Journal of Chromatography A, 1152 (2007) 199–214

Review

Automated, on-line membrane extraction

Kamilah Hylton, Somenath Mitra ∗Department of Chemistry and Environmental Science, New Jersey Institute of Technology,

University Heights, Newark, NJ 07104, USA

Available online 20 December 2006

Abstract

Over the last few years, membranes have been used to develop new approaches in analytical extraction, concentration and cleanup. An importantadvantage of membrane processes is that the sample and the extraction phase can be continuously brought into contact without physical mixing,and may be directly interfaced to an analytical instrument. This provides the basis for automated, real-time monitoring. Membrane extraction hasbeen applied to a wide range of organic and inorganic analytes, and has been directly interfaced with chromatography, spectroscopy and massspectrometry. Implementations of membrane extraction are diverse, encompassing different types of membranes, module designs and configurations.This review highlights some of these, and particularly the unique capabilities in automated, and on-line measurements.© 2006 Elsevier B.V. All rights reserved.

Keywords: Membrane extraction; Pervaporation; Automated sample preparation; LPME

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2002. Principles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200

2.1. Permeation through membrane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2002.2. Membrane classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201

2.2.1. Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2012.2.2. Morphology and structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2022.2.3. Geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2032.2.4. Membrane material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204

2.3. Membrane extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2052.3.1. Liquid–liquid membrane extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2052.3.2. Supported liquid membrane extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2052.3.3. Pervaporation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206

3. Automated, on-line analysis of volatile organic compounds (VOCs) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2064. Automated analysis of semi volatile organic compounds (SVOCs). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208

4.1. On-line LLME. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2084.2. On-line membrane concentration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2094.3. Development of total analytical system based on membrane separation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2104.4. Automated SLME. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2104.5. Automated hollow fiber based microextraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211

5. Automated metal analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2126. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213

∗ Corresponding author. Tel.: +1 973 5965611; fax: +1 973 5963586.E-mail address: Mitra@njit.edu (S. Mitra).

0021-9673/$ – see front matter © 2006 Elsevier B.V. All rights reserved.doi:10.1016/j.chroma.2006.12.047

2 mato

1

oatpapcaToap

atocttsbeto

hdttmcccnTmtbiogsss[a

mbbeias

fb

ewtttmc

2

2

gfwtpiit

l

J

wc

tIntegration of Fick’s law gives:

J = D(cis − cil)

L(2)

00 K. Hylton, S. Mitra / J. Chro

. Introduction

The analysis of trace level analytes involves the discreet stepsf extraction, concentration and clean up prior to detection byn instrument. Traditionally, extraction has been achieved byechniques such as head space analysis, purge and trap, solid-hase microextraction (SPME), liquid–liquid extraction (LLE)nd solid-phase extraction (SPE) [1–5]. Rotary evaporation, gasurging and the Kuderna Danish method are usually used tooncentrate the analytes, while the use of silica gel, gel perme-tion and FlorisilTM columns are used to achieve cleanup [2].hese methods are often time consuming, use large amountsf sample and solvents, require extensive sample handling, andnalysis cost per unit sample is high. Moreover, most of theserocesses cannot be carried out on-line in a continuous fashion.

Over the last few decades, membranes have found manypplications in various separation processes, such as, desalina-ion, dialysis, ultrafiltration, gas separation, dehumidification,smosis, reverse osmosis and electrodialysis [6]. Thus, it isonceivable that they can be used to achieve extraction, concen-ration and cleanup in an analytical application. Being selectiveo a particular species, the membrane primarily functions as aeparator of two bulk phases, and controls the mass transferetween them. This allows the enrichment of the species of inter-st and their removal from the sample matrix. The movement ofhe analytes of interest may be driven by a chemical, pressurer an electrical potential gradient [7,8].

In recent years, the use of membranes for sample preparationas in many instances become a preferred option. This is largelyue to the fact that they facilitate extraction without the mixing ofwo phases, thus eliminating problems such as emulsion forma-ion and high solvent usage. A very important advantage of the

embrane processes is that the sample and the extractant can beontinuously brought into contact, thus providing the basis for aontinuous, real-time process leading to automation and on-lineonnection to instruments [9]. Consequently membrane tech-iques have advanced to solve numerous analytical problems.hese techniques allow the simultaneous extraction and enrich-ent of analytes, and typically facilitate selective extraction at

race levels while consuming small amounts of solvents. Mem-rane extraction has been applied to a wide range of analytesncluding biological molecules [10–13], metals [14–19] andrganic pollutants [20–22], and has been directly interfaced withas chromatography (GC), liquid chromatography (LC), masspectrometry (MS), ion chromatography (IC), atomic absorptionpectroscopy (AAS), inductively coupled plasma atomic emis-ion spectroscopy (ICP-AES) and capillary electrophoresis (CE)23–31]. Moreover, there exist numerous other possibilities thatre yet to be explored.

Membrane extractions are not limited to environmentaledia, such as water. They are becoming quite popular in

iomedical applications with matrices, such as urine, blood andlood plasma, to analyze drugs and their metabolites. Jonsson

t al. [10] were able to detect sub-ppb concentrations of an estern urine flowing at rates as low as 50 �L/min. Such matricesre especially complex and usually require tedious and multipleample preparation steps. Additionally, sample volumes needed

Fp

gr. A 1152 (2007) 199–214

or membrane processes, particularly liquid membranes, tend toe quite small.

Applications of membrane extraction are quite diverse,ncompassing different types of membrane, their design asell as applications. As mentioned before, membranes offer

he unique opportunity for bringing a sample and an extrac-ant into contact in a continuous fashion. This review highlightshis unique capability and focuses on automated analysis using

embranes, with particular attention to on-line techniques forontinuous analysis.

. Principles

.1. Permeation through membrane

A membrane is a selective barrier through which differentases, vapors and liquids move at varying rates. The membraneacilitates the two phases coming into contact with each otherithout direct mixing. Molecules move through membranes by

he process of diffusion and are driven by a concentration (�C),ressure (�P) or electrical potential (�E) gradient. The processs demonstrated in Fig. 1. The interesting aspect of this techniques that both the donor and acceptor can flow continuously leadingo the development of real-time monitoring techniques.

This diffusion-based transport can be expressed by Fick’s firstaw of diffusion:

= −Ddc

dx(1)

here J is the rate of transfer (or flux) (g/cm2 s), D the diffusionoefficient (cm2/s) and dc/dx is the concentration gradient.

Diffusion is a slow process and so it is necessary to con-rol certain parameters to ensure that the flux is significant.

ig. 1. Permeation across a membrane. �C, �P, �E are the concentration,ressure and electrochemical gradients, respectively, down which analytes move.

mato

wct

irspcadd

D

wη

flb

K

wi

sdtpb

wiog

A

a

α

wa

tt[

2

om

2

l(pawflloaubamtpa

F[

K. Hylton, S. Mitra / J. Chro

here cis is the concentration of i at outer membrane surface,il the concentration of i in the lumen and L is the membranehickness.

