G Model

JIEC-1501; No. of Pages 11

Journal of Industrial and Engineering Chemistry xxx (2013) xxx–xxx

Contents lists available at ScienceDirect

Journal of Industrial and Engineering Chemistry

journal homepage: www.elsevier.com/locate/jiec

Mathematical model and experimental validation of the synergistic

effect of selective enantioseparation of (S)-amlodipine from
pharmaceutical wastewater using a HFSLM
Niti Sunsandee a, Prakorn Ramakul b, Milan Hronec c, Ura Pancharoen a,*,
Natchanun Leepipatpiboon d,**
a
Department of Chemical Engineering, Faculty of Engineering, Chulalongkorn University, Bangkok 10330, Thailand
b
Department of Chemical Engineering, Faculty of Engineering and Industrial Technology, Silpakorn University, Nakhon Pathom 73000, Thailand
c
Department of Organic Technology, Faculty of Chemical and Food Technology, Slovak University of Technology, Bratislava, Slovak Republic
d
Chromatography and Separation Research Unit, Department of Chemistry, Faculty of Science, Chulalongkorn University, Patumwan, Bangkok 10330,
Thailand

A R T I C L E I N F O A B S T R A C T

Article history: A case study on the synergistic enantioseparation of (S)-amlodipine from pharmaceutical wastewater
Received 20 January 2013 by using hollow fiber supported liquid membrane (HFSLM) was examined. A chiral reaction flux
Accepted 1 August 2013 mathematical model was applied. Optimum conditions achieved the highest percentages of extraction
Available online xxx
and stripping viz. 84% and 80%, respectively. Relevant parameters affecting the enantioseparation
efficiency of (S)-amlodipine were determined. Standard deviation percentages were 2.31% for
Keywords: extraction and 1.26% for stripping. It was found that the mathematical model proved to be in good
Mathematical model
agreement with the experimental data.
Synergistic effect
(S)-Amlodipine
ß 2013 The Korean Society of Industrial and Engineering Chemistry. Published by Elsevier B.V. All rights
Pharmaceutical wastewater reserved.
HFSLM

1. Introduction
Chemical synthesis-based pharmaceutical wastewater con-
tains a variety of chemical contaminants e.g. solvents, additives,
reactants and high-value pharmaceutically active compounds
[1,2]. Although such compounds may be present in the
environment at low concentrations, these drugs can neverthe-
less have an adverse effect on aquatic organisms [3]. These
effects are chronic rather than acutely toxic [4]. Further, high-
value drug substances can also be recycled for use in the
pharmaceutical industry [5]. Currently, the recycling and
reduction of waste materials has drawn much attention. One
of the main purposes of wastewater treatment is the stripping of
pharmaceutically active compounds [6–8]. In recent years, the
use of a liquid membrane in wastewater treatment has become
increasingly important. It offers several advantages, such as high
selectivity, high efficiency of separation, high enrichment and
* Corresponding author. Tel.: +66 2 2186891; fax: +66 2 2186877.
** Corresponding author. Tel.: +66 2 2187608; fax: +66 2 2541309.
E-mail addresses: ura.p@chula.ac.th (U. Pancharoen), natchanun.l@chula.ac.th
(N. Leepipatpiboon).
less use of the organic phase, compared with the classical
solvent extraction process [9].
The aim of this research is to design a HFSLM system for the
separation and wastewater treatment processes in the pharmaceu-
tical industry. HFSLM is an effective simultaneous process to extract
and recover compounds from a very dilute solution of components
in the feed phase by a single-unit operation [10]. In the past few
years, researchers have been working on the enantioseparation for
(S)-amlodipine using HFSLM [11–14]. HFSLM technique follows
certain rules in the choice of a separation system [15–17]. This
method has potential for large-scale production [18]. However, in
order for it to be applied to industries, reliable mathematical
formulae are required. These help to provide guidelines for mass
transfer which describes the transport mechanism of the target
species through HFSLM. A theoretical model of HFSLM system is
urgently needed for the completion of an efficient stripping process.
The prime objective of this work is to provide a chiral
mathematical model of the HFSLM process. Its aim is also to
investigate the effect of operating parameters on selective
enantioseparation of (S)-amlodipine from pharmaceutical waste-
water. Prediction equations for enantioseparation of (S)-amlodi-
pine via HFSLM are presented and validated by comparing
theoretical results with experimental results.

1226-086X/$ – see front matter ß 2013 The Korean Society of Industrial and Engineering Chemistry. Published by Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jiec.2013.08.008

Please cite this article in press as: N. Sunsandee, et al., J. Ind. Eng. Chem. (2013), http://dx.doi.org/10.1016/j.jiec.2013.08.008
G Model
JIEC-1501; No. of Pages 11

