Anal Bioanal Chem
DOI 10.1007/s00216-013-7418-8
RESEARCH PAPER
Stability and efficiency of supported liquid membranes
in electromembrane extraction—a link to solvent properties
Knut Fredrik Seip & Moheba Faizi & Cristina Vergel &
Astrid Gjelstad & Stig Pedersen-Bjergaard
Received: 26 June 2013 / Revised: 25 September 2013 / Accepted: 4 October 2013
# Springer-Verlag Berlin Heidelberg 2013
Abstract The current work presents a large systematic
screening of 61 possible organic solvents used as supported
liquid membranes (SLM) in electromembrane extraction
(EME). For each organic solvent, recovery, current across
the SLM, and stability considerations have been investigated
and correlated to relevant solvent properties through partial
least square regression analysis. The five unpolar basic drugs
pethidine, haloperidol, methadone, nortriptyline, and
loperamide were used as model analytes. Efficient EME sol-
vents were found to have a low water solubility (<0.5 g L−1)
and belonged to cluster 2 of a Kamlet–and–Taft-based solvent
classification system (high dipole moments and proton accep-
tor properties). These parameters were especially found in
nitroaromatic compounds and ketones. Small molecules with
low log P value and high water solubility were unsuitable, as
they tended to give unstable extractions, caused by a high
current across the SLM. This was often combined with sub-
stantial solvent-related interferences and the generation of an
electroosmotic flow across the SLM, with resulting acceptor
solution expansion. Large molecules with a high log P value
were classified as inefficient. For these solvents, no current
was measured across the SLM and no analytes were extracted.
Published in the topical collection Microextraction Techniques with guest
editors Miguel Valcárcel Cases, Soledad Cárdenas Aranzana and Rafael
Lucena Rodríguez.
Electronic supplementary material The online version of this article
(doi:10.1007/s00216-013-7418-8) contains supplementary material,
which is available to authorized users.
K. F. Seip : M. Faizi : C. Vergel : A. Gjelstad :
S. Pedersen-Bjergaard (*)
School of Pharmacy, University of Oslo, P.O. Box 1068,
Blindern, 0316 Oslo, Norway
e-mail: stig.pedersen-bjergaard@farmasi.uio.no
S. Pedersen-Bjergaard
Department of Pharmacy, Faculty of Health and Medical Sciences,
University of Copenhagen, 2100 Copenhagen, Denmark
This is the first time systematic knowledge on the SLM in
EME has been gathered and investigated, and the presented
results could be highly beneficial for future development and
optimization of EME.
Keywords Electromembrane extraction . Supported liquid
membrane . Basic drugs . Sample preparation .
Organic solvents
Introduction
The use of supported liquid membranes (SLM) in sample
preparation has rapidly increased in popularity during the last
decade. The combination especially of microextraction tech-
niques and membrane-based techniques has inspired modern
forms of sample preparation that give a high degree of sample
cleanup with minimal amounts of organic solvents used [1, 2].
Other benefits include low cost, possibilities to work with
limited sample volumes, and ease of automation.
An important principle for membrane-based microextraction
techniques is liquid–liquid extraction systems based on hollow
fiber membranes. This was introduced in 1999, under the name
hollow fiber liquid phase microextraction (HF-LPME) [3]. The
principle is based on the extraction of analytes from an aqueous
sample solution, through the wall of a hollow fiber impregnated
with an organic liquid and into an organic (two-phase system) or
aqueous (three-phase system) acceptor solution. Because of its
efficient sample cleanup and possible high enrichment from
limited sample volumes, as well as being reasonably priced,
and environmentally friendly, HF-LPME has been utilized in
several applications during the later years [2, 4–6]. However, the
technique suffers from relatively long extraction times
(15–60 min) and limitations in recovery [5, 7].
To improve the technique in terms of time consumption
and recovery, electromembrane extraction (EME) was

