review extraction and transport of metal ions and small ...longn/doc/journal papers/extraction...

A

obcoearamwmmsastfbo©

K

C

0d

Journal of Membrane Science 281 (2006) 7–41

Review

Extraction and transport of metal ions and small organiccompounds using polymer inclusion membranes (PIMs)

Long D. Nghiem a,b,1, Patrick Mornane a,b, Ian D. Potter c, Jilska M. Perera b,Robert W. Cattrall b,c, Spas D. Kolev a,∗

a School of Chemistry, The University of Melbourne, Vic. 3010, Australiab Department of Chemical and Biomolecular Engineering, The University of Melbourne, Vic. 3010, Australia

c Department of Chemistry, La Trobe University, Vic. 3086, Australia

Received 11 January 2006; received in revised form 16 March 2006; accepted 26 March 2006Available online 3 April 2006

bstract

The stability of polymer inclusion membranes (PIMs) relative to other liquid membranes is amongst the major reasons for the recent rejuvenationf interest in carrier-mediated transport for selective separation and recovery of metal ions as well as numerous organic solutes. This is reflectedy an increasing number of PIM investigations reported in the literature over the last two decades. Given the outstanding performance of PIMsompared to other types of liquid membranes particularly in terms of membrane lifetime, it has been predicted that practical industrial applicationsf PIMs will be realized in the near future. This review provides a comprehensive summary of the current knowledge relevant to PIMs for thextraction and transport of various metal ions and small organic solutes. PIM studies reported to date are systematically summarized and outlinedccordingly to the type of carriers used, i.e. basic, acidic and chelating, neutral or solvating, and macrocyclic and macromolecular. The papereviews the various factors that control the transport rate, selectivity and stability of PIMs. The transport phenomena observed by various authorsre related to the membrane characteristics, physicochemical properties of the target solutes as well as the chemistry of the aqueous solutionsaking up the source and receiving phases. The results from these studies reveal an intricate relationship between the above factors. Furthermore,hile the interfacial transport mechanisms in PIMs are thought to be similar to those in supported liquid membranes (SLMs), the bulk diffusionechanisms in PIMs governing their permeability and selectivity require better understanding. This review also delineates two mathematicalodeling approaches widely used in PIM literature: one uses a set of assumptions that allow the derivation of analytical solutions valid under

teady-state conditions only; the other takes into account the accumulation of the target species in the membrane during the initial transport statend therefore can also be applied under non-steady-state conditions. The latter is essential when the interfacial complexation reaction kinetics islow. It involves more complex mathematics and requires the application of numerical techniques. The studies included in this review highlighthe potential of PIMs for various niche applications on a practical scale. The discussions provided, however, also emphasize the need for more
undamental research before any such practical applications of PIMs can be realized. This is specifically important for small organic compoundsecause to date scientific investigation involving the extraction and transport of these compounds remains limited. Transport mechanisms of smallrganic compounds are less well understood and are likely to be more complex than those observed with the transport of metal ions.
2006 Elsevier B.V. All rights reserved.

eywords: Polymer inclusion membranes (PIMs); Extraction; Liquid membranes; Carrier-mediated transport; Mineral processing; Metal recovery

ontents

1. Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2. Base polymers for membrane preparation . . . . . . . . . . . . . . . . . . . . . . . .3. Carriers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

∗ Corresponding author. Tel.: +61 3 8344 7931; fax: +61 3 9347 5180.E-mail address: [email protected] (S.D. Kolev).

1 Current address: School of Civil, Mining and Environmental Engineering, The U

376-7388/$ – see front matter © 2006 Elsevier B.V. All rights reserved.oi:10.1016/j.memsci.2006.03.035

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

niversity of Wollongong, NSW 2522, Australia.

mailto:[email protected]

dx.doi.org/10.1016/j.memsci.2006.03.035

9. The future of PIM research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

. . . . .

1

ctsttoetaibheAft

apns[iSm(st[t

camiiotl

ecacoitatpwIalmbPtup

Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. Introduction

In recent years, membrane-based processes have attractedonsiderable attention as a valuable technology for many indus-ries. This significant gain in momentum is driven in part bypectacular advances in membrane development, wider accep-ance of the technology as opposed to conventional separa-ion processes, increased environmental awareness and mostf all stricter environmental regulations and legislation. How-ver, despite a recent market boom in all other membrane sec-ors including membrane filtration and electrodialysis, practicalpplications of liquid membranes remain largely limited. Thisncludes bulk liquid membranes (BLMs), emulsion liquid mem-ranes (ELMs) and supported liquid membrane (SLMs). BLMsave low interfacial surface areas and mass transfer rates whilemulsion breakage is the main problem associated with ELMs.major drawback associated with SLMs is poor stability. These

actors have severely rendered liquid membranes mostly imprac-ical for many large-scale applications [1,2].

Nevertheless, given the essential need for metal ion recoverys well as for the extraction of numerous small organic com-ounds over the last two decades in hydrometallurgy, biotech-ology and in the treatment of industrial wastewater, significantcientific effort has been expended to understand [3] and improve4] the stability of liquid membranes. The number of scientificnvestigations devoted to this topic has been rising steadily [1].uch dedicated works have resulted in a novel type of liquidembranes, commonly called polymer inclusion membranes

PIMs) [5], although a number of other names are also being used
uch as polymer liquid [6,7], gelled liquid [8], polymeric plas-icized [9–11], fixed-site carrier [12–14] or solvent polymeric15,16] membranes. PIMs are formed by casting a solution con-aining an extractant, a plasticizer and a base polymer such as
e

im

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

ellulose triacetate (CTA) or poly(vinyl chloride) (PVC) to formthin, flexible and stable film. The resulting self-supportingembrane can be used to selectively separate the solutes of

nterest in a similar fashion to that of SLMs. In several stud-es [8,17,18], PVC has been used to simply gel the liquid phasef an SLM to stabilize it within the pores of an inert support. Inhese cases, the PVC concentration of the membrane was muchower than that used for a self-supporting membrane.

PIMs retain most of the advantages of SLMs while exhibitingxcellent stability and versatility. The lower diffusion coeffi-ients often encountered in PIMs can be easily offset by creatingmuch thinner membrane in comparison to its traditional SLMounterpart. In several cases, PIMs with higher fluxes than thosef SLMs have been reported [5,19,20]. In contrast to SLMs, its possible to prepare a PIM with negligible carrier loss duringhe membrane extracting process [5,11,19,20]. In addition, themount of carrier reagent can be greatly reduced, hence creatinghe possibility of using more expensive extractants, which in theast could only be used for high value metals or organics. Thisill no doubt create a wider range of applications for PIMs.

t is also noteworthy that the mechanical properties of PIMsre quite similar to those of filtration membranes. The techno-ogical advancements achieved with filtration membranes for

anufacturing, module design and process configuration wille particularly useful for the large-scale practical realization ofIMs [21]. Consequently, this will enable PIM-based systems

o exhibit many advantages such as ease of operation, minimumse of hazardous chemicals and flexibility in membrane com-osition to achieve the desired selectivity as well as separation

8 L.D. Nghiem et al. / Journal of Membrane Science 281 (2006) 7–41

4. Plasticizers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114.1. The role of plasticizers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114.2. Plasticizer concentration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124.3. Plasticizer viscosity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134.4. Dielectric constant . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

5. Membrane characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145.1. Morphology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145.2. Permeability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155.3. Selectivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165.4. Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

6. Extraction and transport studies in PIMs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186.1. Basic carriers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196.2. Acidic and chelating carriers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226.3. Neutral or solvating carriers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246.4. Macrocyclic and macromolecular carriers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

7. Transport mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 317.1. Interfacial transport mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 317.2. Bulk transport mechanisms. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

8. Mathematical modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

fficiency.It is interesting to note that PIMs have been used in chem-

cal sensing for more than 30 years in the form of polymerembrane ion-selective electrodes (ISEs) [22]. In 1970 [23], it

Memb

wldahpbl

ebttiidcorcop

tisppomaaf

2

sctffpm

opbdimmhp

taismpiio[tirmcMohthrtee

pppac

TP

P

PCC

n

L.D. Nghiem et al. / Journal of

as demonstrated that the organic liquid of a calcium selectiveiquid membrane ISE could be immobilized into PVC to pro-uce a polymer film with identical calcium sensing propertiesnd selectivity as the organic liquid itself. Since that time thereave been numerous PVC-based membranes developed for theotentiometric sensing of various cations and anions. Such mem-ranes for use in potentiometry have also been termed “gellediquid membranes” and “entangled liquid membranes” [22].

About the same time as this first reported use in ISEs, Blocht al. [21] demonstrated that PVC-based membranes could alsoe used for metal ion separation although the requirements forhe membrane characteristics were somewhat different for thewo applications [21,22]. In sensing, fast ion exchange or metalon complexation is required at the sample solution/membranenterface to rapidly establish the interfacial electrical potentialifference but there should be negligible transport of the metalontaining species through the membrane within the timescalef the measurement. In separation, fast interfacial reactions areequired but in this case, high diffusion coefficients of the metalontaining species within the membrane are also desirable inrder to achieve mass transport from the source to the receivinghase within a reasonable timeframe.

This review aims at providing a comprehensive summary ofhe current knowledge relevant to PIMs for the extraction of var-ous metal ions and small organic solutes. Membrane stability,electivity and transport rates are discussed in relation to thehysicochemical properties of the base polymers, carriers andlasticizers as well as the characteristics of the target metal ionsr the organic solutes. Transport mechanisms and their mathe-atical models are also delineated. It should be emphasized that

ll the research on PIMs carried out so far has been conducted onlaboratory scale and the transition of this research to a pilot or

ull-scale application presents a major challenge for the future.

. Base polymers for membrane preparation

The base polymers play a crucial role in providing mechanicaltrength to the membranes. Despite a vast number of polymersurrently used for many engineering purposes, it is surprisinghat PVC and CTA have been the only two major polymers used
or most of the PIM investigations conducted so far. Although theeasibility of several cellulose derivatives (i.e. cellulose acetateropionate (CAP) and cellulose tributyrate (CTB)) as base poly-ers for PIMs has recently been studied [24], a large number
almh

able 1hysical properties of three polymers most frequently used in PIMs

olymer MW used in PIMs (kDa) MWc (kDa) Tg (◦

oly(vinyl chloride) (PVC) 90–180a 12.7c 80d

ellulose triacetate (CTA) 72–74b 17.3c naellulose tributyrate (CTB) 120b 47.4c na

a: not available.a Ref. [59].b Ref. [24].c Ref. [112].d Ref. [28].e Ref. [25].

rane Science 281 (2006) 7–41 9

f polymers that can possibly be used for PIMs remains unex-lored. To some extent, this is because both PVC and CTA cane used to prepare a thin film with a relatively simple proce-ure based on dissolution in an organic solvent. Another factors the dearth of information regarding the role of base poly-

ers in mechanically supporting the membranes, enhancing theembrane stability and at the same time creating a minimal

indrance to the transport of metal ions and small organic com-ounds within the membranes.

Polymers that make up the skeleton of a PIM are thermoplas-ic [25]. They consist of linear polymer strands and because therere no cross-links between these strands, they can be dissolvedn a suitable organic solvent, where the polymer strands becomeeparated. The mechanical strength of a thermoplastic thin filmembrane is a combination of intermolecular forces and the

rocess of entanglement [26]. The former determines the flex-bility of the material with high intermolecular forces resultingn a rigid membrane. The latter is the result of random diffusionf the flexible polymer strands in a sol as the solvent evaporates26]. Consequently, a very stable thin film can be formed, despitehe absence of any intermolecular covalent bonds even thought should be noted that there is a disentanglement process occur-ing over a very long time scale. It is, however, essential that theolecular weight (MW) of the polymer used is larger than the

ritical entanglement molecular weight (MWc) of that polymer.Wc values together with the glass transition temperature (Tg)

r the melting temperature (Tm) of several base polymers whichave been used in PIMs are presented in Table 1. It is noteworthyhat most, if not all, of the PIM studies available in the literatureave used base polymers with a MW much higher than the cor-esponding MWc values. Above these MWc values, variation inhe base polymer MW is speculated to exert a negligible influ-nce on the membrane mechanical strength and performance asxperimentally illustrated by Rais et al. [27].

Although both PVC and CTA have been widely used to pre-are PIMs, understanding the effect of the properties of theseolymers on the performance of PIMs is still limited. CTA is aolar polymer with a number of hydroxyl and acetyl groups thatre capable of forming highly orientated hydrogen bonding. Inontrast, the C–Cl functional group in PVC is relatively polar
nd non-specific dispersion forces dominate the intermolecu-ar interactions. Consequently, PVC is an amorphorous poly-
er with a small degree of crystallinity whereas CTA is oftenighly crystalline [28]. Furthermore, while CTA can be slightly

C) Tm (◦C) Polymeric characteristics

na Slightly crystalline, mostly amorphorousd

302e Infusible, high degree of crystallinity, excellent strengthd

207d Infusible, high degree of crystallinity, excellent strengthd

1 Memb

hthpdntCAtecaf

pimbpaaeIdtimmt

aartii

3

tiltriita

carecp

TE

T

B

A

N

M

0 L.D. Nghiem et al. / Journal of

ydrated [29], PVC is virtually not. This hydration characteris-ic of CTA and other cellulose derivatives makes them prone toydrolysis, particularly in an acidic environment [24,30]. Theolarity and crystalline nature of the CTA polymer may ren-er it incompatible with high concentrations of hydrophobicon-polar carriers. For example, Gherrou et al. [13] reportedhe formation of crystalline layers of crown ether within theTA domain at a sufficiently high crown ether concentration.s a result, metal ion transport at such a high carrier concentra-

ion becomes ineffective [13]. However, it is noteworthy that thexcellent mechanical strength of CTA is mostly attributed to itsrystalline domain [28]. In addition, cellulose-based polymersre highly infusible [28], which makes them particularly usefulor PIM applications.

While the base polymers merely provide mechanical sup-ort to the membrane, their bulk properties appear to be anmportant factor in governing metal ion transport through the

embrane. To date, it is not possible to accurately predict theulk properties of a polymer based on a finite set of physicalarameters. However, the glass transition temperature (Tg) forn amorphorous polymer or the melting temperature (Tm) forcrystalline polymer are often used to characterize the inher-

nt polymer flexibility and its microstructural characteristics.t should be pointed out that both amorphorous and crystallineomains exist in any thermoplastic polymer. Below the glassransition temperature (Tg), the polymer is rigid and glassy and
ndividual polymer strands are unable to change their confor-
ations. Since this condition is thought to be unfavorable foretal ion transport in membranes, plasticizers are often added

o the polymer to lower its Tg value and create more flexible

feti

able 2xamples of PIM carriers reported in the literature and their typical target solutes (al

ype of carriers Examples

asicQuaternary amines Aliquat 336

Tertiary amines TOA, other tri-alkyl aminesPyridine and derivatives TDPNO

cidic and chelatingHydroxyoximes LIX® 84-IHydroxyquinoline Kelex 100�-Diketones Benzoylacetone, dibenzoylacetone,

benzoyltrifluoracetoneAlkyl phosphoric acids D2EHPA, D2EHDTPA

Carboxylic acids Lauric acid, Lasalocid A

eutral or solvatingPhosphoric acid esters TBPPhosphonic acid esters DBBPOthers CMPO, TODGA, TOPO,

polyethylene glycolacrocyclic and macromolecularCrown ethers and calix arenes DC18C6, BuDC18C6

Others Bathophenanthroline, bathocuproine

rane Science 281 (2006) 7–41

nd less brittle membranes. In fact, in a pure polymer withoutplasticizer, the Tg or Tm value is usually much higher than

oom temperature (Table 1). Consequently, all PIMs reported inhe literature contain some form of plasticizer unless the carriertself can also act as a plasticizer. The significance of plasticizersn enhancing metal ion fluxes is discussed in Section 4.

. Carriers

Transport in PIMs is accomplished by a carrier that is essen-ially a complexing agent or an ion-exchanger. The complex oron-pair formed between the metal ion and the carrier is solubi-ized in the membrane and facilitates metal ion transport acrosshe membrane. The well-known classes of solvent extractioneagents namely basic, acidic and chelating, neutral or solvat-ng, and macrocyclic and macromolecular have all been studiedn PIMs. The types of carriers used in PIM research as reported inhe literature along with the target metal ions or organic solutesre summarized in Table 2.

