Separation and Purification Technology 53 (2007) 216–223

Separation of glycyrrhizic acid and liquiritin from Glycyrrhiza uralensis

Fisch extract by three-liquid-phase extraction systems
Shufeng Shen a,b , Zhidong Chang a , Ji Liu a , Xinghua Sun a , Xin Hu a , Huizhou Liu a,∗
a Laboratory of Separation Science and Engineering, State Key Laboratory of Biochemical Engineering, Institute of Process Engineering,
Chinese Academy of Sciences, P.O. Box 353, Beijing 100080, PR China
b Graduate School of the Chinese Academy of Sciences, Beijing 100080, PR China

Received 6 January 2006; received in revised form 1 July 2006; accepted 11 July 2006

Abstract
Some stable three-liquid-phase extraction systems have been used for isolation and separation of glycyrrhizic acid (GA) and liquiritin (LQ) in
aqueous Glycyrrhizia uralensis Fisch extract. Partition behaviors of two compounds were investigated at various conditions. Organic extractants and
the pH of solution were two main factors to affect their partitioning among the phases and separation strategies. Different extraction profiles showed
that the stronger the electron-donating capability of extractants, the higher the upper-limit of pH in the region for co-extracting GA and LQ, and the
higher the lower-limit of pH in the region for stripping GA or LQ from the organic phase. Of all the investigated systems, the three-liquid-phase
system containing poly(ethylene glycol-ran-propylene glycol) (EOPO2500) and tri-butyl phosphate (TBP) was the most suitable for separation
and purification of licorice extract. FTIR and fluorescence spectroscopic investigations showed that hydrogen bond interaction and hydrophobic
interaction among the phases were the main driving forces to affect the partition behaviors of two compounds.
© 2006 Elsevier B.V. All rights reserved.

Keywords: Three-liquid-phase extraction; Separation; Glycyrrhizic acid; Liquiritin; Glycyrrhizia uralensis Fisch; FTIR

1. Introduction extract, it can be roughly classified into four principal categories:

Licorice root has been extensively used as a traditional medic-
inal herb in China for over 2000 years. Among the about 200
chemical constituents, triterpenoids and flavonoids have been
demonstrated the major ingredients and have several important
pharmacological activities, including anti-viral, anti-oxidant,
anti-inflammatory, anti-ulcer, anti-cancer and anti-HIV [1–3].
The liquid extract from Glycyrrhizia uralensis root with a
water-soluble organic solvent or an aqueous alkaline solution is
usually produced by several methods such as room temperature
extraction, ultrasonic extraction, microwave-assisted extraction,
etc. [4,5]. Among the multi-component extract, glycyrrhizic acid
(GA) and liquiritin (LQ) are two dominant components and their
structures are shown in Fig. 1 [2]. GA and LQ are polyhydroxyl
compounds and GA contains three carboxyls group, which
demonstrate their different characteristics. After reviewing tech-
nologies for separating bioactive compounds from licorice
∗ Corresponding author. Tel.: +86 10 62554264; fax: +86 10 62554264.
E-mail address: hzliu@home.ipe.ac.cn (H. Liu).
precipitation by acidification [6]; chromatographic separation
[7–9] such as high performance liquid chromatography (HPLC),
high-speed counter-current chromatography (HSCCC); adsorp-
tion separation with different adsorbents [6,10–13]; extraction
technologies such as solvent extraction [4,14], foam separation
[15], aqueous two phase extraction [16], and so on. These tech-
nologies are very often time consuming, having low recovery
and high cost, and require relatively high quality demand for
licorice extract. Moreover, it is surprisingly found that a large
amount of work has been done concentrating on the separation
and purification of GA. But there are few studies referring to
simultaneous recovery and separation of GA and flavoniods.
In recent years, their pharmacological activities have been more
and more concerned about. Some clinical data demonstrated that
flavoniods would not give rise to side effects even for a long-term
intake. Hence, some new alternative processes or technologies
to separate GA and LQ from each other as well as impurities are
being sought.
Third-phase extraction, first observed in the classical two-
phase solvent extraction, is a promising technology for sep-
aration and analysis of complex systems containing metals