introduced in 2006 [8]. This principle differs from HF-LPME
by the addition of an electric field across the membrane,
adding electrokinetic migration as a method for mass transfer.
Short extraction times and good recoveries for several drug
substances have been achieved by using EME [9–11], and the
technique has also been able to perform exhaustive extractions
for some analytes [12]. EME has been carried out from
various matrices ranging from highly organic aqueous sam-
ples and wastewater, to various body fluids like undiluted
whole blood or breast milk, where it has been used to extract
drug substances, heavy metals, and peptides [2, 10, 13–23].
One of the main factors controlling the extraction process
in EME is the SLM. Recently, a mathematical model for the
extraction process was presented and showed that the mass
transfer has both a distributive component, governed by an
electrically enhanced partition coefficient, as well as an elec-
trophoretic component, governed by the electric field [24]. In
both components, the type of organic solvent in the SLM
plays an important role by affecting the distribution constant
and by controlling the electric field across the membrane. A
few published articles have given insight into some useful
organic solvents to use as an SLM for various analytes
[25–31]; however, whereas other important extraction param-
eters have been thoroughly tested [8, 29, 32], systematic
knowledge in this area is currently not available.
To increase this knowledge, this paper presents a large and
systematic screening of 61 organic solvents and their use as
SLM in EME for basic drug substances of different polarities.
Through partial least square regression (PLS) and thorough
observation, relevant solvent properties of the organic sol-
vents have been linked to extraction recovery, current across
the SLM (SLM current), and EME stability. The resulting
systematic knowledge of the SLM could be highly beneficial
for future development and optimization of EME.
Experimental
Chemicals
Model analytes
Haloperidol hydrochloride, loperamide hydrochloride, meth-
adone hydrochloride, nortriptyline hydrochloride, and pethi-
dine hydrochloride were all obtained from Sigma-Aldrich
(Steinheim, Switzerland).
SLM solvents
Acetophenone (≥98 %), 4-allyl-2-methoxyphenol (eugenol,
≥98 %), benzaldehyde (≥99 %), cinnamonitrile (97 %),
2-decanone (98 %), dibenzylamine (97 %), dibutyl phosphate
(≥97 %), dibutyl phtalate (99 %), 2,6-dimethyl-4-heptanone
Seip et al.
(80 %), (2,2-dimethyl-1-propyl)benzene (≥97 %), dodecyl ace-
tate (≥97 %), di-sec-butylphenol (85 %), 2-ethylhexyl-
4-(dimethylamine)benzoate (98 %), 1-ethyl-2-nitrobenzene
(≥99 %), 2-ethylphenol (99 %), heptaldehyde (95 %), isophorone
(≥97 %), 5-isopropyl-2-methylphenol (98 %), 2-isopropyl phe-
nol (98 %), linoleic acid (≥99 %), 2-methoxy-4-methylphenol
(≥98 %), 3′-methylacetophenone (98 %), 2-nitroacetophenone
(95 %), nitrocyclohexane (97 %), 2-nitrophenylethyl alcohol
(97 %), 2-nitrophenyl octyl ether (≥99 %), 2-nitrotoluene
(≥99 %), nonanal (95 %), nonanoic acid (≥97 %), 1-nonanol
(98 %), 2-nonanone (≥99 %), n-octanoic acid (caprylic
acid, >99.5 %), 1-octanol (99 %), 2-octyl-1-dodecanol (97 %),
pentylbenzene (99 %), 2-phenylethanol (≥99 %), (+)-santolina
alcohol (≥98.5 %), α-tetralone (97 %), 3-tolunitrile (99 %),
tributyl phosphate (≥98 %), 1,2,4-trifluoro-5-nitrobenzene
(99 %), triisopropyl phosphate (95 %), 3,3,5-trime-
thylcyclohexanone (98 %), tris(2-butoxyethyl)phosphate
(94 %), tri-tert-butylborate (98 %), undecanal (97 %), 1-
undecanol (99 %), and 6-undecanone (97 %) were purchased
from Sigma-Aldrich (St. Louis, MO, USA). 2,4-Dimethyl-1-
nitrobenzene (≥99 %), ethoxybenzene (phenetole, ≥98 %), m-
cresol (≥99 %), nitrobenzene (99.5 %), 2-nitrophenyl pentyl ether
(≥99 %), 2-octanone (≥97 %), triethyl phosphate (≥98 %),
trihexylamine (≥99 %), tris(2,4-dimethylphenyl)phosphate
(trixylyl phosphate, ≥98 %), and tris(2-ethylhexyl)phosphate
(≥99 %) were obtained from Fluka (Buchs, Switzerland). 1-
Isopropyl-4-nitrobenzene was purchased from Tokyo Chemical
Industry (Zwijndrecht, Belgium). Benzyl methyl ketone (pure)
was obtained from Koch-Light laboratories (Haverhill, UK).
Benzyl chloride (>99 %) and octylamine (98 %) were purchased
from Merck (Darmstadt, Germany).
Other chemicals
Formic acid (>98 %) was obtained from Sigma-Aldrich
(Steinheim, Switzerland). Acetonitrile (HPLC-grade) and so-
dium hydroxide (NaOH) were purchased from VWR
International (Leuven, Belgium). Concentrated hydrochloric
acid (HCl, 37 %), Ortho-phosphoric acid (85 %), and sodium
dihydrogen phosphate monohydrate (≥99 %) were obtained
from Merck (Darmstadt, Germany). Deionized water for all
experiments was supplied from a milli-Q integral 3 water
purification system (Millipore, Billerica, MA, USA).
Samples
Sample solutions were prepared daily by diluting stock solu-
tions containing 1 mg mL−1 of each model drug in methanol,
stored at 4 °C, and protected from light. Dilutions were
accomplished with 10 mM HCl, to achieve the desired con-
centration of 1 μg mL−1 before extraction. The model analytes
and their chemical structure, log P, and pKa values are pre-
sented in Table 1.
Stability and efficiency of supported liquid membranes in EME

Table 1 Name, chemical structure, log P, and pKa values for the five model analytes

a
Log P and pKa values were obtained from SciFinder web resource (American Chemical Society)

Procedure for electromembrane extraction
Equipment and setup
The setup for EME has been depicted previously [23] and
consisted of a device made up from a sample vial containing
the sample solution, a porous hollow fiber with an organic
liquid immobilized in the pores of the walls, and an accep-
tor solution located in the lumen of the hollow fiber. A glass
vial of the type 2-SV with screw cap, with a volume of 2 mL
(Chromacol, Welwyn Garden City, UK) served as the sam-
ple compartment during the extractions. The screw cap was
perforated and penetrated by a porous hollow fiber of the
type PP Q3/2 polypropylene hollow fiber, with a wall

