3180 Ind. Eng. Chem. Res.

2007, 46, 3180-3187

Reaction Kinetics of the Catalytic Esterification of Citric Acid with Ethanol

Aspi K. Kolah, Navinchandra S. Asthana, Dung T. Vu, Carl T. Lira, and Dennis J. Miller*
Department of Chemical Engineering & Materials Science, Michigan State UniVersity,
East Lansing, Michigan 48824

Reaction kinetics are presented for the reversible esterification reaction of citric acid with ethanol to form
tri-ethyl citrate via mono-ethyl and di-ethyl citrates. The reaction was studied in batch isothermal experiments,
self-catalyzed homogeneously by citric acid and the formed mono- and di-ethyl citrates, and heterogeneously
catalyzed by macroporous Amberlyst-15 ion-exchange resin catalyst. Experimental data were obtained between
78 and 120 °C at different mole ratios of ethanol to citric acid and catalyst concentrations up to 5 wt %
ion-exchange resin. The kinetics of ethanol etherification to form di-ethyl ether were included in the
investigation. Kinetic modeling was performed using a pseudo-homogeneous UNIQUAC-based activity model,
taking into consideration the rate of self-catalyzed esterification and the side reaction to form diethyl ether.
The activity coefficients for the tri-ethyl citrate-ethanol and tri-ethyl citrate-water binary pairs were obtained
from experimental vapor-liquid equilibrium data. Kinetics of the di-ethyl citrate to tri-ethyl citrate reaction
limit the overall tri-ethyl citrate formation rate, as citric acid and mono-ethyl citrate are esterified rapidly to
their equilibrium compositions. Higher temperatures lead to faster reaction kinetics but significantly increase
the production of the undesired byproduct di-ethyl ether. The kinetic model developed is useful for the design
and simulation of processes such as reactive distillation for tri-ethyl citrate formation.

1. Introduction operation. For reviews up to 1995 on the use of ion-exchange

resins as catalysts, the reader is referred to the work of Olah et
There is ever-increasing interest in the production and use al.,2 Chakrabarti and Sharma,3 and Sharma4 For a more recent
of biorenewable chemicals, fueled in part by rising global crude review, the reader is referred to the work of Harmer and Sun.5
oil prices and a growing desire to reduce dependence on
petroleum. Bio-based chemicals typically are environmentally Prior studies on the esterification of citric acid with ethanol
friendly, possess low toxicity, and have favorable biodegrad- or n-butanol are found mainly in the Chinese and German patent
ability, making them prime candidates for replacements of literature. Schröter et al.6 describe a process for synthesis of
petroleum-based products. The esters of bio-based organic acids TEC via a three-stage batch process using methanesulfonic acid
fall into the category of benign or green solvents and are as catalyst. Tao7 discusses the synthesis of TEC in the presence
promising replacements for halogenated petroleum-based sol- of p-toluenesulfonic acid as catalyst and with continuous
vents in a wide variety of applications. removal of the formed water. Frappier et al.8 discuss a process
Citric acid (2-hydroxy-1,2,3-propanetricarboxylic acid) is for synthesis of TEC from fermentation broths containing citric
produced commercially via fungal fermentation of glucose. acid. Some recent attempts to synthesize tri-butyl citrate have
Citric acid can be esterified with alcohols such as ethanol and been described by Nong9 using an aluminum phosphotungstate
n-butanol through a series of reactions to yield tri-ethyl citrate supported catalyst, Shi et al.10 using a solid super acid catalyst,
(TEC) and tri-n-butyl citrate (Figure 1). Tri-ethyl citrate and Zheng et al.11 using tetra-butyl titanate, Deng et al.12 using
tri-butyl citrate are used as nontoxic plasticizers in toys, medical sodium hydrogen sulfate, Song et al.13 using an aluminophos-
products (e.g., as enteric coatings for controlled release drug phate solid acid catalyst, Liu et al.14 using a dealuminated USY
delivery systems), printing ink coatings, cosmetics, and other zeolite, Meng et al.15 using a nanosolid super acid SO42-/Fe2O3,
applications. These plasticizers are also suitable as food additives Zhao et al.16 using a WO3-TiO2-SO42- super acid catalyst,
such as whipping agents for dried egg whites, food flavorings, Meng et al.17 using a p-toluenesulfonic acid catalyst, and Fu et
or food packaging materials. Citrate esters rapidly metabolize al.18 using a SO42- modified zirconium crosslinked clay catalyst.
in the body via liver and blood serum enzymes to liberate the The esterification of citric acid is an equilibrium-limited
citrate ion, which is disposed of through the usual biochemical reaction. In order to overcome the equilibrium limitation, it is
pathways. The global plasticizers market has been estimated at necessary either to carry out esterification in multiple stages or
around 11 billion pounds per year; according to 2003 statistical to use a process such as reactive distillation, in which chemical
data, the U.S. share of this market is 2.4 billion pounds.1 reaction and distillation occur in a single vessel. Design of
In a conventional process to synthesize TEC, citric acid and reactive distillation processes often involves equilibrium-based
ethanol are combined in a batch or continuous stirred reactor models in which both phase and chemical equilibrium are
using a homogeneous catalyst such as sulfuric acid. Use of achieved on each stage.19 Such equilibrium models suffice when
heterogeneous solid catalysts can eliminate many of the reaction kinetics are very fast, but for slow chemical reactions
disadvantages involved with use of homogeneous catalysts; such as citric acid esterification, the kinetics must be taken into
heterogeneous catalysts allow easy separation of the catalyst account in order to achieve a reasonable design. For a review
from the reaction media by decantation or filtration, reduce or on kinetic-based models useful in reactive distillation processes,
eliminate corrosion problems, and facilitate continuous process the reader is referred to the work of Taylor and Krishna19 and
Sundmacher and Qi.20
* To whom correspondence should be addressed. Tel.: 1-517-353- Because no information is available in the open literature
3928. Fax: 1-517-432-1105. E-mail: millerd@egr.msu.edu. describing the kinetics of citric acid esterification over ion-
10.1021/ie060828f CCC: $37.00 © 2007 American Chemical Society
Published on Web 12/20/2006
 Ind. Eng. Chem. Res., Vol. 46, No. 10, 2007 3181