It can be seen that reducing membrane thickness and creat-ng large chemical potential gradients across the membrane willesult in increased flux. It is important to note that the diffu-ion coefficient is a function of concentration. Thus, theoreticalredictions in analytical applications are a difficult task, whereoncentration varies by orders of magnitude. The flux of analytescross the membrane is also affected by temperature because theiffusion coefficient depends upon it. For liquids, this can be bestescribed by the Stokes–Einstein equation:

= kT

6πaη(3)

here k is the Boltzmann’s constant, a the radius of solute andis the solution viscosity.So at increased temperatures, D increases leading to a higher

ux. The other important factor, especially for non-porous mem-ranes is the partition coefficient, K, such that

= cm

cw(4)

here cm is the analyte concentration in the membrane and cws the analyte concentration in water (sample).

Permeation across non-porous dense membranes compriseseveral steps: (1) partitioning of the analyte in the membrane, (2)iffusion under a concentration gradient, and finally (3) parti-ioning into the extractant phase. There are also boundary layersresent on the membrane surfaces that provide an additionalarrier to mass transfer. These steps are illustrated in Fig. 2.

It is important to note that K decreases with temperature,hile D increases under the same conditions. Consequently

ncreasing temperature does not always increase flux, and an

ptimum temperature must be determined where the flux isreatest.

A key parameter that depends on both K and D is selectivity.fter all, the purpose of extraction is the selective transport of

ef

o

ig. 2. Concentration profile in an extraction process, where Cw, Cm and Cs refer to an59].

gr. A 1152 (2007) 199–214 201

nalytes of interest. The selectivity, α, is measured as:

= ai/aj

di/dj

(5)

here ai and aj are the concentration of i and j in the acceptornd di and dj are the concentration of i and j in the donor.

Therefore, the use of a particular “supported liquid” enhanceshe membrane selectivity by improving its ability to more readilyransport the analytes of interest compared to other components6].

.2. Membrane classification

Synthetic membranes may be classified based on a numberf properties including but not limited to geometry, function andorphology [32]. A basic classification is highlighted in Fig. 3.

.2.1. FunctionMembranes can be used in various separation functions as

isted in Fig. 3. The method selected depends on the samplee.g. aqueous, non-aqueous, biological, air, etc.,) and also theroperties of the analytes, such as, size and presence/absence ofcharge. Dialysis refers to the process in which urea and otheraste products are removed from blood by a saline solutionowing counter-current to the blood flowing in the membrane’s

umen. Cellulose membranes are most commonly used. Reversesmosis (RO), ultrafiltration (UF) and microfiltration (MF) arell pressure driven water separation processes. They dependpon the pore size of the membranes used. Pore size in MF mem-ranes ranges between 1000 A and 10 �m, and these membranesre used to filter particles, such as, bacteria, salts and sugars. UFembrane pores range between 20 and 1000 A and separate pro-

eins and some dyes. RO is primarily used for desalination withore size around 50 A. Separation here is based on the solubilitynd mobility of ions. Gas separation separates gas mixtures, for

xample, nitrogen removal from air, and the separation of CO2rom methane in the energy industry.

The membrane used in each function depends on its morphol-gy, media and structure. For instance, polypropylene composite

alyte concentration in water, membrane and the extractant phases, respectively

202 K. Hylton, S. Mitra / J. Chromatogr. A 1152 (2007) 199–214

ations

mfipa

2

bwwTlanstcttatt

a(rnamast

imbdmdfiTr

ti[ss

provide lower fluxes but higher selectivity [33]. In a solid mem-brane, the analyte must first partition and then diffuse undera concentration gradient. In a porous membrane, as shown inFig. 4, anything that can permeate through the pores migrates

Fig. 3. Classific

embrane would be useful for pervaporation but not for ultra-ltration, where a porous membrane would be necessary. Thisaper will discuss select automated analysis techniques and notll methods mentioned above will be covered here.

.2.2. Morphology and structureMembrane morphology refers to the quantity, size and distri-

ution of pores throughout the membrane structure. Membraneshich have no pores in their structure are known as non-porous,hile those which possess pores are referred to as porous.he size, shape and distribution of pores in a membrane are

argely dependent on the processes by which they are madend play a significant role in their mode of separation. As theame suggests, porous membranes have openings through whichelect molecules pass. Movement through these membranes isherefore by size exclusion and the membranes are used in appli-ations such as nanofiltration and dialysis. During extraction,wo liquid phases meet at the pores and during pervaporation,he analytes vaporize at these sites. Non-porous membranesre solid (pore-free) structures and the molecules must movehrough them via diffusion, and therefore the compatibility withhe analyte is critical.

Structure refers to the degree of uniformity of the poress well as the membrane material. For instance, homogenousisotropic) membranes are uniform throughout while asymmet-ic (anisotropic) and composite thin-films are not. It should beoted that so called uniform membranes have varying pore sizes,nd the size quoted is usually an average. For some porous

embranes, such as the Loeb–Sourirajan membrane, the vari-

tion throughout the membrane is quite significant with theub-surface layer being as much as 10–15 times more poroushan the top or skin layer. Isotropic membranes, shown in Fig. 4,

Fht

of membranes.

nclude microporous, non-porous dense and electrically chargedembranes. Separation in microporous membranes (pore size

etween 101 and 104 nm) is a function of particle and pore sizeistribution, and these membranes are used for processes such asicrofiltration. In non-porous dense membranes, transport is via

iffusion, and hence separation is influenced by partition coef-cient as well as diffusivity of components in the membrane.hese types of membranes are commonly used for extraction,

everse osmosis and pervaporation [33].Anisotropic membranes refer to those in which the material,

he porosity and pore size vary throughout the structure andnclude thin-film composites and Loeb–Sourirajan membranes34]. The composite membrane, shown in Fig. 5, usually con-ists of different polymers where the surface layer determineselectivity, while the porous layer serves as a support.

Homogenous solid membranes, such as silicone, tend to

ig. 4. Top view of isotropic membranes. The microporous membrane shownere illustrates pore uniformity while the non-porous the absence of pore struc-ures.

K. Hylton, S. Mitra / J. Chromatogr. A 1152 (2007) 199–214 203

nes ar

abTttptpo(i

2

stdfltbae

atTtrsbv

pdtsru

op

FT

Fig. 5. Examples of anisotropic membranes. These membra

cross. As a result, the porous membranes provide higher fluxut lower selectivity. Composite membranes are a compromise.he porous part provides for a high flux, while the solid layer on

op provides selectivity. For example, a 1 �m silicone layer onop of a polypropylene composite provides high VOCs flux whilereventing large amounts of water from permeating through. Forhin-film composites, the thin surface layer represents a smallercentage of the overall membrane but is responsible for muchf the membrane’s selectivity. Scanning electron microscopeSEM) images of porous and composite membranes are shownn Fig. 6.