thickness of 200 μm, internal diameter of 1.2 mm, and a
pore size of 0.2 μm (Membrana, Wuppertal, Germany). A
2.4 cm piece of this hollow fiber was mechanically sealed in
the lower end, whereas the upper end was attached to the
end of a 2.1 cm end piece of a pipette tip (Finntip 200 Ext,
Thermo Electron, Vantaa, Finland) by heat. Before extrac-
tion, the fiber was impregnated with an organic solvent,
comprising the SLM in the extraction setup. The lumen of
the hollow fiber served as the compartment for the acceptor
solution, making a three-phase system when placed into the
sample solution. Platinum wires with a diameter of 0.5 mm
(K. A. Rasmussen, Hamar, Norway) were used as elec-
trodes and placed in the acceptor solution (in the lumen of
the hollow fiber), and in the sample solution, through the lid
of the sample compartment. The electrodes were connected
to a power supply of the model ES 0300–0.45 from Delta
Power Supplies (Delta Electronika, Zierikzee, The
Netherlands), with programmable voltage in the range 0–
300 V and current output from 0–450 mA. An Eppendorff
thermomixer comfort (Eppendorff, Hamburg, Germany)
was used to agitate the system during the extractions. The
SLM current during the extractions was measured using a
custom-built device for measuring micro-currents. This device
was connected to a computer, running LabVIEW 8.2 software
(National Instruments, Austin, TX, USA), resulting in a plot of
SLM current over time for each extraction.
Extraction procedure
All extractions were performed by adding a desired amount of
a standard drug mix, containing the chosen model analytes, to
10 mM HCl. A final concentration of 1 μg mL−1 for each
model analyte and a total sample volume of 1.0 mL were used.
The hollow fiber was immersed in an organic solvent for
approximately 5 s, resulting in the SLM. Any excess solvent
was wiped off with a medical wipe. By the use of a micro-
syringe, the lumen of the hollow fiber was filled with an
acceptor solution, comprised of 25 μL 10 mM HCl. A cathode
and anode was then placed in the acceptor and sample
solution, respectively, and the hollow fiber, containing the
SLM and the acceptor solution, was inserted, through the
screw cap of the sample vial, and into the sample solution.
The extraction device was then placed on an agitator, which
ensured thorough convection of the sample at 900 rpm
during the entire extraction. An electrical potential that
varied according to the organic solvent used in the SLM,
was applied, and the extractions lasted for 5 min.
Immediately after extraction, the hollow fiber containing
the acceptor solution was removed from the sample com-
partment, and the acceptor solution was transferred to sam-
ple vials for further analysis on HPLC.
Seip et al.
Instrumentation
HPLC
All extracts and samples containing loperamide, haloperidol,
methadone, pethidine, and nortriptylin as model analytes were
analyzed on a Dionex Ultimate 3000 HPLC-system (Dionex
Corporation, Sunnyvale, CA, USA), comprised of a degasser
(SRD-3200), pump (HPG-3200 M), autosampler (WPS-
3000SL), column oven (FLM-3100), and a UV detector
(VWD-3400), running Chromeleon (v. 6.80 SP2 Build 2212)
software (Dionex Corporation). A Gemini C18 (150 mm L×
2.00 mm i.d., 5 μm particle size, and 110 Å pore size) column
from Phenomenex (Torrance, CA, USA) was used for separa-
tion. A gradient elution was used with mobile phases
consisting of 20 mM formic acid and acetonitrile in a
95:5v/v ratio for mobile phase A and a 5:95v/v ratio for
mobile phase B. The gradient increased linearly from 0 to
40 % mobile phase B over 15 min, followed by a plateau at
40 % mobile phase B for 2 min. After this, a washing
procedure was performed with 80 % mobile phase B for
2 min, before re-equilibration with mobile phase A for
6 min. The system was operated at 60 °C column temper-
ature, with a flow rate of 0.4 mL min−1, injection volume of
10 μL, and detection at 214 and 235 nm, producing good
signals for all the analytes tested, with minimal background
noise. Total analysis time was 25 min per sample.
PLS analysis
All PLS analyses were performed using The Unscrambler
v.9.8 software (Camo Software Inc., Woodbridge, NJ, USA).
All physical chemical properties used for the PLS calculations
were obtained from the web resources chemicalize.org
(ChemAxon Ltd., Budapest, Hungary) and chemspider.com
(Royal Society of Chemistry, London, UK). All data were
centered, normalized using a 1/SD weighting, and validated at
95 % confidence interval, using full cross-validation. The
optimal number of PLS components were used in each case,
and clear outliers and insignificant contributions to the PLS
model were removed, based on calculations by the software.
This produced an optimized PLS-model in each case, with
minimal contribution from random noise.
Calculation of recovery
The extraction recovery after EME was calculated for each
analyte using the following equation:
nA;final C A;final *V A
R¼ *100% ¼ *100% ð1Þ
nD;initial C D;initial *V D
Stability and efficiency of supported liquid membranes in EME
where n A,final is the amount of analyte present in the acceptor
solution at the end of the extraction, and n D,initial is the amount
of analyte initially present in the sample. V A and V D are the
volumes of the acceptor and sample solution, respectively.
C A,final is the concentration of analyte in the acceptor solution
after extraction, while C D,final is the initial concentration in the
sample.
Results and discussion
Selection of organic solvents for SLMs
In this work, 61 commercially available organic solvents were
evaluated as potential liquid membranes (SLM) for
electromembrane extraction of non-polar basic drugs. The or-
ganic solvents were selected to represent a broad range of
chemical functionalities, and four criteria were important for
the selection. First, the organic solvent should be liquid at room
temperature in order to serve as an SLM. Secondly, the boiling
point should be high enough to avoid evaporation of organic
solvent during extraction. In the present work, the lower limit
for the boiling point was set at 150 °C at 760 mmHg. Thirdly,
the solvent should be reasonably priced, and finally, the solvent
should be water-immiscible, in order to avoid dissolution into
the aqueous donor and acceptor solutions.
Table 2 lists the 61 different organic solvents that were
selected based on the above criteria. The solvents were
subjected to a large screening to examine their suitability for
EME. EME was accomplished with the non-polar basic drugs
pethidine, nortriptyline, methadone, haloperidol, and
loperamide as model analytes. For each solvent, recoveries
were measured by HPLC; the SLM current over
time was measured according to section Procedure for
electromembrane extraction, and the volume of acceptor so-
lution was measured after EME with a micro-syringe. All the
solvents were classified into either efficient EME solvents
(category 1), inefficient EME solvents (category 2), or sol-
vents that were unsuitable for EME (category 3). The catego-
ry 1 solvents all provided a stable EME system during 5 min
of operation and provided extraction recoveries above 5 % for
the model analytes. The category 2 solvents also provided a
stable EME system, but the system was inefficient and pro-
vided very low recoveries for the model analytes. The solvents
that were considered unsuitable (category 3) all provided an
unstable EME system and were further divided into three sub-
categories (3a, 3b, and 3c). The first sub-category consisted of
solvents with current related problems (category 3a). These
solvents typically resulted in high SLM current. This in turn
resulted in unstable extractions and variable recoveries. In
addition, expansion of the acceptor solution volume was often
observed with these solvents. Thus, while 25 μL acceptor
solution was injected into the hollow fiber prior to EME, an
acceptor solution volume exceeding 25 μL was collected after
the end of the experiment. The second category (category 3b)
comprised solvents where substantial solvent-related interfer-
ences were detected in the chromatograms. Thus, even with
EME from pure 10 mM HCl, a substantial number of back-
ground peaks were observed with the category 3b solvents. The
last category (category 3c) described solvents which were not
impregnated in the wall of the hollow fiber. Thus, although the
dry hollow fiber was dipped for more than 30 s in the solvent,
the category 3c solvents were unable to penetrate the pores of
the hollow fiber. The solvents and their classification will be
further discussed below.
Unsuitable solvents with current related problems
(category 3a)
During EME, a constant electrical potential was sustained over
the SLM, and the associated SLM current was measured con-
tinuously. In all cases of successful EME, with the 16 different
category 1 solvents (Table 2), SLM currents between 3 and
50 μAwere measured with voltages in the range 15–300 V. The
low SLM current with the category 1 solvents was principally
due to charge transfer associated with the mass transport of the
charged analytes, but also to some extent due to mass transfer
of background ions across the SLM [32]. The category 1
solvents are discussed in further detail in section “Efficient
and inefficient EME solvents (categories 1 and 2)”.
For 27 different solvents (category 3a, Table 2), the SLM
current was high and exceeded 50–100 μA when extractions
were performed with voltages above 3 V. Such high and
unstable SLM current tends to give higher standard deviation,
bubble formation due to electrolysis, and less control of the
extraction process [8]. Based on the PLS analysis (see
Electronic supplementary material, Fig. S2), the category 3a
solvents appeared to be small molecules with some water
solubility (>0.5 g L−1) and low log P. A few amines deviated
apparently from this pattern. This was probably due to the fact
that their calculated water solubility was incorrect in relation
to the acidic environment on both sides of the SLM. The
category 3a solvents were prone to water penetration during
extraction. Significant amounts of water were dissolved in the
SLM during the extraction. Thus, the membrane conductivity
and the transfer of background ions increased, which resulted
in high SLM current. This was a time-dependent process as
illustrated in Fig. 1 where the SLM current gradually in-
creased with time for 2-isopropyl phenol as an example.
With voltages at 3 V and below, it was possible to do EME
at acceptable levels of SLM current with the category 3a
solvents. However, for most of these solvents, an increase in
acceptor solution volume was observed during extraction. In
several cases, this increase was substantial. A solvent like 2-
 Seip et al.