Much of the research on PIMs has been conducted usingommercially available solvent extraction reagents as carrierslthough some papers report on the use of newly synthesizedeagents. However, in most cases the physicochemical prop-rties of these new compounds are not well documented. Thehemical reactions that are involved in the extraction and strip-ing of target solutes using PIMs are essentially the same as
or the corresponding solvent extraction systems. However, thessential difference between the two systems is associated withhe transport of the target solutes through the membrane andt is this aspect that has formed the focus of much of the
l abbreviations are explained in the Glossary)

Target solutes References

Au(III), Cd(II), Cr(VI), Cu(II),Pd(II), Pt(IV), small saccharides,amino acids, lactic acid

[6,7,9,33,47,55,59,83–85,91]

Cr(VI), Zn(II), Cd(II), Pb(II) [33,39,58,64,88,89]Ag(I), Cr(VI), Zn(II), Cd(II) [34,70,100]

Cu(II) [43]Cd(II), Pb(II) [41]Sc(III), Y(III), La(III), Pr(III),Sm(III), Tb(III), Er(III), Lu(III)

[79]

Pb(II), Ag(I), Hg(II), Cd(II), Zn(II),Ni(II), Fe(III), Cu(II)

[8,42,92,98]

Pb(II), Cu(II), Cd(II) [11,93,94]

U(VI) [21,46,69]As(V) [99]Pb(II), Ce(III), Cs+, Sr(II) [27,53,66]

Na+, K+, Li+, Cs+, Ba(II), Sr(II),Pb(II), Sr(II), Cu(II), Co(II), Ni(II),Zn(II), Ag(I), Au(III), Cd(II), Zn(II),picrate

[10,13–15,19,20,30–33,54,65,68,93,101,103–106,113]

Lanthanides [15,45,67,97,114]

Memb

rmes

etstttmsaihabacehmPr[cottmra

attbmclictAims

beoepwce

imhldtdwZlsTaBtcpatimtwm[

4

4

btsoitsmctitta

pP2mmm


esearch on PIMs. The main objective of PIM research is toaximize the membrane fluxes whilst retaining the extraction

fficiency and selectivity of the corresponding solvent extractionystem.

The actual transport phenomena in PIMs and SLMs are, how-ver, quite complex and can be strongly influenced by bothhe physicochemical properties of the carrier and the targetolute and to a lesser extent by the chemical composition ofhe membrane phase as well as the source and receiving solu-ions. Although several dedicated investigations have discussedhe significance of these physicochemical properties [31–34],

ore research is still needed to elucidate the intricate relation-hips between these factors and the membrane permeabilitynd selectivity. Understanding these relationships is particularlymportant given the diversity of carriers being studied whichave significant differences in their physicochemical propertiesnd transport modes. It is noteworthy that while there are a num-er of commercial carriers available, there are many more thatre being synthesized and reported. Basic, acidic and solvatingarriers are common reagents in solvent extraction and have beenxtensively studied and used on a large industrial scale in manyydrometallurgy applications [35–37]. Macrocyclic and macro-olecular compounds are also of particular interest to manyIM researchers due to their specific host–guest complexationegimes which often result in excellent separation selectivity38]. They can, however, be quite expensive for large-scale appli-ations and their environmental implications as well as manyther characteristics remain unknown at this stage. Carriers inhis group have been used in almost half of the PIM investiga-ions reported to date in the literature (Table 2). Furthermore,

acrocyclic and macromolecular compounds, unlike other car-iers, have been shown to be useful for the separation of thelkali metals.

In addition to membrane selectivity, transport efficiency isnother critical consideration for PIMs. It can be expected thathe molecular structure of the carrier can markedly influencehe rate of transport of the target solutes across the mem-rane. Walkowiak et al. [33] studied the transport of alkalietal cations across PIMs containing a sym-(alkyl)-dibenzo-16-

rown-5-oxyacetic acid with side arms of different alkyl chainengths geminally attached to the lariat ether as the carrier. Max-mal flux was found when the alkyl side arm contained ninearbon atoms with considerably lower transport efficiency whenhe alkyl chain was longer or shorter [33]. In another study,guilar et al. [32] demonstrated that both membrane selectiv-

ty and transport efficiency could be optimized by designing aacrocyclic carrier with a careful combination of ring size and

ubstituent groups.Due to the lack of experimental data, it is currently not possi-

le to correlate in a systematic way the selectivity and transportfficiency with the carrier properties, although some importantbservations have been established. For example, basic carri-rs with a lower basicity constant have been reported to have a
oorer selectivity but higher transport efficiency [39]. It is note-orthy that different types of carriers are expected to exhibit
onsiderably different transport efficiencies because of differ-nces in their complexation mechanisms.

ptaa

rane Science 281 (2006) 7–41 11

The carrier molecular structure and the chemistry involvedn the complexation and transport processes are amongst the

ost important factors governing the membrane selectivity. Itas been shown that the carrier molecular structures can be tai-ored to achieve a specific selectivity. For example, hydrophiliciazadibenzocrown ether was reported to have a higher selec-ivity for Pb(II) over Cd(II) and Zn(II) than a more hydrophobicerivative [32]. Basic carriers with a higher basicity constantere also reported to have a better selectivity for Cr(VI) overn(II) and Cd(II) [39]. Basic and neutral carriers often exhibit

ow selectivity for metals. The corresponding selectivity con-tants vary within only about two orders of magnitude [33,39].he metal ion reacts with the former via an ion-pair mech-nism and with the latter via a solvating mechanism [40].oth of these mechanisms are governed by electrostatic interac-

ions and are non-specific in nature. The selectivity of acidicarriers is also relatively low and is normally controlled byH [41,42]. In contrast, certain chelating carriers can offermuch better selectivity due to their specific and conforma-

ional interactions with metal ions [42,43]. PIMs using chelat-ng agents as the carrier can selectively transport the target

etal ions [41,43] whilst the flux of other metal ions is vir-ually zero. Generally, excellent selectivity can be achievedith macrocyclic and macromolecular carriers, although itay vary significantly depending on their chemical structures

38].

. Plasticizers

.1. The role of plasticizers

The individual molecular chains in PIMs are held togethery a combination of various types of attractive forces. Amongsthem, van der Waals forces are abundant but are weak and non-pecific, while polar interactions are much stronger but can onlyccur at polar centers of the molecule [44]. The latter often resultn a rigid non-flexible thin film with a three-dimensional struc-ure within its polymeric matrix [28,44]. This three-dimensionaltructure rigidity is, however, unfavorable for a diffusive flux ofaterial within the polymer matrix [28]. Consequently, plasti-

izers are often used to increase the metal species, flux as well ashe membrane softness and flexibility. The role of a plasticizers to penetrate between polymer molecules and “neutralize”he polar groups of the polymer with its own polar groups oro merely increase the distance between the polymer moleculesnd hence reduce the strength of the intermolecular forces [44].

While there is a large number of commercially availablelasticizers, few of them have been tested for applications inIMs. Amongst them, 2-nitrophenyl octyl ether (2-NPOE) and-nitrophenyl pentyl ether (2-NPPE) have been used in theajority of successful PIM studies currently available. Theolecular structures of several examples of plasticizers com-only studied with PIMs are shown in Fig. 1. Most of these

lasticizers seem to have been chosen because of their applica-ion in ISE membranes rather than because of their commercialvailability at low cost or because of their potential industrialpplications.


lastic

cottapiSmagaoatItw

bdgaTctpnNpglefrwa

Cncp

4

caewFPgattictwtc

p

pbieem

Fig. 1. Chemical structures of p

As can be seen in Fig. 1, plasticizers are generally organicompounds containing a hydrophobic alkyl backbone with oner several highly solvating polar groups. The former governshe compatibility of the plasticizer with the membrane whilehe latter interact with the polar groups of the base polymernd hence “neutralize” them. Therefore, a balance between theolar and non-polar portions of the plasticizer molecule is anmportant factor as has been illustrated in an early study byugiura [45]. In this study, lanthanide ion flux was evaluated withembranes made with polyoxyethylene alkyl ethers of different

lkyl chain length and different numbers of polar oxyethyleneroups. The results indicate that the optimum number of carbontoms in the alkyl chain is 12 and that of the polar groups is 2r 3 in this homologous series. An increase in the length of thelkyl chain results in a more hydrophobic, viscous plasticizerhat eventually suppresses the polar properties of the plasticizer.n contrast, an increase in the number of polar groups decreaseshe viscosity and increases the hydrophilicity of the plasticizer,hich eventually renders it unusable.Given their vital role in the plastics industry, plasticizers have

een the subject of extensive scientific investigations for manyecades [44]. Although findings resulting from such investi-ations may also to some extent be applicable to PIMs, manyspects of the plasticization process remain poorly understood.he relationship between the membrane performance and plasti-izer concentration as well as the physicochemical characteris-ics of the plasticizers is complex and to a large part remainsoorly understood. This is further complicated by the largeumber of essential properties often required of a plasticizer.otable amongst these are a good compatibility with the baseolymer, low volatility, low viscosity, high dielectric constant,ood resistance to migration from the base polymer, low cost andow toxicity. Nevertheless, considerable effort has been made tolucidate the influence of the plasticizer on the membrane per-
ormance. Results reported in such studies underline the vitalole of the plasticizer as a key ingredient of PIMs. It is note-orthy that several carriers such as quaternary ammonium salts
nd phosphoric acid esters can also play the role of a plasticizer.

pspb

izers commonly used in PIMs.

onsequently in such cases, no additional plasticizers may beecessary [6,7,21,46,47]. The influence of the plasticizer con-entration and physicochemical characteristics on membraneerformance is discussed below.

.2. Plasticizer concentration

Low plasticizer concentration is undesirable since it mayause the membrane to become more rigid and brittle due tophenomenon commonly referred to as the “anti-plasticizing”

ffect [44]. The minimum plasticizer concentration variesidely depending on both the plasticizer and the base polymer.or PVC, this concentration can be in a range of up to 20% (w/w).ioneering research by Barshtein and Kotlyarevskii [48,49] sug-ested that it was possible to determine for a given plasticizer themount necessary for all polar groups on the polymer backboneo be neutralized or shielded from each other by a monolayer ofhe plasticizer. This later became a commonly used parametern the plastics industry, often denoted as phrmin (parts of plasti-izer per 100 parts of polymer by mass). The phrmin depends onhe molecular weight (MW) of the plasticizer and the moleculareight of one helical unit of the polymer. For a PVC membrane,

he molecular weight of one helical unit is 875 g/mol and phrminan be calculated as

hrmin = MW of plasticizer

875× 100 (1)

This expression is particularly useful to determine a suitablelasticizer concentration for ISE membranes using PVC as thease polymer. In a series of investigations, where the mechan-cal properties and dielectric characteristics of plasticized PVClectrode membranes were systematically examined using sev-ral plasticizers, Gibbons and Kusy [50–52] demonstrated thatembrane performance could indeed be improved by using a

lasticizer concentration closer to its phrmin value. This is muchmaller than the empirical amount (two parts plasticizer per oneart polymer) commonly used in ISE membranes, which haseen proven to be rather excessive.

L.D. Nghiem et al. / Journal of Membrane Science 281 (2006) 7–41 13

Fig. 2. Permeability of Cu(II) as a function of the plasticizer TBEP concentrationiCw

bmsptmbcrrcwmflitd2nitbtsctpfa

4

dIii

Fig. 3. The effect of plasticizer viscosity on Cr(VI) transport through CTA andPVC membranes. (a) The membranes contained 1.0 M TOA (based on plasticizervolume). Source phase: 2 × 10−2 M Cr(VI), 1.0 M HCl; receiving phase: 0.1 MNaOH. (b) The membranes contained 20% (w/w) Aliquat 336, 40% (w/w) plas-t −6

Sf2

cittcrifldFfltOitd2bit

n CTA membranes containing LIX® 84-I as the carrier. Source phase: 20 mg/Lu(II), 0.025 M AcH/AcNa, pH 5.0; receiving phase: 1 M H2SO4. (Reproducedith permission from Ref. [43]. Copyright 2005 Elsevier Science.)

In fact, excessive plasticizer concentration is problematicecause the excess plasticizer could migrate or exude to theembrane/aqueous interface and form a film on the membrane

urface, which would create an additional barrier to the trans-ort of metal ions across the membrane. Exudation depends onhe compatibility between the plasticizer and the base poly-

er and above the compatible concentration level exudationecomes more severe [29]. Furthermore, excessive plasticizeran significantly reduce the thin film mechanical strength, henceendering it unusable in a practical situation. Several studieseported an increase in metal ion transport as the plasticizer con-entration increased [10,53,54]. However, when considering aider concentration range, it appears that there exists an opti-um plasticizer concentration at which a maximum metal ionux occurs and beyond that, the flux starts to decrease. In an early

nvestigation, Sugiura [45] observed that the lanthanide ion fluxhrough a CTA membrane increased to an optimum value thenecreased as the plasticizer concentration increased from 0 toM. Although a plausible explanation for this observation wasot given, it is consistent with similar observations in recent stud-es by Fontas et al. [9] and de Gyves et al. [43]. In the latter study,he permeability of Cu(II) as a function of the plasticizer tri-n-utoxyethyl phosphate (TBEP) concentration follows a similarrend (Fig. 2) to the curve reported by Sugiura [45]. The authorspeculated that the decrease in Cu(II) permeability as the plasti-izer concentration exceeded the optimum point was attributedo an increase in viscosity, which was unfavorable for the trans-ort of Cu(II) [43]. This explanation, however, contradicts theact that beyond the anti-plasticizing range, plasticizer additionlways results in a lower Tg and hence a less viscous medium.

.3. Plasticizer viscosity

Metal ion transport through PIMs is accomplished by
iffusion, which follows the Stokes–Einstein relationship [55].t is therefore not surprising that the viscosity of the plasticizers an important parameter, governing the rate of transportn PIMs. Kozlowski and Walkowiak [39] reported a linear
4

c

icizer. Source phase: 1.8 × 10 M Cr(VI), pH 2; receiving phase: 1 M NaNO3.ee the glossary list for the plasticizer names. (Reproduced with permissionrom Ref. [55], Copyright 2005 Elsevier Science and from Ref. [39], Copyright004 Taylor & Francis.)

orrelation between the plasticizer viscosity and the chromiumon flux through both PVC and CTA membranes containingri-n-octylamine (TOA) as the carrier. This is consistent withhe observation by Scindia et al. [55] who also investigated thehromium ion flux as a function of plasticizer viscosity. Theseesults are presented in Fig. 3. A similar correlation can also benferred from a study by Sugiura and Kikkawa [56], where theuxes of zinc across CTA membranes prepared from twelveifferent plasticizers were reported. However, it is notable fromig. 3b that the use of 2-NPOE results in a higher chromiumux than that of the other plasticizers even though one of them,

ris(2-ethylhexyl)phosphate (T2EHP), has a similar viscosity.n the basis of the Stokes–Einstein relationship alone this

s an unexpected result and the authors point to the fact thathese two plasticizers have a considerable difference in theirielectric constants. The dielectric constants are 24 and 4.8 for-NPOE and T2EHP, respectively. This argument is supportedy an independent study by Mohapatra et al. [54] who alsonvestigated the influence of these two plasticizers on Sr(II)ransport through CTA membranes.

.4. Dielectric constant

It appears in addition to the viscosity that the dielectriconstant of the plasticizer plays an important role in the dif-

1 Membrane Science 281 (2006) 7–41

fobtTntrfisabprcsaer

5

5

mrthfsmorb

Crcasmaaw6sfamTwre

mt

Fig. 4. Morphology of the thin films with various membrane constituents. ThetN2

ctqaatarctt3c[ccrae

a


usion process. In fact, the dependence of copper ion transportn plasticizer concentration as reported in Fig. 2 can possiblye explained by the dielectric constant effect. In high dielec-ric constant media, ion-pairs can dissociate more readily [57].he individual ions have a higher diffusion coefficient than aeutral and bulky ion-pair consisting of the target solute andhe carrier [57]. In addition, this could also allow for ions toelocate more readily between active sites of the neighbouringxed (or partially mobile) carriers. The detail of this fixed-ite jumping mechanism is discussed in Section 7.2. Kozlowskind Walkowiak [58] have reported a characteristic correlationetween the chromium ion flux and the dielectric constant of thelasticizer. Furthermore, the curve reported previously in Fig. 2esembles very well the dependence of the thin film dielectriconstant on the plasticizer concentration, which also takes thehape of a convex curve [29,44]. Several other researchers havelso attributed the success of nitrophenyl alkyl ether plasticiz-rs in PIMs to their high dielectric constants in addition to theirelatively low viscosities [9,54,56].

. Membrane characteristics

.1. Morphology

One important aspect of PIMs is the microstructure of theembrane materials, which determines the distribution of car-

iers in the polymer matrix and ultimately affects the membraneransport efficiency. Consequently, considerable research effortas been devoted to clarifying this issue. While a variety of sur-ace characterization techniques have been employed in thesetudies, scanning electron microscopy (SEM) and atomic forceicroscopy (AFM) have been the most frequently used. Results

btained from SEM and AFM studies consistently indicate aemarkable influence of the polymeric composition on the mem-rane morphology.