1383-5866/$ – see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.seppur.2006.07.003
 S. Shen et al. / Separation and Purification Technology 53 (2007) 216–223
Fig. 1. Chemical structures of glycyrrhizic acid and liquiritin: (a) glycyrrhizic acid; (b) liquiritin.
and organic compounds [17]. Due to different physicochemical
properties of three liquid phases, they could offer new possibili-
ties of separating two or more compounds by single extraction. In
the last few years, some new three-liquid-phase systems, which
were composed of organic solvent, polymers, inorganic salt and
water, were developed and applied to the fields of pharmaceuti-
cal industry and wastewater treatment [18–20].
The application of three-liquid-phase extraction on active
ingredients of aqueous licorice extract will offer new practi-
cal approach to separate or fractionate different compounds or
groups of nature product. In this study, some stable three-liquid-
phase extraction systems were used for separation of GA and
IQ in aqueous G. uralensis extract. Partition behaviors of two
compounds were investigated at various experimental condi-
tions, such as the pH of solution, kinds and concentrations of
extractants and phase-forming components. Moreover, their par-
titioning mechanisms among phases were studied by FTIR and
fluorescence spectroscopy.
2. Materials and methods
2.1. Chemicals and materials
Sample slices (thickness about 3–5 mm) of G. uralensis
were provided by Inner Mongolia Yili Science and Technology
Industry Co. Ltd. (China). Glycyrrhizic acid monoammonium
salt trihydrate (C42 H65 O16 ·3H2 O, MW 894.03, 98%) was pur-
chased from the Acros Organics (USA). Liquiritin (C21 H22 O9 ,
MW 418.39, 98%) was purchased from National Pharmaceu-
tical Engineering Center for Solid Preparation in Chinese
Herbal Medicine (China). Polyethylene glycol 2000 (PEG2000),
Polyethylene glycol 4000 (PEG4000) and Polyethylene gly-
col 6000 (PEG6000) were purchased from Tianjin Tiantai
Fine Chemial Co. Ltd., PR China. Poly(ethylene glycol-
ran-propylene glycol), Average Mn. ca.2500 (EOPO2500),
was purchased from Sigma–Aldrich Inc. Cyanex923 (TRPO,
MW = 340–350, 93%) was a mixture of four trialkylphosphine
oxides (R3 P O, R2 R P O, RR 2 P O, R 3 P O; where R is
hexyl and R is octyl), which was kindly supplied by Canada
Cytec Inc. and diluted with kerosene without any further purifi-
cation. Pyrene (99%) was purchased from Sigma and was recrys-
217
tallized twice from ethanol. Acetonitrile, acetic acid, methanol
and water were of HPLC grade. Tri-butyl phosphate (TBP), 2-
ethyl hexanol and other chemicals were of analytical reagent
grade.
The pKa values and octanol/water partition coefficient (log P)
of GA and LQ were calculated at 25 ◦ C using the ACD/pKa /log P
DB software developed by Advanced Chemistry Development,
Inc. (Canada). All the contents of polymers, salt and other com-
positions were given as percentage of weight (wt.%). Stock
solutions of ammonium sulphate ((NH4 )2 SO4 ) with concen-
tration of 40% and polymers with concentration of 50%, were
prepared. A stock solution of 1 mM pyrene in methanol was
prepared.
2.2. Preparation of aqueous licorice extract by ultrasonic
extraction
Dried licorice slices (20.0 g) were extracted twice, each with
150 ml of aqueous ammonia solution (0.5 vol.%) for 45 min by
sonication using an ultrasonic cleaning instrument. The residue
was filtrated and washed with 50 ml of distilled water. The com-
bined extracts were used as stock solution.
2.3. Three-liquid-phase extraction
Three-liquid-phase systems contained four components that
were organic solvent, inorganic salt, polymer and the treated
licorice extract. The selected phase-forming agents were shown
in Table 1.
Extraction experiments were carried out in beakers with mag-
netic stirring. All phase-forming components, aqueous licorice
Table 1
Phase-forming components in three-liquid-phase systems
Phase-forming Selected chemicals Percent range
components (wt.%)
Organic solvent 2 2-Ethyl hexanol, TBP, TRPO 20–25
Polymer PEG2000, PEG4000, PEG6000, 7.5–15
EOPO2500
Inorganic salt (NH4 )2 SO4 8.5–14
218 S. Shen et al. / Separation and Purification Technology 53 (2007) 216–223

extract were added at presetted weight from stock solutions

and the systems were prepared with a final total mass of about
38.5 g at room temperature (about 25 ◦ C). Then, the suitable
pH of the mixture solution was adjusted with sulphuric acid
(2 mol/l) or sodium hydroxide solution (1 mol/l) controlled by
Microprocessor pH meter (pH 211, Hanna, Italy). After mixing,
the phase separation was enhanced by centrifugation for 5 min at
2000 rpm. There liquid phases, an upper organic phase, a middle
copolymer-rich phase and a lower aqueous phase, were formed.
Three-phase volumes were measured by graduated pipette and
the content of constituents in each phase was analyzed by HPLC.