Figure 1. Esterification of citric acid

exchange resin catalysts, we have undertaken a study of citric Q. The column oven was subject to a temperature program
acid esterification kinetics in order to develop a rate model that involving heating from 413 K (after a 2 min hold) to 493 K
will be useful in designing reactive distillation processes for (and held for 6 min) at a rate of 20 K/min. n-Butanol was used
tri-alkyl citrate formation. Both resin-catalyzed and citric acid- as an internal standard. High purity helium (99.999% pure) was
catalyzed (“self-catalyzed”) reactions are included, and ethanol used as carrier gas at a flow rate of 20 mL/min. The injector
dehydration to form di-ethyl ether (DEE) is included as part of and detectors were maintained at 493 K.
the reaction system. Experimental vapor-liquid equilibrium Samples were analyzed for ethanol and DEE using a Perkin-
(VLE) studies were performed on the citric acid-ethanol and Elmer Sigma-2000 gas chromatograph equipped with a flame
citric acid-water binary pairs in order to obtain the UNIQUAC ionization detector (FID) and a bonded-phase fused-silica
activity coefficients. A pseudo-homogeneous activity-based capillary column (SPB-5, 30 m × 0.53 mm). The column oven
kinetic model is presented for correlation of the experimental was subject to a temperature program involving heating from
data. 313 K (after a 7 min hold) to 473 K (and held for 5 min) at a
rate of 2 K/min. Anhydrous toluene was used as an internal
2. Experimental standard. High purity helium (99.999% pure) was used as carrier
gas at a flow rate of 10 mL/min. The injector and detectors
2.1. Materials. Anhydrous citric acid crystals were obtained
were maintained at 493 K.
from Aldrich Chemical Co. Absolute ethanol (99% purity) and
high-performance liquid chromatography (HPLC) grade water Analysis of samples from the VLE experiments was per-
were obtained from J. T. Baker, Inc. The strong acid cation- formed with a Varian 3400 gas chromatograph with both an
exchange resin catalyst Amberlyst-15 (Rohm and Haas, Phila- FID and a TCD. Column packing was 10% OV-101 on
delphia, PA) was obtained in H+ form and was used without Chromosorb W-HP 80/100 with a helium flow rate at 20 mL/
modification. The purity of all chemicals was checked by gas min.
chromatography or HPLC. 2.3. Batch Kinetic Experiments. Esterification reactions at
For the VLE experiments, water (HPLC grade) was obtained 78 °C were performed in a 2 × 10-4 m3 jacketed glass reactor
from J.T. Baker, Inc. Ethanol (200 proof) and TEC (99% purity) equipped with a recirculating constant temperature oil bath. The
were purchased from Sigma-Aldrich. Chemicals were used as reaction volume was maintained between 100 and 110 mL. A
received. spiral coil condenser, open to the atmosphere, was placed on
2.2. Analysis. The presence of citric acid, mono-ethyl citrate top of the reactor. The glass reactor was equipped with
(MEC), di-ethyl citrate (DEC), and TEC was first confirmed temperature and stirrer speed monitoring devices and a sampling
by gas chromatography-mass spectroscopy (GC-MS) analysis port. In operation, measured quantities of ethanol and citric acid
of their trimethylsilyl (TMS) derivatives. For reaction samples, were added to the reactor and heating and stirring were started
citric acid and its ethyl esters (MEC, DEC, and TEC) were simultaneously. Once the desired temperature was achieved,
quantitatively analyzed on a Hewlett-Packard 1090 HPLC using usually in about 15 min, catalyst (Amberlyst-15 ion-exchange
a reversed phase C18 column (Novapak, 3.9 mm × 150 mm) resin) was added for the case of resin-catalyzed reactions and
held at 40 °C. Water/acetonitrile (ACN) mixtures, buffered at the stirring speed was increased to 800 rpm. This point in time
pH ) 1.3, were used as mobile phase (1.0 mL/min) in a gradient was considered as the zero reaction time. Samples were
mode (0% ACN (t ) 0) to 60% ACN (t ) 20 min) to 90% withdrawn at specific time intervals and immediately transferred
ACN (t ) 25 min) to 0% ACN (t ) 28 min)), and species were to an ice bath (prior to analysis) in order to ensure that no further
quantified by UV detection (Hitachi L400H) at a wavelength reaction took place.
of 210 nm. Citric acid and TEC were identified and quantified For reaction temperatures of 90 °C and above, esterification
by comparing HPLC retention time and peak area with their was performed in a 1 × 10-4 m3 stainless steel autoclave (5000
respective calibration standards. Standards for MEC and DEC Multi-reactor System, Parr Instrument Co.) equipped with
could not be obtained commercially. On a mass basis, the temperature and stirrer speed monitors and a sampling port. In
response factors for citric acid and TEC were found to be same; operation, measured quantities of ethanol, citric acid, and catalyst
therefore, MEC and DEC were each assigned the same response for the resin-catalyzed reaction cases were added to the reactor
factor as TEC and citric acid. Using this response factor, the and heating was started with slow stirring. The total reaction
carbon balance for each reaction sample, based on citric acid volume was maintained between 55 and 60 mL. The desired
and its esters, was in the range of (10%. temperature was achieved in about 15 min, at which time the
Reaction samples were analyzed for water content using a stirring rate was increased to 740 rpm. This time was considered
Varian 3700 gas chromatograph equipped with a thermal as the zero reaction time. Samples were withdrawn at specific
conductivity detector (TCD) and a stainless steel column (4 m time intervals through a cooled metal tube and immediately
× 3.25 mm) packed with a liquid stationary phase of Porapak transferred to an ice bath in order to ensure that no further
3182 Ind. Eng. Chem. Res., Vol. 46, No. 10, 2007