.2.3. GeometryIn the context of membranes, geometry is synonymous with

hape. The two main classes are flat and hollow fiber. The lat-er have a tubular geometry. Each shape has its advantages andisadvantages. The membrane module designs (shell and tube,at sheet and microfluidic) are based on the membrane geome-

ry. Typical hollow fibers are 100–500 �m in diameter and maye porous, solid or composite. They are often used in a shellnd tube module as shown in Fig. 7. Multiple parallel fibers arencased in a large tubing to provide high packing density. For

moaa

ig. 6. SEM of thin-film composite (polyamide surface layer supported by polyprechnology, NJIT).

e heterogeneous in terms of pore size and pore distribution.

nalytical purposes, the module may consist of a single fiber andhe encasement may be made of glass, polyethylene or metal.he ends may be fitted with a tee and sealed. The sample enters

he lumen and the extract is collected on the shell side. As aesult the sample and extract do not mix. It is also used in pres-ure driven separations. These modules offer the advantage ofeing able to accommodate a much higher surface area per unitolume compared to their flat counterparts.

A simple flat sheet module, shown in Fig. 8a, could be com-rised of a cell with a single flat sheet dividing the acceptor andonor. A more complicated structure is the one referred to ashe plate and frame design. It essentially consists of flat sheetstacked on top of each other with an interspersed support mate-ial [35]. A generic design is shown in Fig. 8b. It is typicallysed in applications such as UF, MF and NF.

Microfluidics involves miniaturization with the eventual goalf developing lab-on-a-chip devices. Typically, the flow takeslace in microchannels etched in silicon, glass or a polymer

atrix as shown in Fig. 9. Their small size offers the advantages

f a large surface to volume ratio, inexpensive mass productionnd low solvent usage. The channels are machined on each wafernd then the membrane is sandwiched between them to complete

opylene) and microporous polypropylene (courtesy of Center for Membrane

204 K. Hylton, S. Mitra / J. Chromatogr. A 1152 (2007) 199–214

Fig. 7. Shell and tube module. This module facilitates high membrane area in a relatively small space.

Fig. 8. (a) Simple flat sheet membrane module an

Fig. 9. A microfluidic membrane module using flat sheet membrane [36].

tanv

ftmoet

2

tmsrahegh

d (b) schematic of plate and frame design.

he supported liquid membrane (SLM) module. Channels canlso be prepared by using lithography or laser ablation. Chan-el dimensions may be fine-tuned to facilitate optimal sampleolumes and extraction efficiency.

Microfluidic modules using a flat sheet have also beenabricated [36]. Although analytical operations, such as elec-rophoresis, have been quite successful, extraction on a

icrofluidic device has been a challenge. Membrane extractionffers a plausible solution [36] as shown in Fig. 10. The systemssentially consisting of microsyringe pumps can be interfacedo other microfluidic or conventional devices.

.2.4. Membrane materialThe material from which the membrane is made is referred

o as the membrane material. Typical polymeric hollow fiberembranes used in analytical chemistry are polypropylene,

ilicone, polytetrafluoroethylene (PTFE), polyvinylidenefluo-ide (PVDF) and polysulfone (PS) which are resistant to mostcids, bases and other chemicals, are thermally stable and are

ydrophobic. Membranes can also be made of metal. Commonxamples are Pd alloys [37,38], which have a high hydro-en flux, are chemically resistant and are typically used inydrogen purification. Ceramic membranes are usually tubu-

K. Hylton, S. Mitra / J. Chromatogr. A 1152 (2007) 199–214 205

ted liq

lacrZmiL

2

ui(Tmdafeebt

E

watt

E

wao

2

sa

eawws

K

wt

pocpilwaia

2When a liquid is immobilized in the pores of a porous mate-

rial, via capillary action, that liquid can serve as the membrane,and the latter functions solely as a support as shown in Fig. 12[6,42]. This is referred to as supported liquid membrane and

Fig. 10. Continuous microfluidic suppor

ar and are resistant to high temperature/pressure, corrosion andbrasion. They have been used in the food/beverage, pharma-eutical and biotechnology industries [39]. Lobo et al. [40]ecently reported on the use of Carbosep (active layer ofrO2/TiO2 on carbon) to remove contaminants, such as oil, frometal working fluids before disposal. Other ceramic membranes

nclude Anodisc (�-Al2O3 with a polypropylene support) anda1 − xSrxCo1 − yFeyO3 − α (LSCF) [41].

.3. Membrane extraction

Although many types of extractions have been carried outsing membranes, the two most common approaches in analyt-cal separations are: (1) supported liquid membrane extractionSLME) and (2) liquid–liquid membrane extraction (LLME).hese are used for the analysis of SVOCs, ionic compounds andetals. The third type of membrane extraction, pervaporation,

iffers from the above. Here, the permeate is extracted into a gas,nd this method is mainly used for the analysis of VOCs, andor sample preconcentration. The enrichment factor (EF) andxtraction efficiency (EE) are the two major parameters used tovaluate the effectiveness of a particular extraction. The EF maye defined as the ratio of analyte concentration in the extract tohat in the initial donor:

F = Ca

Cd(6)

here Ca and Cd represent the analyte concentration in thecceptor after extraction and in the initial donor. The EE referso the fraction of analyte that is extracted into the acceptor suchhat:

E = ma

md= Ca × Va

Cd × Vd= EF

Va

Vd(7)

here ma and md represent the total mass of the analyte in thecceptor and donor, respectively, and Vd and Va are the volumesf the donor and extract.

.3.1. Liquid–liquid membrane extractionIn LLME, the analyte is typically extracted from an aqueous

olution into an organic one and hence it is a two-phase organic-queous system as shown in Fig. 11. It is essentially liquid–liquid

Fma

uid membrane extraction (SLME) [36].

xtraction with the phases physically separated by a membrane,nd in contact only at the membrane pores. It is easily interfacedith normal phase HPLC and GC. The efficiency of this systemill therefore be largely dependent on partition coefficient, Kp

uch that:

p = Co

Cw(8)

here Co and Cw represent the equilibrium analyte concentra-ion in the organic and aqueous phases, respectively.

The extractant should have low solubility in the aqueoushase and low volatility. The enrichment factor for non-polarrganic compounds is usually higher than polar and chargedompounds. This is mainly because the solubility of these com-ounds tends to be higher in the aqueous phase. Here enrichments driven by the concentration gradient of the analyte and isimited by its partition coefficient. This means that in the casehere the acceptor is stagnant, the greater hydrophobicity of any

nalyte translates into increased efficiency because of the driv-ng force to reach equilibrium. Consequently having a mobilecceptor tends to increase the enrichment and EE.

.3.2. Supported liquid membrane extraction

ig. 11. Schematic representation of LLME where the analyte diffuses across aicroporous membrane while in contact with an organic acceptor. Kp is defined

s the partition coefficient of the analyte between the aqueous and organic phase.

206 K. Hylton, S. Mitra / J. Chromato

cs

mmtibipsoTbferwae

2

frspsgovpt

FTdB

Savv

J

wao

mmmaspmf

3c

acadodrep

stiand the permeate is pulled by vacuum into the ion source. Mem-

Fig. 12. Schematic representation of supported liquid membrane.

an be prepared simply by immersing a porous membrane in theupporting solvent.