Table 2 Classification of all potential solvents for SLMs used in the screening process, according to their main functional group, and into efficient,
inefficient, or unsuited solvents

Solvent name Category 1 Category 2 Category 3

Efficient solvents Inefficient Unsuitable solvents

solvents
Category 1a, Category 1b, <5 % recovery Category 3a, Category 3b, Category 3c,
>25 % 5–25 % current-related substantial solvent-related insufficient
recovery recovery problems interferences impregnation

Alcohols
1-Nonanol X
1-Octanol X
2-Octyl-1-dodecanol X
2-Phenylethanol X
Santolina alcohol X X
1-Undecanol X
Aldehydes
Benzaldehyde X X
Heptaldehyde X X
Nonanal X
Undecanal X
Amines
Dibenzylamine X X
2-Ethylhexyl-4-(dimethylamino)benzoate X X
Octylamine X X
Trihexylamine X
Aromatic hydrocarbons
(2,2-Dimethyl-1-propyl) benzene X
Pentylbenzene X
Carboxylic acids
Linoleic acid X
Caprylic acid X
Nonanoic acid X
Esters
Dibutyl phtalate X
Dodecyl acetate X
Ethers
Phenetole X
Ketones
Acetophenone X
Benzyl methyl ketone X X
2-Decanone X
2,6-Dimethyl-4-heptanone X
Isophorone X
3′-Methylacetophenone X
2-Nonanone X
2-Octanone X
α-Tetralone X X
3,3,5-Trimethylcyclohexanone X
6-Undecanone X
Nitriles
Cinnamonitrile X X X
3-Tolunitrile X
Nitroaromatics
2,4-Dimethyl-1-nitrobenzene X
1-Ethyl-2-nitrobenzene X
Stability and efficiency of supported liquid membranes in EME