Table 1. Summary of Kinetic Experiments and Average Prediction Errors

run figure mole ratio resin catalyst temp average relative error (Frel) (eq 17) average absolute error (Fabs) (eq 16)
no. no. EtOH:CA loading (wt %) (oC) CA MEC DEC TEC EtOH water CA MEC DEC TEC EtOH water
1 2/S1 15:1 5 78 8.3 10.6 34.8 27.9 1.8 19.0 0.16 0.17 0.29 0.08 1.55 0.92
2 S2 15:1 5 90 13.5 10.7 37.4 57.5 2.5 24.4 0.35 0.21 0.40 0.16 1.41 0.77
3 3/S3 15:1 5 100 13.2 8.9 24.2 44.6 2.8 19.2 0.26 0.16 0.43 0.34 2.55 1.80
4 S4 15:1 5 110 9.3 20.9 19.7 39.8 2.2 13.4 0.11 0.28 0.35 0.32 1.75 1.35
5 4/S5 15:1 5 120 12.8 14.1 12.5 27.5 2.7 8.8 0.06 0.18 0.30 0.33 2.06 1.13
6 S6 15:1 3 120 12.1 16.8 6.1 27.9 0.9 3.8 0.08 0.23 0.15 0.26 0.70 0.41
7 5/S7 15:1 2 120 12.6 13.1 11.9 38.6 1.0 17.8 0.09 0.20 0.16 0.23 0.86 1.00
8 S8 15:1 1 120 13.4 10.1 26.4 48.7 5.6 23.6 0.18 0.20 0.50 0.49 4.34 2.67
9 S9 15:1 3 78 19.5 7.6 14.2 43.4 1.3 8.3 0.24 0.17 0.08 0.06 1.10 0.31
10 6/S10 15:1 2 78 21.4 8.4 13.6 40.0 0.8 4.7 0.37 0.21 0.06 0.05 0.67 0.15
11 S11 15:1 1 78 13.9 6.0 16.4 13.2 1.4 6.1 0.35 0.15 0.09 0.01 1.20 0.32
12 S12 20:1 5 120 25.8 12.9 13.5 16.9 0.5 3.1 0.08 0.09 0.22 0.16 0.43 0.30
13 S13 10:1 5 120 11.9 15.3 6.6 28.7 2.3 6.8 0.19 0.25 0.20 0.48 1.60 1.18
14 7/S14 5:1 5 120 12.5 27.1 7.4 35.7 9.2 10.1 0.34 0.98 0.43 1.14 4.02 3.31
15 S15 20:1 0 120 14.8 6.7 11.1 30.2 0.5 6.9 0.12 0.09 0.15 0.16 0.40 0.37
16 8/S16 15:1 0 120 14.6 14.3 13.2 17.1 3.1 10.3 0.18 0.37 0.46 0.10 2.51 1.28
17 S17 10:1 0 120 15.2 2.1 10.6 9.1 1.7 5.6 0.11 0.05 0.39 0.05 1.23 0.84
18 S18 5:1 0 120 6.9 4.4 11.1 6.0 4.2 5.2 0.09 0.18 0.82 0.12 2.12 1.51
19 S19 15:1 0 78 4.1 15.3 40.1 22.6 1.0 20.5 0.18 0.22 0.10 0.00 0.88 0.30
20 S20 15:1 0 90 5.7 6.1 22.8 37.3 0.9 11.1 0.15 0.11 0.12 0.02 0.76 0.33
21 S21 15:1 0 100 2.9 6.0 22.1 29.2 2.0 13.8 0.05 0.15 0.32 0.06 1.68 0.92
22 S22 15:1 0 110 8.0 4.5 15.4 19.7 1.2 8.2 0.08 0.10 0.25 0.05 1.06 0.52

reaction took place before analysis. All samples were analyzed 120 °C; Figures 2 and 6 show results for the same loadings at
using the method described in section 2.2. 78 °C. The full set of graphs is available in the Supporting
2.4. Vapor-Liquid Equilibrium Experiments. The primary Information (Figures S5-S8 for 120 °C and Figures S1 and
volatile species in this reaction system are ethanol and water, S9-S11 for 78 °C). Analysis of the initial reaction rate shows
so the vapor-liquid equilibria behavior of these components that the esterification rate is linearly dependent on catalyst
with TEC was characterized in order to obtain TEC-ethanol loading.
and TEC-water binary pair thermodynamic parameters for
reliable process design. A P-x-y apparatus, described in detail
in a previous paper from the authors’ laboratory,21 was used in
the investigation.