2.4. Determination of GA and LQ

Concentrations of GA and LQ were analyzed by HPLC

(HP1100 Agilent Technologies, USA) with UV at 248 and
275 nm. A C18 column (250 mm × 4.6 mm × 5 ␮m, Agilent
Technologies, USA) was used along with a mobile phase consist-
ing of acetonitrile (A) and 3.0% aqueous acetic acid (B) using a
gradient elution of 15–45% at 0–18 min, 45–65% at 18–20 min,
65% at 20–22 min and 65–15% at 22–24 min at a flow rate of
1.0 ml/min. Retention time of LQ and GA were about 9.8 min
and 21.7 min, respectively. Calibration curves of GA and LQ
showed good linearity (r2 > 0.999) over the range from 0.2 to
9.9 ␮g and from 0.1 to 6.0 ␮g, respectively. The analytical oper-
ation could be completed in 28 min.
The mass fraction of compounds in each phase was calculated
as follows:
C j × Vj
wj =
C0 × V 0
where j denotes the top (t) or middle (m) or bottom (b) phase in
the system and wj represents the mass fraction of compound in
the j phase. Cj and Vj are the concentration and phase volume for
the compound in the j phase, respectively. C0 is the concentration
of the compound in the stock solution and V0 is the volume of
stock solution added to the system.
Fig. 2. Partition profiles of GA and LQ in the three-liquid-phase system with
2-ethyl hexanol as extractant at 25 ◦ C. The system contained 22.0 wt.% 2-ethyl
hexanol, 13.6 wt.% PEG2000 and 9.4 wt.% (NH4 )2 SO4 . GA: , wt ; 䊉, wm ; ,
wb ; LQ: , wt ; , wm ; , wb .
3. Results and discussion
3.1. Three-liquid-phase extraction profiles of GA and LQ
with different extractants
Partition behaviors of GA and LQ were investigated in three-
liquid-phase systems with different extractants. In this part, the
compositions of components in the tested systems were as fol-
low: 22.0 wt.% extractant, 13.6 wt.% PEG2000 and 9.4 wt.%
(NH4 )2 SO4 and about 30.0 wt.% aqueous extract of licorice.
(1)
Effects of pH on mass fraction of GA and LQ in each phase with
2-ethyl hexanol, TBP, TRPO and 10 vol.% TRPO (kerosene as
diluent) were typically shown in Figs. 2–4, respectively.
In the investigated pH range, partition behavior of GA and
LQ in the three-liquid-phase system with 2-ethyl hexanol could
be discussed in two regions, A region (pH < 4.3) and B region
(pH > 4.3), as shown in Fig. 2. For LQ, its partition behavior

2.5. FTIR spectroscopy

FTIR spectra with a resolution of 2 cm−1 were recorded

on a Bruker Vecter 22 FTIR spectrometer in the region from
800 to 4000 cm−1 using a DTGS (deuterotriglycine sulphate)
detector at room temperature (25 ◦ C). FTIR spectra of the liq-
uid samples were recorded by scanning 32 times and using a
BaF2 cell. The OPUS spectroscopic software was used for data
handling.