2 N. Sunsandee et al. / Journal of Industrial and Engineering Chemistry xxx (2013) xxx–xxx

of the aqueous phase. The chiral organic selector or extractant is

Nomenclature
trapped in polymeric hydrophobic microporous fibers of the
hollow fiber module by capillary action [18]. The supported
A cross-sectional area (cm3) liquid membrane lies between the aqueous solution which
A, B, C, D target enantiomer (A), Chiral selector (B), Chiral initially contains the racemic feed solution and the aqueous
complex (C) and Chiral stripping compound (D) stripping solution. The feed phase, consisting of pharmaceutical
a, b, c, d stoichiometric coefficients wastewater containing (S)-amlodipine, together with the strip-
C concentration (mmol/L) ping solution are fed counter-currently into the tube and shell
D data results sides of the module, respectively. The transport mechanism of
do external diameter of the hollow fiber tube (cm) (S)-amlodipine in the microporous hollow fiber is presented
schematically in Fig. 1. In the liquid membrane, the synergistic
HFSLM hollow fiber supported liquid membrane
enantioseparation of (S)-amlodipine by a mixture of the chiral
J flux (mg cm2 min1)
selector ((+)-DBTA) and the achiral selector D2EHPA occurs via
Kex equilibrium constant of the interfacial reaction in two extraction reactions in the system.
the feed phase The first reaction is the reaction between the chiral selector
Ks equilibrium constant of the interfacial chemical and the achiral selector to form the extractant complex (+)-
reaction in the stripping phase DBTA–D2EHPA [19–21], as shown in Eq. (1). The second reaction
Kf dimensionless mass transfer coefficient in the feed is the reaction between the extractant complex and (S)-
layer amlodipine, as shown in Eq. (2) [20]. The reaction takes place
ke,f rate constant of extraction reaction in the feed in the presence of proton transfer-chiral interactions i.e. (+)-
phase (min1) DBTA–D2EHPA. This produces the derivative complex (S)-
amlodipine-(+)-DBTA–D2EHPA. Thus, complex chiral com-
km rate of mass transfer coefficient in the membrane
pounds are formed and transported across the membrane phase
phase (cm min1)
to the other side. No transport of (S)-amlodipine passes this
ks,f rate constant of stripping reaction in the stripping
interface. At the interface, between the organic membrane phase
phase (min1) and the stripping phase, the complex chiral compounds react
L length of the hollow fiber (cm) with the stripping solution and release (S)-amlodipine to the
m order of stripping reaction stripping phase.
N numbers of hollow fibers in the module
k1
n order of extraction reaction ½ðþÞ-DBTAm þ ½D2EHPAm @ ½ðþÞ-DBTAD2EHPAm (1)
k2
Qf volumetric flow rate of feed solution (mL min1)
2½ðSÞ-amlodipine f
R universal gas constant
k3
R2 coefficient of determination þ ½ðþÞ-DBTAD2EHPAm @ ½½ðSÞ-amlodipine2 (2)
k4
S synergistic coefficient  ½ðþÞ-DBTAD2EHPAm
ri internal radius of the hollow fiber tube (cm)
where k1 and k2 are the apparent rate constants of formed
ro external radius of the hollow fiber tube (cm)
extractant and unformed extractant, and k3 and k4 are the apparent
rlm log-mean radius of the hollow fiber
rate constants of feed–membrane and membrane–feed interfacial
re,f rate of extraction reaction in the feed phase transport, respectively, of the amlodipine enantiomer. The suffixes
rs,f rate of stripping reaction in the stripping phase f and m are defined as the feed phase and membrane phase,
T absolute temperature (K) respectively.
The presence of b-cyclodextrin in the stripping phase produces
Greek letters higher stripping results by creating a complex with (S)-enantiomer
t tortuosity of the liquid membrane via various electrostatic forces and hydrogen bonding [22,23]. The
e porosity of the membrane
m viscosity of the membrane (kg/(s m))

Subscripts
A target compound
aq aqueous
C chiral complex
Exp experimental
f feed phase
m membrane phase
Mod model-calculated value
org organic
s stripping phase

2. Theoretical background

2.1. Transport mechanisms of synergistic extraction of (S)-amlodipine

Transportation of (S)-amlodipine occurs as a consequence of

the concentration driving force between the two opposite sides Fig. 1. Schematic of transport flux within HFSLM.