3. Results and Discussion

Several batch kinetic experiments were carried out to study
the effects of reaction temperature, catalyst loading, and initial
reactant molar ratio on the heterogeneously catalyzed esterifi-
cation of citric acid with ethanol. Table 1 shows the reaction
conditions and summarized results for all of the experimental
studies performed in this work. We observed from initial
experiments and comparison with prior work that external mass-
transfer resistances were negligible at stirring speeds above 500
rpm. Hence, all kinetic experiments were performed at 800 rpm. Figure 2. Esterification of citric acid catalyzed by ion-exchange resin. The
reaction conditions are as follows: mole ratio ethanol:citric acid, 15:1;
The influence of internal mass transfer resistances was neglected catalyst loading, 5 wt %; reaction temperature, 78 °C (-9-, CA; -b-,
for reactions catalyzed by Amberlyst-15.22,23 MEC; -2-, DEC; -×-, TEC; --∆--, DEE).
3.1. Effect of Reaction Temperature. The effect of increas-
ing the reaction temperature from 78 to 120 °C on the
esterification of citric acid with ethanol at a catalyst loading of
5 wt % and an initial mole ratio of ethanol:citric acid of 15:1
was studied in the present work. Figures 2-4 give a full
description of product evolution at 78, 100, and 120 °C,
respectively. The full set of graphs is located in the Supporting
Information (Figures S1-S5). It can be observed from Figures
2-4 that the rate of conversion of citric acid, MEC, and DEC
increases with the increasing reaction temperature. Experimental
quantification of DEE formation was carried out at reaction
temperatures of 100, 110, and 120 °C. At lower temperatures,
negligible quantities of DEE were formed.
3.2. Effect of Catalyst Loading. The effect of varying the
catalyst loading from 1 to 5 wt % (of the reaction solution) on
Figure 3. Esterification of citric acid catalyzed by ion-exchange resin. The
citric acid esterification with ethanol at an initial mole ratio of reaction conditions are as follows: mole ratio ethanol:citric acid, 15:1;
ethanol:citric acid of 15:1 was examined. Figures 4 and 5 give catalyst loading, 5 wt %; reaction temperature, 100 °C (-9-, CA; -b-,
results for 5 and 2 wt % catalyst loading, respectively, at MEC; -2-, DEC; -×-, TEC; --∆--, DEE).
 Ind. Eng. Chem. Res., Vol. 46, No. 10, 2007 3183

Figure 4. Esterification of citric acid catalyzed by ion-exchange resin. The Figure 7. Esterification of citric acid catalyzed by ion-exchange resin. The
data at the right edge of the graph represent the liquid-phase composition data at the right edge of the graph represent the liquid-phase composition
at the end of reaction (t ) 1600 min). The reaction conditions are as at the end of reaction (t ) 1600 min). The reaction conditions are as
follows: mole ratio ethanol:citric acid, 15:1; catalyst loading, 5 wt %; follows: mole ratio ethanol:citric acid, 5:1; catalyst loading, 5 wt %; reaction
reaction temperature, 120 °C (-9-, CA; -b-, MEC; -2-, DEC; -×-, temperature, 120 °C (-9-, CA; -b-, MEC; -2-, DEC; -×-, TEC;
TEC; --∆--, DEE). --∆--, DEE).

Figure 5. Esterification of citric acid catalyzed by ion-exchange resin. The Figure 8. Self-catalyzed esterification of citric acid. The reaction conditions
data at the right edge of the graph represent the liquid-phase composition are as follows: mole ratio ethanol:citric acid, 15:1; reaction temperature,
at the end of reaction (t ) 1600 min). The reaction conditions are as 120 °C (-9-, CA; -b-, MEC; -2-, DEC; -×-, TEC; --∆--, DEE).
follows: mole ratio ethanol:citric acid, 15:1; catalyst loading, 2 wt %;
reaction temperature, 120 °C (-9-, CA; -b-, MEC; -2-, DEC; -×-, 3.4. Self-Catalyzed Reactions. The self-catalyzed reaction
TEC; --∆--, DEE). of citric acid with ethanol at temperatures from 78 °C to
120 °C and initial mole ratios of ethanol:citric acid from 5:1 to
20:1 was examined; a representative graph is shown in Figure
8 with full results available in the Supporting Information
(Figures S15 - S22). Comparing the self-catalyzed rate with
that of the ion-exchange resin-catalyzed reactions, it is evident
that the self-catalyzed rate contributes significantly to esterifi-
cation at low catalyst loadings, especially at the higher reaction
temperature of 120 °C.
3.5. Vapor-Liquid Equilibria. Experimental P-x data for
TEC (1) + ethanol (2) were collected at 40 °C as shown in
Figure 9 and fit (solid line in Figure 9) using the UNIQUAC
model to obtain binary interaction parameters. Experimental
P-x data for TEC (1) + water (2) at 25 and 60 °C are shown
Figure 6. Esterification of citric acid catalyzed by ion-exchange resin. The in Figure 10 and are similarly fit. Calculated vapor-phase
reaction conditions are as follows: mole ratio ethanol:citric acid, 15:1; compositions are also included in the figures. The r and q values
catalyst loading, 2 wt %; reaction temperature, 78 °C (-9-, CA; -b-, used in the UNIQUAC model are 5.9585 and 4.808 for CA,
MEC; -2-, DEC; -×-, TEC; --∆--, DEE). 7.1912 and 6.380 for MEC, 8.4674 and 7.424 for DEC, 9.7436
3.3. Effect of Initial Reactant Mole Ratio. The effect of and 8.468 for TEC, 2.1055 and 1.972 for ethanol, 0.92 and 1.4
varying the initial ethanol:citric acid mole ratio is shown in for water, and 3.3949 and 3.016 for DEE. The TEC-water
Figures 4 and 7 for an initial ethanol:citric acid mole ratio of binary system exhibits liquid-liquid immiscibility (horizontal
15:1 and 5:1, respectively. All reactions were conducted with line in Figure 10) with one phase nearly pure in water and the
5 wt % catalyst loading at 120 °C. Complete results are given other phase approximately equimolar in water and TEC. ASPEN
in the Supporting Information (Figures S5 and S12-S14). The Plus (version 12.0) was unable to converge on an improved
equilibrium extent of conversion to TEC increases with the parameter set. It is not possible to discern the vapor-liquid
increasing initial ethanol to citric acid molar ratio. behavior of the system near the pure water side from the
3184 Ind. Eng. Chem. Res., Vol. 46, No. 10, 2007