Table 2 (continued)
Solvent name Category 1 Category 2 Category 3

Efficient solvents Inefficient Unsuitable solvents

solvents
Category 1a, Category 1b, <5 % recovery Category 3a, Category 3b, Category 3c,
>25 % 5–25 % current-related substantial solvent-related insufficient
recovery recovery problems interferences impregnation

1-Isopropyl-4-nitrobenzene X
2-Nitroacetophenone X X
Nitrobenzene X X X
2-Nitrophenylethyl alcohol X
2-Nitrophenyl octyl ether X
2-Nitrophenyl pentyl ether X
2-Nitrotoluene X
1,2,4-Trifluoronitrobenzene X
Phenols
Di-sec-butylphenol X
2-Ethylphenol X X
Eugenol X
5-Isopropyl-2-methylphenol X
2-Isopropyl phenol X X
m-Cresol X X
2-Methoxy-4-methylphenol X X
Phosphates
Dibutyl phosphate X X
Tributyl phosphate X
Triisopropyl phosphate X X
Tris(2-butoxyethyl) phosphate X X
Tris(2-ethylhexyl) phosphate X
Trixylyl phosphate X
Others
Benzyl chloride X X
Nitrocyclohexane X
Tri-tert-butylborate X

Solvents with more than one functional group have been sorted according to their most influential group for EME. The recovery values are based on the
average value for the five model analytes

isopropylphenol showed a volume increase of 4 μL (from 25 was transferred across the SLM during extraction. This was
to 29 μL) after 5 min of EME, corresponding to a 16 % justified by the observation that the acceptor solution always
volume expansion. This expansion was due to water, which appeared as a single phase after EME, and the magnitude of

Fig. 1 The SLM current over

time during a 5-min EME
extraction at 5 V, using
2-isopropyl phenol as organic
solvent
 Seip et al.

Fig. 2 Chromatograms showing substantial solvent-related interfer- as SLM. c Resulting chromatogram after LLE between triisopropyl
ences. a Resulting chromatogram after EME with 6-undecanone as phosphate and 10 mM HCl
SLM. b Resulting chromatogram after EME using triisopropyl phosphate