2.6. Fluorescence spectroscopy

Stock solution of pyrene, 3 ␮l, was added to 10 ml of samples

of three phases and mixed together, respectively. Fluorescence
emission spectra of all samples were recorded on a Perkin-Elmer
LS-55 fluorescence spectrophotometer with excitation wave-
length at 335 nm. Excitation and emission slit widths were set
as 5.0 and 2.5 nm, respectively.
Fig. 3. Partition profiles of GA and LQ in the three-liquid-phase system with
TBP as extractant at 25 ◦ C. The system contained 22.0 wt.% TBP, 13.6 wt.%
PEG2000 and 9.4 wt.% (NH4 )2 SO4 . GA: , wt ; 䊉, wm ; , wb ; LQ: , wt ; ,
wm ; , wb .
 S. Shen et al. / Separation and Purification Technology 53 (2007) 216–223
Fig. 4. Partition profiles of GA and LQ in the three-liquid-phase system with
TRPO as extractant at 25 ◦ C. The system contained 22.0 wt.% TRPO, 13.6 wt.%
PEG2000 and 9.4 wt.% (NH4 )2 SO4 . GA: , wt ; 䊉, wm ; , wb ; LQ: , wt ; ,
wm ; , wb .
almost did not change in the two regions. About 98% of LQ were
partitioned into the middle phase and only trance was found in
the other two phases. But for GA, there was an obvious difference
between two regions. In the A region, mass fraction of GA in the
top and middle phase almost equaled (about 50 wt.%) at around
pH 2.0. But mass fraction of GA in the top phase was gradually
reduced with the increasing pH. Most of them were transferred
into the PEG2000-rich middle phase from the top phase when
the pH was around 4.0. With further increasing pH into the B
region, most of them were retained in the middle phase along
with the LQ.
Different from 2-ethyl hexanol as extractant, partitioning of
GA and LQ in the three-liquid-phase systems with TBP was
very complex but interesting phenomena. Mass fraction of GA
and LQ had a great change between the top and middle phase
in these three-liquid-phase systems. Both of them almost were
not detected in the bottom phase. Five different regions of par-
tition profiles, A region (pH below 3.3), B region (pH 3.3–6.1),
C region (pH 6.1–8.8), D region (pH 8.8–11.0) and E region
(pH above 11.0), were shown in Fig. 3 respectively. In the A
region, two compounds, above 95% of GA and 85% of LQ,
could be unevenly partitioned into the top phase. Only about
10% of LQ and trance of GA were kept in the middle phase.
With the increasing pH (B region), GA was transferred into the
middle phase from the top phase, but most of LQ were still par-
titioned into the top phase. In the C region, around the neutral
pH range, two compounds were partitioned into the middle and
top phase, respectively. Therefore, two compounds could be iso-
lated and separated from each other. At the following D and E
regions, there was no obvious change about the partitioning of
GA. However, LQ in the top phase was gradually transferred into
the middle phase at the higher pH (D region). In the E region,
almost all of LQ were partitioned into the middle phase. These
behaviors would offer a possibility of stripping the LQ from the
top TBP phase. Therefore, different purification strategies for
GA and LQ could be considered and the extractant could also
be recovered and reused.
219
Fig. 4 showed partitioning of GA and LQ in three-liquid-
phase systems with TRPO as extractant. Similarly, partition
profiles of two compounds were divided into four main regions
by three important pH points (5.0, 7.0 and 9.5). In the C region,
GA and LQ could still achieve efficient separation. However,
compared to the aforementioned systems using TBP as extrac-
tant, there are four different characteristics about the curves.
Firstly, mass fraction of LQ in the top phase could be up to
about 100% when the pH was below 9.5, which was higher than
the systems with TBP. Secondly, the minimal pH to simultane-
ously extract two compounds into the organic phase was raised
to about 5.0 from 3.3. Thirdly, the B region (pH 3.3–6.1) was
shortened and moved to the higher pH range (pH 5.0–7.0), where
GA could be inversely partitioned into the middle phase from
the top phase with the increasing pH. Lastly, the D region was
enlarged and the E region disappeared in the same investigated
pH range. At pH 11.69, about 40% of LQ were still kept into
the middle phase. Therefore, LQ in the top phase might proba-
bly be transferred into the middle phase at the higher pH than
pH 12.0, which would probably destroy the bioactivities of LQ.
These features showed the TRPO had much stronger affinity or
extraction ability than TBP under the same conditions.
For the diluted extractant, 10 vol.% TRPO, their partition
profiles showed different characteristics from those in the afore-
mentioned three systems. As shown in Fig. 5, extraction profiles
could also be divided into four regions, A region (pH below
3.0), B region (pH 3.0–5.3), C region (pH 5.3–7.0) and D region
(pH above 7.0), respectively. In the whole investigated range,
about 70–90% of LQ were partitioned into the middle phase,
and its mass fraction had a little increase with the increasing pH