Table 2. UNIQUAC Interaction Parameters for Binary Component

Pairs in the Form τij ) exp(Aij + Bij/T)
component pairs Aij Aji Bij (K) Bji (K)
ethanol-water 2.0046 -2.4936 -728.971 756.948
ethanol-TEC 70.5 -301.6
TEC-water -501.8 82.56
CA-ethanol -139.839 54.177
CA-TEC 90.604 -172.585
CA-water 92.644 53.676
CA-MEC 28.751 -33.13
CA-DEC -19.605 -0.897
ethanol-MEC -79.836 17.574
ethanol-DEC -135.446 60.371
TEC-MEC -154.776 107.843
TEC-DEC -95.921 77.814
Figure 9. P-x data for TEC(1)-ethanol(2) at 40 °C compared with the Water-MEC -447.773 263.187
UNIQUAC fit. The lower line represents the calculated vapor-phase Water-DEC -336.304 178.184
composition. MEC-DEC 28.784 -34.376
ethanol-DEE -3.7063 5.3512 1175.781 -1893.535
water-DEE -50.888 -611.06
CA-DEE 47.749 -457.830
TEC-DEE 103.507 -171.685
MEC-DEE 84.542 -315.967
DEC-DEE 105.529 -244.055

cation, the model is based on the law of mass action as first

order with respect to catalytic acid concentration, reacting acid
concentration, and ethanol concentration. An activity-based
model is preferred over the conventional concentration-based
model, since water and ethanol are highly polar in comparison
to other components of the system and thus strong nonideal
behavior exists. Moreover, kinetic models used in process
simulation and design using programs such as Aspen Plus are
best written in terms of activities.22,24-27
Figure 10. P-x data for TEC(1)-water(2) at 25 and 60 °C compared
with the UNIQUAC fit (-[-, 25 °C; --2--, 60 °C). The lower lines The UNIQUAC model was used to determine activity
represent the calculated vapor-phase compositions. coefficients in the liquid phase. The UNIQUAC parameters for
the TEC-ethanol and TEC-water pairs were determined using
measurements, because the solubility of TEC in water is very the experimental VLE data presented in section 3.5 of this paper.
low. The UNIQUAC parameters used in fitting of the experi- The UNIQUAC parameters for the ethanol-water, DEE-
mental VLE data sets are shown in Table 2. VLE data for MEC ethanol, and DEE-water pairs were obtained from literature
and DEC were not collected because they are not available data reported in the ASPEN Plus databank. The rest of the
commercially and were not isolated experimentally. UNIQUAC parameters were obtained by fitting the UNIQUAC
model to the UNIFAC predictions. Table 2 shows the UNI-
4. Kinetic Model QUAC parameters used in the present modeling work.
4.2. Reaction Equilibrium Constants. Chemical reaction
4.1. Kinetic Pathways. Reactions 1-4 below describe the equilibrium constants are given by
esterification of citric acid with ethanol and the formation of
side product DEE. Ka,i ) ∏aiν ) Kx,i.Kγ,i ) ∏xiν ‚∏γiν
i i i
(5)
k1
CA + EtOH {\
k /K
} MEC + W (1) The values of the equilibrium constants Ka,i for esterification
1 eq,a,1 reactions were determined by analysis of the experimental data
k2
at long reaction times (e.g., approaching equilibrium) and were
MEC + EtOH {\
k /K
} DEC + W (2) found to be 6.35, 2.72, and 3.78 for the formation of MEC,
2 eq,a,2
DEC, and TEC, respectively. These constants were taken to be
independent of temperature. The values of Kx,i are technically
k3
DEC + EtOH {\
k /K
} TEC + W (3) not constant but, over the range of typical reaction conditions,
3 eq,a,3 were found to be approximately 6.35, 1.82, and 0.54 for the
k4
formation of MEC, DEC, and TEC, respectively. The values
2EtOH 98 DEE + W (4) of Kγ,i, also not constant, were averaged from γi determined
from UNIQUAC at the extrapolated equilibrium concentrations
The formation of DEE (eq 4) via dehydration of two and were found to be approximately 1.0, 1.5, and 7.0 for the
molecules of ethanol is significant at relatively high temperatures formation of MEC, DEC, and TEC, respectively.
and high catalyst loadings. Dehydration of ethanol to DEE is 4.3. Kinetic Model for Self-Catalyzed Esterification. The
considered irreversible, since the equilibrium constants for such rate of self-catalyzed esterification is a significant fraction of
reactions are known to be very high.24 Formation of DEE as a the ion-exchange resin-catalyzed rate at low catalyst loadings
zero-order reaction was considered only in the presence of ion- and high temperatures. The self-catalyzed reactions are thus
exchange resin. modeled using a pseudo-homogeneous model based on the law
On the basis of the above reactions, a pseudo-homogeneous, of mass action as first order with respect to reactants and to
activity-based kinetic model has been developed. For esterifi- catalytic acid concentration, denoted as xacid in the equation
 Ind. Eng. Chem. Res., Vol. 46, No. 10, 2007 3185

below. The reaction rate for the self-catalyzed esterification is Table 3. Kinetic Model Parameters for Self-Catalyzed Reactions
written in generalized form as parameters units values