the expansion could not be explained by the water solubility of Unsuitable solvents with insufficient impregnation
the organic solvents. The transfer was also voltage-dependent in the hollow fiber (category 3c)
and was insignificant at 0 V, which pointed towards a small
electroosmotic flow of water across the SLM. This observa- Category 3c (Table 2) represents the solvents that were unable
tion is new and shows that the category 3a solvents should be to penetrate into the pores of the hollow fiber. Allocation into
avoided in future EME applications. this category was based on visual inspection, where the or-
ganic solvent was seen as droplets or a thin film on the surface
of the hollow fiber, after being immersed in the organic
Unsuitable solvents causing substantial solvent-related solvent. The PLS analysis revealed that this behavior was seen
interferences (category 3b) primarily with small molecules with a high polar surface area
(>45). The reason for this is most probably that highly polar
Some solvents (category 3b, Table 2) were found to be unsuitable molecules will have difficulties in wetting the polypropylene
for EME because several solvent related interferences were hollow fiber, since the surface tension will be too high for the
observed by the HPLC-UV method used. This was the case even solvent to penetrate the pores.
with new solvents especially acquired for this project. The influ-
ence from these interferences might, however, be negligible by
using other detector principles, such as mass spectrometry. Two Efficient and inefficient EME solvents (categories 1 and 2)
examples are illustrated in Fig. 2, where 6-undecanone (category
1 solvent) and triisopropyl phosphate (category 3a and 3b sol- Among the 61 investigated solvents, 36 solvents were classified
vent) were used as SLM. The lower trace (a) illustrates the pure as unsuited (category 3a, 3b, and 3c, Table 2) as discussed in
background chromatogram, with no solvent related interferences sections “Selection of organic solvents for SLMs,” “Unsuitable
from EME with 6-undecanone. The middle trace (b) illustrates
the resulting chromatogram from EME with triisopropyl phos-
phate, where substantial solvent related interferences were de- Table 3 Four of the efficient category 1a solvents with their correspond-
tected. The interferences were clearly related to the organic ing recovery, extraction voltage, and average SLM current during the
extraction
solvent. This is exemplified in Fig. 2c, where a liquid–liquid
extraction between triisopropyl phosphate and 10 mM HCl SLM Voltage Average SLM Average
provided the same pattern of interferences, as compared with current recovery
EME with triisopropyl phosphate as SLM (b). The category 3b
NPOE 250 V 8.6 μA 67 %
solvents were highly correlated with the category 3a solvents,
2-nonanone 40 V 5,6 μA 61 %
and functional groups especially, like phenols, phosphates, and
6-undecanone 200 V 3,1 μA 51 %
aldehydes showed substantial solvent-related interferences. The
2,4-dimethyl-1-nitrobenzene 20 V 7,4 μA 71 %
category 3b solvents were all of relatively hydrophilic character
(low log P), and consequently impurities present in the organic The recovery was measured as the average recovery between the five
solvent were also slightly water-soluble and leaked into the model analytes (n =3)
acceptor solution (and sample solution). NPOE 2-nitorophenyl octyl ether
Stability and efficiency of supported liquid membranes in EME
solvents with current related problems (category 3a),”
“Unsuitable solvents causing substantial solvent-related interfer-
ences (category 3b),” and “Unsuitable solvents with insufficient
impregnation in the hollow fiber (category 3c).” The remaining
25 solvents all provided a stable EME system but differed
significantly in terms of extraction efficiency. Eleven of these
solvents gave recoveries exceeding 25 % (category 1a, Table 2),
five with recoveries in the range 5–25 % (category 1b, Table 2),
and nine were inefficient solvents with recoveries less than 5 %
(category 2, Table 2).
As seen from Table 2, all the category 1a solvents were
nitroaromatic compounds or ketones (see also Electronic sup-
plementary material, Fig. S1). Among the successful
nitroaromatic compounds, 2-nitrophenyl octyl ether (NPOE),
1-isopropyl-4-nitrobenzene, 1-ethyl-2-nitrobenzene, and 2-
nitrophenyl pentyl ether have already been established as
good EME solvents in the literature [31]. On the other hand,
2-nitrotoluene and 2,4-dimethyl-1-nitrobenzene have been
used for the first time in the current work and further support-
ed the high efficiency of nitroaromatic solvents. In addition,
five ketones were found to be efficient, and except for 2-
octanone [8], the use of pure ketones in EME has not been
reported earlier. In a next step, the category 1a solvents were
investigated in relation to a solvent classification system based
on the Kamlet and Taft solvatochromic parameters [33]. Very
interestingly, both the nitroaromatics and the ketones
belonged to the same solvent cluster (cluster 2) in this classi-
fication system. Solvent cluster 2 is characterized by high
dipole moment, high proton acceptor properties, and low
proton donor properties. This suggested that dipole–dipole
interactions, combined with interactions between the proton
donor properties of the protonated basic drugs and the proton
acceptor properties of the organic solvent, occurred. Among
the solvents tested in Table 2, a few aldehydes and nitriles also
belonged to cluster 2 in the solvent classification system.
Several of these solvents actually provided good recoveries;
however, they were all classified into group 3b because of
substantial solvent-related interferences.
Among the five category 1b solvents, three were aliphatic
alcohols and thus belonged to cluster 5 of the solvent classifi-
cation system. Although these solvents all provided high dipole
moments and proton acceptor properties, their efficiency were
probably suppressed by their strong proton donor properties.
The remaining two category 1b solvents (dibutyl phthalate and
tributyl phosphate) had properties in the vicinity of cluster 2.
For the nine different category 2 solvents, no significant
recoveries were observed. With those solvents, no SLM cur-
rent (<1 μA) was observed even though the power supply was
operated at maximum at 300 V. For the category 2 solvents,
the EME system was inefficient, and neither charged analytes
nor background ions were transferred across the SLM. A PLS
analysis showed that these solvents appeared to be relatively
large molecules with low water solubility and high log P
(>4.0).
Comprehensive data for four of the category 1a solvents are
summarized in Table 3. As seen from the table, the new SLM
consisting of 2,4-dimethyl-1-nitrobenzene provided compara-
ble recovery data to NPOE. Also, the ketones in general
appeared to give slightly lower recoveries than the
nitroaromatics, as seen from the recoveries of 2-nonanone
and 6-undecanone. With all four solvents, replicated extrac-
tions (n =3) were acceptable and within 15 % RSD. As also
seen from the table, the more hydrophobic solvents were
operated at higher voltages without exceeding 10 μA as the
SLM current. This current was within the acceptable limits for
category 1 solvents as discussed in the section “Unsuitable
solvents with current-related problems (category 3a).” The
recovery values provided by the best nitroaromatics
approached the reported values in optimized SPE and LLE
methods of the same substances [34–38]. It should, however,
be noted that most EME parameters has not been thoroughly
optimized for each SLM in this work.
Conclusion
The present work has for the first time introduced a systematic
approach towards selection of solvents for the supported
liquid membrane in electromembrane extraction of unpolar
basic drugs. The best solvents were found to have a low water
solubility (<0.5 g L−1), high dipole moment, high proton
acceptor properties, and very low proton donor properties.
The solvents belonged to cluster 2 in a Kamlet–Taft-based
solvatochromic classification system, and nitroaromatics and
ketones were especially efficient. Solvents with a high molec-
ular surface area and high log P values were inefficient and
provided no extraction with voltages up to 300 V. Solvents
with a low molecular surface area, low log P, and water
solubility exceeding approximately 0.5 g L−1 tended to give
high SLM current, migration of water through the SLM, and
substantial solvent-related interferences and were thus unsuit-
able for EME.
References
1. Chimuka L, Michel M, Cukrowska E, Buszewski B (2011) Advances
in sample preparation using membrane-based liquid-phase
microextraction techniques. Trac-Trend Anal Chem 30(11):1781–
1792. doi:10.1016/j.trac.2011.05.008
2. Bello-Lopez MA, Ramos-Payan M, Ocana-Gonzalez JA, Fernandez-
Torres R, Callejon-Mochon M (2012) Analytical applications of
hollow fiber liquid phase microextraction (HF-LPME): a review.
Anal Lett 45(8):804–830. doi:10.1080/00032719.2012.655676