Please cite this article in press as: N. Sunsandee, et al., J. Ind. Eng. Chem. (2013), http://dx.doi.org/10.1016/j.jiec.2013.08.008
G Model
JIEC-1501; No. of Pages 11
N. Sunsandee et al. / Journal of Industrial and Engineering Chemistry xxx (2013) xxx–xxx
Fig. 2. Schematic for transport of the plug flow reaction in the feed phase and stripping phase of HFSLM.
complex formation depends largely on the size, shape and polarity of
the (S)-enantiomer [24]. This means that the configuration of (S)-
amlodipine should be well compatible with the cavity of b-
cyclodextrin. Furthermore, the stability of the generated complex is
subject to the shape, polarity and side groups of (S)-amlodipine. b-
cyclodextrin reacts with [[(S)-amlodipine]2-(+)-DBTA–D2EHPA]m to
recover (S)-amlodipine into the stripping phase:
½½ðSÞ-amlodipine2  ½ðþÞ-DBTAD2EHPAm
k5
þ 2½b-cyclodextrins @ 2½ðSÞ-amlodipine-b-cyclodextrins
k6
þ ½ðþÞ-DBTAD2EHPAm
where k5 and k6 are the apparent rate constants of membrane–strip
and strip–membrane interfacial transport, respectively, of the
amlodipine enantiomer. The suffixes, m and s, are defined as the
membrane phase and stripping phase, respectively.
2.2. Mathematical modeling of a hollow fiber supported liquid
membrane (HFSLM)
The mathematical model assumes only reaction flux for the
component transport within the hollow fiber tube (feed phase) and
shell annulus (stripping phase). Across the micropore sites of the
hollow fibers (membrane phase), the transport model assumes
diffusion flux for the chiral complex.
2.2.1. The chiral extraction reaction flux (CERF) model for transport in
the HFSLM tube (feed phase)
The chiral extraction reaction shown in Eq. (2) can be simplified
as in Eq. (4):
ke; f
aA þ bB @ cC
ke;b
where A is (S)-amlodipine in the feed phase, B is the extractant
complex ((+)-DBTA–D2EHPA) in the membrane phase, and C is
the complex species of the chiral extraction complex ((S)-
amlodipine)2-((+)-DBTA–D2EHPA)) in the membrane phase. The
Please cite this article in press as: N. Sunsandee, et al., J. Ind. Eng. Chem. (2013), http://dx.doi.org/10.1016/j.jiec.2013.08.008
3
symbol a, b, and c are stoichiometric coefficients of A, B, and C,
respectively.
The chiral extraction reaction is assumed to be a pseudo-
irreversible reaction. Rate constant of forward reaction (ke,f) is
higher than the rate constant (ke,b). Thus, the reaction rate of (S)-
amlodipine in feed phase re,f can be described by Eq. (5):
r e; f ¼ ke; f CAn ðx; tÞ (5)
where re,f is the extraction reaction rate (mmol/L/min), kf is the
extraction reaction rate constant ((L/mmol)n1/min), CA(x,t) is the
concentration of (S)-amlodipine (mmol/L) at time t (in min), and n
(3) is the reaction order.
The chiral extraction reaction flux (CERF) model is developed by
considering the conservation of mass whereby components are
lost from the feed phase through the influence of the chiral
extraction reaction. The concept is utilized from the first reaction
flux model, as shown in the previous work of Pancharoen et al. [25].
The transport schematic for the plug flow reaction in the HFSLM
tube (feed phase) is shown in Fig. 2. The key assumption for the
model is perfect mixing within the small cross-sectional area of
the inner hollow fiber tube. Therefore, there is no component
concentration variation at the cross-sectional area. Mass flux exists
only in the axial direction due to the very small membrane
thickness and the reaction flux along the length of the hollow fiber
tube.
Based on the above assumption, the equation for calculation of
(S)-amlodipine concentration in the case of n = 1 is expressed as
[25]:
C̄A ðL; tÞ ¼ expðt f ke; f Þ  C̄A ð0; t  t f Þ  uðt  t f Þ (6)
With the initial condition;
(4) !
ke; f A f
C A ðL; 0Þ ¼ C A ð0; 0Þ  exp L (6a)
qf
where L is effective length of the hollow fiber (cm) and uðt  t f Þ is a
unit function
Given:
G Model
JIEC-1501; No. of Pages 11
4 N. Sunsandee et al. / Journal of Industrial and Engineering Chemistry xxx (2013) xxx–xxx
C̄A ðx; tÞ ¼ C A ðx; tÞ  C A ðx; 0Þ
AfL
tf ¼
qf
A f ¼ 35; 000  pðr i Þ2
2.2.2. The chiral stripping reaction flux (CSRF) model for transport in
the HFLSM shell and tube annulus (stripping phase)
The chiral stripping reaction as shown in Eq. (3) can be
simplified as in Eq. (7). The stripping reaction is basically the
reverse of the extraction reaction in which the stripping solution
agent (D) reacts with (C) and reverts it back to (E).
ks; f
cC þ dD @ bB þ eE
ks;b
where C is the complex species of chiral extraction complex ((S)-
amlodipine)2  ((+)-DBTA–D2EHPA)) in membrane phase, D is b-
cyclodextrin in the stripping phase, B is the extractant complex
((+)-DBTA–D2EHPA) in membrane phase and E is the complex
species of chiral stripping complex ((S)-amlodipine-b-cyclodex-
trin) in stripping phase. The symbol b, c, d, and e are stoichiometric
coefficients.
The chiral stripping reaction is assumed to be a pseudo-
irreversible reaction. The forward reaction rate constant (ks,f) is
higher than the backward reaction rate constant (ks,b). The rate of
complex species of the chiral extraction complex ((S)-amlodi-
pine)2-((+)-DBTA–D2EHPA)) as in the membrane phase rs,f can be
described by Eq. (8):
0
r s; f ¼ ks; f CCm ðx̄; tÞ ¼ ks; f CEm ðx̄; tÞ
Let:
x̄ ¼ L  x and 0 < x̄ < L
where rs,f is the stripping reaction rate in mmol/L/min, ks,f is the
stripping reaction rate constant (L/mmol)m1/min. C C ðx̄; tÞ is the
concentration of the chiral extraction complex ((S)-amlodipine)2-
((+)-DBTA–D2EHPA)) in mmol/L at time t (min.) and m is the
stripping reaction order.
Based on the concept of the chiral extraction reaction flux
(CERF) model, the same concept is developed for the model of the
chiral stripping reaction flux. The chiral stripping reaction flux
concept is applied from the recovery reaction flux model. The
modeling of the chiral stripping reaction flux (CSRF) is performed
under the following assumptions in the previous work of
Chaturabul et al. [26], In the case of m = 1, the concentration of
the target complex species of chiral stripping complex ((S)-
amlodipine-b-cyclodextrin) can be calculated. The equation used
is shown below [26]:
C̄E ðL; tÞ ¼ expðt s ks; f Þ  C̄E ð0; t  t s Þ  uðt  t s Þ
With the initial condition;
 