Arous et al. [14] reported distinct differences between pureTA membranes, plasticized CTA membranes without any car-

iers and plasticized CTA membranes with different macrocyclicarriers (Fig. 4). SEM pictures of a pure CTA membrane revealedhighly porous polymer matrix with a relatively uniform pore

ize in the sub-micrometer range, typical of that for filtrationembranes made of cellulose acetate. These pores vanished anddense membrane was formed as the plasticizer 2-NPOE was

dded. The addition of macrocyclic carriers resulted in a thin filmith distinctive separate layers. In the case of dibenzo 18-crown-(DB18C6) as the carrier, a well-orientated fibrous structure of

ub-micrometer size was clearly visible in the SEM image. Theormation of multilayers at high crown ether concentrations haslso been reported by other researchers who have observed pooretal ion transport associated with this morphology [10,13].his effect was attributed to the discrete distribution of carriersithin the membrane polymer matrix [10,13,14]. This possibly

eflects the high degree of crystallization of both CTA and crown
thers as discussed previously in Section 2.
Xu et al. [59] studied the morphology of PVC/Aliquat 336embranes at various Aliquat 336 concentrations. In contrast to

he phenomena reported above, PVC membranes with a low con-

Bode

hin films consist of: (A) pure CTA; (B) CTA and 2-NPOE; and (C) CTA, 2-POE, and DB18C6. (Reproduced with permission from Ref. [14]. Copyright004 Elsevier Science.)

entration of the Aliquat 336 carrier were characterized as densehin films with no apparent porosity. As the concentration of Ali-uat 336 increased above 40% (w/w), the authors [59] reportedclear porous membrane structure with irregular shape pores

nd pore sizes of a few micrometers or less. It was speculatedhat this transformation in the interior structure could explain thepparent increase in metal ion transport through the membraneseported by several other studies [6,60] when the Aliquat 336oncentration reached 40% (w/w). The authors, however, cau-ioned that these SEM results did not provide direct evidence aso whether these micro pores were previously filled with Aliquat36 nor did they provide any evidence of the existence of microhannels of Aliquat 336 within the membrane polymer matrix59]. It is noteworthy that membrane samples must be dry andoated with a thin layer of conductive material such as gold,hromium or carbon prior to SEM analysis. Moreover, due to aelatively low resolution, it is often difficult to discern featurest the nanometer scale or smaller. For this reason, AFM is oftenmployed in conjunction with SEM to improve resolution.

Several studies using AFM have been reported and the resultsre in good agreement with those obtained in SEM studies.
ecause PIMs are usually cast on a glass surface, the morphol-gy of the membrane surface on the glass side can be quiteifferent to that of the membrane surface exposed to air. Wangt al. [6] reported a smoother surface at the glass interface when

Memb

tmcAArtsWtortrsaBmsco

cfdctteetRrmtlr

5

dndn�rt2fltpistat

cpma

sSrotsadcfslohiomc

maapwToiwsa[rtt

ioatcvcSricpl


hey examined the surface morphology of PVC/Aliquat 336embranes. This difference is, however, alleviated as the con-

entration of Aliquat 336 increases above 50% (w/w). Given thatliquat 336 can play the role of a plasticizer and the fact thatliquat 336 preferentially migrates to the membrane/air surface,

esults reported in these studies suggest a possible influence ofhe plasticizer on the membrane surface roughness. This is con-istent with studies by Walkowiak et al. [33] and Kozlowski andalkowiak [58], in which a marked difference was shown in

he surface roughness of CTA/TOA membranes with and with-ut a plasticizer. A combination of CTA and TOA resulted in aelatively smooth membrane surface while addition of the plas-icizer 2-NPPE dramatically increased the membrane surfaceoughness. In another study, Wang and Shen [61] examined theurface topography of PVC/Aliquat 366 membranes before andfter extraction experiments with Cu(II) as the target metal ion.ased on AFM images, they reported a slight increase in theembrane surface roughness after a 2-week exposure to a Cu(II)

olution. The authors attributed this to the redistribution of thearrier within the membrane polymer matrix. However, valuesf the surface roughness were not reported in this study.

Although both SEM and AFM techniques are versatile andan provide good visual confirmation of the membrane sur-ace and to some degree the membrane interior structure, toate, studies employing these techniques have not been able tolearly elucidate the distribution of carrier and plasticizer withinhe membrane. Consequently, more advanced material charac-erization techniques have been attempted. Tripathi et al. [62]xamined the distribution of carrier within the membrane by pre-xtracting Cs+ and Ag(I) into the membrane and then mappinghe distribution of these high atomic mass metal ions using theutherford backscattering spectrometry (RBS) technique. Their

esults suggested a uniform distribution of organic carrier at aicroscopic scale. Although this conclusion is in contradiction

o the hypothesis about micro channels filled with carrier out-ined above, it is prudent to note that evidence available to dateemains too limited to conclusively eliminate this speculation.

.2. Permeability

The rate of metal ion transport through PIMs is arguably aecisive factor influencing the commercialization of this tech-ology. Not surprisingly, most of the PIM studies available toate have reported on this crucial parameter, although oftenot in a consistent form. While permeate fluxes up to severalmol m−2 s−1 can be commonly found in studies cited in this

eview, a permeate flux as high as 45 �mol m−2 s−1 for Cr(VI)hrough a CTA membrane containing TOA and plasticized with-NPPE ether has been reported [39]. In most cases, the permeateux for PIMs is marginally lower than that for SLMs. Although

he actual transport mechanisms can be quite different, trans-ort in both PIMs and SLMs is governed by a number of factorsncluding the membrane morphology, membrane composition,
olution chemistry in the source and receiving phases as well ashe temperature. Because many aspects of PIMs remain uncleart this stage, interpretation and comparison of the current litera-ure data will require a careful consideration of the experimental
Ivfl[

rane Science 281 (2006) 7–41 15

onditions involved. As we delineate in the section below, a com-rehensive understanding of the influence of these factors on theembrane transport rate is particularly important for the design

nd implementation of PIMs.PIMs and SLMs are distinctively different in their interior

tructures as well as in their surface morphologies. BecauseLMs consist of a porous supporting layer impregnated with car-ier and diluent, metal ions can be transported through a networkf fluid-filled micro channels. Consequently, transport is subjecto the tortuosity of this network and is limited by the availableurface area. In contrast, PIMs can be viewed as non-porous ands discussed in the previous section, there is no convincing evi-ence of a micro channel network in which metal ion transportan take place. As a result, the whole membrane is availableor transport. Although the diffusion coefficients of the targetolutes in PIMs are generally one or two orders of magnitudeower than those in SLMs, several cases of comparable [9,11,55]r even higher diffusion coefficient values [5,19,20] in PIMsave been reported when both types of membranes have beennvestigated under similar extraction conditions. Taking accountf the fact that transport can be further improved by preparing auch thinner membrane, PIMs have the potential to offer a very

ompetitive transport rates.Transport through PIMs can be strongly influenced by the

embrane morphology. In certain cases, a high concentration ofcrown ether carrier may result in a crystalline thin film char-

cterized by distinctively separate layers [10,13,14]. This mor-hology was found to be unfavorable for transport in PIMs andas often associated with poor target solute fluxes [10,13,14].he membrane surface roughness is also an important morphol-gy parameter. Wang et al. [6] reported a slight but discerniblencrease in metal ion transport when the rougher side of the PIMsas exposed to the source solution. This is consistent with other

tudies, where a positive correlation between metal ion perme-bility and the membrane surface roughness was also observed55,58]. However, it should be noted that an increase in surfaceoughness can be attributed to the addition of a plasticizer andherefore the membrane surface roughness can also be relatedo the membrane composition.

Results from several PIM studies indicate a considerablenfluence of the membrane composition on the transport ratef the target solutes. However, the data remain quite scatterednd the variation in permeate flux following any changes inhe membrane composition depends quite substantially on theharacteristics of each constituent involved. As discussed pre-iously, there exists an optimum concentration of plasticizeroncentration at which a maximum rate of transport is achieved.imilar observations for the carrier concentration have also beeneported for several macrocyclic carriers [10,13,14,63], whichs possibly due to their high hydrophobicity and crystallizationapacity. In contrast, other researchers reported an increase in theermeate flux with increasing concentration of TOA [64], Lasa-ocid A [11] and Aliquat 336 [9] in CTA-based membranes.
n the first two cases, the permeate flux approached a plateaualue, while in the latter case, a linear increase in the permeateux was reported. In another study, Kozlowski and Walkowiak58] compared the transport efficiency between CTA-based and

1 Memb

PcvbPmCsotadTcf

csiti[tNiusmtt

bpch

F

misrdtielptttaca

ttetmNaWitmd

5

maqeimaa[

dttrcstaitdiu

bmboeoIitawwa


VC-based membranes containing a relatively high TOA con-entration in the range from 0.9 to 1.45 M (based on plasticizerolume) and plasticized with 2-NPPE. At 0.9 M TOA, the CTA-ased membrane produced a higher permeate flux than that ofVC. However, the Cr(VI) permeate flux of the PVC-basedembrane increased at a considerably higher rate than that ofTA as the TOA concentration in the membrane increased. Con-

equently, at a TOA concentration of 1.45 M, the permeate fluxf both membranes was quite comparable. The authors attributedhis to the hydrophobicity difference between CTA and PVC andrgued that a high concentration of carrier was needed for a highegree of compatibility within the membrane polymer matrix.he results summarized here emphasize the significance andomplexity of this issue, which will remain an important topicor further research in the years to come.

The driving force in a PIM system is principally the con-entration gradient across the membrane of either the metallicpecies itself or another species known as the coupled-transporton [12,58,65–68]. Consequently, the ionic compositions of bothhe source and the receiving solutions play a vital role in govern-ng the metal ion transport process. Kozlowski and Walkowiak58] studied the effect of the source solution pH on Cr(VI)ransport across a CTA-based membrane containing TOA and 2-PPE as plasticizer. They observed that the Cr(VI) permeate flux

ncreased as the pH of the source phase decreased. In this partic-lar case, the difference in pH between the source and receivingolutions produces a proton concentration gradient across theembrane and consequently, the transport of protons through

he membrane results in an uphill transport of HCrO4− in order

o maintain electroneutrality in both liquid phases.The driving force in a PIM system can also be influenced

y the mobility of the coupled-transport ion within the organichase of the membrane [68]. In general, the efficiency of theoupled-transport ions increases in the opposite order of theirydrated radii as follows:

− < Cl− < Br− < NO3− < SCN− < I−

< IO4− < ClO4

− < picrate−

This is evidenced in several studies where transport experi-ents were carried out with several different coupled-transport

ons under a similar concentration gradient [12,65–68]. In thesetudies, better transport of the target metal ion was consistentlyeported when lower hydrated energy ions were employed as theriving ion. Ions are hydrated in aqueous solutions and becausehe membrane phase is hydrophobic, transport of such hydratedons is thought to be restricted, depending on their hydrationnergy or the number of attached water molecules [12]. Simi-arly, Levitskaia et al. [68] demonstrated that permeability can beredicted based on the Gibbs energy of ion partitioning betweenhe source and membrane phases, which is essentially related tohe hydrated energy of the ion. However, it must be emphasized
hat most of the coupled-transport ions can act as complexinggents. Consequently, the phenomena discussed here are furtheromplicated by the fact that the types of metallic species presentre often concentration-dependent [69]. In addition, transport of
tZ8T

rane Science 281 (2006) 7–41

he target metal ion may also be influenced by the competi-ion with other ions present in the aqueous phase. Wionczykt al. [70] found that the effect of H2SO4 concentration inhe source solution on Cr(VI) transport across a CTA-based

embrane containing a basic carrier and plasticized with 2-POE can be described by a convex parabolic relationship. This

ppears to contradict the observation reported by Kozlowski andalkowiak [58] discussed previously because the driving force

s also the proton concentration gradient. However, accordingo Wionczyk et al. [70] the sulfate ion competed with the chro-

ate ion for transport and as a result Cr(VI) transport started toecrease at a sufficiently high sulfate ion concentration.

.3. Selectivity

Selectivity is undoubtedly an important issue in the imple-entation of PIMs for a number of reasons. In environmental

pplications, the concentration of the target metal ions can beuite low and reasonably high selectivity is essential for anffective treatment. In hydrometallurgical applications, puritys a key factor in determining the commodity price of a given

etal, particularly for those of high value [71]. Impurities canlso affect downstream processing such as electrowinning oper-tions where only a certain level of impurity can be tolerated72].

In a solvent extraction process the selectivity for metal ionsepends almost entirely on the difference in the lipophilicity ofhe corresponding metal complexes. Consequently, the selec-ivity is usually low and numerous extraction stages are oftenequired to increase the selectivity [73]. This flexibility in pro-ess design is, however, not available in membrane extractionystems. Fortunately, data available in the SLM and PIM litera-ure consistently indicate that a membrane extraction process canchieve an appreciably higher selectivity than that encounteredn solvent extraction although the underlying mechanisms forhis phenomenon have not been adequately studied to date. Theiscussion below is restricted to selectivity amongst metal ionsnvolving the same complexation mechanisms with the carrierssed.

It is important to point out two fundamental differencesetween solvent extraction and PIMs as well as other liquidembranes used for separation [41,69]. Firstly, while there can

e an excess of extractant in solvent extraction, reactive sitesr the availability of carrier in a PIM are restricted. Secondly,xtraction and back extraction in PIMs occur simultaneously aspposed to their consecutive arrangement in solvent extraction.n fact, it would be wrong to view the mass transfer phenomenan PIMs (or in fact membrane extraction in general) as analogs ofhose in solvent extraction [69]. In a recent study by de Gyves etl. [43] only Cu(II) transport was observed with a PIM preparedith LIX® 84-I, CTA, and tris(2-n-butoxyethyl) phosphatehen the source solution contained a 10-fold excess of Fe(III)

nd Zn(II). The source solution in this experiment was adjusted
o pH 5.0, at which pH considerable extraction of Fe(III) andn(II) would occur based on the extraction isotherm for LIX®
4-I which can be found in the solvent extraction literature.his is consistent with observations reported in two separate

Memb

iI1tssafteitmpfssdhsackaurtslsrdhu

5

ig

TIfpmmawiaghclbTsmibwbTwclSsfldwfbbm

TR

M

CC

C

C

CC

CCC


nvestigations by Ulewicz et al. [42] and Aguilar et al. [41].n the former study, the selectivity for Pb(II) over Cd(II) was40 with PIMs using Kelex 100, significantly higher than theheoretical selectivity (ca. three-fold) obtained in a single-stageolvent extraction process at the same extraction pH. In the lattertudy [41], when employing PIMs consisting of CTA, 2-NPOE,nd di(2-ethylhexyl) phosphoric acid (D2EHPA), the selectivityor Zn(II), Cu(II) and Co(II) was considerably different despitehe fact that solvent extraction isotherms indicated that theirxtraction coefficients would be comparable at the same pH usedn the PIM extraction experiments. Further data showing similarrends have also been reported, although it is not possible to

ake a direct comparison to a corresponding solvent extractionrocess in these studies [32,33,39,74,75]. Such data underscoreundamental differences in transport mechanisms betweenolvent and membrane extraction processes. Aguilar et al. [41]uggested a direct competition for active transport sites in PIMsue to their limited availability. However, to date, no attemptsave been made to relate complexation stability constants andelectivities amongst competing metal ions. Furthermore, it haslso been proposed that the high selectivity observed in PIMsould possibly be attributed to differences in complexationinetics [69]. Given the difference in structure between PIMsnd SLMs, they are expected to exhibit different selectivitiesnder similar experimental conditions. For example, PIMs wereeported to have a better selectivity for Pt(IV) over Pd(II) thanhat of SLMs when exposed to identical source and receivingolution conditions [9]. Unfortunately, further data from theiterature to substantiate this premise are not available at thistage. It is prudent to note that the discussion presented hereemains quite speculative because of the lack of comprehensiveata related to these important issues. However, this fact clearlyighlights the need for further fundamental research to fullynderstand the transport phenomena observed with PIMs.

.4. Stability

A major reason for the limited use of SLMs on a largendustrial scale is the membrane stability or lifetime, which, ineneral, is far too low for commercial applications [1,2,76–78].

ub

t

able 3eported PIM lifetimes under continuous operation

embranes (base polymer/carrier/plasticizer) Reported lifetim

TA/calix[6]arene/2-NPOE Small flux declTA/Lasalocid A/2-NPOE No sign of flux

days. Stable aftTA/acyclic polyether bearing amide/2-NPOE-TBEP Small flux decl

plasticizer lossTA/calix[4]arene/2-NPOE Small flux decl

plasticizer lossTA/calix[4]arene/2-NPOE Stable flux afteTA/DC18C6/2-NPOE, TBEP Flux decline be

and plasticizerTA/Aliquat 336/2-NPOE, DOS, DOTP, or DOP Flux decline anTA/Aliquat 336/T2EHP Flux decline beTA/t-BuDC18C6/2-NPOE-DNNS Stable for sever

rane Science 281 (2006) 7–41 17

his is possibly a major motivation for the development of PIMs.n SLMs, the capillary force or interfacial tension is responsibleor the binding of the membrane liquid phase to the supportingores [76,77]. This form of adhesive force is, however, weak andembrane breakdown can easily occur via several destabilizingechanisms including lateral shear forces, emulsion formation

nd leaching of the membrane liquid phase to the aqueous phase,hich can be worsened by an osmotic flow [76,77]. In contrast,

n PIMs, as we have discussed previously, the carrier, plasticizernd base membrane are well integrated into a relatively homo-eneous thin film. Although several FTIR studies [10,13,14]ave revealed no signs of covalent bond formation between thearrier, plasticizer and the base membrane skeleton, it is mostikely that they are bound to one another by a form of secondaryonding such as hydrophobic, van der Waals or hydrogen bonds.hese secondary bonds are much stronger than interfacial ten-ion or capillary forces. Consequently, PIMs are considerablyore stable than SLMs as clearly evidenced in the PIM stud-

es discussed in this review, where a comparison of the stabilityetween these two types of membranes has been a focus of theork [11,19,20,55]. Kim et al. [19,20] have investigated the sta-ility of PIMs and SLMs under similar experimental conditions.hey reported no flux decline or evidence of material lossesithin 15 days of continuous transport experiments with PIMs

ontaining CTA, 2-NPOE and macrocyclic carriers. In contrast,eakage of the organic material became clearly evident in theLM counterparts after 48 h agitation in aqueous solutions. Theuperior stability of PIMs over SLMs has also been reportedor membranes using Aliquat 336 as carrier [55]. Under simi-ar experimental conditions, stable performance of PIMs for 30ays was recorded while organic material leakage from SLMsas reported after only 7 days. Table 3 summarizes the results

rom several PIM studies, in which membrane lifetimes haveeen reported. It is noteworthy that while leakage of the mem-rane liquid is usually used to assess the stability of a SLM, theembrane lifetime reported in Table 3 for PIMs was evaluated

sing flux stability because no carrier or plasticizer losses coulde observed in most of these studies.