( )
aEsteraWater k°self,1 1/s 8.37 × 106
rself,i ) kself,ixacid aReacting_acidaAlcohol - (6) k°self,2 1/s 9.82 × 106
Ka,i k°self,3 1/s 5.00 × 106
EA,self,1 kJ/kmol 70 800
EA,self,2 kJ/kmol 72 000
The catalytic acid concentration xacid is taken to be propor- EA,self,3 kJ/kmol 72 400
tional to the hydrogen ion (proton) activity, which can arise Ka,1 6.35
not only from citric acid but also from MEC and DEC as well. Ka,2 2.72
Because pKa values for the partial esters are not available in Ka,3 3.78
the literature, we assume that hydrogen ion activity is propor-
tional to the summed weighted mole fractions of acidic species. 4.4. Determination of Rate Constants. To determine values
This is equivalent to assuming that the acid strength of all of the six adjustable parameters for the self-catalyzed esterifi-
-COOH groups in citric acid and in the partial esters are the cation (pre-exponential factors k°self,1, k°self,2, and k°self,3 and
same and that the mean ionic activity coefficient is equal for activation energies EA,self,1, EA,self,2, and EA,self,3), the rate
all species in the range of concentrations studied. The constants expressions were numerically integrated via a fourth-order
that quantify these assumptions are lumped into the pre- Runge-Kutta method using ordinary differential equation solver
exponential factors for the self-catalyzed reactions. The catalytic ode23 in Matlab 7.0. Starting with an initial set of estimated
acid concentration in eq 6 can thus be written as rate constants, the liquid-phase mole fractions of all species over
the course of reaction were calculated and compared with the
2 1 experimental data collected in runs 15-22. The rate constants
(
xAcid ) xCA + xMEC + xDEC
3 3 ) (7)
were then sequentially incremented in order to minimize the
sum of the mean square differences Fmin2, given by
The rate of formation of each species in the reaction mixture
is described by eqs 8-13 given below:
∑ (xj,cal - xj,expt)2

( )
samples
dxCA aMECaW Fmin2 ) (15)
- ) kself,1xacid aCAaEtOH - (8) nsamples
dt Ka,1

-
dxMEC
dt (
) kself,2xacid aMECaEtOH -
aDECaW
Ka,2
+ ) The values of the self-catalyzed kinetic parameters which best
describe the data are reported in Table 3. Predicted mole fraction

( )
curves are shown as continuous lines in Figure 8 in the text;
aMECaW the model fits for other self-catalyzed reactions are located in
kself,1xacid - aCAaEtOH (9)
Ka,1 the Supporting Information (Figures S15-S22). The correlation
between the experimental data and the calculated trend lines is

-
dxDEC
dt (
) kself,3xacid aDECaEtOH -
aTECaW
Ka,3
+ ) reasonable.
After determining a final set of rate constants, the calculated

( )
mole fractions of each component were compared to the
aDECaW experimental values by calculating a mean deviation of all data
kself,2xacid - aMECaEtOH (10)
Ka,2 points for an experiment, represented both on an absolute and
on a relative (percentage) basis as shown below
-
dxTEC
dt
) kself,3xacid
aTECaW
Ka,3 (
- aDECaEtOH ) (11)
∑
samples
|xj,cal - xj,expt|

( )
Fabs ) × 100% (16)
dxEtOH aMECaW nsamples
- ) kself,1xacid aCAaEtOH - +
dt Ka,1

( )
aDECaW xj,cal - xj,expt
kself,2xacid aMECaEtOH -
Ka,2
+ ∑ | xj,expt
|

( )
samples
aTECaW Frel ) × 100% (17)
kself,3xacid aDECaEtOH - (12) nsamples
Ka,3

( )
The values of the absolute mole fraction error Fabs and relative
dxW aMECaW
- ) kself,1xacid - aCAaEtOH + mole fraction error Frel are reported as in Table 1 for the self-
dt Ka,1 catalyzed reactions (runs 15-22). Residual errors are highest

kself,2xacid
aDECaW
Ka,2( - aMECaEtOH + ) for TEC at reaction times up to about 300 min; there are also
some minor inconsistencies in the experimental concentrations

( )
for DEC, for reasons not clear at this time.
aTECaW 4.5. Kinetic Model for Combined Resin-Catalyzed and
kself,3xacid - aDECaEtOH (13)
Ka,3 Self-Catalyzed Esterification. Kinetics of the ion-exchange
resin-catalyzed esterification of citric acid (runs 1-14 in Table
where for the ith reaction 1) have been described using a pseudo-homogeneous model that

( )
includes the self-catalyzed reactions and DEE formation. The
-EA,self,i inclusion of self-catalyzed reactions in esterification has also
kself,i ) k°self,i exp (14)
RT been previously presented by Omata et al.28
3186 Ind. Eng. Chem. Res., Vol. 46, No. 10, 2007

The rate of formation of each species in the reaction mixture

is described by eqs 18-24 below:

-
dxCA
dt (
) (wcatkcat,1 + xacidkself,1) aCAaEtOH -
aMECaW
Ka,1
(18) )
-
dxMEC
dt (
) (wcatkcat,2 + xacidkself,2) aMECaEtOH -
aDECaW
Ka,2
+ )
(wcatkcat,1 + xacidkself,1) (
aMECaW
Ka,1
- aCAaEtOH (19) )
-
dxDEC
dt (
) (wcatkcat,3 + xacidkself,3) aDECaEtOH -
aTECaW
Ka,3
+ ) Figure 11. Comparison of activity-based and mole fraction-based model

( )
fits to the experimental data. The data at the right edge of the graph represent
aDECaW the liquid-phase composition at the end of reaction (t ) 1600 min). The
(wcatkcat,2 + xacidkself,2) - aMECaEtOH (20) reaction conditions are as follows: mole ratio ethanol:citric acid, 5:1; catalyst
Ka,2 loading, 5 wt %; reaction temperature, 120 °C (9, CA; b, MEC; 2, DEC;
×, TEC; ∆, DEE; s, calculated profiles using the activity model; -‚-‚-,

-
dxTEC
dt
) (wcatkcat,3 + xacidkself,3) (
aTECaW
Ka,3
- aDECaEtOH ) calculated profiles using the mole fraction model).