ks; f As
C E ðL; 0Þ ¼ C E ð0; 0Þ  exp L (9a)
qs
where L is effective length of the hollow fiber (cm) and uðt  t f Þ is a
unit function
Given:
C̄E ðx̄; tÞ ¼ C E ðx̄; tÞ  C E ðx̄; 0Þ
AL
ts ¼ s
qs
Please cite this article in press as: N. Sunsandee, et al., J. Ind. Eng. Chem. (2013), http://dx.doi.org/10.1016/j.jiec.2013.08.008
pffiffiffi !
3 2 pro2
As ¼ 35; 000 d 
4 o 2
2.2.3. Chiral complex transport across the hollow fiber tube
(membrane phase)
According to the assumptions of the complex transport within
hollow fiber liquid membrane, chiral extraction complex species
(C) are the only species able to be transported within micropore
sites in a radial direction under the following assumptions:
(1) The system is considered to be at pseudo-steady state.
(2) Only the chiral extraction complex species (C) which occur
from the reaction, not chiral molecule (A), can diffuse through
the micropore sites in the membrane phase.
(7) (3) The chiral extraction complex species (C) can only diffuse in
radial direction across the liquid membrane from feed to
stripping phase.
(4) The axial transport direction of the chiral extraction complex
species (C) is neglected.
(5) There is only the diffusion mass transport across the membrane
phase. The chiral extraction complex species (C) did not have
any reaction in membrane phase.
The mass transport of chiral complexes permeating through
fiber micropores was described by Fick’s law. There is no interface
between the aqueous and membrane phases. The concentration
gradient can be determined directly from the concentration of
chiral complex on the feed side (CC,f) and on the stripping side (CC,s).
The flux equation based on Fick’s law is represented by Eq. (11)
and is illustrated by the transport schematic as shown in Fig. 3.
(8)
J m ¼ km ðC C; f  C C;s Þ (10)
Rewriting Eq. (10) in order to express the measurement of the
concentration of the chiral molecule (CA) gives [26]:
a c
J m ¼ km ðC A ðx; tÞ  C A ðx þ Dx; tÞÞ  km ðC E ðx̄ þ Dx; tÞ
c e
 C E ðx̄; tÞÞ (11)
Mass transfer rates in the membrane will be equal to the
concentration flux in the feed and stripping phases in the pseudo-
steady-state operated system, as described in Eq. (12):
J f ¼ Jm ¼ Js (12)
Along the considered segment (Dx), the fluxes in the feed phase
(Jf) and in stripping phase (Js) are largely functions of the
concentration gradient.
The flux in the feed phase can be expressed as:
qf
Jf ¼  ½C A f ðx; tÞ  C A f ðx þ Dx; tÞ (13)
Af
(9)
The flux in the stripping phase can be expressed as:
qs
Js ¼  ½C As ðx̄ þ Dx; tÞ  C As ðx̄; tÞ (14)
As
From Eqs. (12)–(14), the mass transfer coefficient in the
membrane phase (km) can be determined accordingly. The values
of the membrane mass-transfer coefficient (km) and the aqueous
feed mass-transfer coefficient (kf) were established based on
previous work [11]. The values of kf and km were found to be
4.87  102 and 2.89  102 cm/s, respectively.
2.3. The validation of the mathematical model
The validation of the mathematical model for the hollow fiber
module was investigated using two indicators. First, calculated
G Model
JIEC-1501; No. of Pages 11
N. Sunsandee et al. / Journal of Industrial and Engineering Chemistry xxx (2013) xxx–xxx
Fig. 3. Chiral complex transport across the membrane phase.
values of extraction and stripping were compared with experi-
mental data in order to determine the percentage of absolute
deviation. Second, the percentage of standard deviation – S.D. (%)
between calculated and measured results – was investigated for
the validity of the mathematical model [27].
Extraction efficiency of (S)-amlodipine in feed solution for the
CERF model and stripping efficiency of (S)-amlodipine in stripping
solution for the CSRF model, as given by the simulation, was
compared with data from experimental runs. Absolute error
percentage obtained by experimental runs and model simulation is
defined as:
jDEx p  DMod j
A:E: ð%Þ ¼  100
DMod
where A.E. (%) is the absolute error, DExp is the experimental data,
DMod is the model-calculated value from the mathematical model.
The validity of the mathematical model was based on the
percentage of standard deviation S.D. (%) as determined by the
values of the mathematical model and the experimental results
viz.
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Pn
i¼1 fðDEx p =DMod Þ  1g
S:D: ð%Þ ¼ 100 
N1
where DExp is the experimental data, DMod is the model calculated
value from the mathematical model and N is the number of
experimental points.
3. Experiment
3.1. Chemicals and reagents
The aqueous phase was pharmaceutical wastewater contain-
ing racemic amlodipine (Government Pharmaceutical Organiza-
tion (GPO), Bangkok, Thailand). (R)-Amlodipine, (S)-amlodipine
and racemic amlodipine, all of pharmaceutical grade, were also
Please cite this article in press as: N. Sunsandee, et al., J. Ind. Eng. Chem. (2013), http://dx.doi.org/10.1016/j.jiec.2013.08.008
5
provided by the GPO. The organic phase was a solution of the
synergistic extractant (+)-DBTA + D2EHPA in 1-decanol, analyti-
cal reagent grade, obtained from Merck (Darmstadt, Germany).
(+)-DBTA as a chiral selector and D2EHPA as a cationic extractant
were obtained from Acros Organics (Geel, Belgium). The
stripping solution, b-cyclodextrin (Sigma–Aldrich, St. Louis,
USA), was dissolved in deionized water (Millipore1, USA). The
solvents, i.e. N,N-dimethylformamide, cyclohexane, 1-decanol
and 1-propanol, were all of analytical reagent grade and were
obtained from Merck.
3.2. Apparatus
(15)
The hollow fiber supported liquid membrane (HFSLM) system
(Liqui-Cel1 Extra-flow 2.5 in.  8 in. membrane contactor) was
manufactured by Celgard (formerly Hoechst Celanese), USA. The
module uses Celgard1 microporous polypropylene fibers that are
woven into fabric and wrapped around a central tube feeder that
supplies the shell-side fluid (Table 1).
3.3. Procedures
2
(16)
The single-module operation is shown in Fig. 4. The selected
organic extractant (+)-DBTA, D2EHPA and synergistic extractant
((+)-DBTA + D2EHPA) were individually dissolved in 1-decanol
(500 mL). The extractant was simultaneously pumped into the
tube and shell sides of the hollow fiber module for 40 min to
ensure that the extractant was entirely embedded in the
micropores of the hollow fibers [11]. Subsequently, 5 L (each)
of feed solution and stripping solution were fed counter-currently
into the tube side and shell side of the single-module operation,
respectively.
The concentration of the selected organic extractant (+)-DBTA,
D2EHPA and synergistic extractant ((+)-DBTA + D2EHPA) in the
liquid membrane was deliberately varied in order to find the
optimum value for (S)-amlodipine separation. The volumetric flow
G Model
JIEC-1501; No. of Pages 11