To enhance the selectivity of the liquid membrane, a carrierolecule with a high affinity for the analyte is used. A carrierediated transport that involves the reversible complex forma-

ion between the carrier and the analyte may also be used. SLMEs suitable for polar and ionic compounds such as organic acids,ases and metals. As shown in Fig. 13, it is a three-phase systemn which an organic phase is sandwiched between two aqueoushases. For instance, in the analysis of acids, the pH of the donorolution must be such that the compounds are in their neutralr uncharged forms, thus allowing them to enter the membrane.he pH of the acceptor is maintained such that once in the mem-rane, the analytes are extracted into the acceptor in a chargedorm and cannot be back-extracted into the donor. The pH gradi-nt therefore provides the driving force. This technique usuallyesults in high enrichment factors (thousands) and is compatibleith reversed phase HPLC. Consequently, SLME offers distinct

dvantages such as high selectivity, donor/acceptor ratio andxtraction efficiency when compared to LLME.

.3.3. PervaporationWhile LLME and SLME deal with the extraction of analytes

rom one liquid into another, volatile organics are better sepa-ated by extracting into a gas phase. Pervaporation refers to theeparation of a liquid mixture by partial vaporization through aorous, or a non-porous membrane. The sample flows on oneide of the membrane, and the volatile species permeate as aas to the other side, which is maintained under low pressure,

r just purged with a stream of carrier gas. Theoretically, per-aporation can be viewed as a three-step process comprising ofartitioning from the donor to the membrane, diffusion throughhe membrane followed by evaporation into the gas phase [43].

ig. 13. Schematic representation of the extraction of organic acids by SLM.he bases cannot be extracted, and the neutral species can only transfer byiffusion. Equivalent chemistry is implemented for bases and metals. HA, acid;+, base; N, neutral species; A−, ionized acid.

bn[

gr. A 1152 (2007) 199–214

eparation is due to the differences in the partitioning coefficientnd vaporization of the donor components. Flux through a per-aporation membrane can be expressed in terms of the partialapor pressure on either sides of the membrane such that:

a = PGa

Pao − Pal

L(9)

here Ja is the flux, L the membrane thickness, PGa the gas sep-

ration permeability coefficient, Pao the partial vapor pressuref donor and Pal is the partial vapor pressure in acceptor.

Apart from factors, such as chemical potential gradients andembrane thickness, another important consideration is the per-eation through the boundary layer. Typically, there is poorixing at the membrane/aqueous phase interface. Consequently,stagnant aqueous boundary layer is formed on the membrane

urface that impedes permeation. For continuous monitoring,ermeation must be complete before another injection can beade. Much effort has gone into reducing the boundary layer to

acilitate faster permeation and instrument response time.

. Automated, on-line analysis of volatile organicompounds (VOCs)

Characterized by their low boiling points, high vapor pressurend low molecular weights, VOCs are usually extracted andoncentrated using techniques such as head space analysis, purgend trap and solid-phase microextraction. These techniques areiscrete processes that do not readily facilitate on-line analysisr automation. Additionally, some are not very reproducible ando not exhibit low detection limits. In general, on-line methodseduce sample handling and hence the probability of analyticalrrors and sample loss. The following sections will look at therogressive development in membrane extraction of VOCs.

Membrane separation has also been interfaced with masspectrometers in a technique referred to as membrane introduc-ion mass spectrometry (MIMS). In this configuration, shownn Fig. 14, the sample is constantly introduced to the membrane

rane media such as latex, polyethylene, Teflon, polyimide anditrile have been investigated for use in environmental analyses44,45]. Recently, Thompson et al. [46] used polydimethlysilox-

Fig. 14. Schematic representation of MIMS [50].

K. Hylton, S. Mitra / J. Chromatogr. A 1152 (2007) 199–214 207

Fc

atfocuta

mt

dm[b

tussIa

paepwosfi

T(oitat

Fc

ig. 15. Interfacing of membrane extraction with on-line GC [50]. Sorbent trapsoncentrate the VOCs to minimize broad peaks.

ne (PDMS) hollow fiber membranes in a MIMS configurationo determine VOCs and SVOCs in air at ppt levels. MIMS allowsor direct analysis of VOCs in air and water and has been usedn compounds such as dichloromethane, chlorobenzene, 1,1-hloroethene and acetone [47]. This technique has also beensed in the food industry for real-time monitoring of bioreduc-ions by bakers yeast [48], and for concentrating o-nitrotoluenend methyl salicylate in air [49].

GC interfacing of membrane extraction, shown in Fig. 15, isore complicated, because a positive pressure is maintained on

he permeate side to facilitate the flow of gas.Pervaporation can be used for VOC analysis and may be

esigned to operate in an automated fashion. It has been imple-ented in several different ways and may be classified as follows

50]: (1) membrane in sample (MIS) and (2) sample in mem-rane (SIM). These are shown in Fig. 16.

In the first case, a hollow fiber is dipped into the water con-aining the VOCs. A gas flowing through the membrane picksp the VOCs. SIM uses a shell and tube module. The aqueous

ample is passed through a hollow fiber membrane. The VOCselectively migrate across the membrane into an inert gas stream.n either case, the organics are concentrated in a sorbent beforenalysis by GC for fast on-line analysis.

adas

ig. 17. Gas injection membrane extraction (GIME) for on-line VOCs monitoring [arrying them to the microtrap.

Fig. 16. HFM configurations in VOCs monitoring [50].

As mentioned before, because of the boundary layer effects,ermeation may take a relatively long time to reach steady state,nd any measurement during the transitional period providesrroneous results. The time taken to complete the permeationrocess can be the limiting factor in analysis. Consequently itas realized that the best approach may be a non-steady statene. A pulse introduction process using a flow-injection typeample introduction was developed thus eliminating the needor the system to equilibrate. This led to a significantly quickernstrument response [51].

In another version, a gas was used to inject a liquid sample.he system referred to as gas injection membrane extraction

GIME) is shown in Fig. 17. GIME involves the introductionf an aqueous sample by a N2 stream which injects the samplento the membrane. The membrane serves as a selective barrierhrough which organic analytes permeate. On the permeate side,counter-current gas stream strips the organics and transports

hem to a microtrap. The VOCs are trapped and concentrated by

microtrap in front of the GC column. The retained VOCs areesorbed from the microtrap by an electrically generated temper-ture pulse. Rapid heating generates a concentration pulse whicherves as an injection for chromatographic separation [52]. Con-

55]. Nitrogen flows countercurrent to the sample stream and removes VOCs,

208 K. Hylton, S. Mitra / J. Chromatogr. A 1152 (2007) 199–214

FV

tiofsm

mrpbctca

4c

aant

Fwn

nat

4

apRF

[twti

ig. 18. Chromatogram of water sample, containing �g/L levels of variousOCs, obtained using GIME [55].

inuous monitoring is achieved by making a series of pulses (ornjections) and corresponding to each pulse a chromatogram isbtained [53,54]. The system shown below in Fig. 17 can be usedor the analysis of individual samples by discrete injections, ashown in the chromatogram in Fig. 18, or for continuous on-lineonitoring by sequentially injecting a series of samples.The advantage of gas injection is that the gas cleans the

embrane and destroys the boundary layer on its surface. Theesponse time decreased dramatically and tailing in permeationrofiles was eliminated. The overall membrane extraction ofenzene was found to be completed in 2 min by gas injectionompared to 8 min by liquid. This method is also simpler inerms of instrumentation and operational procedures. The pro-ess, illustrated by Kou et al. [55] used a composite membrane,nd the detection limits were at low ppb (�g/L) levels.

