewig, G.E. Ham, Dow Chemical
Co., US Patent 3,634,341, 1972.
[12] K. Jerabek, K. Setnek, J. Polym. Sci. Part A: Polym.
Chem. 28 (1990) 1387.
[13] K. Jerabek, V. Homola, V. Musil, J. Novotny, Czech. Pat.
278 (1993) 463.
[14] E.J. Pressman, S.J. Shafer, Method for making bisphenol,
US Patent 5,414,151, 1995.
Z. Prokop et al. / Reactive & Functional Polymers 60 (2004) 7783 83