in the A and D region. Correspondingly, its mass fraction in the
top phase gradually decreased. At about pH 8.4, its mass frac-
tion could reach above 95% in the middle phase. Only the A–C
regions were correspondingly moved to the lower pH range,
where the critical points of pH decreased about two points.
From the partition behaviors of GA and LQ in the aforemen-
tioned three-liquid-phase systems, it can be clearly seen that the
Fig. 5. Partition profiles of GA and LQ in the three-liquid-phase system with
10 vol.% TRPO as extractant at 25 ◦ C. The system contained 22.0 wt.% extrac-
tant, 13.6 wt.% PEG2000 and 9.4 wt.% (NH4 )2 SO4 . GA: , wt ; 䊉, wm ; , wb ;
LQ: , wt ; , wm ; , wb .
220 S. Shen et al. / Separation and Purification Technology 53 (2007) 216–223
pH and different extractants can greatly affect their partitioning.
These extraction characteristics could be explained by consid-
ering hydrogen bond interaction, hydrophobic interaction and
physicochemical properties of two compounds.
GA (pKa1 = 2.76 ± 0.70; pKa2 = 2.81 ± 0.70; pKa3 = 4.71 ±
0.70) is a weak acid with three carboxyl and five hydroxyl
groups, and LQ (pKa1 = 7.79 ± 0.40; pKa2 = 12.79 ± 0.70) is
also a polyhydroxyl compound. Under strong acidic condi-
tions (pH < 2.5), most of GA and LQ exist as molecular form.
However, dissociated form of GA predominates under pH > 6.0.
At the higher pH (e.g. pH > 9.0), phenolic hydroxyl groups in
LQ can also dissociate to form H+ and their corresponding
anions. Molecular form is more hydrophobic than the dissoci-
ated form. Organic extractants, 2-ethyl hexanol, TBP and TRPO,
have polar functional groups (C–O or P O bond), which could
form hydrogen bond interaction with the molecular form of
GA and LQ. Strength of bond interaction depends strongly on
the electron-donating capability of extractants and the molecu-
lar state of compounds. As we have known, dipole moment of
P O bond (D = 2.7) was larger than that of C–O band (D = 0.7).
Hence, electron-donating capability of phosphate-containing
extractants was much greater than that of aliphatic alcohol. But
for TBP and TRPO, with the alkoxy groups and alkyl groups
(R = hexyl or R = octyl) linking to the phosphorus atom respec-
tively, P O bond in TRPO became a much stronger electron
donor through inductive effects [21]. Therefore, the sequence
order of electron-donating capability of different extractants
could be TRPO > TBP > 2-ethyl hexanol. These bond interac-
tions probably resulted in the corresponding changes in different
regions. The weaker the hydrogen bonding interaction between
extractants and two compounds, the lower or upper limit of the
pH region would be close to the pKa value. The stronger the
electron-donating capability of extractants, the higher the upper
limit of pH in the region for co-extracting GA and LQ, and the
higher the lower limit of pH in the region for stripping GA or
LQ from the organic phase.
In the other hand, the partition behaviors of GA and LQ would
be explained from the viewpoint of hydrophobic interaction.
log P is a good measure of the hydrophobicity and hydrophilic-
ity for neutral compounds. A high log P indicates a compound
will preferentially partition into organic phase rather than water,
and the compound shows a high hydrophobicity. As shown in
Table 2, molecular form of GA was more hydrophobic than that
of LQ. The hydrophobicity sequence order of different extrac-
Table 2
Octanol–water partition coefficient (log P) of solutes and organic extractants
log P (calculated data) log P (experimental
dataa )
GA 4.64 ± 0.75 2.80
LQ 0.61 ± 0.38 –
2-Ethyl hexanol 2.82 ± 0.19 –
TBP 4.26 ± 0.30 4.00
TRPO (R = C6, hexyl) 6.21 ± 0.55 –
TRPO (R = C8, octyl) 7.80 ± 0.55 –
a Data from ACD/log P Software Database developed by Advanced Chemistry
Development, Inc.
tants could be TRPO > TBP > 2-ethyl hexanol. Moreover, the
PEG2000-rich middle phase was more hydrophobic than the
salt-rich bottom phase, but much less hydrophobic than the top
phase. Molecular forms of two compounds are favorable for
keeping in the more hydrophobic microenvironment. Therefore,
the different hydrophobic characteristics would probably offer
another driving force to partition two compounds into the three
phases. The affinity of two compounds to the bottom phase was
much lower than that to the other two phases. The reason could
possibly due to the strong salting-out interaction of ammonium
sulphate.
3.2. Effects of phase-forming polymers on partition
behaviors
Different phase-forming polymers, PEG2000, PEG4000,
PEG6000 and EOPO2500, were used to form three-liquid-phase
systems at room temperature. TBP was used as organic extrac-
tant. System compositions of polymers, ammonium sulphate and
extractant were 10.9, 9.4 and 22.0 wt.%, respectively. Aqueous
licorice extract was added from the stock solution. Sulphuric
acid solution was added to maintain pH at 4.45 under the mixing
condition. Effects of phase-forming polymer on mass fraction
of GA and LQ in each phase were illustrated in Figs. 6 and 7,