In general, PIMs are highly resistant to carrier and plas-icizer leakage. This has probably prompted a recent inves-

e and membrane performance Reference

ine after 30 days [106]decline or carrier and plasticizer losses after 10er 10 months storage in air

[11]

ine after 15 days but no evidence of carrier and [20]

ine after 20 days but no evidence of carrier and [19]

r 1 month [30]gan slowly after 100 days but no evidence of carrierloss

[5]

d carrier/plasticizer loss began after 30 days [55]gan after 18 days [55]al weeks [31]

1 Memb

tutaCbaepba(wtamtsmrt

fstwspc[tmmrSarccctaphsppdicwddeosc

cFbtniaiccbn

6

obsibiHstmaUbrpcing phases with respect to each type of carrier are distinctivelydifferent. This is evident in the following sections where thePIM studies reported to date are systematically reviewed anddiscussed on the basis of the carrier type.


igation focusing instead on the hydrolysis of base polymersnder extreme conditions. Gardner et al. [24] have studiedhe stability of PIMs prepared from a crown ether, 2-NPOEnd various cellulose polymer derivatives including CTA, CAP,AB and CTB. The authors have demonstrated that dura-ility of the base polymer increases as longer alkyl chainsre added to the cellulose glucoside backbone units. How-ver, the membrane permeability was found to decrease pro-ortionally. Furthermore, they reported that the membranesroke down quickly under caustic conditions (3 M KOH) whilemuch longer lifetime was reported under acidic conditions

3 M HNO3). Hydrolysis of the CTA-based membrane occurredithin 2.9 days under caustic conditions. Under acidic condi-

ions, the CTA-based membrane was stable for 12.3 days. Innother study [30], it has also been reported that the CTA-basedembranes quickly decomposed when the source phase con-

ained 1 M LiOH. Stability of PVC-based PIMs has not beenystematically studied. However, on the basis of their poly-eric structure, PVC-based membranes are expected to be more

esistant to hydrolysis under extreme caustic or acidic condi-ions.

Nevertheless, it is necessary to emphasize that carrier lossrom PIMs under certain conditions has been reported in severaltudies [7,31,79]. Although PIMs are typically solid thin films,he carriers remain in a quasi-liquid state and are in contactith the aqueous phase on both the source and receiving

ides. Consequently, hydrophobicity and water solubility arerobably amongst the most critical parameters governing thearrier leaching behavior [7,30,32]. Nazarenko and Lamb31] reported a noticeable loss of the DC18C6 carrier fromheir CTA/2-NPOE-DNNS (dinonylnaphthalenesulfonic acid)

embranes. However, this problem was resolved by using aore hydrophobic carrier (t-BuDC18C6) and the membranes

emained stable after several weeks of experimentation [31].imilarly, while a CTA-based membrane containing 2-NPOEnd 1,3-calix[4]arene-biscrown-6 became unstable after a fewepetitive transport experiments, membrane lifetime underontinuous operation of over a month was reported when thearrier was substituted with 1,3-bis(dodecyloxy)calix[4]arene-rown-6, which is more hydrophobic due to the addition ofwo dodecyl chains [30]. In fact, most of the leaching problemsre associated with hydrophilic carriers with an octanol–waterartitioning coefficient (log Kow) of less than 5 [30,32]. Theydrophobicity of the carrier and its water solubility can betrongly influenced by the solution chemistry of the aqueoushase. For example, �-diketone loss at a high source solutionH has been reported [79]. This is because �-diketones canissociate at high pH and therefore become much more solublen water [79]. Similarly, Argiropoulos et al. [7] reportedonsiderable Aliquat 336 loss from a PVC-based membranehen the membrane was submerged in distilled water for 10ays. However, negligible Aliquat 336 loss (within 2.5% in 10ays) was observed when a 2.5 M HCl solution was used. The
vidence reported in this study indicates a possible influencef the solution pH on the water solubility of Aliquat 336 andubsequently on the stability of the membranes using thisarrier.
rane Science 281 (2006) 7–41

Of particular note is that macrocyclic and macromoleculararriers were used in all but one of the studies reported in Table 3.urther studies will be needed to evaluate the lifetime of mem-ranes involving the use of other types of carriers, particularlyhose that are currently commercially available. While there iso universal benchmark for a satisfactory membrane lifetime,t is clear that lifetimes of PIMs will continue to be extendeds additional base polymers, carriers and plasticizers are beingnvestigated and as the operating conditions and the membraneomposition are being optimized. Nevertheless, even with theurrent expected lifetime, it has been predicted that PIMs woulde commercially useful in a range of niche applications in theear future [78].

. Extraction and transport studies in PIMs

To date, all research on PIMs has been conducted on a lab-ratory scale. For extraction studies, beaker experiments haveeen used where the PIM is immersed in a solution of the targetpecies and samples of the solution are taken at various timentervals for analysis. Some extraction experiments have alsoeen carried out in a two-compartment transport cell of a sim-lar design to that used in a typical SLM experiment (Fig. 5).owever, this type of cell is more commonly used in transport

tudies. The transport process across PIMs involves essentiallyhe exchanging of ionic species between these two compart-

ents via the membrane phase which separates them. The over-ll transport process is quite similar for all four types of carriers.phill transport of the target solute across the membrane cane achieved with a suitable ionic composition in the source andeceiving compartments. However, because of different com-lexation mechanisms involved, the transport characteristics andhoices of the ionic compositions for both the source and receiv-

Fig. 5. A typical PIM or SLM experimental set up.

Memb

6

ostnqcgimar

cfamwmtscgpc

i(sittut

fttf

Fo(Rr(

7

R

wr

cptbh

scwgc5

e20p

•

•


.1. Basic carriers

Basic carriers include high molecular mass amines, e.g. tri-n-ctylamine (TOA). In addition, some weakly basic compoundsuch as alkyl derivatives of pyridine N oxides also belong tohis group, e.g. 4-(1′-n-tridecyl)pyridine N-oxide (TDPNO). Aumber of researchers [80–82] have classified fully substituteduaternary ammonium compounds (e.g. Aliquat 336) as basicarriers though they do not have a lone electron pair at the nitro-en atom. The reason for this classification, which is also usedn the current review, is based on the similarity in the extraction

echanism involving amines and fully substituted quaternarymmonium compounds. Chemical structures of these basic car-iers are shown in Fig. 6.

In the case of fully substituted quaternary ammoniumompounds, the carrier in the PIM reacts as an anion-exchangerorming an ion-pair with a metal anion complex from thequeous phase. While in the case of amines and the weak basesentioned above, the carrier must be protonated first to reactith the metal anion complex or react directly with a protonatedetal anion complex [40]. In addition, amines can be used

o extract simple mineral and organic acids as well as smallaccharides [83–85]. With the exception of the saccharides thatan form hydrogen bonds with the carrier via their hydroxylroups, the extraction of the other solutes with amines takeslace at a source solution pH below the pKb value of thearrier.

Basic carriers have been used quite extensively in PIMs stud-es for the extraction and transport of precious and heavy metalsTable 2). The majority of these studies make use of the fact thatuch metals readily form anionic complexes with the chlorideon. While no PIMs containing basic carriers have been reportedo date for the separation of actinides such as uranium and plu-onium, it is noteworthy that these carrier reagents have beensed quite extensively in solvent extraction for the separation ofhese metals [86].

Argiropoulos et al. [7] have studied the extraction of Au(III)
rom hydrochloric acid solutions using a PVC membrane con-aining Aliquat 336 chloride (R4N+Cl−). For a membrane con-aining 50% (w/w) Aliquat 336, gold was completely extractedrom a 2.5 M HCl solution containing 100 �g/L of gold within
ig. 6. Chemical structures of several basic carriers. Note: Aliquat 336 consistsf a mixture of up to five quaternary ammonium compounds with alkyl chainsR) varying from C8 to C10 [115]. In the following discussion, we are using

4N+Cl− to represent Aliquat 336 for simplicity. It should be noted that someesearchers refer to Aliquat 336 simply as tri-n-octylmethyl ammonium chlorideTOMAC).

pgs

aftidPoct

bie3cf

rane Science 281 (2006) 7–41 19

5 h. The following reaction mechanism was proposed:

4N+Clmem− + AuCl4 aq

− � R4N+AuCl4 mem− + Claq

−

(2)

here mem and aq refer to the membrane and aqueous phases,espectively.

Transport studies were carried out on this system using a two-ompartment cell with a solution of thiourea as the receivinghase. However, it was found that thiourea was transported tohe source phase thus stopping the gold transport after 50% hadeen transported. The transport of thiourea using this membraneas been the subject of a separate study [87].

This system was aimed at the recovery of gold from electroniccrap and was tested on a sample of scrap containing 96% (w/w)opper, 0.13% (w/w) gold and a low concentration of iron. Thisas dissolved in Aqua Regia and the resulting solution diluted toive a HCl concentration of 2.5 M. It was found that all the goldould be extracted by the membrane with a selectivity factor of× 105 for gold over copper.

The transport of platinum(IV) has been studied by Fontast al. [9] using a CTA membrane containing Aliquat 336 and-NPOE as plasticizer. The source phase consisted of Pt(IV) in.5 M NaCl at pH 2 and the receiving phase was 0.5 M NaClO4 atH 2. The authors proposed the following transport mechanism:

Source phase:

PtCl6 aq2− + 2(R4N+Cl−)mem

� (R4N+)2PtCl6 mem2− + 2Claq

− (3)

Receiving phase:

(R4N+)2PtCl6 mem2− + 2ClO4 aq

−

� 2R4N+ClO4 mem− + PtCl6 aq

2− (4)

A suggestion was made that the driving force for the trans-ort of Pt(IV) was provided by the high ClO4

− concentrationradient. However, this was not verified by analyzing the sourceolution for the perchlorate ion.

The authors also conducted a comparison between the PIMnd the corresponding SLM. It was found that the flux of Pt(IV)or the PIM system exceeded that of the SLM when a certainhreshold concentration (250 mM) of the carrier was reachedn the PIM [9]. Further, in the case of the SLM there was noifference in selectivity between Pt(IV) and Pd(II) while theIM showed higher selectivity for Pt(IV). This was explainedn the basis of the higher hydrophobicity of the Pt complexompared to the Pd complex making the transport faster for Pthan for Pd in the PIM.

The extraction of Pd(II) from hydrochloric acid solutions haseen studied by Kolev et al. [47] using PVC membranes contain-ng up to 50% (w/w) Aliquat 336. It was found that the extraction
quilibrium constant (Kex) values for 40 and 50% (w/w) Aliquat36 membranes were of the same order of magnitude as for theorresponding solvent extraction systems. Pd(II) was extractedaster than Au(III) and was preferentially extracted over Au(III),

2 Memb

Cf

CmeCstmfAaTwedsttth

oWbhHcpflbcww

CdtshHCosro

ftwatpH

t

ap0tebdtbctgacs

b2Cr

T

tN[c(ffwt

•

•

fsol(Cts


d(II) and Cu(II). At high Pd(II) loadings a brown precipitateormed at the membrane surface.

Wang et al. [6] have studied the extraction of Cd(II) andu(II) from 2 M hydrochloric acid solutions using a PVC-basedembrane containing 30, 40 and 50% (w/w) Aliquat 336. The

xtraction of Cd(II) was found to be much higher than that foru(II) and the authors suggested this was because the CdCl3−

pecies was involved, which was present in higher concentra-ion than the CuCl3− species. In 2 M HCl solution Cu(II) exists

ainly as CuCl42−. An important observation the authors madeor this system was that the Kex values for 30 and 40% (w/w)liquat 336 membranes were close to each other whereas for50% (w/w) membrane the Kex value was more than double.hey suggested that the properties of the membrane had alteredhen the Aliquat 336 concentration exceeded 40% (w/w). This

ffect is the subject of another paper by Xu et al. [59] and isiscussed in Section 5.1 of this review. Wang et al. [6] have alsotudied the transport of Cd(II) and Cu(II) from HCl solutionso a receiving phase containing a much lower HCl concentra-ion using a two-compartment cell (Fig. 5) and have found thatransport is faster for Cd(II) than for Cu(II) in keeping with theigher extraction constant for Cd(II).

Similar results showing the preferential transport of Cd(II)ver Zn(II) from acid chloride solutions have been reported byalkowiak et al. [33] and Kozlowski et al. [34,88] for CTA-

ased PIMs containing TOA and the plasticizer 2-NPPE. Theyave also shown that the selectivity coefficients decrease as theCl concentration in the source phase is increased. In these

ases, ammonium or sodium acetate was used in the receivinghase. In one of the papers mentioned above [34], the initialuxes for Cd(II) and Zn(II) have been calculated for the mem-rane containing TOA and for another membrane containing thearrier TDPNO. It was shown that the initial fluxes for TDPNOere higher than for TOA whereas the selectivity coefficientsere higher for TOA.The competitive transport of Cd(II) and Pb(II) using a

TA/NPOE-based PIM containing TOMAC as carrier has beenescribed by Hayashita [89]. This work reported that the selec-ivity depended on the membrane surface area. For a smallurface area (0.8 cm2), Cd(II) was transported preferentially,owever for a large surface area (15 cm2) this was reversed.ayashita [89] explained this on the basis of accumulation ofd(II) in the small surface area membrane that consumed mostf the carrier thus inhibiting the transport of Pb(II). For the largeurface area membrane containing a much larger amount of car-ier, this did not occur and preferential transport of Pb(II) wasbserved.

Extensive research has been conducted on the use of PIMsor the extraction and transport of Cr(VI) which is extremelyoxic. Consequently, new methods for its removal from naturalaters and from industrial effluents arising from electroplating

nd mining are of considerable interest. Cr(VI) in aqueous solu-ions at pH 2 exists as HCrO4

− and as Cr2O72− at pH 8 and most

apers on this subject involve the extraction and transport of theCrO4

− species using basic extractants as discussed below.Walkowiak et al. [33] have studied the transport of Cr(VI)

hrough CTA-based membranes plasticized with 2-NPPE using

obt

rane Science 281 (2006) 7–41

series of tertiary amines and Aliquat 336 as carriers. The sourcehase for this work was HCl at pH 2 and the receiving phase.1 M NaOH. It was found the flux decreased with an increase inhe hydrocarbon chain length of the amines used with the high-st flux being obtained for TOA. This was further confirmedy Kozlowski and Walkowiak [58] who showed that the fluxecreased linearly with an increase in the log Kow values of theertiary amines. They reported that the flux for a similar PVC-ased PIM was lower than for the CTA-based membrane andould be attributed to the fact that PVC was more hydrophobichan CTA. In their study, Walkowiak et al. [33] also investi-ated the competitive transport between Cr(VI) and Cr(III) usingcidic chloride and sulfate solutions and the selectivity coeffi-ient was found to be considerably higher for chloride than forulfate.

Kozlowski and Walkowiak [64] have also proposed a methodased on the use of a PIM of composition 41% (w/w) CTA,3% (w/w) TOA and 36% (w/w) 2-NPPE for the removal ofr(VI) from acidic chloride solutions according to the following

eaction:

OAmem + Haq+ + HCrO4 aq

− � TOAH+HCrO4 mem− (5)

In another study Wionczyk et al. [70] have described theransport of Cr(VI) from a sulfuric acid solution into 0.1 MaOH using a PIM similar to that used by Walkowiak et al.

33] but containing TDPNO as the carrier. The actual membraneomposition was 20% (w/w) CTA, 70% (w/w) 2-NPPE and 10%w/w) TDPNO (0.5 M based on the plasticizer volume). It wasound that the initial flux reached a maximum for 0.3 M sul-uric acid. They proposed the following transport mechanism,here both HCrO4

− and H+ were concurrently transported tohe receiving phase:

Source phase:

TDPNOmem + HCrO4 aq− + Haq

+

� TDPNOH+HCrO4 mem− (6)

Receiving phase:

TDPNOH+HCrO4 mem− + 2OHaq

−

� TDPNOmem + CrO4 aq2− + 2H2Oaq (7)

Kozlowski et al. [34] have compared the Cr(VI) fluxesor CTA-based PIMs containing TOA and TDPNO with aource phase containing 0.5 M HCl and a receiving phasef 0.5 M sodium acetate (pH 8.0) and found a mucharger flux for TDPNO (16.01 �mol m−2 s−1) than for TOA6.62 �mol m−2 s−1). However, selectivity coefficients forr(VI) with respect to species such as Cd(II) and Zn(II) tend

o be higher for TOA. The authors have suggested that the rea-on for this was associated with the higher basicity of TOA.