6 N. Sunsandee et al. / Journal of Industrial and Engineering Chemistry xxx (2013) xxx–xxx

Fig. 4. Schematic representation of the counter-current flow diagram for batch-mode operation in HFSLM [13]: (1) feed reservoir, (2) gear pumps, (3) inlet pressure gauges, (4)
outlet pressure gauges, (5) the hollow-fiber module, (6) flow meters, (7) stripping reservoir, (8) stirrer with temperature controller and (9) temperature control box.

rates of feed and stripping solutions, and the number of separation
cycles of the HFSLM were investigated. The operating time was
50 min. The concentration of (S)-amlodipine in samples from both
feed and stripping solutions was determined by high-performance
liquid chromatography (HPLC) in order to calculate the percen-
tages of extraction and stripping.
3.4. Analytical instruments and chromatographic conditions
The quantification of amlodipine enantiomers in the aqueous
phase was performed following U.S. Patent No 6646131 B2 [28].
The chromatographic procedure was carried out using an Ultron
ES-OVM ovomucoid chiral column (5 mm, 4.6 mm  150 mm) –
Agilent Technologies, Palo Alto, CA, USA. The latter consisted of
an Agilent 1100 Compact LC series. This was equipped with a
built-in solvent degasser, quaternary pump, column compart-
ment, photodiode array detector with variable wavelength, and
autosampler.
Data analysis was carried out using ChemStation1 version
B.04.01 software (Agilent1). The column temperature was
controlled at 298.15 K. The mobile phase was prepared by
mixing disodium hydrogen phosphate buffer (20 mmol/L) and
Table 1
Physical characteristics of hollow fiber module.
Properties
Material
Inside diameter of hollow fiber
Outside diameter of hollow fiber
Effective length of hollow fiber
Number of hollow fibers
Average pore size
Porosity
Effective surface area
Area per unit volume
Module diameter
Module length
Contact area
Tortuosity factor
Operating temperature
acetonitrile (80:20%, v/v). The flow rate of the mobile phase
was 0.3 mL/min. and the injection volume was 20 mL. Retention
time for (R)-amlodipine was about 4.8 min and for (S)-
amlodipine about 5.6 min as detected by a HPLC system with
a photodiode array detector set at UV 237 nm. Analysis time was
set at 20 min per sample to eliminate potential interference from
late eluting peaks. The chromatogram is shown in Fig. 5 [14].
The pH of the aqueous phase was measured using a SevenMul-
tiTM pH meter with modular expansion (Mettler-Toledo,
Greifensee, Switzerland).
4. Results and discussion
4.1. Optimization of process conditions for mathematical modeling of
HFSLM studies
The condition parameters of HFSLM are most important. These
influence the overall mass transfer of the components diffusing
across HFSLM. Mass transfer is controlled by several parameters
viz. pH, concentration and flow rate of feed phase, chiral selector
concentration, phase concentration and flow rate of the stripping
solution, as well as the number of separation cycles through the
hollow-fiber module. Optimal condition is shown in Table 2 [13].
The parameters above can be determined by examining the
physical properties of the compounds. Optimized pH of feed phase
was pH 5.0 [11]. Concentration of the feed phase was approximately
Descriptions
4 mmol/L. The concentration ratio between (+)-DBTA:D2EHPA in
Polypropylene the membrane phase was 4 mmol/L:4 mmol/L. The stripping phase
240 mm
(b-cyclodextrin) concentration was also 4 mmol/L. The flow rates of
300 mm
15 cm feed and stripping solution were 100 mL/min. [13]. Optimized
35,000 temperature was 273.15 K which achieved the highest percentage of
0.03 mm extraction and stripping at 84% and 80%, respectively. Enantiomeric
30%
excess (% e.e.) of (S)-amlodipine was 70%.
1.4  104 cm2
29.3 cm2/cm3
6.3 cm 4.2. Determination of the reaction order and the reaction rate
20.3 cm constant for (S)-amlodipine extraction
30%
2.6
The extraction reaction order (n) and the extraction reaction
273.15–333.15 K
rate constant (ke,f) for (S)-amlodipine extraction for (S)-amlodipine

Please cite this article in press as: N. Sunsandee, et al., J. Ind. Eng. Chem. (2013), http://dx.doi.org/10.1016/j.jiec.2013.08.008
G Model
JIEC-1501; No. of Pages 11

N. Sunsandee et al. / Journal of Industrial and Engineering Chemistry xxx (2013) xxx–xxx 7

the (S)-amlodipine concentration. Consequently, extraction reac-

tion rate constant (ke,f) of 0.0228 min1 was also obtained.

4.3. Determination of the reaction order and the reaction rate

constant for (S)-amlodipine stripping

Fig. 5. Chromatogram of resolution of amlodipine by HPLC in pharmaceutical
wastewater containing racemic amlodipine [14]: (1) (R)-amlodipine and (2) (S)-
amlodipine.
The stripping reaction order (m) and the stripping reaction rate
constant (ks,f) of (S)-amlodipine stripping for the model input from
Eq. (8) were verified by integration and graphical methods [29].
The best-fit result was obtained from the integrated first-order
rate law at the semi natural logarithm plot between (S)-
amlodipine concentration and time. The linear curve was drawn
tangentially along the plot, giving a calculated coefficient of
determination (R2) which was higher than 0.99 for the chiral
stripping reaction. Results are shown in Fig. 7 and Table 4. The
stripping reaction rate constant (ks,f) for (S)-amlodipine stripping
was 0.0373 min1.