respectively.
It can be clearly seen that effects of molecular weight and
type of polymer on partitioning of GA was greater than that of
LQ. Increase in the molecular weight of polymer would drive
two compounds to partition more towards the middle phase.
Moreover, partitioning of two compounds was very similar in
the systems containing PEG4000 and EOPO2500 respectively.
These partitioning might be attributed to different hydropho-
bic properties of the middle phase. The greater molecular
weight of polymer, the more hydrophobic the microenviron-
ment in the middle phase. The order of hydrophobicity would
be PEG2000 < PEG4000 < PEG6000. Due to the introduction
of hydrophobic segments (PPO), EOPO copolymers have more
hydrophobic than PEG with the same molecular mass. At pH
Fig. 6. Effect of phase-forming polymers on partitioning of GA at pH 4.45.
These systems contained 22.0 wt.% TBP, 10.9 wt.% polymers and 9.4 wt.%
(NH4 )2 SO4 .
 S. Shen et al. / Separation and Purification Technology 53 (2007) 216–223
Fig. 7. Effect of phase-forming polymers on partitioning of LQ at pH 4.45.
These systems contained 22.0 wt.% TBP, 10.9 wt.% polymers and 9.4 wt.%
(NH4 )2 SO4 .
4.45, most of molecular form of GA and LQ were partitioned
into the organic phase. Only small percent (about 5%) were
kept in the PEG2000-rich middle phase. With the increasing
hydrophobicity of microenvironment, more molecules of two
compounds were transferred into the middle phase. As shown in
Figs. 6 and 7, mass fraction of GA and LQ in the middle phase
enriched in PEG6000 could be up to 30% and 20%, respectively.
3.3. Effects of concentration of phase-forming components
on partitioning
In this section, all the partitioning investigations were per-
formed at pH 2.98. Six different concentration combination of
EOPO2500 and ammonium sulphate were studied in the three-
liquid-phase systems. 2-Ethyl hexanol was used as extractant
and its weight percent in these systems was 22.1%.
As shown in Figs. 8 and 9, concentration variation of
EOPO2500 or ammonium sulphate had a little influence on
Fig. 8. Effects of concentrations of phase-forming components on partitioning of
GA at pH 2.98. 1, 2, 3, 4, 5, 6 represented three-liquid-phase systems containing
different concentration combination of EOPO2500/(NH4 )2 SO4 , 11.7%/8.7%,
11.7%/11.2%, 11.7%/14.0%, 7.8%/10.9%, 10.9%/10.9%, 14.0%/10.9%,
respectively. The extractant was 2-ethyl hexanol and its percent of weight in
these systems was 22.1%.
221
Fig. 9. Effects of concentrations of phase-forming components on partitioning
of LQ at pH 2.98. 1, 2, 3, 4, 5, 6 represented three-liquid-phase systems con-
taining different concentration combination of EOPO2500/(NH4 )2 SO4 , 11.7%/
8.7%, 11.7%/11.2%, 11.7%/14.0%, 7.8%/10.9%, 10.9%/10.9%, 14.0%/10.9%,
respectively. The extractant was 2-ethyl hexanol and its percent of weight in
these systems was 22.1%.
the partition behaviors of two compounds compared to other
parameters such as pH, extractants. With the increasing concen-
tration of ammonium sulphate or EOPO2500, mass fraction of
GA decreased gradually in the top phase and increased corre-
spondingly in the middle phase. Its mass fraction in two phases
varied between 40% and 55%. However, above 90% of LQ was
partitioned into the middle phase in all the investigated systems.
Its mass fraction was not obvious change. These results were
probably attributed to extraction efficiency of extractant and
different hydrophobic properties in the middle phase. 2-Ethyl
hexanol had much weaker hydrogen bond interaction than TBP
and TRPO, which showed low extraction efficiency and capacity.
Therefore, change of hydrophobicity in the middle phase would
have great effect on portioning between the top and the middle
phase. With the increasing mass fraction of copolymer or ammo-
nium sulphate, the concentration of EOPO2500 in the middle
phase increased and more water was forced to pass into the bot-
tom phase. Accordingly, the more hydrophobic environment of
the middle phase would have more GA and LQ transferred into
the middle phase from the top phase. Moreover, due to possi-
ble aggregation of GA molecules at lower pH resulting in the
increasing viscosity of solution, two compounds would also be
inclined to partition into the polymer-rich middle phase or inter-
phase.
3.4. FTIR measurements of organic phase
FTIR spectroscopy is a chemical-microenvironment-
sensitive and functional-group-characteristic technique. The
representative spectra of organic phase, TBP and TRPO, at
different loading conditions were shown in Figs. 10 and 11.
The bands from 1200 to 1350 cm−1 are attributed to the
P O stretching vibration of TBP [22]. As shown in Fig. 10,
peaks around 1282 and 1272 cm−1 were assigned to the unhy-
drated and the hydrated P O stretching vibration respectively
in the blank three-liquid-phase system. After three-liquid-phase
222 S. Shen et al. / Separation and Purification Technology 53 (2007) 216–223