A recent paper by Scindia et al. [55] examined three typesf membranes for Cr(VI) transport, an SLM, a pore-filled mem-rane (PFM) and CTA-based PIMs. Aliquat 336 was used ashe carrier and various plasticizers were studied for the PIM. In

L.D. Nghiem et al. / Journal of Memb

this study, source phases at pH 2 (HCrO4− transport) and pH

8 (Cr2O72− transport) with 1 M sodium nitrate as the receiving

phase were investigated. Coupled-diffusion with Cr(VI) beingtransported to the receiving phase and the nitrate ion to thesource phase has been proposed as the transport mechanism.The authors presented IR spectral evidence that the HCrO4

−ion dimerizes to Cr2O7

2− in the membrane phase. They sug-gested the PIM was effective for the recovery of Cr(VI) frommunicipal water and seawater.

Promising results have also been reported on the use of PIMsfor the extraction and transport of organic species. Basic carri-ers were used in most of these studies. Smith and co-workers[83,84] have investigated the facilitated non-competitive trans-port of glucose, fructose and sucrose by PIMs containing20% (w/w) CTA, 40% (w/w) plasticizer (2-NPOE or TBEP)and 40% (w/w) carrier (various tetraalkylammonium salts).Permeate flux of the disaccharide sucrose was significantlylower than that of the monosaccharide glucose. The PVC/2-NPOE/TOMAC membrane also exhibited selectivity for fruc-tose over glucose. When both the source and receiving solutionswere buffered at pH 7.3 by using 100 mM sodium phosphateand the source solution contained 300 mM saccharide, the per-meate fluxes for fructose and glucose were reported to be 10and 5.7 �mol m−2 s−1, respectively. An improved fructose fluxof 31.8 �mol m−2 s−1 was noted when using TBEP as plasti-cizer. Various anionic salts of TOMAC and other tetraalkyl-ammonium salts were also investigated as likely carriers inalternate membranes [83]. Anion exchange was used to pre-pare the bromide, tetraphenylborate, 4-toluensulfonate, bis(2-ethylhexyl)phosphonate, 4-methylbenzoate and diphenylphos-phinate salts of TOMAC [84]. The authors reported that thebromide salt outperformed the others, but still only returned afructose flux of two-thirds of that for TOMAC chloride indi-cating that the carrier size played an important role in therate of mass transfer. Of the other tetraalkylammonium saltstested, tetrahexylammonium chloride performed significantlybetter than tetrabutylammonium chloride, tridodecylmethylam-monium chloride or tetraoctylphosphonium bromide, but thefructose flux was still only one-third of that for TOMAC chlo-ride. Notably, all alternate membranes tested exhibited a sig-nificant decline in fructose flux. Some combinations of carrierand plasticizer did not form a stable plastic membrane. All theoriginal and alternate membranes tested exhibited a markedlybetter flux when TBEP was used, compared to 2-NPOE, asplasticizer, although, no explanation was offered for this obser-vation. A fixed-site jumping mechanism was proposed for thefacilitated membrane transport where the hydroxyl groups ofthe neutral carbohydrate were attracted to the “fixed” chlorideions of TOMAC [84]. This mechanism is explained in moredetail in Section 7.2. Evidence to support this mechanism camefrom experiments that showed negligible transport at a carrierconcentration of less that 20% (w/w). Transport rates increasedexponentially above this value, referred to as the percolationthreshold, as the carrier concentration was increased to a max-imum of 50%. Carrier percolation thresholds were consistentlyobserved for all the saccharides and carriers tested [83]. Thedecrease in carrier percolation threshold from 20 to 17% (w/w)

Temcn

scCcbTawaTsi

sssawfiabt

sttjmctsasaTracat

aheatasacp

rane Science 281 (2006) 7–41 21

OMAC for the transport of sucrose compared to fructose wasxplained on the basis of solute size. As sucrose is the largerolecule it does not need the same extent of “close packing” of

arrier molecules to initiate a solute jumping transport mecha-ism.

White et al. [83] have also studied the effect of the carriertructure on the transport mechanism. The cation content of thearrier was varied by changing the length of the alkyl chain from8H17 in TOMAC to C12H25 in tridodecylmethylammoniumhloride (TDMAC). The anion content was further changedy preparing various phosphate and carboxylate analogues ofOMAC. An order of magnitude decrease in glucose diffusionnd flux resulted when TDMAC was used rather than TOMAC,ith diffusion coefficients of 6 × 10−12 and 0.6 × 10−12 m2 s−1

nd flux values of 188 and 16 �mol m−2 s−1, respectively [83].hese results indicate that saccharide diffusion is faster with amaller carrier and imply that diffusion of the carrier is involvedn the rate determining transport step [83].

All carriers examined gave similar extraction constants butubstantially different diffusion coefficients for glucose anducrose [83]. For example, when using TOMAC, extraction con-tants of 0.2 and 0.5 M−1 and diffusion coefficients of 6 × 10−12

nd 0.7 × 10−12 m2 s−1 for glucose and sucrose, respectively,ere reported [83]. The order of magnitude lower diffusion coef-cients for the larger disaccharide sucrose compared to glucose,monosaccharide, is consistent with diffusion of the saccharideeing involved in the rate determining step in the membraneransport process.

The effect on saccharide diffusion by varying the size of theaccharide and of the cation and anion of the carrier indicateshat the ion-pair carrier is mobile throughout the membrane. Onhe basis of these results, the authors proposed a mobile-siteumping transport mechanism [83]. A diffusion only transport

echanism of the saccharide–ion-pair complex or of the sac-haride ion does not take into account the observed percolationhreshold at higher carrier concentration. Alternatively, a fixed-ite transport mechanism involving the carrier ion-pair fixednd the saccharide jumping, or the carrier cation fixed and theaccharide–anion complex jumping would not be significantlyffected by diffusion of either the anion or cation of the carrier.he experiments did not indicate the form in which the saccha-

ide was transported, that is, whether the saccharide jumped tocarrier anion, or if a saccharide–anion complex jumped to a

arrier cation. The authors proposed that saccharide moleculesnd carrier anions formed hydrogen bonded complexes that werehen weakly associated with the carrier cations [83].

Using the same CTA/2-NPOE/TOMAC PIMs with 20, 40,nd 40% (w/w) carrier, respectively, Munro and Smith [85]ave studied the facilitated non-competitive transport of sev-ral amino acids including l-phenylalanine, l-leucine, l-alaninend dopamine. In this investigation, the source solution con-ained 100 mM of the individual target solute and both the sourcend receiving solutions were buffered at pH 7.3 using 100 mM
odium phosphate. While negligible transport occurred in thebsence of carrier, at 40% TOMAC (w/w) the authors observedonsiderable selectivity amongst these amino acids and reportedermeate fluxes for l-phenylalanine, l-leucine, l-alanine and

2 Memb

dtaitctSoiptcrtnltsaaAotept1tflr

f(aNrfta[iot

tTptTmbffsts

6

cpsrttpD

och�aa

c1atmM

r

Fc


opamine to be 15.9, 5.2, 2.9 and 0.25 �mol m−2 s−1, respec-ively. A fixed-site jumping transport mechanism was proposeds the flux was minimal at less than 20% (w/w) carrier andncreased dramatically, in a non-linear manner, at values abovehis concentration. The authors suggested that the 20% (w/w)arrier concentration represented a percolation threshold forhis system. Flux results were compared to those for similarLM experiments using a microporous polypropylene supportf Celgard 2500TM and a liquid membrane phase of TOMACn NPOE [85]. The SLM results indicated a small, but linear l-henylalanine flux up to 40% (w/w) carrier concentration, andhen a substantial non-linear increase in the flux as the carrieroncentration was raised. The 40% (w/w) carrier concentrationepresents a percolation threshold for the SLM system. Belowhis value, the transport occurs by a carrier-diffusion mecha-ism that is favored by the lower viscosity of the polymer freeiquid membrane phase. Above the percolation threshold, theransport is significantly enhanced by contributions from a fixed-ite jumping mechanism. At pH 7.3 all the amino acids studiedre significantly ionised and fixed-site jumping of the aminocid anions occurs by attraction to the “fixed” carrier cation.lthough not mentioned by the authors, the low dopamine fluxbserved could be due to its significantly higher pKa of 8.9 [90]hat limits the amount of ionised dopamine available. Furthervidence to support the transport of ionised species came from l-henylalanine flux studies at different pH. The flux increased ashe pH was raised from 5.5 to 10 and was five times greater at pH0 than at pH 7.3. Altering the carrier counter-anion also affectedhe flux. Changing from chloride to phosphate reduced theux by 30%. The lipophilic bis(2-ethylhexyl)phosphate anionesulted in a complete absence of l-phenylalanine transport.

Matsumoto et al. [91] have investigated the potential of PIMsor commercial applications using a 20% (w/w) CTA, 40%w/w) 2-NPOE and 40% (w/w) TOMAC membrane for the sep-ration of lactic acid from a source solution containing someaCl at pH 6 to a receiving solution of 3 M NaCl. The authors

eported permeabilities of the same order of magnitude as thoseor similar SLM experiments. They proposed a carrier-diffusionransport mechanism where at pH 6 the lactic acid was ionisednd transported as the trioctylmethylammonium/lactate ion-pair
91]. TOMAC leakage from the membrane was identified as anmportant criterion for accessing the viability of further devel-pment of an in situ separation method. TOMAC is knowno be toxic to the microbe Lactobacillus rhamnosus used in
odfc

ig. 7. Chemical structures of some acidic carriers. Note: In the discussion in the tearrier, respectively.

rane Science 281 (2006) 7–41

he production of lactic acid by fermentation. Addition of freeOMAC to the fermentation mixture resulted in zero lactateroduction. A 70% decline in lactate production was consis-ently observed for PIMs containing 6, 12, 40 and 57% (v/v)OMAC. This suggested that leakage of TOMAC from theembranes reached a steady-state value independent of mem-

rane composition. Areas that were highlighted as requiringurther investigation before a commercial PIM extraction systemor lactate could be considered were the design of a membraneeparator to improve diffusion of lactate from the source solu-ion, and selectivity experiments to determine any co-extractedpecies.

.2. Acidic and chelating carriers

There are several types of compounds in solvent extractionhemistry that are classified as acidic carriers. These includehosphorus and thiophosphorus acid esters, carboxylic acids andulfonic acids. Examples of the phosphorus group are phospho-ic, phosphonic and phosphinic acids such as D2EHPA, di(2,4,4-rimethylpentyl)phosphinic acid (Cyanex 272) and di(2,4,4-rimethylpentyl)monothiophosphinic acid (Cyanex 302). Exam-les of carboxylic acids are naphthenic and versatic acids whileNNS is an example of a sulfonic acid carrier.In addition to the above acidic carriers, there is another group

f compounds that are also acidic but, in addition, show stronghelating properties. These are the �-hydroxyoximes and the �-ydroxyaryloximes (LIX reagents), quinolines (Kelex 100) and-diketones. Of course, in many circumstances the phosphoruscids, particularly D2EHPA, can also act as bidentate chelatinggents.

Of the above list of acidic carriers, some of the commercialarriers that have been used in PIMs to date are LIX® 84-I, Kelex00 and D2EHPA (Table 2). This limited use is surprising ascidic reagents have proven to be of particular importance forhe separation of a wide range of metals including transition

etals, rare earth metals and the actinides in solvent extraction.olecular structures of these carriers are shown in Fig. 7.Extraction and transport of a metal cation by an acidic car-

ier is governed by the exchange of the metal ion for protons
f the carrier. Consequently, counter-transport of protons is theriving force and is achieved by maintaining a suitable pH dif-erence between the source and receiving solutions. In addition,areful pH control in the source solution can result in good selec-
xt HR and R are used to represent the neutral and deprotonated forms of the

Memb

tr

tCt(

PaaptscatDwaisFm

eAwppo1CmPfC

P

C

C

uae1pntpr

bwa

carasHdtpoudo1

iaisatcatP

LeF

tsipmisac1t

ibmaration was achieved by pH control of the source phase with1.0 M HCl as the receiving phase. Best selectivity for Zn(II)was achieved using the source phase at pH < 2, which was inaccordance with the solvent extraction properties of D2EHPA


ivity as is the case in solvent extraction systems using acidiceagents.

Most studies with PIMs using acidic carriers have involvedhe transport of heavy metals such as Pb(II), Cd(II), Zn(II),u(II), Ag(I) and Hg(II) (Table 2) and the focus of much of

he work has been on the clean-up of process effluent streamse.g. the electroplating industry).

Salazar-Alvarez et al. [92] have studied the transport ofb(II) using a CTA-based membrane containing D2EHPA withsource phase containing 10 mM NaNO3 at a pH between 2.85nd 3.45 and a 1.5 M nitric acid receiving phase. This systemroduced a high Pb(II) flux of 3.5 �mol m−2 s−1 which was ofhe same order of magnitude as reported for the analogous SLMystem by these authors. The stoichiometry of the transportedomplex in the PIM system was reported to be PbR2·2HR. Theuthors found that if the pH of the source phase was greaterhan pH 3.45, the permeability decreased and suggested that2EHPA micelle formation was involved. Membrane stabilityas a problem in this system and Pb(II) transport decreased by

bout 30% after 5 cycles of 3 h each. However, membrane stabil-ty improved after addition of ethanol to the membrane castingolution although no ethanol was found in the membrane byTIR. The mechanism of how ethanol interacts with the CTAembrane is unclear.A highly selective PIM for the transport of Pb(II) in the pres-

nce of Cd(II) (separation factor of 140) has been reported byguilar et al. [41] using Kelex 100 as the carrier. Selectivityas achieved through pH control of the source solution. Trans-ort was studied using a source phase containing 0.1 M NaCl atH 6.0 for Pb(II) and pH 8.5 for Cd(II) and a receiving phasef 0.1 M NaCl at pH 3.0. The CTA-based membrane contained7% (w/w) CTA, 48% (w/w) 2-NPOE and 2.25 M Kelex 100.omparison of the PIM with the analogous SLM indicated aetal ion permeability about half of that for the SLM. In theb(II) case, it is suggested the transport mechanism involves theormation of a single complex in the membrane (Eq. (8)) but ford(II), two complexes are involved Eqs. (9) and (10):

baq2+ + 2HRmem � PbR2 mem + 2Haq

+ (8)

daq2+ + Claq

− + HRmem � CdRClmem + Haq+ (9)

daq2+ + Claq

− + 2HRmem � CdHR2Clmem + Haq+ (10)

The transport of Pb(II) has also been studied by Lee et al. [93]sing a synthesized lipophilic acyclic polyether dicarboxyliccid as the carrier of structure similar to the antibiotic mon-nsin. The membrane composition was 76% (w/w) 2-NPOE,8% (w/w) CTA and 6% (w/w) carrier. Using a source phase atH 4.5–5.5, 100% of the Pb(II) could be transported to a 0.1 Mitric acid receiving phase with little transport of Cu(II) and noransport of Zn(II), Co(II) and Ni(II). Negligible decrease of theermeability properties of the membrane was observed over 10epeated experiments.

Copper transport using a CTA membrane has been studiedy de Gyves et al. [43]. The other components of the membraneere LIX® 84-I (optimum mass percentage of 50–60% (w/w))

nd TBEP (optimum mass percentage of 30% (w/w)) as plasti-

rane Science 281 (2006) 7–41 23

izer. The source phase contained copper(II) sulfate or chloridet pH 4–6 in acetate buffer with 1 M H2SO4 or 1 M HCl as theeceiving phase. Maximum permeability for Cu(II) was obtainedt pH 5 with no difference between sulfate and chloride for theource phase. However, higher permeability was obtained using2SO4 in the receiving phase rather than HCl. This is most likelyue to the higher concentration of H+ in 1 M H2SO4 comparedo 1 M HCl. There was a considerable decrease in membraneermeability within only a few runs but some improvement wasbtained by adding ethanol to the casting solution. The authorssed ‘slope analysis’ in membrane extraction experiments toetermine the complex stoichiometry as CuR2·2HR. They alsobtained satisfactory separation of Cu(II) in the presence of a0-fold excess of Zn(II) and Fe(III).

A comparison between carrier-facilitated transport of Cu(II)n a CTA-based PIM and a SLM has been reported by Paugamnd Buffle [94]. The carrier used was lauric acid with TBEP act-ng as plasticizer for the PIM and as the diluent for the SLM. Theource phase was buffered to pH 6.0 using N-morpholinoethanend the receiving phase was 5 × 10−4 M cyclohexanediaminete-raacetic acid (CDTA) at pH 6.4. The authors suggested a stoi-hiometry for the extracted complex of CuR2. They found thatfter normalization by correcting for membrane porosity andortuosity, the diffusion coefficient was 22 times lower for theIM than for the SLM.

The transport of Cd(II) across a CTA-based membrane usingasalocid A (Fig. 8) as the carrier has been described by Tayebt al. [11]. The chemical structure of this carrier is shown inig. 8.

The membrane also contained 2-NPOE as plasticizer andransport in this system was compared with the analogous SLMystem. The source phase was buffered to pH 8 and the receiv-ng phase contained HCl at pH 2. The authors found the PIM (orolymeric plasticized membrane, PPM, as they termed it) wasore efficient for Cd(II) transport at high carrier concentration

n the membrane than the SLM system and had good long-termtability. Again, as others have observed [13,65,83,84], theseuthors found that transport in the PIM did not occur until thearrier concentration reached a threshold value (0.031 mg percm2 of membrane) which was close to the carrier concentra-

ion necessary to reach maximum Cd(II) flux for the SLM.The selective removal of Zn(II) from other transition metal

ons such as Co(II), Ni(II), Cu(II) and Cd(II) using PIMs haseen studied by Ulewicz et al. [42]. Their system used a CTAembrane containing D2EHPA and 2-NPOE as plasticizer. Sep-

Fig. 8. Chemical structure of Lasalocid A.