. Automated analysis of semi volatile organicompounds (SVOCs)

SVOCs refer to compounds that are not readily volatilized

nd include compounds such as PAHs, PCBs, biomolecules,cids, phenols and pesticides. Both SLME and LLME tech-iques can be applied to these compounds [56–58]. Methodshat have been successfully developed integrating these tech-

d

[p

Fig. 19. Interfacing on-line memb

ig. 20. Series of chromatograms obtained from continuous monitoring of aater stream where acetonitrile was used as the extractant [59]. In, injectionumber n.

iques include online analysis of non-polar and polar SVOCs,nd automated HF-protected dynamic liquid phase microextrac-ion (LPME).

.1. On-line LLME

This is essentially, automated liquid–liquid extraction acrossmembrane, which allows faster sample throughput, less sam-le handling and continuous monitoring of non-polar analytes.egardless of the analyte, the basic set-up is as illustrated inig. 19.

The use of both polymeric hollow fibers [59] and flat sheet60] has been reported. On-line LLME has been demonstratedo effectively monitor the concentration of different SVOCs inater with the time between injections limited by the separation

ime. Fig. 20 illustrates the results obtained from a sequence ofnjections, with enrichment factors as high as 62 and ppb level

etection limits [59].

In on-line LLME, the extractant may be static or flowing16,61]. Highest EFs are typically obtained when the acceptorhase is mobile. Sandahl et al. [60] used on-line LLME for moni-

rane extraction with HPLC.

K. Hylton, S. Mitra / J. Chromatogr. A 1152 (2007) 199–214 209

alyte

tabpdeGtE[pibti

ceetatst

4

itgttetti

rTstdib

Fig. 21. Simultaneous extraction and concentration [64]. The an

oring Vinclozolin (a fungicide) and obtained EFs up to 118 anddetection limit of 1 ppb. Automated on-line LLME has also

een reported for PCB determination [62], the fungicide thio-hanate methyl [26] and anesthetics in blood [63]. For the PCBsetermination, Barri et al. [62] designed a miniature membranextraction card (referred to as the ESy card) connected to theC injector via an electromechanical installation which con-

rols pretreatment and triggers the GC. This technique exhibitedFs between 33 and 40 for PCBs in river water. Shen et al.

63] used a sample processor system consisting of an auotsam-ling injector, dilutors and a six-port valve connected to the GCnjector loop to obtain EF up to 50 for some local anesthetics inlood plasma. In fact Sandahl et al. [26] designed a parallel sys-em which could analyze both non-polar and polar compoundsn one run.

On-line LLME could be further improved by simultaneousoncentration, Kou and Mitra [92] demonstrated that furthernrichment was possible by selectively eliminating some of thextractant. This phenomenon is shown in Fig. 21. It was foundhat solvent loss was maximum for water soluble solvents such

s methanol, acetonitrile and isopropyl alcohol. Enrichment fac-or was directly related to solvent loss. This was an effective, yetimple technique for combining extraction and concentration inhe same step.

adab

Fig. 22. Hollow fiber membrane concentrator. It can be

is selectively enriched while the solvent is selectively removed.

.2. On-line membrane concentration

For trace analysis, once the analytes have been extractednto a solvent, often a concentration step is necessary. Tradi-ionally, this is carried out off-line using a rotary evaporator,as purging and Kuderna Danish apparatus. These methods areime consuming, cumbersome and do not allow for automa-ion. While there has been much attention placed on on-linextraction techniques, a concurrent development in concentra-ion procedures has not occurred. With the push to developotally automated systems, concentration procedures must bentegrated.

A membrane based, on-line concentration technique has beeneported by Bishop and Mitra and is shown in Fig. 22 [65].his is based on the principle of pervaporation. Instead of theelective permeation of the solute (as mentioned before), selec-ive solvent permeation leads to analyte preconcentration. Theilute solution is injected into a shell and tube module, and annert gas such as nitrogen flows on the permeate side. The mem-rane preferentially allows migration of the solvent across it and

concentrated solution results in the lumen. The process was

emonstrated using both polar and non-polar membranes fornalytes such as, atrazine, pentachlorophenol, naphthalene andiphenyl. The process is illustrated below. The instrumentation

interfaced with HPLC for on-line analysis [65].

210 K. Hylton, S. Mitra / J. Chromatogr. A 1152 (2007) 199–214

action

is

(c(fruti

wcm(p3ahwfam

4m

cotTttoMaas

4

bcasilmIcwcttHcarsolvent showed an EF of 50 and EE of around 50% [72]. TheSLM had a fairly long lifetime and can be re-supported on-line.

Automated SLME has also been used for the determination oftriazines in oil [73], thiophanate methyl and its metabolites [74]

Fig. 23. Interfacing of membrane extr

s simple and can be automated to concentrate either multipleamples or interfaced with chromatography.

The solvents tested were hexane (non-polar) and methanolpolar). The choice of membrane depended on the solvent. Theombination of hexane and a non-polar composite membranepolypropylene with a thin layer of siloxane) provide enrichmentactors close to 20 in less than 30 s. Equivalent concentration in aotary evaporator would take hours. A NafionTM membrane wassed for polar solvents, such as methanol, and the concentrationime was same as above. This is clearly a significant developmentn an on-line concentration procedure.

In addition, on-line coupling of pervaporation and HPLCas applied to the continuous monitoring of trace pharmaceuti-

als in a process stream [66]. A Nafion hollow fiber membraneodule was used for monitoring 2,6-dichlorophenylacetic acid

DCPA), naphthylacetonitrile (NA), 4-chloro-3-nitrobenzo-henone (CNBP), 1,2-diphenylhydrazine (DPH) and 2-chloro-,4-dihydroxyacetophenone (CDHAP) in methanol. Analysisnd detection was via HPLC-UV and enrichment factors asigh as 7.9 with 91% solvent reduction were observed. On thehole, the advantage of this approach is that it can provide

ast (30–60 s) preconcentration of discrete samples for off-linenalysis, and can also be performed on-line for continuousonitoring.

.3. Development of total analytical system based onembrane separation

The concept of a total analytical system (TAS) refers to theoupling of different sample preparation and analytical steps intone integrated system. Membrane based extraction and concen-ration techniques presented so far can be coupled to form aAS. Fig. 23 shows the coupling of two membrane modules. Inhe first, the analytes are extracted by LLME, and in the secondhey are concentrated via pervaporation, which is followed byn-line HPLC detection. This was demonstrated by Wang and

itra [67] for naphthalene and biphenyl, where EF was as high

s 192, and ppt level detection limits were obtained. Conceptu-lly, this idea serves as the basis for developing other integratedystems.

F81

to pervaporative concentration [67].