4.4. Experimental data for validation

extraction were determined as shown in Table 3. In this work,
extraction reaction order (n) and extraction reaction rate constant
(ke,f) for (S)-amlodipine extraction as in Eq. (5) were verified by
integration and graphical methods [29]. The integral concentra-
tions with respect to zero, first and second orders were plotted
against time at optimum conditions, as shown in Table 3. The
linear line of each reaction order was presented by the coefficient
of determination (R2). Results, from the plots between the initial
concentration of (S)-amlodipine in the feed solution versus time,
are shown in Fig. 6. These form a linear line of first-order reaction
(n = 1). Thus, the rate of extraction reaction depends linearly upon
The validity of Eqs. (6) and (9) has been amply demonstrated
by the experiments. (S)-Amlodipine at concentrations of 1, 2, 3, 4
and 5 mmol/L in the initial feed were studied. By using optimal
condition parameters, the experimental results were investigat-
ed (Table 5). Experimental data showed promising agreement
with the chiral extraction reaction flux (CERF) model. Further-
more, the chiral stripping reaction flux (CSRF) model provided an
equally satisfactory result for prediction of (S)-amlodipine
concentration. The percentage of absolute relative deviation
A.E. (%) is presented in Table 5. The definition of A.E. (%) is given
by Eq. (15).

Table 2
Optimized operation using HFSLM for enantioseparation.

Phase Chemical reagent Concentration Flow rate

Feed (R,S)-Amlodipine pKa = 8.6 at 298.15 K 4 mmol/L 100 mL/min

Membrane (+)-DBTA:D2EHPA 4 mmol/L:4 mmol/L –
Strip b-Cyclodextrin 4 mmol/L 100 mL/min

Table 3
Analysis of extraction reaction order and rate constants.

Rate order Extraction

Determination

Relationship Rate constant R2 Remark

a 1 1
0 CA vs. time 0.0534 mmol L min 0.9625
1 ln(CA,0/CA) vs. timea 0.0228 min1 0.9958 Best fit
2 1/CA vs. timea 0.0105 L mmol1 min1 0.9830
n = 1.28 ln(@CA/@t) vs. ln CA b 0.0160 mmol1n Ln1 min1 0.6571
a
Integral analysis.
b
Differential analysis.

Table 4
Analysis of recovery reaction order and rate constants.

Rate order Recovery

Determination

Relationship Rate constant R2 Remark

a 1 1
0 CE vs. time 0.0359 mmol L min 0.9374
1 ln(CE,0/CE) vs. timea 0.0373 min1 0.9988 Best fit
2 1/CE vs. timea 0.0471 L mmol1 min1 0.9482
m = 1.13 ln(@CE/@t) vs. ln CE b 0.0399 mmol1m Lm1 min1 0.8833
a
Integral analysis.
b
Differential analysis.

Please cite this article in press as: N. Sunsandee, et al., J. Ind. Eng. Chem. (2013), http://dx.doi.org/10.1016/j.jiec.2013.08.008
G Model
JIEC-1501; No. of Pages 11

8 N. Sunsandee et al. / Journal of Industrial and Engineering Chemistry xxx (2013) xxx–xxx

Table 5
Experimental data for validation.

Time (t) CERF model CSRF model

DMod (mmol/L) DExp (mmol/L) A.E.(%) DMod (mmol/L) DExp (mmol/L) A.E.(%)

(a) Initial (S)-amlodipine concentration, 1 mmol/L

5 0.8913 0.9012 1.11 0.1021 0.1025 0.39
10 0.7945 0.8012 0.84 0.1176 0.1159 1.45
15 0.7082 0.7077 0.07 0.1525 0.1504 1.38
20 0.6313 0.6112 3.18 0.1755 0.1749 0.34
25 0.5627 0.5509 2.10 0.2128 0.2107 0.99
30 0.5016 0.5107 1.81 0.2624 0.2701 2.93
35 0.4471 0.4289 4.07 0.3121 0.3115 0.19
40 0.3985 0.3899 2.16 0.3758 0.3903 3.86
45 0.3552 0.3357 5.49 0.4625 0.4605 0.43
50 0.3166 0.3094 2.27 0.5454 0.5419 0.64

S.D. (%) 2.31 1.26

(b) Initial (S)-amlodipine concentration, 2 mmol/L
5 1.7827 1.8230 2.26 0.2045 0.2089 2.15
10 1.5890 1.6563 4.24 0.2356 0.2376 0.85
15 1.4165 1.4379 1.51 0.3095 0.3013 2.65
20 1.2626 1.2367 2.05 0.3511 0.3505 0.17
25 1.1255 1.1647 3.48 0.4256 0.4212 1.03
30 1.0032 1.0046 0.14 0.5208 0.5205 0.06
35 0.8942 0.8689 2.83 0.6252 0.6201 0.82
40 0.7971 0.7886 1.07 0.7532 0.7598 0.88
45 0.7105 0.7178 1.03 0.9256 0.9347 0.98
50 0.6333 0.6374 0.65 1.0951 1.0942 0.08

S.D. (%) 1.93 0.97

(c) Initial (S)-amlodipine concentration, 3 mmol/L
5 2.6741 2.6464 1.04 0.3075 0.3016 1.92
10 2.3836 2.4560 3.04 0.3523 0.3615 2.61
15 2.1247 2.1626 1.78 0.4574 0.4611 0.81
20 1.8939 1.8260 3.59 0.5250 0.5259 0.17
25 1.6882 1.6574 1.82 0.6377 0.6357 0.31
30 1.5048 1.5479 2.86 0.7836 0.7875 0.50
35 1.3414 1.3464 0.37 0.9313 0.9356 0.46
40 1.1957 1.1936 0.18 1.1259 1.1325 0.59
45 1.0658 1.0263 3.71 1.3875 1.3789 0.62
50 0.9500 0.9463 0.39 1.6350 1.6554 1.25

S.D. (%) 1.88 0.92

(d) Initial (S)-amlodipine concentration, 4 mmol/L
5 3.5655 3.5259 1.11 0.4131 0.4144 0.31
10 3.1782 3.2261 1.51 0.4725 0.4684 0.87
15 2.8330 2.8326 0.01 0.6143 0.6236 1.51
20 2.5252 2.4364 3.52 0.7374 0.7362 0.16
25 2.2510 2.2363 0.65 0.8537 0.8746 2.45
30 2.0065 2.0049 0.08 1.0443 1.0547 1.00
35 1.7885 1.6831 5.89 1.2437 1.2420 0.14
40 1.5942 1.6351 2.57 1.5722 1.5737 0.10
45 1.4211 1.4431 1.55 1.8572 1.8446 0.68
50 1.2667 1.2853 1.47 2.1827 2.1842 0.07