Table 4. Kinetic Model Parameters for Resin-Catalyzed Reactions

(21) parameters units values

( )
dxEtOH aMECaW k°cat,1 kgsol/(kgcat s) 3.34 × 108
- ) (wcatkcat,1 + xacidkself,1) aCAaEtOH - + k°cat,2 kgsol/(kgcat s) 1.87 × 109
dt Ka,1 k°cat,3 kgsol/(kgcat s) 1.99 × 107

( )
k°cat,4 kgsol/(kgcat s) 3.35 × 108
aDECaW
(wcatkcat,2 + xacidkself,2) aMECaEtOH - + (wcatkcat,3 + EA,cat,1 kJ/kmol 76 900
Ka,2 EA,cat,2 kJ/kmol 83 100

( )
EA,cat,3 kJ/kmol 73 200
aTECaW EA,cat,4 kJ/kmol 102 000
xacidkself,3) aDECaEtOH - + 2wcatkcat,4 (22) Ka,1 6.35
Ka,3
Ka,2 2.72

( )
Ka,3 3.78
dxW aMECaW
- ) (wcatkcat,1 + xacidkself,1) - aCAaEtOH +
dt Ka,1 from Figures 2-7 that the correlation between the experimental

(wcatkcat,2 + xacidkself,2) (
aDECaW
Ka,2 )
- aMECaEtOH + (wcatkcat,3 +
data and the calculated trend lines is good. The residual errors,
both absolute and relative (eqs 16 and 17), are shown in Table

( )
1. For citric acid, the error in mole fraction is highest in the
aTECaW region when the citric acid concentration is very low. Large
xacidkself,3) - aDECaEtOH - wcatkcat,4 (23)
Ka,3 errors are also observed for TEC in the initial reaction period
(<300 min), where its concentration is low.
dxDEE The citrate esterification system under consideration never
- ) -wcatkcat,4 (24) really achieves equilibrium in a true sense, since irreversible
dt
DEE formation (a zero-order reaction at high ethanol concentra-
where tion) continues to proceed by converting the ethanol to DEE
even after other reactions approach equilibrium. If this reaction
2 1
(
xacid ) xCA + xMEC + xDEC
3 3 ) (25) were allowed to proceed for very long reaction times, it would
ultimately initiate hydrolysis of the various ethyl citrates,

( )
liberating ethanol which would further etherify to DEE as a
-EA,cat,i terminal product.
kcat,i ) k°cat,i exp (26)
RT Finally, fitting of the reaction kinetic data was performed

( )
using a mole fraction-based model to determine whether
-EA,self,i nonideal solution behavior played a significant role. The mole
kself,i ) k°self,i exp (27)
RT fraction-based model did not describe the kinetic data as
satisfactorily as an activity-based model. Figure 11 gives fits
Eight adjustable parameters for the resin-catalyzed reactions for the two models for reaction at 120 °C, 5:1 initial mole ratio
(the pre-exponential factors k°cat,1, k°cat,2, k°cat,3, and k°cat,4 and of ethanol:citric acid, and 5% ion-exchange resin catalyst loading
activation energies EA,cat,1, EA,cat,2, EA,cat,3, and EA,cat,4) have been using the combined resin-catalyzed and self-catalyzed esterifi-
fitted to the experimental data of runs 1-14 by minimizing Fmin2 cation kinetic model described in section 4.5. The fit is clearly
for those runs. The constants for self-catalyzed reactions superior with the activity-based model.
determined earlier were used without alteration in determining
the kinetic constants. The final values of the kinetic parameters 5. Conclusions
are shown in Table 4.
Representative predicted mole fractions are given as continu- Experimental kinetic studies of citric acid esterification were
ous lines in Figures 2-7; the full set of predicted mole fraction performed at reaction temperatures from 78 to 120 °C, initial
curves for resin-catalyzed (runs 1-14) are reported in the mole ratios of ethanol:citric acid from 5:1 to 15:1, and
Supporting Information (Figures S1-S14). It can be observed Amberlyst-15 catalyst concentrations up to 5 wt %. The kinetics
 Ind. Eng. Chem. Res., Vol. 46, No. 10, 2007 3187