2 Memb

fata(

aeP(pfwNwau

m�etYibrtiwC

taNswhrolttslmoet

mfli11ctf

rmft

ocaoasuwg

bowhaawmatw(t3amfbCbapaZ

6

pphate (TBP), tri-n-octyl phosphine oxide (TOPO), and dibutylbutyl phosphonate (DBBP) (Fig. 9). However, this group caninclude other neutral organic reagents with a strong solvat-ing capacity due to Lewis acidic centers or hydrogen bonding


or these metal ions [95]. In another study Baba et al. [96] havelso demonstrated that Cu(II) can be preferentially transported inhe presence of Pd(II) across a PVC-based PIM using D2EHPAs the carrier and plasticizer from a source phase at pH 3–4acetate buffer) to a receiving phase of 1.0 M HCl.

Some work on membrane transport used PVC as a gellinggent in an attempt to stabilize SLMs. For example, Brombergt al. [8] used a tetrahydrofuran (THF) solution containingVC (up to 10% (w/w)), di(2-ethylhexyl)dithiophosphoric acidD2EHDTPA) and 2-NPOE to impregnate a Celgard 2400 sup-ort membrane. On evaporation of the THF, a membrane filmormed within the pores of the support phase. This membraneas used to study the transport of Ag(I), Hg(II), Zn(II), Cd(II),i(II) and Fe(III). For Ag(I) and Hg(II), thiourea in 1 M H2SO4as used in the receiving phase, while stripping of Zn(II), Cd(II)

nd Ni(II) required 10 M HCl. Fe(III) could only be strippedsing 1 M oxalic acid.

Sugiura et al. [79] have studied the transport of rare earthetal ions through CTA-based membranes containing a series of-diketones as carriers and with 2-NPOE and TBEP as plasticiz-rs. Selectivity was controlled by the pH of the source phase andhey found that Sc(III) was transported at a lower pH than both(III) and La(III) which was most likely due to the difference in

onic radii. At pH values 5.1 and 6.1, high fluxes were obtainedut there was no significant difference in transport amongst theare earth metal ions except for Sc(III). Also, the solubility ofhe �-diketones in the aqueous phase increased with increasen pH. In general, fluxes were around 0.6 �mol m−2 s−1, whichas of a similar order of magnitude to values reported for otherTA-based PIM studies as discussed previously in Section 5.2.

Sigiura has published two further papers [45,67] on the PIMransport of lanthanides using hinokitiol (HIPT) and flavonols carriers. The membranes were CTA-based and contained 2-POE and TBEP as plasticizers. HIPT is a derivative of the

even-membered ring compound cyclohepta-2,4,6-trien-1-oneith an isopropyl-group in the 4-position in the ring and aydroxo-group in the 2-position. In the first study, Sigiura [67]eported on the effect of the anion and pH of the source phasen the hydrogen ion-coupled transport. Fluxes for the series ofanthanides from Sm(III) to Lu(III) were higher than for La(III)o Nd(III). Also, the addition of perchlorate or thiocyanate ionso the source phase increased the lanthanide flux. In the secondtudy, Sugiura [45] reported on the effect of adding polyoxyethy-ene n-alkyl ethers (POEs) of different alkyl chain lengths to the

embrane mixture. In general, the POEs increased the fluxesf the lanthanides for the series from Sm(III) to Lu(III) butxtremely low fluxes were obtained for the series from La(III)o Nd(III).

In an extension to this research on HIPT using POEs in theembrane, Sugiura [97] reported on the effect on the lanthanideux when two pyrazolone derivatives were used as carriers

nstead of HIPT. The two carriers were 4-benzoyl-3-methyl--phenyl-5-pyrazolone (BMPP) and 4-trifluoroacetyl-3-methyl-
-phenyl-5-pyrazolone (TMPP) and were used under the sameonditions as HIPT. The fluxes for the lanthanides were foundo be higher for BMPP than for TMPP but there was no dif-erentiation between the various lanthanides. Sugiura [97] then
rane Science 281 (2006) 7–41

eplaced the POEs with a series of quaternary ammonium bro-ides with C12 to C6 alkyl chains and again found higher fluxes

or BMPP than for TMPP and the fluxes were much higher forhe quaternary ammonium bromides with higher MW.

Rais et al. [27] have studied the use of the chlorinatedrganoboron compound cobalt dicarbollide (bis(dicarbollyl)obaltate) in a PVC-based membrane for the extraction of 137Csnd 90Sr from high ionic strength solutions containing nitric acidr sodium hydroxide. The plasticizer used was 2-NPOE. Theuthors claimed the use of this PIM was an improvement on theolvent extraction system for these radioactive elements, whichsed nitrobenzene as the diluent, even though the PIM extractionas slower. The PVC-based PIM also contained polyethylenelycol as an effective synergistic reagent.

A very recent and novel development in PIM systems haseen described by Resina et al. [98]. These workers have devel-ped a hybrid membrane using sol–gel technology combinedith the procedure for making CTA-based membranes. Theyave carried this out by mixing a CTA matrix with polysilox-nes containing plasticizers and extractants, mixing vigorouslynd curing the mixture at 80 ◦C for 20 h. One aim of this workas to improve the stability of PIMs. Membranes made by thisethod have been studied for the transport of Zn(II), Cd(II)

nd Cu(II) with D2EHPA and D2EHDTPA as carriers. Theransport properties of the hybrid membranes were comparedith normal PIMs and with activated composite membranes

ACMs). In all cases, the source phase was 0.5 M NaCl andhe receiving phase was 0.5 M HCl. Flux values of the order of.4–6.2 �mol m−2 s−1 were found for all three membrane typeslthough slightly higher values were obtained for the hybridembrane. Using mixtures of Zn(II), Cd(II) and Cu(II), it was

ound that with D2EHPA Zn(II) was transported but the mem-rane was impermeable to Cd(II) and Cu(II). With D2EHDTPA,u(II) was extracted preferentially but could not be strippedecause of strong complexation within the membrane and thisccumulation of Cu(II) within the membrane impeded the trans-ort of Zn(II) and Cd(II). The use of a 5:1 mixture of D2EHPAnd D2EHDTPA in the membrane resulted in high transport ofn(II), no transport of Cd(II) and a limited transport of Cu(II).

.3. Neutral or solvating carriers

Most commercially available neutral or solvating carriers arehosphorus-based extraction reagents such as tri-n-butyl phos-

Fig. 9. Chemical structures of several neutral carriers.

Memb

(baawtauHli

cTaarsmnItc[tfnttbifatntttssir[wtiaoi[

ttiFsb

wtbclth

nfact3r

•

•

saeam

lwc

Cnbotwtaftoeoa


e.g. amides). Solvating reagents such as TBP and TOPO haveeen used extensively in solvent extraction for the processing ofctinides such as uranium and lanthanides as well as for ceriumnd other metal ions commonly found in low radioactivity levelastewaters [86]. As mentioned in the introduction section of

his review, the use of PIMs can be dated back to almost 40 yearsgo, when Bloch et al. [21] reported on a pioneering attempt tose PIMs containing TBP as a carrier for uranium recovery.owever, probably because the use of neutral carriers has been

argely restricted to radioactive metal ions, only a few PIM stud-es using this type of carrier have been reported since then.

The membranes used by Bloch et al. [21,69] were prepared byasting a thin film consisting of 25% (w/w) PVC and 75% (w/w)BP on paper of high wet strength or on a glass plate. Cyclohex-none was used as the solvent and the membranes were curedt 140 ◦C for 60 s. The thickness of the resulting thin film waseported to be 40 �m. By varying the nitrate concentration in theource solution, the authors [21,69] demonstrated that the per-eate flux closely correlated with the distribution ratio of uranyl

itrate between the aqueous phase and the TBP/PVC membrane.n these experiments, the receiving solution was pure water andhe source solution contained 0.02 M UO2(NO3)2 with the con-entration of HNO3 varying in the range from 2 to 8 M. Bloch69] observed that the use of an anion, which can permeate acrosshe membrane such as nitrate might be an inherent disadvantageor PIMs using solvating carriers. A rather unexpected phe-omenon occurred where the uranyl ion was transported fromhe source to the receiving solution and then transported backo the source as nitrate continued to diffuse through the mem-rane until an equal concentration of nitrate and the uranyl ionn both compartments was attained. This was explained by theact that nitrate permeated through the membrane not only inssociation with UO2

2+ but also in association with a proton inhe form of HNO3. Because of the high initial concentration ofitrate in the source solution, the uranyl ion was transported tohe receiving solution. As long as the nitrate concentration inhe receiving solution remained low, this transport process con-inued even when the uranyl ion concentration in the receivingolution was higher than in the source solution. Then, when aufficient amount of nitrate had been transported to the receiv-ng solution, back diffusion of uranium occurred until the systemeached equilibrium. To confirm this explanation, Bloch et al.21] showed that the permeation of nitrate through the membraneas insignificant when nitrate was introduced to the source solu-

ion in the form of a suitable nitrate salt such as sodium nitratenstead of nitric acid. However, the authors also reported thatsmall but discernible passage of nitrate occurred when ferricr aluminum salts were used. This was attributed to a decreasen pH in the source solution due to the hydrolysis of these salts21].

Bloch [69] also emphasized that the rate of complex forma-ion and dissociation could, in certain circumstances, governhe membrane selectivity. Using a membrane prepared with an
mmobilized chelating agent known to complex slowly withe(III), namely, thenoyl trifluoroacetone, the author demon-trated that there was a significant difference in permeabilityetween Fe(III) and Cu(II). The membrane selectivity factor
coab

rane Science 281 (2006) 7–41 25

as 200 times for Cu(II) over Fe(III), which was much higherhan expected based purely on the distribution ratio differencesetween these two metals in solvent extraction. Bloch [69]ompared this result with observations often encountered in bio-ogical systems where the differences in dissociation rates whenranslated into membrane permeabilities yielded extraordinarilyigh selectivities.

Matsuoka et al. [46] have studied the transport of uranylitrate using a CTA-based membrane containing 0.5 mL of TBPor every 0.3 g of the base polymer CTA. The authors reportedn uphill transport of uranyl nitrate from a source solutionontaining 3.5 mM of UO2

2+ and 1 M HNO3 to a receiving solu-ion containing an initially equal concentration of UO2

2+ (i.e..5 mM) and 1 M Na2CO3 [46]. The transport mechanism isepresented by Eqs. (11) and (12):

Source phase:

UO2 aq2+ + 2NO3 aq

− + 2TBPmem

� UO2(NO3)2 · 2TBPmem (11)

Receiving phase:

UO2(NO3)2 · 2TBPmem + 3Na2CO3 aq

� Na4UO2(CO3)3 aq + 2TBPmem + 2Naaq+ + 2NO3 aq

−

(12)

In contrast, they did not observe any uranyl transport when theource solution containing 3.5 mM UO2

2+ and 1 M of Na2CO3nd the receiving solution contained 1 M HNO3. This wasxplained by the formation of the bulky sodium uranyl carbon-te complex, which could not effectively be extracted into theembrane.It should also be noted here that the authors observed a severe

eakage of TBP from their membranes and the aqueous phasesere saturated with TBP in all of their experiments to prevent

arrier loss.Kusumocahyo et al. [53] have developed PIMs consisting of

TA as the base polymer, 2-NPOE as the plasticizer and twoovel solvating carriers namely octyl(phenyl)-N,N-diisobutyl-arbamoylmethyl phosphine oxide and N,N,N’,N’-tetraoctyl-3-xapentanediamide denoted as CMPO and TODGA, respec-ively, for the removal of Ce(NO3)3 from low radioactivityastewater. The source solution used in their experiments con-

ained 200 mg/L Ce(NO3)3, 0.05 M HNO3, and 2.95 M NaNO3nd the receiving solution was pure water. The distribution ratiosor Ce(III) between an aqueous phase of the same composi-ion as the source solution and the membrane containing CMPOr TODGA were reported to be 550 and 28, respectively. Byxamining the variation of the distribution ratio as a functionf the carrier concentration in the membrane, Kusumocahyo etl. [53] have successfully determined the stoichiometry of the
arrier/cerium nitrate complex to be 2:1. Complete transportf Ce(III) from the source to the receiving solution could bechieved within 60 min or more depending on the actual mem-rane composition and experimental conditions. The authors

2 Memb

atrtwa

cudttTSour

cbrsiAHtrpwtHept2cTo

6

PcttstmbTiscnai

wtcIpow

CahsephtZw

PsbIAdhZpOs

CsammC

bpTttdt“a[(twc


lso presented several sets of data relating to the diffusion ofhe cerium nitrate complex across the membrane. The transportate was reported to increase as the plasticizer or carrier concen-ration increased to certain levels. In addition, this transport rateas also found to correlate well with the membrane thickness

nd with the reciprocal of the temperature.Lamb and Nazarenko [66] have examined the influence of the

ounterion on the sorption and transport processes for Pb(II)sing CTA-based PIMs containing TOPO and 2-NPOE. Fourifferent anions (i.e. I−, SCN−, Br− and NO3

−) were investiga-ed. They reported a correlation between the Pb(II) transport andhe hydrophobicity or the hydration energy of the counterion.he permeability increased in the order NO3

− ≤ Br− <CN− ≤ I−. This also agrees well with the extraction constantf membranes containing TOPO when these anions have beensed in the solution matrix [66]. The authors, however, did noteport data relating to the transport of sodium or the counterions.

The transport of the arsenate ion (AsO43−) by a solvating

arrier has been investigated by Ballinas et al. [99] using a PIMased on CTA and dibutyl butyl phosphonate (DBBP) as the car-ier. Passive transport was almost complete after 6 h when theource solution contained 2800 mg/L AsO4

3− and the receiv-ng solution contained 2 M LiCl. The authors suggested thatsO4

3− was transported through the membrane in the form of3AsO4[DBBP]2. However, when the source solution also con-

ained 2.2 M H2SO4, the arsenate transport rate was dramaticallyeduced and almost 27 h were needed for the complete trans-ort of AsO4

3− while complete transport of the HSO4− ion

as achieved within 6 h in both cases. The authors speculatedhat this was due to the formation of the additional complex,

3AsO4[DBBP]H2SO4, in a coupling phenomenon in the pres-nce of a sufficient concentration of H2SO4 [99]. The same ex-lanation was also used to describe the uphill transport of 10% ofhe AsO4

3− observed after 5 h when the source phase contained800 mg/L AsO4

3− and 2 M H2SO4 and the receiving solutionontained an equal concentration of AsO4

3− and 2 M LiCl [99].he authors did not explain the mechanisms for the formationf the H3AsO4[DBBP]2 and H3AsO4[DBBP]H2SO4 species.

.4. Macrocyclic and macromolecular carriers

A considerable amount of the research reported to date onIMs has used macrocyclic and macromolecular carriers such asrown ethers as the extracting reagents in PIMs. This is likely dueo their high complexing selectivity for metal ions as a result ofhe presence of specifically tailored encapsulating coordinationites in their structures and their low solubility in aqueous solu-ions. In addition, the use of expensive macrocyclic reagents in

embranes is more feasible than in traditional solvent extractionecause of the considerably smaller reagent inventory required.he large number of macrocyclic carriers that has been stud-

ed are summarized in Table 4. Even though crown ethers andimilar macrocyclic ligands have traditionally been used in the
omplexation of the alkali and alkaline earth metal ions, a largeumber of PIM studies are related to the transport of radioactivend heavy metal ions. This is possibly due to the strong interestn the use of membrane separation in the treatment of nuclear
[fop

rane Science 281 (2006) 7–41

astes and contaminated waters and toxic sludge. The potentialo selectively extract and recover organic ions using macrocyclicarriers has also been demonstrated in an early PIM study [15].n PIMs using macrocyclic and macromolecular carriers, trans-ort is facilitated and the mechanism involves the co-transportf the target cation along with an associated anion and generallyater is used as the receiving phase.Lamb et al. [100] have studied the transport of Ag(I) through

TA/2-NPOE membranes containing a series of pyridino-nd bipyridino-podands (A, Table 4). These carriers haveydrophilic palmitoyl tails making them insoluble in aqueousolutions and highly soluble in the membranes thus conveyingxcellent homogeneity and stability. These podands have threeyridine nitrogen atoms in tridentate coordinating positions andave strong binding properties for the Ag(I) ion. Thus, only Ag(I)ransport from a perchlorate solution in the presence of Cd(II),n(II), Co(II), Ni(II), Pb(II) and Cu(II) to a water receiving phaseas observed.A similarly high transport selectivity for Ag(I) compared to

b(II) and Cd(II) has been observed by Kim et al. [101] for aeries of calix[4]azacrown ether derivatives (B, Table 4) immo-ilized in a CTA-based PIM plasticized with 2-NPOE and TBEP.n another study, Kim et al. [20] also investigated the transport ofg(I) in PIMs and SLMs using acyclic polyether ligands havingiamide and diamine end-groups (C, Table 4). They found againighly selective transport of Ag(I) in the presence of Cd(II),n(II), Co(II), Ni(II), Pb(II), and Cu(II). In addition, the trans-ort of Ag(I) was considerably faster in PIMs than in SLMs.nly one carrier (the amide where R = H (C, Table 4)) showed

ome transport of a metal ion (i.e. Pb(II)) other than Ag(I).A study [102] has been reported on the complexation of

o(II), Ni(II), Cu(II), Zn(II), Cd(II), Ag(I) and Pb(II) with aeries of N-benzylated macrocycles incorporating O2N2-, O3N2-nd O2N3-donor sets. Part of this work has involved testing theseacrocycles in solvent extraction and membrane systems. Oneacrocycle (D, Table 4) showed high selectivity for Ag(I) in aTA/2-NPOE-based PIM.