3. Pedersen-Bjergaard S, Rasmussen KE (1999) Liquid–liquid–liquid
microextraction for sample preparation of biological fluids prior to
capillary electrophoresis. Anal Chem 71(14):2650–2656
4. Sarafraz-Yazdi A, Amiri A (2010) Liquid-phase microextraction.
Trac-Trend Anal Chem 29(1):1–14. doi:10.1016/j.trac.2009.10.003
5. Rasmussen KE, Pedersen-Biergaard S (2004) Developments in hol-
low fibre-based, liquid-phase microextraction. Trac-Trend Anal
Chem 23(1):1–10. doi:10.1016/S0165-9936(04)00105-0
6. Ho TS, Pedersen-Bjergaard S, Rasmussen KE (2002) Recovery,
enrichment and selectivity in liquid-phase microextraction compari-
son with conventional liquid–liquid extraction. J Chromatogr A
963(1–2):3–17
7. Halvorsen TG, Pedersen-Bjergaard S, Rasmussen KE (2001) Reduc-
tion of extraction times in liquid-phase microextraction. J
Chromatogr B 760(2):219–226
8. Pedersen-Bjergaard S, Rasmussen KE (2006) Electrokinetic migra-
tion across artificial liquid membranes. New concept for rapid sample
preparation of biological fluids. J Chromatogr A 1109(2):183–190.
doi:10.1016/j.chroma.2006.01.025
9. Ramos-Payan M, Villar-Navarro M, Fernandez-Torres R, Callejon-
Mochon M, Bello-Lopez MA (2013) Electromembrane extraction
(EME)—an easy, novel and rapid extraction procedure for the HPLC
determination of fluoroquinolones in wastewater samples. Anal
Bioanal Chem 405(8):2575–2584. doi:10.1007/s00216-012-6664-5
10. Eibak LE, Gjelstad A, Rasmussen KE, Pedersen-Bjergaard S (2010)
Kinetic electro membrane extraction under stagnant conditions—fast
isolation of drugs from untreated human plasma. J Chromatogr A
1217(31):5050–5056. doi:10.1016/j.chroma.2010.06.018
11. Fotouhi L, Yamini Y, Molaei S, Seidi S (2011) Comparison of
conventional hollow fiber based liquid phase microextraction and
electromembrane extraction efficiencies for the extraction of ephed-
rine from biological fluids. J Chromatogr A 1218(48):8581–8586.
doi:10.1016/j.chroma.2011.09.078
12. Eibak LE, Gjelstad A, Rasmussen KE, Pedersen-Bjergaard S (2012)
Exhaustive electromembrane extraction of some basic drugs from
human plasma followed by liquid chromatography-mass spectrometry.
J Pharm Biomed Anal 57(1):33–38. doi:10.1016/j.jpba.2011.08.026
13. Balchen M, Reubsaet L, Pedersen-Bjergaard S (2008)
Electromembrane extraction of peptides. J Chromatogr A 1194(2):
143–149. doi:10.1016/j.chroma.2008.04.041
14. Davarani SS, Moazami HR, Keshtkar AR, Banitaba MH, Nojavan S
(2013) A selective electromembrane extraction of uranium (VI) prior
to its fluorometric determination in water. Anal Chim Acta 783:74–
79. doi:10.1016/j.aca.2013.04.045
15. Jamt RE, Gjelstad A, Eibak LE, Oiestad EL, Christophersen AS,
Rasmussen KE, Pedersen-Bjergaard S (2012) Electromembrane ex-
traction of stimulating drugs from undiluted whole blood. J
Chromatogr A 1232:27–36. doi:10.1016/j.chroma.2011.08.058
16. Strieglerova L, Kuban P, Bocek P (2011) Electromembrane extrac-
tion of amino acids from body fluids followed by capillary electro-
phoresis with capacitively coupled contactless conductivity detec-
tion. J Chromatogr A 1218(37):6248–6255. doi:10.1016/j.chroma.
2011.07.011
17. Rezazadeh M, Yamini Y, Seidi S (2011) Electromembrane extraction
of trace amounts of naltrexone and nalmefene from untreated biolog-
ical fluids. J Chromatogr B Anal Technol Biomed Life Sc 879(15–
16):1143–1148. doi:10.1016/j.jchromb.2011.03.043
18. Payan MR, Lopez MA, Torres RF, Navarro MV, Mochon MC (2011)
Electromembrane extraction (EME) and HPLC determination of
non-steroidal anti-inflammatory drugs (NSAIDs) in wastewater sam-
ples. Talanta 85(1):394–399. doi:10.1016/j.talanta.2011.03.076
19. Kuban P, Strieglerova L, Gebauer P, Bocek P (2011)
Electromembrane extraction of heavy metal cations followed by
capillary electrophoresis with capacitively coupled contactless con-
ductivity detection. Electrophoresis 32(9):1025–1032. doi:10.1002/
elps.201000462
Seip et al.
20. Seidi S, Yamini Y, Baheri T, Feizbakhsh R (2011) Electrokinetic
extraction on artificial liquid membranes of amphetamine-type stim-
ulants from urine samples followed by high performance liquid
chromatography analysis. J Chromatogr A 1218(26):3958–3965.
doi:10.1016/j.chroma.2011.05.002
21. Hasheminasab KS, Fakhari AR, Shahsavani A, Ahmar H (2013) A
new method for the enhancement of electromembrane extraction