.4. Automated SLME

SLME has been used extensively in a variety of laboratoryased off-line extraction techniques [68–71]. These are not dis-ussed here because this article mainly focuses on automatednalysis. An important application of automated SLME analy-is reported recently, is for monitoring haloacetic acids (HAAs)n water. Conventional methods for HAA analysis involves, off-ine extraction and derivatization followed by GC-ECD. On-line

onitoring of HAA was tried using both LLME and SLME.n the SLME mode, the HFM was soaked in supporting liquidomprising 5% trioctylphosphine oxide (TOPO) in dihexyl etherhich was held in the pores of the hydrophobic membranes by

apillary forces. The pH of the sample solution was kept low sohat the HAAs were in their uncharged form. The HAAs werehen extracted by a high pH acceptor and analyzed by on-linePLC. The set-up was like the one shown in Fig. 18. A typi-

al series of chromatograms of the HAAs generated by on-linenalysis is shown in Fig. 24. EFs were as high as 500 and EEeached 54%. Automated LLME using MTBE as the extraction

ig. 24. Chromatogram of continuous SLME of reagent water spiked with0 ng/mL (ppb) nine haloacetic acids (HAAs). Injections were made every5 min [72].

K. Hylton, S. Mitra / J. Chromatogr. A 1152 (2007) 199–214 211

he m

am[mtetStibia

4

mli2iw[

bdrdarELtt1u

rG(tdpsa

dtarlHn490 and RSDs below 7.5% [84]. While this technique providedhigh enrichment factors it is not a continuous or on-line process,and is automated by a series of batch operations.

Fig. 25. Simplified schematic of simultaneous SLME and LLME. T

nd Ropivacaine metabolites in urine [75] and for the enrich-ent of triazine herbicides in both urine and surface water

76]. Ropivacaine was analyzed using a microfluidic membraneodule in tandem with ion-pair chromatography [75] while the

riazine studies involved a flat sheet block module coupled toither a diode array spectrophotometer or to HPLC [73]. Sincehiophanate methyl, is non-polar and its metabolites are polar,andahl et al. [74] developed an on-line sample preparation sys-

em that used parallel SLME and LLME. The system, shownn Fig. 25, consisted of flat porous PTFE membrane enclosedetween blocks of microchannels, syringe pumps and multiportnjection valves which allowed interface of the membrane unitnd HPLC-UV.

.5. Automated hollow fiber based microextraction

This technique has also been called hollow fiber liquid phaseicroextraction (HF-LPME). It refers to the use of a single hol-

ow fiber strand filled with a few microliters of an extractant. Its placed in an aqueous sample and extraction is carried out over0–40 min. It can be used in the LLME or SLME mode depend-ng on the analyte being investigated. It has been used for aide range of organics including PAHs, drugs and pesticides

77–79].Hollow fiber liquid phase microextraction (HF-LPME) may

e carried out in two modes: (1) static HF-LPME and (2)ynamic HF-LPME. In static mode, the acceptor solutionemains in the lumen throughout the entire extraction. Inynamic mode however, the acceptor is repeatedly introducednd withdrawn from the lumen. In 2002, Zhao and Lee [80]eported a semi-automated technique for PAHs which exhibitedF of approximately 35 and 70 for static and dynamic HF-PME, respectively. Other studies [77,78,81] have also shown

hat dynamic HF-LPME typically gives higher enrichment fac-ors. HF-LPME is capable of achieving high EE and EF within5–45 min and is compatible with matrices such as plasma,rine, saliva, breast milk, pond water, seawater and soil slur-

Fi

embrane module was fabricated on PTFE or titanium blocks [74].

ies [61]. Basheer et al. [81,82] used LPME in conjunction withC–MS and reported EF up to 100 for organochlorine pesticides

OCPs) and PAHs in rainwater and detection limits comparableo EPA method 508 for OCPs in seawater. Hou et al. [77] in 2003eveloped an automated HF dynamic LPME for organochlorineesticides. A programmable syringe pump was attached to theyringe, as shown in Fig. 26, and parameters such as refill speednd sample volume could be automatically controlled.

Following extraction, the extract filled syringe was injectedirectly into a GC–MS. This technique provided enrichment fac-ors of 480 and upper RSDs of 9.8% compared to 11.3% [83]nd 14% [82] for HF-LPME non-automated. In 2005, Wu et. al.eported another automated LPME called automated dynamiciquid–liquid–liquid microextraction (D-LLLME) coupled withPLC-UV. This was a three-phase system in which chlorophe-oxy herbicides were extracted with enrichment factors up to

ig. 26. Automated dynamic HF-LPME. A programmable pump continuouslynjects the organic solvent into, and then withdraws it from the lumen.

212 K. Hylton, S. Mitra / J. Chromatogr. A 1152 (2007) 199–214

d with

5

isa

tc

TO

T

A

A

A

A

Fig. 27. Schematic of SLM interface

. Automated metal analysis

The most common sample preparation technique for metalsn water is organic extraction via chelation. A chelating agentuch as 8-hydroxyquinoline, dithizone or cupferron along withn organic solvent (such as methyl isobutyl ketone) is added to

[csb

able 1verview of automated, on-line membrane extraction

echnique Membrane used I

utomated VOC analysis • Composite silicone (homogenous siloxanesupported on microporous polypropylene)• Polydimethylsiloxane• Polyethylene

M

utomated SVOC analysis • LLME: microporous polypropylene filled withn-hexane, flat sheet microporous polypropyleneand isooctane as extractant

GG

• SLME: n-decane with microporouspolypropylene support, dihexyl ether or6-undecanone, immobilized in microporouspolypropylene

HLC

utomated concentration • Microporous hydrophobic polypropylene coatedhomogenous siloxane• Nafion [co-polymer of tetrafluoroethylene(Teflon) and perfluoro-3,6-dioxa-4-methyl-7-octene-sulfonic acid]• Flat sheet P84(co-polyimide of 3,3-4,4-benzophenone tetracarboxylic dianhydride and80% methylphenylene-diamine + 20% methylenediamine)

H

utomated metal extraction • Porous PTFE with di-2-ethylhexyl phosphoricacid as membrane liquid

EI

• Polypropylene HFM with Cyanex 471 andAliquat 336 as extractant

AAS for on-line analysis of metals.

he sample and shaken in a separatory funnel [1]. In 1980, Bab-ock et al. reported the use of hollow fiber membrane in SLM

85], and in 1984, Danesi described a simplified model for thearrier-facilitated transport of metal ions through hollow fiberupported liquid membranes [86]. By the late nineties, mem-rane extraction of metals became an acceptable technology for

nterface Compounds extracted Reference

S, GC, GC/MS Polar and non-polar VOCS, e.g.benzene, toluene, chlorobenzene,methanol, acetone, methylenechloride, o-nitrotoluene

[46,49,51,52,94,95]

C-ECD,C-NPD

Polar and non-polar SVOCs, e.g.PAHs, acids, PCBs, pesticides, drugsand their metabolites, anesthetics

[60,62]

PLC-UV,C/ESI-MS,E-UV

[26,63]

PLC-UV Naphthalene, atrazine,pentachlorophenol, isopropanol,2,6-dichlrorophenylacetic acid

[65,66]

TAAS, FAAS,CP-MS

Co, Zn, Cd, Pd, Pt, Rh [71,73,89]

mato

ta

mCitiwTd8tr

6

taatcsMttoaid

R

[[

[

[[[[[[

[[

[[[[

[[

[[

[

[[

[

[

[

[

[[[[[

[

[

[

[[[

[[

[

[

[[[[[[

[

K. Hylton, S. Mitra / J. Chro

he recovery of metals from waste streams [87–91]. Its use innalytical sample preparation has been rather limited.