A comparison of Ag(I) and Cu(II) facilitated transport haseen reported by Arous et al. [14] using a series of macrocyclicolyether cryptands (E, Table 4) in CTA/2-NPOE-based PIMs.ransport experiments showed a seven-fold higher flux for

he PIMs compared to analogous SLMs. The main focus ofhe study was on the structure of the PIMs using FTIR, X-rayiffraction, differential scanning calorimetry and SEM. Theerm used by the authors for these polymer membranes wasfixed-sites membranes (FSMs)” although they are PIMsccording to the definition used in this review. Similar studies10,13] were reported on the use of 18-crown-6 ether derivativesF and G, Table 4) in CTA/2-NPOE-based membranes forhe transport of Ag(I), Cu(II), and Au(III). These studiesere aimed at providing an understanding of the factors that

ontrolled the membrane properties.The transport of Pb(II) has been studied by Aguilar et al.

32]. They synthesized a series of diazadibenzocrown ethers andound two of these (H, Table 4) had high selectivity for Pb(II)ver Cd(II) and Zn(II) using a CTA/2-NPOE membrane of com-osition (w/w) 22.0% CTA, 71.6% 2-NPOE and 6.4% carrier.


Table 4Reported PIM studies using macrocyclic and macromolecular carriers

Macrocyclic carriers Target species Basepolymer/plasticizer

References

A Ag(I) CTA/2-NPOE [100]

B Ag(I) CTA/2-NPOE andTBEP

[101]

C Ag(I) CTA/2-NPOE-TBEP

[20]

D Ag(I) CTA/2-NPOE [102]

E Ag(I), Cu(II) CTA/2-NPOE [14]

F Dibenzo-18-crown-6 (DB18C6) Cu(II) CTA/2-NPOE [10]


Table 4 (Continued )


References

G Dibenzo-18-crown-6 (DB18C6), hexathiol-18-crown-6 (HT18C6), diaza-18-crown-6(DA18C6), hexaza-18-crown-6 (HA18C6)

Ag(I), Cu(II), Au(III) CTA/2-NPOE [13]

H Pb(II) CTA/2-NPOE [32]

I Zn(II), Cd(II), Cu(II) CTA/2-NPOE [103]

J �-Cyclodextrin (�-CD) polymers Cu(II), Co(II), Ni(II), Zn(II) CTA/2-NPPE [104]

K Zn(II), Cu(II), picrate CTA/variousplasticizers

[15,56]

L Sr(II) CTA/2-NPOE [31,54]


Table 4 (Continued )


References

M Ba(II) CTA/PVC/NPOE [105]

N Ba(II) CTA/NPOE-TBEP [106]

O Cs+ CTA/2-NPOE [30,68]

P Cs+, Pb(II) CTA/2-NPOE [63]

Q Na+ CTA/2-NPOE or2-NPPE

[33]

R K+, Li+ Sol–gel [65]

3 Memb

CpIp

KoascgapTea

C(Zospsd

aCpwnC

SuIsempcne

•

•

w

Btt

cCmopo(wb

thttitc

cmttcCfi

eaTeoosf

pimm1stradt


Ulewicz et al. [103] have reported on the transport of Zn(II),d(II) and Cu(II) using CTA-based membranes containing com-lex diphosphaza-16-crown-6 derivatives as carriers (I, Table 4).n this work they studied the effect of the acidity of the sourcehase on the selectivity.

A new type of macrocyclic carriers has been reported byozlowski et al. [104]. This was based on the polymerizationf �-cyclodextrin with alkenyl(nonenyl and dodecenyl)succinicnhydride derivatives (J, Table 4). Several cyclic oligomers wereynthesized and studied in CTA/2-NPPE membranes for theompetitive transport of Cu(II), Co(II), Ni(II) and Zn(II) andave a transport order that increased from Cu(II) to Zn(II). Theuthors suggested that transport involved the formation of ion-airs between hydroxyl groups on the polymer and the metal ion.he addition of DNNS to the membrane produced a synergisticffect that was used effectively for the removal of Cr(VI), Cu(II)nd Cd(II) from industrial wastewaters and municipal sludge.

Sugiura [15] has studied the transport Zn(II) and Cu(II) usingTA membranes plasticized with p-nitrophenyl-n-heptyl ether

NPHE) that contained the carriers bathophenanthroline (forn(II)) or bathocuproine (for Cu(II)) (K, Table 4). Transportf Zn(II) against its concentration gradient was observed forolutions containing nitrate and chloride whereas Cu(II) trans-orted in the presence of chloride but not nitrate. An additionaltudy in this paper involved the transport of the picrate ion usingibenzo-18-crown-6 as carrier.

The transport of Sr(II) has been studied by Mohapatra etl. [54] using several cyclohexano-18-crown-6 derivatives inTA/2-NPOE PIMs. They found that t-BuDC18C6 (L, Table 4)rovided selective transport of Sr(II) from synthetic nuclearaste solutions containing 1 and 2 M nitric acid and a largeumber of other metal ions. At 3 M nitric acid hydrolysis ofTA was observed.

Nazarenko and Lamb [31] have studied the transport ofr(II) along with Pb(II) from synthetic nuclear waste solutionssing DC18C6 and t-BuDC18C6 in CTA/2-NPOE membranes.n their initial experiments, they added dialkyl-naphthalene-ulfonic acid to the membrane containing DC18C6 and obtainedxcellent recovery of Sr(II), however, the membrane perfor-ance deteriorated due to loss of the carrier to the aqueous

hase. This problem was solved by using the more hydrophobicarrier t-BuDC18C6. In this mixed carrier system, the mecha-ism involved counter-transport as illustrated by the followingquations and the receiving phase had to be acidic.

Source phase

Maq2+ + 2HLmem + t-BuDC18C6mem

� [M(t-BuDC18C6)L2]mem + 2Haq+ (13)

Receiving phase

[M(t-BuDC18C6)L2]mem + 2Haq+

� Maq2+ + 2HLmem + t-BuDC18C6mem (14)

here HL is DNNS.

(iu2

rane Science 281 (2006) 7–41

Nazarenko and Lamb [31] also found that the use of t-uDC18C6 on its own in the membrane gave excellent selec-

ivity for Pb(II). In this case, the mechanism was facilitatedransport and water could be used as the receiving phase.

Elshani et al. [105] synthesized some di-[N-(X)sulfonylarbamoyl]polyethers (M, Table 4) and studied them inTA/PVC/2-NPOE membranes as carriers for the alkaline earthetal ions. They also studied the solvent extraction properties

f these new carriers in chloroform. The competitive PIM trans-ort experiments showed high Ba(II) transport compared to thether alkaline earth metal ions with an order for the carrierscompound 1 > 2 > 4 > 3 (see M, Table 4)) that was associatedith the type of the X group attached to the compound back-one molecule.

Kim et al. [106] have also reported on a new carrier with highransport selectivity for Ba(II). Amongst several calix[6]arenesaving both carboxylic acid and carboxamide groups studied byhese authors, 1,3,5-tricarboxylic acid-2,4,6-tricarboxamide-p-ert-butylcalix[6]arene (N, Table 4) exhibit the highest selectiv-ty ratio for Ba(II) over Mg(II) (i.e. 40:1). Much lower selec-ivities ratio for Ba(II) over Mg(II) were reported for the otherompounds.

Levitskaia et al. [30] have employed some synthesizedalix[4]arene-crown-6 carriers (O, Table 4) in CTA/2-NPOEembranes and have reported highly efficient and selective

ransport of Cs+. Of the 17 different metal ions present in a syn-hetic nuclear wastewater sample containing a high nitric acidoncentration only K+ and UO2

2+ were transported along withs+. In a separate study, Levitskaia et al. [68] have examined the

undamental thermodynamic parameters that underpin transportn PIMs using these Cs+ selective carriers.

PIMs transport of Cs+ has also been investigated by Kimt al. [19] using calix[4]arene-crown-6, calix[4]arene-crown-7nd calix[4]arene-crown-8 in CTA/2-NPOE/TBEP membranes.hey found that the calix[4]arene-crown-6 compound gave anxcellent selectivity and transport rate for Cs+ compared to thether alkali metal ions and attributed this to the close matchingf the sizes of the crown ring cavity and the Cs+ ion. The authorsuggested this PIM system might be useful for separating Cs+

rom radioactive wastewaters [19].In another report on the use of PIMs for selective Cs+ trans-

ort, Lee et al. [63] have demonstrated that a self-assembledsoguanosine (IsoG) structure can be used in a CTA/2-NPOE

embrane to separate Cs+ from Na+ which is also present inost nuclear waste solutions. The authors have shown that IsoG(5′-(tert-butyldimethylsilyl)-2′,3′-O-isopropylene isoguano-

ine, see P, Table 4) self assembles into a decamer sandwich inhe presence of Cs+. The Cs+/IsoG 1 decamer dissociates at theeceiving phase/membrane interface in the presence of acid and,ccordingly, the transport of Cs+ has been found to increase withecreasing pH of the receiving phase. In the absence of Cs+ inhe solution, the membrane produced high selectivity for Ba(II).

Walkowiak et al. [33] have studied a series of sym-
alkyl)dibenzo-16-crown-5-oxyacetic acid carriers (Q, Table 4)n the competitive transport of alkali metal ions. The membranessed were CTA-based and were plasticized by either 2-NPOE or-NPPE. The source phase contained 0.2 M of the alkali metal

Memb

ipmflean

hoSptTLtae

actoptr1Tcrpam

rapccZbasst

7

7

tbttmb

SffaorfTaome

tcfdaatttostdqottfisasfistpbtistdopwc

pca


on studied and the receiving phase was 0.1 M HCl. Due to theresence of the acetic acid functionality in these carriers, theechanism involved proton-coupled metal ion transport and theux depended on the chain length of the alkyl chain. These work-rs found that maximum flux was obtained for the C9 alkyl chainnd that this carrier was very selective for Na+ with practicallyo transport of the other alkali metal ions.

An important study of alkali metal ion transport using PIMsas been reported by Lacan et al. [65]. They used a hybridrganic–inorganic membrane derived from a sol–gel process.uch heteropolysiloxane membranes are relatively easy to pre-are and can contain a covalently bound carrier. In this paperhe carrier was 4′-N-butylcarboxamidobenzo-15-crown-5 (R,able 4) and the membrane gave good selectivity for Na+ overi+ and high diffusion rates. The presence of the picrate ion in

he membrane was found to increase the transport rate and theuthors suggested that the transport mechanism involved cationxchange between the picrate ion and carrier sites.

The extraction of the organic picrate ion was studied inn early work by Sugiura [15]. In this study, the membranesonsisted of PVC as the base polymer, NPHE as the plas-icizer and DC18C6 as the carrier. A concentration gradientf potassium ions across the membrane was used to trans-ort picrate against its concentration gradient with an ini-ial flux of 5.7 × 10−2 �mol m−2 s−1. Optimum transport waseported when both the source and receiving solutions contained0−4 M potassium picrate and buffered at pH 8.3 using 0.01 Mris–H2SO4. Additionally, the source and receiving solutionsontained 0.5 M potassium sulfate and 0.5 M lithium sulfate,espectively. The author postulated that an ion-pair was trans-orted through the membrane and in this case potassium acteds a counter-ion for the transport of picrate anion across theembrane [15].A most recent paper [107] that was in press just as this

eview was in the final stages of completion describes thepplication of PIMs containing some macrocyclic metal com-lexes and various plasticizers for the transport of anions. Thearriers were essentially bis(pyridylmethyl)amine compoundsoordinated to the transition metals ions Fe(III), Cu(II) andn(II). One carrier was based on a resorcinarene bonded to fouris(pyridylmethyl)amines. Because of the presence of the met-ls in the carrier complex, some deviation from the ‘Hofmeister’equence of anions was observed. The PIMs showed excellentelectivity for the ReO4

− ion and the suggestion was made thathey might also show high selectivity for the TcO4

− ion [107].

. Transport mechanisms

.1. Interfacial transport mechanisms

Both SLMs and PIMs involve the selective transport of aarget solute from one aqueous solution to another via the mem-rane that separates them as can be seen in Fig. 5 [1]. This overall
ransport consists of two processes, namely the transfer of thearget solute across the two interfaces and diffusion through the
embrane. The former process is similar for both types of mem-ranes, however, because PIMs are distinctively different from

spib

rane Science 281 (2006) 7–41 31

LMs in their composition and morphology, the actual bulk dif-usion mechanisms within the membrane phase can be quite dif-erent. Consequently, the overall transport mechanisms of SLMsnd PIMs are not identical. Nevertheless, fundamental findingsbtained from SLM studies, which have been comprehensivelyeviewed by several authors [1,2,108], can be particularly use-ul to elucidate the transport mechanisms observed with PIMs.his section discusses the interfacial transport mechanisms withfocus on the chemistry of the aqueous phase and phenomenabserved at the membrane–aqueous interface. The bulk diffusionechanisms occurring within the membrane phase are delin-

ated in the next section (Section 7.2).Danesi [108] describes the permeation of the target solute

hrough an SLM as a single-stage extraction in simultaneousombination with a back extraction stage. The transport, there-ore, occurs under non-equilibrium conditions. De Gyves ande San Miguel [2] provide a more detailed analysis taking intoccount the diffusion of the target solute through a stagnant layert the membrane/aqueous solution interface and also consider forhe transport of co- or counter-ions. However, these authors notehat when suitable hydrodynamic conditions are maintained nearhe membrane/aqueous solution interface by constant agitationr tangential flow, the diffusion process through this aqueoustagnant layer is relatively fast and can be ignored. Consequently,he interfacial transport description proposed by de Gyves ande San Miguel [2] for a typical membrane extraction system isuite similar to that described by Danesi [108]. On the basisf these reviews, the three major steps which characterize theransport of a target solute from the source to the receiving solu-ion in PIMs are schematically illustrated in Figs. 10a–d. Thesegures depict uphill metal transport which will occur in the finaltages of the separation process. In the first step, the target solutefter diffusing through the aqueous stagnant layer at the sourceolution/membrane interface, reacts with the carrier at this inter-ace to form a complex, which is then transported across thisnterface and replaced by another molecule of the carrier. In theecond step, the complex diffuses across the membrane towardhe receiving solution. The detailed nature of this bulk diffusionrocess is discussed in the next section. Finally, at the mem-rane/receiving solution interface, the complex dissociates andhe target solute is released into the receiving solution, whichs essentially the reverse of the process occurring at the sourceolution/membrane interface. While Figs. 10a–d are represen-ative of most PIM systems encountered in the literature, theyo not provide any information about the speciation processesccurring in the aqueous phases. Consequently, the aqueoushase metal ion concentration is its total analytical concentrationhich is the sum of the concentrations of all chemical species

ontaining this metal ion.It is obvious that the distribution ratio (often referred to as the

artition constant or coefficient, Kp) of the target solute/carrieromplex between the organic phase of the membrane and thequeous solution, Ks

p (superscript ‘s’ representing the source
olution), must be as high as possible to favor the extractionrocess. In contrast, Kr
p (superscript ‘r’ representing the receiv-ng solution) at the receiving site must be sufficiently low to favorack extraction of the target solute from the membrane phase.


Fig. 10. Schematic description of coupled transport of a positively charged (M+) or negatively charged (M−) species through a PIM. C represents the carrier and X isan aqueous soluble coupled-transport ion. [M+], [M−], [X−] and [X+] represent the total analytical concentrations of the respective solute in the bulk aqueous phases.( transpw urrena

CtadgittttsccsimMtttht

t1Ncw

tbiicditcdbc

a) The target solute is a cation and is concurrently transported with a coupled-ith a coupled-transport cation; (c) the target solute is an anion and is counter-c

nd is concurrently transported with a couple transport cation.

onsequently, within the membrane phase, there is a concen-ration gradient of the target solute/carrier complex or ion-paircting as a driving force for its transport across the membrane,espite the fact that the total analytical concentration of the tar-et solute in the source solution can be substantially lower thann the receiving solution. In other words, uphill transport onlyakes place with respect to the total analytical concentration ofhe solute, while in the membrane phase it is actually downhillransport regarding the actual chemical species diffusing acrosshe membrane. In practice, such a difference in Kp between theource and receiving solutions can be maintained with a suitablehemical composition of these solutions. For example, a strongomplexing reagent is often used in the receiving solution totrip the target ion from the membrane. Amongst several stud-es in which observations in support of this interfacial transport

echanism have been reported, those by Bloch et al. [21] andatsuoka et al. [46] can be used as examples. The former argued

hat permeate flow continues until Ksp · Cs

i = Krp · Cr

i (where Ci is
he concentration of the target solute in the aqueous solution),hat is until the concentration gradients within the membraneave vanished. This is also consistent with the work of the lat-er authors [46], who examined the transport of the uranyl ion
atbe

ort anion; (b) the target solute is a cation and is counter-currently transportedtly transported with a coupled-transport anion; (d) the target solute is an anion

hrough a CTA/TPB membrane and reported a Ksp/Kr

p ratio of04 between the source and receiving solutions containing 1 MaNO3 and 1 M Na2CO3, respectively. Consequently, a nearly

omplete uphill transport of the uranyl ion has been observedith this system.Another driving force for the uphill transport phenomenon is

he potential gradient of a coupled-transport ion across the mem-rane. In a typical PIM process, the target solute is transportedn association with this ion to maintain electroneutrality. Thiss known as the coupled-transport phenomenon, which can beounter-transport (Figs. 10b and c) or co-transport (Figs. 10a and), depending on the transport direction of the coupled-transporton with respect to that of the target solute. Typical examples forhis can be seen by considering PIM systems using acidic orhelating carriers [41,42,92]. In such cases, the potential gra-ient of protons maintained by adjusting the solution pH cane seen as the driving force for the uphill transport of a metalation across the membrane. In fact, studies involving the use of
cidic and chelating carriers have consistently revealed a charac-eristic correlation between permeability and the pH differenceetween the source and receiving solutions [41,42,92]. How-ver, it should be noted that the distribution ratio of the target

Memb

sipka

dtmottmtvdtanah

7

pibtlmHagsasFria(rmlbb

“picmecms

SvdtdsbbpPbtsCciprcb

mctbtatifc[pCIrpaw[sffd

aombbapt


olute between the aqueous solution and the membrane phases also related to the solution pH. While Kp values at variousH values are not usually reported in PIM literature, it is wellnown in solvent extraction that the distribution ratios for acidicnd chelating extractants are strongly pH-dependent.