Fig. 10. Representative FTIR spectra of TBP phase in P O stretching vibra-
tion region (1360–1180 cm−1 ) at different conditions: (a) TBP; (b) TBP phase
loading GA; (c) TBP phase loading GA and LQ; (d) TBP phase loading LQ.
extraction, TBP phase loading LQ could lead to shift of the P O
stretching vibration from 1282 to 1267 cm−1 (d line). But, if
the TBP phase was loaded only GA or both GA and LQ, inten-
sity of the P O stretching vibration decreased greatly and the
peak shifted to the lower wavenumber around 1263 cm−1 (b and
c line). These results above indicated that there had an obvious
hydrogen bond interaction between TBP and two compounds
and the interaction between TBP and GA was much stronger,
rendering the peak of P O stretching vibration shifted to a much
lower wavenumber.
It can be clearly seen from the Fig. 11 that the P O stretch-
ing vibration peak of pure TRPO was located at 1150 cm−1 ,
but after dilution to 10 vol.% with kerosene, the P O stretch-
ing vibration peak of TRPO moved to 1168 from 1150 cm−1 .
This demonstrated the existence of a strong interaction among
TRPO molecules in pure TRPOs, probably due to hydrogen
bond formation between the O atoms in the P O bond and
the H atom in the alkyl chain. The addition of kerosene might
Fig. 11. Representative FTIR spectra of TRPO in P O stretching vibration
region (1210–1100 cm−1 ) at different conditions: (a) TRPO in blank system;
(b) TRPO phase loading GA and LQ; (c) 10 vol.% TRPO in blank system; (d)
10 vol.% TRPO phase loading GA and LQ.
weaken the hydrogen bonds between two TRPO molecules,
which resulted in the P O stretching vibration peak being moved
to a higher wavenumber and the absorbance intensity of peak
greatly decreased. But, after loading GA and LQ for pure TRPO
and 10% TRPO, the P O stretching vibration peak moved to
1152 and 1144.5 cm−1 , respectively. These shifts demonstrated
the hydrogen bond formation between the P O group and the
hydroxyl or carboxyl groups in LQ and GA.
3.5. Fluorescence spectroscopy investigation on polarity of
three phases
Pyrene is a well-characterized polarity-sensitive probe and
its fluorescence emission spectrum consists mainly of five bands
(I1 − I5 ) from shorter to longer wavelengths [23]. The I1 /I3 inten-
sity ratio of this vibrational fine structure depends strongly on the
polarity of medium. The larger the intensity ratio, the more polar
the medium. The micropolarity of three phases were investigated
using pyrene as a fluorescent probe at 25 ◦ C in different three-
liquid-phase systems which contained 22.0 wt.% TBP, 9.4 wt.%
(NH4 )2 SO4 and 10.9 wt.% polymers. M-1, M-2, M-3 and M-
4 represent the middle phase enriched in PEG2000, PEG4000,
PEG6000 and EOPO2500, respectively. Ratio of vibronic band
intensities, R = I1 /I3 , in various solutions of were depicted in
Fig. 12. The data obtained from the middle phase were com-
pared to those obtained in the top organic phase (T) and the
salt-rich bottom phase (B).
It can be clearly seen that the I1 /I3 ratios followed the
order I1 /I3 (T)  I1 /I3 (M) < I1 /I3 (B). The micropolarity of
top phase was much lower than that of other two phases.
Hence, molecular form of GA and LQ should have strong
affinity to the more hydrophobic top phase. However, for
those systems containing the same extractants, results showed
that the higher molecular weight of polymer in the middle
phase, the lower the I1 /I3 value. The sequence order was I1 /I3
(PEG2000) > I1 /I3 (PEG4000) > I1 /I3 (PEG6000). Due to the
Fig. 12. Ratio of vibronic band intensities, R = I1 /I3 , of pyrene (0.3 ␮M) in vari-
ous solutions of three-liquid-phase systems containing 22.0 wt.% TBP, 9.4 wt.%
(NH4 )2 SO4 and 10.9 wt.% polymers at 25 ◦ C. T: a representative top phase
(TBP); B: a representative bottom phase; M-1, M-2, M-3 and M-4 represent
the middle phase enriched in PEG2000, PEG4000, PEG6000 and EOPO2500,
respectively.
 S. Shen et al. / Separation and Purification Technology 53 (2007) 216–223
effect of the hydrophobic block (PPO), I1 /I3 (EOPO2500) was
similar to I1 /I3 (PEG4000). Therefore, there were different
hydrophobic properties in the three phases, which might pro-
duce hydrophobic interaction with the molecular form of two
compounds. These characteristics could probably result in dif-
ferential partitioning of GA and LQ, which were consistent with
the aforementioned partitioning in different phases of systems.
4. Conclusions
GA and LQ in aqueous extract of G. uralensis Fisch could
be unevenly partitioned into the different phases by three-liquid-
phase extraction. Different extraction profiles or partition behav-
iors were obtained at the various conditions. Organic extractant
and pH were the main factors to affect the partitioning of two
compounds. Of the investigated three-liquid-phase systems, the
system containing EOPO2500 (NH4 )2 SO4 and TBP could offer
a suitable strategy to separate and purify two compounds. GA
and IQ could be unevenly partitioned into the middle phase and