The use of HF-SLME in the cleanup, extraction and enrich-ent of numerous metals including Pb, Cu, Cr, La, Ce, Zn ando has been reported [14,92,93]. This method has usually been

nterfaced with atomic absorption spectroscopy (AAS) or induc-ively coupled plasma mass spectrometry (ICP-MS) as shownn Fig. 27, although Ndungu et al. [14] reported an interfaceith ion pair chromatography for Zn, Ni, Co, Mn and Cd.ypical extractants include di-hexylether or kerosene dilutedi-(2-ethylhexyl) phosphoric acid (DEHPA), Aliquat 336, LIX60-I, Cyanex 302 and tri-n-butyl phosphate (TBP). Extrac-ion efficiencies and enrichment factors up to 95% and 200,espectively, have been reported [92].

. Conclusion

Non-porous and microporous membranes are versatile struc-ures that can be used for the extraction and concentration of

wide range of compounds, such as, VOCs, SVOCs, organiccids, ions and metals. Different processes have been developedo facilitate automation and direct interfacing with analyti-al instrumentation like GC, HPLC, MS and AA. Table 1ummarizes the extraction techniques discussed in this paper.

oreover, there exist numerous other possibilities that are yeto be explored. On the whole membrane extraction and concen-ration offers some unique possibilities beyond the capabilitiesf conventional means, particularly in the field of automatednalysis. The challenge in the future lies in enhancing selectiv-ty, testing diverse analytes and incorporation in lab-on-a-chipevices.

eferences

[1] S. Mitra, B. Kebbekus, Environmental Chemical Analysis, Chapman andHall/CRC, Boca Raton, Florida, 1998.

[2] P. Patnaik, Handbook of Environmental Analysis—Chemical Pollutants inAir, Water, Soil and Solid Wastes, CRC Press, Boca Raton, Florida, 1997.

[3] J.L. Tadeo, C. Sanchez-Brunete, B. Albero, L. Gonzalez, Crit. Rev. Anal.Chem. 34 (2004) 165.

[4] R. Eisert, J. Pawliszyn, Crit. Rev. Anal. Chem. 27 (1997) 103.[5] J. Pawliszyn, Anal. Chem. 75 (2003) 2543.[6] M. Mulder, Basic Principles of Membrane Technology, Kluwer Academic

Publishers, Dordrecht, 1991.[7] K.K. Sirkar, W.S.W. Ho, in: K.K. Sirkar, W.S.W. Ho (Eds.), Membrane

Handbook, Kluwer Academic Publishers, Norwell, MA, 1992, p. 4.[8] S.B. Tuwiner, Diffusion and Membrane Technology, Reinhold Publishing

Corporation, New York, NY, 1962.[9] J.A. Jonsson, L. Mathiasson, TrAC, Trends Anal. Chem. 18 (1999) 318.10] O.B. Jonsson, U. Nordlof, U.L. Nilsson, Anal. Chem. 75 (2003) 3506.11] S. Palmarsdottir, L. Mathiasson, J.A. Jonsson, L.-E. Edholm, J. Chro-

matogr. B: Anal. Technol. Biomed. Life Sci. 688 (1997) 127.12] B. Lindegard, J.A. Jonsson, L. Mathiasson, J. Chromatogr. B: Anal. Tech-

nol. Biomed. Life Sci. 573 (1992) 191.13] C. Basheer, H.K. Lee, J. Chromatogr. A 1047 (2004) 189.14] K. Ndungu, N.-K. Djane, L. Mathiasson, J. Chromatogr. A 826 (1998) 103.

15] C. Yang, E.L. Cussler, J. Membr. Sci. 166 (2000) 229.16] R.-S. Juang, I.-P. Huang, Ind. Eng. Chem. Res. 39 (2000) 1409.17] R.-S. Juang, J.-D. Chen, H.-C. Huan, J. Membr. Sci. 165 (2000) 59.18] I. Ortiz, M.F. San Roman, S.M. Corvalan, A.M. Eliceche, Ind. Eng. Chem.

Res. 42 (2003) 5891.

[

[[

gr. A 1152 (2007) 199–214 213

19] W.S.W. Ho, Ind. Eng. Chem. Res. 41 (2002) 381.20] T. Berhanu, J.-F. Liu, R. Romeroa, N. Megersa, J.A. Jonsson, J. Chro-

matogr. A 1103 (2006) 1.21] M. Khrolenko, P. Dzygiel, P. Wieczorek, J. Chromatogr. A 975 (2002) 219.22] Y. Qin, J.P. Sheth, K.K. Sirkar, Ind. Eng. Chem. Res. 41 (2002) 3413.23] J.A. Jonsson, L. Mathiasson, TrAC, Trends Anal. Chem. 18 (1999) 325.24] K. Luthje, T. Hyotylainen, M.-L. Riekkola, Anal. Bioanal. Chem. 378

(2004) 1991.25] J.A. Jonsson, L. Mathiasson, J. Chromatogr. A 902 (2000) 205.26] M. Sandahl, J.A. Jonsson, L. Mathiasson, J. Chromatogr. A 975 (2002)

211.27] R.-S. Juang, H.C. Kao, W.-H. Wu, J. Membr. Sci. 228 (2004) 169.28] C. Fontas, M. Hidalgo, V. Salvado, E. Antico, Anal. Chim. Acta 547 (2005)

255.29] I. Van de Voorde, L. Pinoy, R.F. De Ketelaere, J. Membr. Sci. 234 (2004)

11.30] O. Arous, A. Gherrou, H. Kerdjoudj, Desalination 161 (2004) 295.31] S. Mitra, N. Zhu, X. Zhang, B. Kebbekus, J. Chromatogr. A 736 (1996)

165.32] http://www.bccresearch.com/membrane/DMD00.html, Accessed on June

6, 2006.33] W. Eykamp, in: R.D. Noble, S.A. Stern (Eds.), Membrane Separations

Technology: Principles and Application, Elsevier Science B.V., Amster-dam, 1995, p. 1.

34] S. Loeb, S. Sourirajan, Sea Water Demineralization by Means of an OsmoticMembrane in Saline Water Conversion. II. Advances in Chemistry Series,American Chemical Society, Washington, DC, 1963, p. 117.

35] http://www.epa.gov/ORD/NRMRL/pubs/625r96009/625r96009.pdf,Accessed on July 6, 2006.

36] X. Wang, C. Saridara, S. Mitra, Anal. Chim. Acta 543 (2005) 92.37] P.M. Thoen, F. Roa, J.D. Way, Desalination 193 (2006) 224.38] P. Kamakoti, D.S. Sholl, J. Membr. Sci. 279 (2006) 94.39] J. Finley, Filtr. Sep. 42 (2005) 34.40] A. Lobo, A. Cambiella, J.M. Benito, C. Pazos, J. Coca, J. Membr. Sci. 278

(2006) 328.41] D. Chen, L.K. Weavers, H.W. Walker, J.J. Lenhart, J. Membr. Sci. 276

(2006) 135.42] P.R. Danesi, in: N.N. Li, H. Strathmann (Eds.), Separation Technology:

Proceedings of the Engineering Foundation Conference, Engineering Foun-dation, New York, NY, 1987, p. 261.