S.D. (%) 1.84 0.73

(e) Initial (S)-amlodipine concentration, 5 mmol/L
5 4.4569 4.4838 0.60 0.5125 0.5172 0.92
10 3.9727 4.0010 0.71 0.5877 0.5827 0.85
15 3.5412 3.5160 0.71 0.7625 0.7626 0.01
20 3.1566 3.1023 1.72 0.8752 0.8726 0.30
25 2.8137 2.7542 2.11 1.0625 1.0626 0.01
30 2.5081 2.5035 0.18 1.3262 1.3172 0.68
35 2.2356 2.2126 1.03 1.5526 1.5527 0.01
40 1.9928 1.9837 0.46 1.8752 1.8941 1.01
45 1.7763 1.7637 0.71 2.3125 2.3116 0.04
50 1.5834 1.5730 0.66 2.7256 2.7161 0.35

S.D. (%) 0.89 0.42

4.5. Prediction of (S)-amlodipine concentration by a mathematical
model
An estimation of the chiral reaction flux (CRF) model for (S)-
amlodipine transport within HFLSM was developed. It was
composed of three flow transport paths. The first path was in
the fiber tube (feed phase). The second path was between the tube
and shell annulus (stripping phase) which focused only on the axial
concentration distribution. Extraction reaction rate constant (ke,f)
was used to determine the concentration profile of the feed phase
in the tube. Similarly, stripping reaction rate constant (ks,f) was
used to calculate the concentration profile in the stripping phase.

Please cite this article in press as: N. Sunsandee, et al., J. Ind. Eng. Chem. (2013), http://dx.doi.org/10.1016/j.jiec.2013.08.008
G Model
JIEC-1501; No. of Pages 11

N. Sunsandee et al. / Journal of Industrial and Engineering Chemistry xxx (2013) xxx–xxx 9

Fig. 6. Integral concentrations of (S)-amlodipine extraction reaction: (a) zero order,

(b) first order and (c) second order. Fig. 7. Integral concentrations of (S)-amlodipine recovery reaction: (a) zero order,
(b) first order and (c) second order.

Fig. 8. Concentration of (S)-amlodipine in the retentate (feed) phase plotted as a function of time at different (S)-amlodipine concentrations.

Please cite this article in press as: N. Sunsandee, et al., J. Ind. Eng. Chem. (2013), http://dx.doi.org/10.1016/j.jiec.2013.08.008
G Model
JIEC-1501; No. of Pages 11
10 N. Sunsandee et al. / Journal of Industrial and Engineering Chemistry xxx (2013) xxx–xxx
Fig. 9. Concentration of (S)-amlodipine in the raffinate (stripping) phase plotted as a function of time at different (S)-amlodipine concentrations.
Table 6
Values of validity of the model-calculated results.
(S)-Amlodipine concentration (mmol/L) Values of validity
CERF model
2
R S.D. (%)
1 0.9981
2 0.9980
3 0.9944
4 0.9958
5 0.9991
The third path concerned the membrane phase. The membrane
mass-transfer coefficient (km) was used to determine the
concentration profile of (S)-amlodipine in the liquid membrane
phase which was calculated by the equal flux relationship as in Eq.
(12) and verified by previous work [12]. The membrane mass-
transfer coefficient (km) was calculated as 2.89  102 cm/s.
The mass transfer reaction of (S)-amlodipine through HFLSM
was assessed by chiral reaction flux (CRF) modeling. The
concentrations of (S)-amlodipine in the retentate (feed) solution,
as a function of time, for initial feed concentrations of 1, 2, 3, 4 and
5 mmol/L were investigated. Theoretical results from the mathe-
matical model were shown to be in good agreement and fitted well
with the experimental data, as represented in Fig. 8.
The diffusion process is explained by Fick’s law of diffusion.
However, mass transfer flux was presented in this model.
Therefore, it could be concluded that the enantiomeric flux model
was satisfactory for extraction of (S)-amlodipine through hollow
fiber supported liquid membrane.
The concentrations of (S)-amlodipine in the raffinate stripping
solution, as a function of time, for initial feed concentrations at 1, 2,
3, 4 and 5 mmol/L were determined. The mathematical model
results (Fig. 9) are shown by the line which fitted well with the
experiment data.
Minimum coefficient of determination (R2) was 0.9944 for the
extraction and 0.9956 for stripping. Maximum percentage of
standard deviation was 2.31% for the extraction and only 1.26% for
stripping. Results are shown in Table 6.
5. Conclusion
The chiral reaction flux (CRF) model was developed in order to
analyze mass transfer rates for selective enantioseparation of (S)-
amlodipine from pharmaceutical wastewater by using a hollow fiber
supported liquid membrane containing a synergistic chiral extrac-
tant. The mathematical model was found to be most useful.
Please cite this article in press as: N. Sunsandee, et al., J. Ind. Eng. Chem. (2013), http://dx.doi.org/10.1016/j.jiec.2013.08.008
CSRF model
R2 S.D. (%)
2.31 0.9956 1.26
1.93 0.9983 0.97
1.88 0.9988 0.92
1.84 0.9979 0.73
0.89 0.9984 0.42
Experimental validation results showed that the proposed model
was able to predict the concentration of the enantiomers throughout
the HFSLM process. Moreover, the chiral reaction flux (CRF) model
could be applied to the enantioseparation of chiral drugs.
Acknowledgements
The authors are sincerely grateful to the Chulalongkorn
University Dutsadi Phiphat Scholarship, the Integrated Innovation
Academic Center: IIAC Chulalongkorn University Centenary
Academic Development Project (RES 560530019) and the Chro-
matography and Separation Research Unit, Department of
Chemistry, Faculty of Science, Chulalongkorn University. Addi-
tional thanks are given to the Separation Laboratory, Department
of Chemical Engineering, Faculty of Engineering, Chulalongkorn
University, Bangkok, Thailand for chemical and apparatus support,
as well as to the Government Pharmaceutical Organization (GPO),
Thailand, for kindly supplying active pharmaceutical ingredients
and all instruments for analysis and measurement.
References
[1] Z. Chen, N. Ren, A. Wang, Z. Zhang, Y. Shi, Water Res. 42 (2008) 3385.
[2] Y.A. Oktem, O. Ince, P. Sallis, T. Donnelly, B.K. Ince, Bioresour. Technol. 99 (2007)
1089.
[3] N. Nakada, K. Kiri, H. Shinohara, A. Harada, K. Kuroda, S. Takizawa, H. Takada,
Environ. Sci. Technol. 42 (2008) 6347.
[4] B.A. Alistair, Eur. Mol. Biol. Org. 5 (2013) 1110.
[5] I. Rodriguez-Donis, V. Gerbaud, A. Arias-Barreto, X. Joulia, Heterogeneous batch
distillation processes for waste solvent recovery in pharmaceutical industry,
Comput. Aided Chem. Eng. 27 (2009) 1119.
[6] D. Samuel, S. Raj, Y. Anjaneyulu, Proc. Biochem. 40 (2005) 165.
[7] M. Breuer, K. Ditrich, T. Habicher, B. Hauer, M. Kesseler, R. Sturmer, T. Zelinski,
Angew. Chem. Int. 43 (2004) 788.
[8] S. Heng, K.L. Yeung, M. Djafer, J.C. Schrotter, J. Membr. Sci. 289 (2007) 67.
[9] P. Ramakul, U. Mooncluen, Y. Yanachawakul, N. Leepipatpiboon, J. Ind. Eng. Chem.
18 (2012) 1606.
[10] N.S. Rathore, J.V. Sonawane, A. Kumar, A.K. Venugopalan, R.K. Singh, D.D. Bajpai,
J.P. Shukla, J. Membr. Sci. 189 (2001) 119.
G Model
JIEC-1501; No. of Pages 11