the top phase at the pH range of 6.1–8.8, respectively. Then,
they could be further purified by extracting into the organic
phase at pH below 3.3 or by stripping into the aqueous phase
at pH above 11.0, respectively. Hydrogen bond interaction and
hydrophobic interaction were the main driving forces resulting
in differential partitioning by FTIR and fluorescence spectro-
scopic investigations. Three-liquid-phase extraction would be a
promising technology to separate different compounds or groups
in the complex licorice systems.
Acknowledgments
This work was financially supported by Innovative Research
Group Science Fund (No. 20221603), National Natural Science
Fund (No. 20406021) and National Natural Science Key Fund
(No. 20236050).
References
[1] J.F. Hu, F.J. Shen, A survey of the studies on chemical constituents of
Glycyrrhiza, Nat. Prod. Res. Dev. 8 (1996) 77–91.
[2] F. Rauchensteiner, Y. Matsumura, Y. Yamamoto, S. Yamaji, T. Tani, Anal-
ysis and comparison of Radix Glycyrrhizae (licorice) from Europe and
China by capillary-zone electrophoresis (CZE), J. Pharm. Biomed. Anal.
38 (2005) 594–600.
[3] Y. Zhou, M.K. Wang, X. Liao, X.M. Zhu, S.L. Peng, L.S. Ding, Rapid iden-
tification of compounds in Glycyrrhiza Uralensis by liquid chromatogra-
phy/Tandem mass spectrometry, Chin. J. Anal. Chem. 32 (2004) 174–178.
[4] G.G. Niu, Y.C. Xie, J.F. Lou, H.Z. Liu, Isolation and purification of
glycyrrhizic acid with solvent extraction, Sep. Purif. Technol. 44 (2005)
189–196.
223
[5] Q.L. Li, M.H. Zhou, J.N. Chen, Application of supercritical fluid extraction
in the analysis of Glycyrrhizic acid, J. Instrum. Anal. 17 (1) (1998) 37–40.
[6] D.H. Fu, A.L. Liu, K.N. Deng, T.J. Zhou, J.D. Xiao, Study on the adsorption
behavior and application of DM-130 resin to purifying glycyrrhizic acid,
Nat. Prod. Res. Dev. 14 (1) (2001) 60–64.
[7] H.B.S. Thomas, W.Z. Howaed, D.B. Alexis, Separation of major compo-
nents in licorice using high-performance liquid chromatography, J. Chro-
matogr. 175 (1979) 350–355.
[8] Y. Jiang, H.T. Lu, F. Chen, Preparative purification of glycyrrhizin extracted
from the root of liquorice using high-speed counter-current chromatogra-
phy, J. Chromatogr. 1033 (1) (2004) 183–186.
[9] Q.E. Wang, F.S.C. Lee, X. Wang, Isolation and purification of inflacoumarin
A and licochalcone A from licorice by high-speed counter-current chro-
matography, J. Chromatogr. A 1048 (1) (2004) 51–57.
[10] S.H. Yu, S.J. Wu, F.X. Jin, Y. Guo, Isolation and purification of Glycyrrhizin
from Licorice with Resin Column, Food Ferment. Ind. 25 (1) (1998) 40–
43.
[11] F.S. Feng, C.S. Xuan, S.W. Chen, Y. Chang, B.Z. Qiao, Application of
an adsorbent AB-8 in extraction and separation of glycyrrhizic acid, Nat.
Prod. Res. Dev. 10 (1) (1997) 60–64.
[12] Q. Liu, L. Yang, X. Wu, J. Chen, Study on purification of glycyrrhizic acid
with macroporous adsorption resin, Zhong Yao Cai. 26 (5) (2003) 357–
359.
[13] B. Fu, J. Liu, H. Li, L. Li, F.S. Lee, X. Wang, The application of macrop-
orous resins in the separation of licorice flavonoids and glycyrrhizic acid,
J. Chromatogr. A 1089 (1/2) (2005) 18–24.
[14] Q.E. Wang, S.M. Ma, B.Q. Fu, F.S. Lee, X.Y. Wang, Development of multi-
stage countercurrent extraction technology for the extraction of glycyrrhizic
acid (GA) from licorice (Glycyrrhiza uralensis Fisch), Biochem. Eng. J.
21 (3) (2004) 285.
[15] J.G. Ma, Z.L. Xiu, D.J. Zhang, L.Y. Jia, Concentration and separation of
glycyrrhizic acid by foam separation, J. Chem. Technol. Biotechnol. 77
(2002) 720–724.
[16] T.W. Tan, Q. Huo, Q. Ling, Purification of glycyrrhizin from Gly-
cyrrhiza uralensis Fisch with ethanol/phosphate aqueous two phase system,
Biotechnol. Lett. 24 (2002) 1417–1420.
[17] M. Mojski, I. Gluch, Characteristics and applications of three-phase extrac-
tion systems, J. Anal. Chem. 51 (4) (1996) 329–342.
[18] S.F. Shen, Z.D. Chang, X.H. Sun, X. Hu, H.Z. Liu, Application and recy-
cling of tri-block copolymer in three-phase extraction system, in: Proceed-
ings of the International Conference of Solvent Extraction (ISEC2005),
Beijing, 2005, pp. 1360–1365.
[19] S.F. Shen, Z.D. Chang, H.Z. Liu, Three-liquid-phase extraction systems
for separation of phenol and p-nitrophenol from wastewater, Sep. Purif.
Technol. 49 (3) (2006) 217–222.
[20] S.F. Shen, Z.D. Chang, X.H. Sun, H.Z. Liu, Process integration for pro-
duction of 6-aminopenicillanic acid from penicillin G cultivation broth,
Process Biochem. 41 (3) (2006) 571–574.
[21] W.B. Cai, S.L. Zhu, X.L. Piao, Extraction equilibria of formic and acetic
acids from aqueous solution by phosphate-containing extractants, J. Chem.
Eng. Data 46 (2001) 1472–1475.
[22] J.Z. Jiang, W.H. Li, H.C. Gao, J.G. Wu, Extraction of inorganic acids with
neutral phosphorus extractants based on a reverse micelle/microemulsion
mechanism, J. Colloid Interface Sci. 268 (2003) 208–214.
[23] T. Nivaggioli, P. Alexandridis, T.A. Hatton, Fluorescence probe studies of
pluronic copolymer solutions as a function of temperature, Langmuir 11
(1995) 730–737.