43] R.W. Baker, Membrane Technology and Applications, John Wiley andSons, England, 2004.

44] A.J. Madden, M.J. Hayward, Anal. Chem. 68 (19961805).45] R. Liu, X. Qiao, T.-S. Chung, Chem. Eng. Sci. 60 (2005) 6674.46] A. Thompson, A. Creba, R. Ferguson, E. Krogh, C. Gill, Rapid Commun.

Mass Spectrom. 20 (2006) 2000.47] M.A. LaPack, J.C. Tou, C.G. Enke, Anal. Chem. 62 (1990) 1265.48] C.D.F. Milagre, H.M.S. Milagre, J. Augusto, R. Rodrigues, L.L. Rocha,

L.S. Santos, M.N. Eberlin, Biotechnol. Bioeng. 90 (2005) 888.49] I. Cotte-Rodrıguez, E. Handberg, R.J. Noll, D.P.A. Kilgour, R.G. Cooks,

Analyst 130 (2005) 679.50] G.C. Slack, N.H. Snow, D. Kou, in: S. Mitra (Ed.), Sample Preparation

Techniques in Analytical Chemistry, John Wiley and Sons, New York, NY,2003, p. 212.

51] X. Guo, S. Mitra, J. Chromatogr. A 826 (1998) 39.52] X. Guo, S. Mitra, Anal. Chem. 71 (1999) 4587.53] Y.H. Xu, S. Mitra, J. Chromatogr. A 688 (1994) 171.54] A. San Juan, X. Guo, S. Mitra, J. Sep. Sci. 24 (2001) 599.55] D. Kou, A. San Juan, S. Mitra, Anal. Chem. 73 (2001) 5462.56] J.-F. Liu, X. Liang, G.-B. Jiang, Y.-Q. Cai, Q.-X. Zhou, G.-G. Liu, Talanta

60 (2003) 1155.57] V. Jung, L. Chimuka, J.A. Jonsson, N. Niedack, P. Bowens, B. Alsanius,

Anal. Chim. Acta 474 (2002) 49.

58] S.H. McMurray, G.J. Griffin, J. Chem. Technol. Biotechnol. 77 (2002)

1262.59] X. Guo, S. Mitra, J. Chromatogr. A 904 (2000) 189.60] M. Sandahl, E. Ulfsson, L. Mathiasson, Anal. Chim. Acta 424 (2000)

1.

2 mato

[

[

[[[[[[[

[

[[[

[

[

[

[

[[

[[

[[

[[

[[

[[[[

14 K. Hylton, S. Mitra / J. Chro

61] K.E. Rasmussen, S. Pedersen-Bjergaard, TrAC, Trends Anal. Chem. 23(2004) 1.

62] T. Barri, S. Bergstrom, J. Norberg, J.A. Jonsson, Anal. Chem. 76 (2004)1928.

63] Y. Shen, J.A. Jonsson, L. Mathiasson, Anal. Chem. 70 (1998) 946.64] D. Kou, S. Mitra, Anal. Chem. 75 (2003) 6355.65] E. Bishop, S. Mitra, J. Chromatogr. A 1046 (2004) 11.66] E. Bishop, S. Mitra, J. Pharm. Biomed. Anal. 37 (2005) 81.67] X. Wang, S. Mitra, J. Chromatogr. A 1068 (2005) 237.68] P. Dzygiel, P. Wierczorck, J. Chromatogr. A 889 (2000) 93.69] M. Knutsson, G. Nilve, L. Mathiasson, J.A. Jonsson, J. Chromatogr. A 754

(1996) 197.70] M. Luque, E. Luque-Perez, A. Rios, M. Valcarcel, Anal. Chim. Acta 410

(2000) 127.71] N. Mergersa, J.A. Jonsson, Analyst 123 (1998) 225.72] X. Wang, D. Kou, S. Mitra, J. Chromatogr. A 1089 (2005) 39.73] R.C. Martinez, E.R. Gonzalo, E.H. Fernandez, J.H. Mendez, Anal. Chim.

Acta 304 (1995) 323.74] M. Sandahl, L. Mathiasson, J.A. Jonsson, J. Chromatogr. A 975 (2002)

211.

75] J.A. Jonsson, M. Anderson, C. Melander, J. Norberg, E. Thordarson, L.

Mathiasson, J. Chromatogr. A 870 (2000) 151.76] N. Megersa, L. Chimuka, T. Solomon, J.A. Jonsson, J. Sep. Sci. 24 (2001)

567.77] L. Hou, G. Shen, H.K. Lee, J. Chromatogr. A 985 (2003) 107.

[

[[[

gr. A 1152 (2007) 199–214

78] L. Hou, H.K. Lee, J. Chromatogr. A 976 (2002) 377.79] M. Ma, S. Kang, Q. Zhao, B. Chen, S. Yao, J. Pharm. Biomed. Anal. 40

(2006) 128.80] L. Zhao, H.K. Lee, Anal. Chem. 74 (2002) 2486.81] C. Basheer, R. Balasubramanian, H.K. Lee, J. Chromatogr. A 1016 (2003)

11.82] C. Basheer, H.K. Lee, J.P. Obbard, J. Chromatogr. A 968 (2002) 191.83] C. Basheer, V. Suresh, R. Renu, H.K. Lee, J. Chromatogr. A 1033 (2004)

213.84] J. Wu, K.H. Ee, H.K. Lee, J. Chromatogr. A 1082 (2005) 121.85] W.C. Babcock, R.W. Baker, K.L. Smith, E.D. Lachapelle, J. Membr. Sci.

7 (1980) 71.86] P.R. Danesi, J. Membr. Sci. 20 (1984) 231.87] C.H. Yun, R. Prasad, A.K. Guha, K.K. Sirkar, Ind. Eng. Chem. Res. 32

(1993) 1186.88] Z.-F. Yang, A.K. Guha, K.K. Sirkar, Ind. Eng. Chem. Res. 35 (1996) 4214.89] C. Fontas, V. Salvado, M. Hidalgo, Ind. Eng. Chem. Res. 41 (2002) 1616.90] R.-S. Juang, H.-L. Huang, J. Membr. Sci. 213 (2003) 125.91] F. Valenzuela, C. Basualto, C. Tapia, J. Sapag, J. Membr. Sci. 155 (1999)

163.

92] N.-K. Djane, I.A. Bergdahl, K. Ndungu, A. Schutz, G. Johannsson, L.

Mathiasson, Analyst 122 (1997) 1073.93] X.-J. Yang, C. Pin, A.G. Fane, Anal. Chim. Acta 369 (1998) 17.94] M.J. Yang, S. Harms, Y.Z. Luo, J. Pawliszyn, Anal. Chem. 66 (1994) 1339.95] X. Guo, S. Mitra, Anal. Chem. 71 (1999) 4407.