of esterification reactions have been correlated using an activity- (5) Harmer, M. A.; Sun, Q. Solid acid catalysis using ion-exchange
based, pseudo-homogeneous model that includes both the resin- resins. Appl. Catal., A. 2001, 221, 45.
(6) Schröter, J.; Konetzke, G.; Weidemann, F.; Klein, T.; Bohnen, H.;
catalyzed and self-catalyzed esterification and diethyl ether Bergrath, K.; Schmidt, K. Method for producing citric acid esters. WO
formation. The rate expressions are applicable over a wide range 2003008369 A1, 2003.
of catalyst concentrations, molar ratios of reactants, and tem- (7) Tao, X. Optimization of conditions for tri-ethyl citrate synthesis.
peratures. The results constitute an accessible, reliable kinetic Huazue Shijie 1998, 39 (6), 302.
model that is useful in the design of reactive distillation and (8) Frappier, E. P.; Davis, J. E.; Grendze, M.; Scriven, E. F. V.; Wyeth,
J. T.; Chiu, K. W. World Patent WO 9851657, 1998.
other systems for citrate ester formation. (9) Nong, L. Synthesis of tri-butyl citrate with aluminium phospho-
tungstate supported on activated carbon. Jingxi Huagong Zhongjianti 2004,
Acknowledgment 34, 50.
(10) Shi, L.; Ding, X.; Zhang, H.; Wu, D. Synthesis of tri-butyl citrate
Financial support from the National Corn Growers Associa- catalyzed by solid superacid S2O8-2/TiO2-SiO2. Yingyong Huagong 2004,
tion and the U.S. Department of Energy is greatly appreciated. 33, 41.
(11) Zheng, Y.; Wang, J.; Tan, M.; Zheng, S. Synthesis of tri-butyl citrate
catalyzed by tetra-butyl titanate. Jingxi Huagong Zhongjianti 2004, 34, 28.
Nomenclature and Units (12) Deng, B.; Huang, H.; Liu, G. The synthesis of tri-butyl citrate
aj ) activity of species j in solution catalyzed by sodium hydrogen sulfate. Jingxi Huagong Zhongjianti 2003,
33, 49.
CA ) citric acid (13) Song, Y.; Wang, H.; Huang, S.; Gao, W.; Zhang, H.; Lian, P.
DEC ) di-ethyl citrate Catalytic synthesis of tri-n-butyl citrate with aluminophosphate solid acid
DEE ) di-ethyl ether catalyst. Huagong Keji 2003, 11, 6.
EA,i ) energy of activation of reaction i, kJ/kmol (14) Liu, H.; Tan, F.; Yin, D. Catalytic synthesis of tri-butyl citrate with
EtOH ) ethanol dealuminated USY zeolite. Hecheng Huaxue 2003, 11, 175.
(15) Meng, X.; Wang, M.; Liu, B.; An, Y.; Kang, Y. Synthesis of tri-
Fmin2 ) absolute sum of mean square differences butyl citrate catalyzed by the nanosolid superacid SO42-/Fe2O3. Hebei Shifan
Fabs ) sum of absolute mean deviations Daxue Xuebao, Ziran Kexueban. 2003, 27, 64.
Frel ) sum of relative mean deviations (16) Zhao, J.; Cheng, Z.; Xing, C. Synthesis of tri-butyl citrate catalyzed
kself,i ) rate constant for self-catalyzed reaction i, 1/s by complex solid superacid WO3-TiO2-SO42-. Huagong Keji 2002, 10
kcat,i ) rate constant for resin-catalyzed reaction i, kgsol/ (5), 11.
(17) Meng, P.; Li, L.; Yang, C.; Zhou, C.; Jia, S. Synthesis of tri-butyl
(kgcat s) citrate with p-toluene sulfonic acid catalyst. Hecheng Shuzhi Ji Suliao 2002,
k°self,i ) pre-exponential factor for self-catalyzed reaction i, 1/s 19 (2), 16.
k°cat,i ) pre-exponential factor for resin-catalyzed reaction i, (18) Fu, L.; Guo, X.; Lin, Q.; Li, J. Synthesis of tri-butyl citrate catalyzed
kgsol/(kgcat s) by SO42- modified zirconium crosslinked solid clay. Jingzi Huagong 2002,
19 (1), 28.
Ka,i ) activity-based equilibrium constant for reaction i
(19) Krishna, R.; Taylor, R. Modelling reactive distillation. Chem. Eng.
Kx,i ) mole-fraction-based equilibrium constant for reaction i Sci. 2000, 55 (22), 33.
Kγ,i ) ratio of liquid-phase activity coefficients for reaction i (20) Sundmacher, K.; Qi, Z. Importance of Reaction Kinetics for
MEC ) mono-ethyl citrate Catalytic Distillation Processes. In ReactiVe Distillation - Status and Future
nsample ) number of data points Directions; Sundmacher, K., Kienle, A., Eds.; Wiley-VCH: Weinheim,
Germany, 2003; pp 97-142.
R ) gas constant, 8.31 kJ/(kmol K) (21) Vu, D. T.; Lira, C. T.; Asthana, N. S.; Kolah, A. K.; Miller, D. J.
T ) temperature, K Vapor-liquid equilibria in the systems ethyl lactate + ethanol and ethyl
TEC ) tri-ethyl citrate lactate + water. J. Chem. Eng. Data. 2006, 51 (4), 1220-1225.
W ) water (22) Gangadwala, J.; Mankar, S.; Mahajani, S. Esterification of acetic
wcat ) catalyst concentration (kgcat/kgsoln) acid with butanol in the presence of ion-exchange resins as catalysts. Ind.
Eng. Chem. Res. 2003, 42, 2146-2155.
xj ) mole fraction of species j in solution (23) Asthana, N.; Kolah, A.; Vu, D.; Lira, C. T.; Miller, D. J. A
Subscripts continuous reactive separation process for ethyl lactate formation. Org.
Process Res. DeV. 2005, 9, 599.
i ) reaction index (24) Song, W.; Venimadhavan, G.; Manning, J. M.; Malone, M. F.;
j ) species in solution Doherty, M. F. Measurement of residue curve maps and heterogeneous
kinetics in methyl acetate synthesis. Ind. Eng. Chem. Res. 1998, 37, 1917-
Greek Letters 1928.
(25) Rehfinger, A.; Hoffmann, U. Kinetics of methyl tertiary butyl ether
γ ) liquid-phase activity coefficient liquid-phase synthesized by ion exchange resin- I. Intrinsic rate expression
in liquid-phase activities. Chem. Eng. Sci. 1990, 45, 1605.
Supporting Information Available: Results of all kinetic (26) Sundmacher, K.; Hoffmann, U. Macrokinetic analysis of MTBE-
synthesis in chemical potentials. Chem. Eng. Sci. 1994, 49, 3077.
experiments along with model fits are presented as Figures S1-
(27) Venimadhavan, G.; Buzard, G.; Doherty, M. F.; Malone, M. F.
S22. This material is available free of charge via the Internet at Effect of kinetics on residue curve maps for reactive distillation. AIChE J.
http://pubs.acs.org. 1995, 41, 2613.
(28) Omata, F.; Dimian, A. C.; Bliek, A. Fatty acid esterification by
Literature Cited reactive distillation. Part 1: Equilibrium-based design. Chem. Eng. Sci. 2003,
58, 3159-3174.
(1) Chem. Market Reporter 2004, 65 (June 14), 12.
(2) Olah, G. A.; Iyer, P. S.; Prakasn, G. K. S. Synthesis 1986, 513. ReceiVed for reView June 28, 2006
(3) Chakrabarti, A.; Sharma, M. M. Cationic ion-exchange resin as ReVised manuscript receiVed October 12, 2006
catalyst. React. Funct. Polym. 1993, 20, 1. Accepted October 18, 2006
(4) Sharma, M. M. Some novel aspects of cationic ion exchange resins
as catalysts. React. Funct. Polym. 1995, 26, 1. IE060828F