N. Sunsandee et al. / Journal of Industrial and Engineering Chemistry xxx (2013) xxx–xxx 11

[11] N. Sunsandee, N. Leepipatpiboon, P. Ramakul, U. Pancharoen, Chem. Eng. J. 180 [19] B. Tan, G.S. Luo, J.D. Wang, Tetrahedron: Asymmetry 17 (2006) 883.
(2012) 299. [20] K.W. Tang, Y.Y. Chen, J.J. Liu, Sep. Purif. Technol. 62 (2008) 681.
[12] N. Sunsandee, P. Ramakul, N. Thamphiphit, U. Pancharoen, N. Leepipatpiboon, [21] B. Tan, G.S. Luo, X. Qi, J.D. Wang, Sep. Purif. Technol. 53 (2007) 330.
Chem. Eng. J. 209 (2012) 201. [22] M. Zandkarimi, A. Shafaati, S.M. Foroutan, Iran. J. Pharm. Res. 7 (2008) 275.
[13] N. Sunsandee, N. Leepipatpiboon, P. Ramakul, T. Wongsawa, U. Pancharoen, Sep. [23] M. Zandkarimi, A. Shafaati, S.M. Foroutan, C.A. Lucy, DARU 17 (4) (2009) 269.
Purif. Technol. 102 (2013) 50. [24] T. O’Brien, L. Crocker, R. Thompson, K. Thompson, P.H. Toma, D.A. Conlon, B.
[14] N. Sunsandee, U. Pancharoen, P. Rashatasakhon, P. Ramakul, N. Leepipatpiboon, Feibush, C. Moeder, G. Bicker, N. Grinberg, Anal. Chem. 69 (1997) 1999.
Sep. Sci. Technol. (2013), http://dx.doi.org/10.1080/01496395.2013.804088. [25] U. Pancharoen, T. Wongsawa, A.W. Lothongkum, Sep. Sci. Technol. 46 (2011)
[15] Z. Wang, C. Cai, Y. Lin, Y. Bian, H. Guo, X. Chen, Sep. Purif. Technol. 79 (2011) 63. 2183.
[16] S. Panja, P.K. Mohapatra, S.C. Tripathi, P.M. Gandhi, P. Janardan, J. Hazard. Mater. [26] S. Chaturabul, K. Wongkaew, U. Pancharoen, Sep. Sci. Technol. 48 (2013) 93.
237/238 (2012) 339. [27] P. Kittisupakorn, W. Weerachaipichaskul, P. Thitiyasook, J. Ind. Eng. Chem. 13
[17] S. Agarwal, M.T.A. Reis, M.R.C. Ismael, M.J.N. Correia, J.M.R. Carvalho, Sep. Purif. (2007) 903.
Technol. 102 (2013) 103. [28] Y.S. Chung, M.C. Ha, Resolution of the enantiomers of amlodipine, US Patent
[18] Z.F. Cui, Y. Jiang, R.W. Field, in: Z.F. Cui, H.S. Muralidhara (Eds.), Membrane Application 2003/6646131 B2 (November 11, 2003).
Technology: A Practical Guide to Membrane Technology and Applications in [29] H.S. Fogler, Elements of Chemical Reaction Engineering, 4th ed., Prentice-Hall,
Food and Bioprocessing, Butterworth-Heinemann, Elsevier, UK, 2010, p. 12. New Jersey, 2005p. 253.

Please cite this article in press as: N. Sunsandee, et al., J. Ind. Eng. Chem. (2013), http://dx.doi.org/10.1016/j.jiec.2013.08.008