efficiency using carbon nanotube reinforced hollow fiber for the
determination of acidic drugs in spiked plasma, urine, breast milk
and wastewater samples. J Chromatogr A 1285:1–6. doi:10.1016/j.
chroma.2013.01.115
22. Dominguez NC, Gjelstad A, Nadal AM, Jensen H, Petersen NJ,
Hansen SH, Rasmussen KE, Pedersen-Bjergaard S (2012) Selective
electromembrane extraction at low voltages based on analyte polarity
and charge. J Chromatogr A 1248:48–54. doi:10.1016/j.chroma.
2012.05.092
23. Gjelstad A, Rasmussen KE, Pedersen-Bjergaard S (2009)
Electromembrane extraction of basic drugs from untreated human
plasma and whole blood under physiological pH conditions. Anal
Bioanal Chem 393(3):921–928. doi:10.1007/s00216-008-2344-x
24. Seip KF, Jensen H, Sonsteby MH, Gjelstad A, Pedersen-Bjergaard S
(2013) Electromembrane extraction: distribution or electrophoresis?
Electrophoresis 34(5):792–799. doi:10.1002/elps.201200587
25. Gjelstad A, Rasmussen KE, Pedersen-Bjergaard S (2006) Electroki-
netic migration across artificial liquid membranes. Tuning the mem-
brane chemistry to different types of drug substances. J Chromatogr
A 1124(1–2):29–34. doi:10.1016/j.chroma.2006.04.039
26. Balchen M, Gjelstad A, Rasmussen KE, Pedersen-Bjergaard S
(2007) Electrokinetic migration of acidic drugs across a supported
liquid membrane. J Chromatogr A 1152(1–2):220–225. doi:10.1016/
j.chroma.2006.10.096
27. Seip KF, Stigsson J, Gjelstad A, Balchen M, Pedersen-Bjergaard S
(2011) Electromembrane extraction of peptides—fundamental stud-
ies on the supported liquid membrane. J Sep Sci 34(23):3410–3417.
doi:10.1002/jssc.201100558
28. Chimuka L, Mathiasson L, Jonsson J (2000) Role of octanol-water
partition coefficients in extraction of ionisable organic compounds in
a supported liquid membrane with a stagnant acceptor. Anal Chim
Acta 416(1):77–86. doi:10.1016/S0003-2670(00)00853-9
29. Middelthon-Bruer TM, Gjelstad A, Rasmussen KE, Pedersen-
Bjergaard S (2008) Parameters affecting electro membrane extraction
of basic drugs. J Sep Sci 31(4):753–759. doi:10.1002/jssc.
200700502
30. Kocherginsky NM, Yang Q, Seelam L (2007) Recent advances in
supported liquid membrane technology. Sep Purif Technol 53(2):
171–177. doi:10.1016/j.seppur.2006.06.022
31. Kjelsen IJ, Gjelstad A, Rasmussen KE, Pedersen-Bjergaard S (2008)
Low-voltage electromembrane extraction of basic drugs from bio-
logical samples. J Chromatogr A 1180(1–2):1–9. doi:10.1016/j.
chroma.2007.12.006
32. Gjelstad A, Rasmussen KE, Pedersen-Bjergaard S (2007) Simulation
of flux during electro-membrane extraction based on the Nernst-
Planck equation. J Chromatogr A 1174(1–2):104–111. doi:10.1016/
j.chroma.2007.08.057
33. de Juan A, Fonrodona G, Casassas E (1997) Solvent classification
based on solvatochromic parameters: a comparison with the Snyder
approach. Trac-Trend Anal Chem 16(1):52–62. doi:10.1016/S0165-
9936(96)00079-9
34. del Mar Ramirez Fernandez M, Wille SM, Samyn N (2012) Quanti-
tative method validation for the analysis of 27 antidepressants and
metabolites in plasma with ultraperformance liquid chromatography-
tandem mass spectrometry. Ther Drug Monit 34(1):11–24. doi:10.
1097/FTD.0b013e31823bf0fd
35. Favretto D, Stocchero G, Nalesso A, Vogliardi S, Boscolo-Berto R,
Montisci M, Ferrara SD (2013) Monitoring haloperidol exposure in
body fluids and hair of children by liquid chromatography-high-
Stability and efficiency of supported liquid membranes in EME
resolution mass spectrometry. Ther Drug Monit 35(4):493–501. doi:
10.1097/FTD.0b013e3182892d11
36. Lin HR, Chen CL, Huang CL, Chen ST, Lua AC (2013)
Simultaneous determination of opiates, methadone,
buprenorphine and metabolites in human urine by superficially
porous liquid chromatography tandem mass spectrometry. J
Chromatogr B Anal Technol Biomed Life Sc 925:10–15. doi:
10.1016/j.jchromb.2013.02.020
37. Wang X, Xiang Z, Cai X, Wu H, Wang X, Li J, Zhang M (2011)
Determination of pethidine in human plasma by LC-MS/MS.
Biomed Chromatogr 25(7):833–837. doi:10.1002/bmc.1524
38. Streel B, Ceccato A, Klinkenberg R, Hubert P (2005) Validation of a
liquid chromatographic-tandem mass spectrometric method for the
determination of loperamide in human plasma. J Chromatogr B Anal
Technol Biomed Life Sc 814(2):263–273. doi:10.1016/j.jchromb.
2004.10.050