In practice, these two driving forces outlined above cannot beistinguished. This is partly due to the complexity of the specia-ion involved at both sides of the membrane as well as within the

embrane itself. In essence, they are usually both integral partsf a complex interfacial transport mechanism. One emphasizeshe distribution ratio (Kp) difference and the other emphasizeshe potential gradient of the coupled-transport ion across the

embrane. It is probably more convenient to use the formero describe the interfacial transport process in PIMs using sol-ating carriers such as TBP. The latter presents a more suitableescription for membranes carrying fixed charged groups withinhe polymer matrix, in other words, PIMs prepared with basic,cidic or chelating carriers. Some authors have related such Don-an behavior to a characteristic correlation between permeabilitynd the dielectric constant of the carriers and plasticizers, whichas been widely observed by several researchers [9,58].

.2. Bulk transport mechanisms

As discussed in the previous section, facilitated trans-ort across a membrane from a source to a receiving phasenvolves also diffusion of the carrier/target complex through theulk membrane in addition to transport across the two solu-ion/membrane interfaces [1,2,78,109]. In the case of a bulkiquid membrane, the carrier, which is assumed to be able to

ove freely within the membrane, plays the role of a shuttle [78].owever, facilitated transport can also occur in ion-exchange

nd other types of membranes, in which the reactive functionalroup (or carrier) is covalently bound to the polymeric backbonetructure [55]. In this case, the carrier is immobilized and it isssumed that the bulk diffusion of the target solute takes place viauccessive relocations from one reactive site to another [78,109].or a typical PIM, as discussed in Section 5.1, although the car-ier is not covalently bound to the base polymer, the membranes essentially a quasi-solid homogeneous thin film and it is nottrue liquid phase [110]. Because, the carriers are often bulky

Figs. 6–9 and Table 4), their mobilities in PIMs are much moreestricted compared to SLMs. Consequently, although the actualechanisms are still a subject of stimulating discussion in the

iterature, the bulk diffusion processes in PIMs are thought toe different from those in SLMs and other types of liquid mem-ranes [78,109].

In a pioneering work, Cussler et al. [109] proposed thechained carrier” theory to describe the facilitated transportrocess in a solid membrane where mobility of the carriers restricted. The authors developed a mathematical model toompare the mobile carrier diffusion and the chained carrierechanisms. Their model explicitly indicates that there are sev-

ral circumstances where the permeate flux of an immobilizedarrier membrane can be comparable to that of a mobile carrierembrane. This is consistent with the results reported in several

tudies [9,11,55] where the permeate fluxes between PIMs and

ctpm

rane Science 281 (2006) 7–41 33

LMs have been found to be comparable, although such obser-ations are uncommon and it is unclear whether the conditionsescribed by Cussler et al. [109] and those of the studies men-ioned above were indeed similar. More importantly, the modelemonstrates that membranes with immobilized carriers mayhow a percolation threshold, i.e. the carrier concentration muste sufficiently high so that a continuous chain across the mem-ranes can be formed. It is noteworthy that this concept of aercolation threshold has later become the foundation for otherIM studies on the bulk diffusion mechanism within the mem-rane phase [13,65,83,84,110,111]. The model also indicateshat facilitated transport can occur only when the carriers them-elves have some local mobility. On the basis of their model,ussler et al. [109] have postulated that the apparent diffusionoefficient (or the apparent rate of transport of the target solute)n immobilized carrier membranes does not reflect the diffusionrocess but rather the chemical kinetics of the complexationeaction. This is consistent with the theory by Bloch [69] dis-ussed earlier in Section 6.3 that selectivity in PIMs may alsoe governed by the rate of complex formation and dissociation.

Although the model of Cussler et al. [109] provides a clearechanistic insight for the understanding of the diffusion pro-

ess in PIMs, later experimental results appear to deviate fromhis model [83,110,111]. In fact, as has been carefully discussedy Cussler et al. [109], a limitation of their model is the assump-ion that free uncomplexed solute cannot enter the membranend the carrier sites must be within reach of one another so thathe transfer of the target solute can take place. Plate et al. [110]nvestigated the transport of Co(II) using TOA as the carrier andound that the percolation threshold also depended on the initialoncentration of the target solute. In another study, White et al.83] reported a higher percolation threshold for fructose as com-ared to the disaccharide sucrose in a PIM investigation usingTA and TOA as the base polymer and carrier, respectively.

t was concluded that the larger disaccharide species did notequire the carrier molecules to be as close together for trans-ort to occur as in the case of smaller monosaccharides. Theuthors also proposed an extended bulk diffusion model, whichas essentially an improvement of the model by Cussler et al.

109]. In addition to the requirement for locally mobile carrierpecies, this new model assumed that the target solute can jumprom one carrier to another [83]. However, physical conditionsor this model were not clearly postulated and a mathematicalerivation was not provided.

Unlike the mobile carrier-diffusion mechanism, whichssumes that complexation formation and dissociation occurnly at the solution/membrane interface, the fixed-site jumpingechanism explicitly includes the complexation reaction

etween the carrier and target solute as an integral part of theulk membrane transport. Both of these mechanisms followFickian diffusion pattern and therefore the overall transport

rocess is similar. However, knowledge of the actual bulkransport mechanism kinetics underlines the significance of the
omplexation kinetics, which in addition to the complexationhermodynamics, may also be a crucial factor governing bothermeability and selectivity in PIMs. Nevertheless, experi-ental results to support this premise remain limited in the

3 Memb

PiTa[coitbb

8

dffcaics

hlKtto(

l

C

wcc

[c(L

l

P

l

C

w

m[dspimf

(

(

icd

A

[

J

wJcPcs

l


IM literature. This is partly due to experimental difficultiesn distinguishing between the two mechanisms outlined above.o date, the fixed-site jumping mechanism was postulatedlmost exclusively on the basis of a percolation threshold13,65,83,84]. However, an increase in the carrier concentrationan lead to a variation in the membrane morphology as pointedut by some researchers [6,59,60], which may ultimatelynfluence the nature of the diffusion process. In fact, becausehe carrier is not covalently bound to the base polymer, it maye assumed that the actual diffusion mechanism is intermediateetween mobile carrier diffusion and fixed-site jumping.

. Mathematical modeling

The development of mathematical models to adequatelyescribe the extraction and transport processes is fundamentalor PIM investigations. Mathematical modeling is a vital toolor an in-depth understanding of the relevant physicochemi-al and transport processes, determining their thermodynamicnd kinetic constants as well as optimizing the correspond-ng membrane separation systems (e.g. membrane and solutionomposition and system dimensions). Not surprisingly, a con-iderable number of PIM studies have addressed this subject.

The stoichiometry of the transported complex across PIMsas been determined in a number of studies [43,53,92] fol-owing the standard approach used in solvent extraction.usumocahyo et al. [53] fitted Eq. (15a) to experimental dis-

ribution data to determine the stoichiometry of complexa-ion between Ce(NO3)3 and the carrier N,N,N′,N′-tetraoctyl-3-xapentanediamide (TODGA) immobilized in CTA-based PIMsEq. (15b)):

og Kp = log Kex + a log[L]mem + 3 log[NO3−]aq (15a)

eaq3+ + aLmem + 3NO3 aq

− � {CeLa(NO3)3}mem (15b)

here Kex is the extraction equilibrium constant, Kp the partitiononstant (Kp = [CeLa(NO3

−)3]mem/[Ce3+]) and L represents thearrier.

Using a similar approach, Salazar-Alvarez et al. (Eq. (16a))92] and de Gyves et al. (Eq. (17a)) [43] determined the stoi-hiometry of the extraction of Pb(II) (Eq. (16b)) and Cu(II) (Eq.17b)) into CTA PIMs incorporating the carriers D2EHPA orIX® 84-I, respectively:

og Kp = log Kex + 12 (2 + a) log[(HL)2]mem + 2pH (16a)

baq2+ + 1

2 (2 + a){(HL)2}mem

� {PbL2aHL}mem + 2Haq+ (16b)

og Kp = log Kex + a log[(HL)2]mem + 2pH (17a)

uaq2+ + a{(HL)2}mem

� {CuL2(2a − 2)HL}mem + 2Haq+ (17b)

here (HL)2 is the dimeric form of D2EHPA or LIX® 84-I.

tAdS

rane Science 281 (2006) 7–41

There have been a number of attempts to mathe-atically model PIM extraction and transport behavior

6,30,41,43,47,60,66,68,87,92,94]. The mathematical modelseveloped can be divided into two main groups according to theimplifying assumptions used. The first group incorporates sim-le steady-state transport (permeation) models involving metalons (e.g. Cs+, Rb+, K+, Na+, Cu(II), Cd(II) and Pb(II)). The

ajority of these models are based on the following six simpli-ying assumptions [30,41,43,66,68,92,94]:

(i) Interfacial and bulk phase reactions are very fast leading toinstantaneous establishment of chemical equilibria in thesystem studied.

(ii) The metal concentration in the membrane phase is negli-gible with regard to the carrier concentration in the mem-brane, thus resulting in constant free carrier concentrationwithin the membrane.

iii) The concentration of the metal–carrier complex at the mem-brane/receiving phase interface is negligible relative to itsconcentration at the membrane/source phase interface. Thisassumption will be valid if the concentration of the metalin the receiving phase remains virtually zero.

(iv) Mass transport within the membrane is the result of Fick-ian diffusion only and the concentration gradient of themetal–carrier complex is linear.

(v) The diffusion in the aqueous stagnant layer at the mem-brane/source phase interface is either much faster than thediffusion of the metal–carrier complex across the mem-brane or is characterized by a linear concentration gradient.

vi) Both the source and receiving phases are ideally mixed.

According to assumptions (ii)–(vi) the depletion of the metalon in the source phase can be described by Eq. (18) with initialondition (18a) while its transport across the membrane can beescribed by Eq. (19):

Js = −Vsd[M]s

dt(18)

M]s = [M]0s at t = 0 (18a)

s = Ps[M]s (19)

here A is the membrane area, V the phase volume, t the time,the flux, and [M] and [M]0 refer to the transient and initial

oncentrations of the metal ion in the source phase, respectively.is the cation permeability coefficient which can be used to

haracterize PIM transport efficiency. Subscript ‘s’ refers to theource phase.

The simultaneous solution of Eqs. (18) and (19) gives

n

([M]s

[M]0s

)= −

(A

Vs

)Pst (20)

Lamb and Nazarenko [66] applied Eq. (20) to describe Pb(II)
ransport through a CTA PIM incorporating TOPO as the carrier.fter substituting Ps with Eq. (21) they were able to estimate theiffusion coefficients of the PbX2·nTOPO complexes (X = I−,CN−, Br− or NO3
−) in the PIMs studied. All values were of

Memb

t

P

wδ

[

P

tlgAsr(eelas

C

abcKu

C

C

P

fTaiCerfPsth

omtPffia

ds

(Pf

l

wr

ipaov8gm

omei

M

wp

eiw

P

wt

(s

−

w

nb

adt


he order of 10−12 m2 s−1:

s =(

D

δ

)Kp (21)

here D is the apparent diffusion coefficient of PbX2·nTOPO,is the membrane thickness and Kp = [PbX2·nTOPO]PIM/

PbX2]s.The plot of log P versus log[TOPO]PIM revealed that n in the

bX2·nTOPO complexes was at least 2.Paugam and Buffle [94] compared the transport of Cu(II)

hrough SLMs and PIMs containing various concentrations ofauric acid as the carrier. The polymeric supports used were Cel-ard 2500 polypropylene for the SLMs and CTA for the PIMs.ssuming a linear concentration gradient in the source phase

tagnant layer and 1:2 complexation between Cu(II) and lau-ic acid (Eq. (22)), the authors derived an equation for the fluxEq. (19)) which was subsequently fitted to both SLM and PIMxperimental flux data calculated by Eq. (18). The PIM appar-nt diffusion coefficient was found to be an order of magnitudeower than the SLM apparent diffusion coefficient under thessumption that the PIM and SLM extraction constants wereimilar:

uaq2+ + 2HLmem � {CuL2}mem + 2Haq

+ (22)

Aguilar et al. [41] described mathematically the extractionnd transport of Cd(II) and Pb(II) in SLM and PIM systemsased on Eq. (20). Solvent extraction data regarding the stoi-hiometry of the corresponding extraction equilibria involvingelex 100 in kerosene as the carrier (Eqs. (23)–(25)) were alsosed [41]:

daq2+ + Claq

− + HLmem � {CdL+Cl−}mem + Haq+ (23)

daq2++Claq

− + 2HLmem�{CdHL2+Cl−}mem + Haq

+ (24)

baq2+ + 2HLmem � {PbL2}mem + 2Haq

+ (25)

Polyvinylidene fluoride (PVDF) was the polymeric supportor the SLMs studied while CTA was used for the PIMs [41].he model takes into account the transport of the metal speciescross the aqueous stagnant layer at the membrane/source phasenterface. Expressions for the permeability coefficients for bothd(II) and Pb(II) were derived and the model was fitted toxperimental data to determine the numerical values of the cor-esponding apparent membrane diffusion coefficients. The dif-usion coefficients of the Cd(II) and Pb(II) complexes within theIMs studied were determined to be of the order of 10−12 m2 s−1

imilar to those obtained by Lamb and Nazarenko [66] whilehe corresponding SLM values were three orders of magnitudeigher.

Salazar-Alvarez et al. [92] followed the modeling approachf Aguilar et al. [41] to describe the transport characteristics (i.e.,aximum flux, thickness and resistance of the stagnant layers,

he apparent activation energy for the facilitated diffusion of
b(II) across the membrane) of a CTA-based PIM system usedor the extraction of Pb(II) with D2EHPA. The diffusion coef-cient found for the transport of the Pb(II)–D2EHPA complexcross the PIMs studied (1.5 × 10−11 m2 s−1) was similar to the
[cew

rane Science 281 (2006) 7–41 35

iffusion coefficient values reported earlier for other Pb(II)/PIMystems [41,92].

De Gyves et al. [43] evaluated both the source phase (Eq.20)) and receiving phase (Eq. (26)) permeability of CTA-basedIMs incorporating LIX® 84-I during the transport of Cu(II)rom chloride and sulfate containing source phases:

n

{1 − [M]r

[M]∞r

}= −

(A

Vr

)Prt (26)

here [M]∞r is the equilibrium metal ion concentration in theeceiving phase at t → ∞.

The results obtained showed evidence of metal accumulationn the membrane in the case of the chloride containing sourcehase. An expression for the permeability coefficient based onssumptions (i)–(vi) was derived and fitted to experimental databtained for PIMs with different LIX® 84-I concentrations. Thealue of the apparent diffusion coefficient of the copper LIX®

4-I complex was estimated as 10−12.2 m2 s−1. This value is inood agreement with those reported for the other PIM systemsentioned earlier [43,66,92].Levitskaia et al. [30,68] developed a model for the transport

f monovalent metal ions, denoted here as M(I), across CTAembranes incorporating calixarene-crown ethers (B) as carri-

rs which formed 1:1 metal–carrier complexes (MB+, Eq. (27))n the membrane:

aq+ + Xaq

− + Bmem � {MB+}mem + Xmem− (27)

here X− is the anion in the metal salt dissolved in the sourcehase.

When a single salt (MX) was present in the source phase, thequation for the transient source phase metal concentration wasdentical to Eq. (20) where the cation permeability coefficientas defined by Eq. (28):

f = (K′ex[B]mem)1/2 D

δ(28)

here K′ex is the formal equilibrium constant of the complexa-

ion reaction described by Eq. (27).The equation for the transient metal concentration derived

Eq. (29)) when the anion concentration was constant differedubstantially from Eq. (20) [66]:

[M+]1/2s = −([M+]

0s )

1/2 + 1

2

(A

Vs

)Ps[X

−]1/2s t (29)

here Ps is defined by Eq. (28).Eq. (29) is also valid when more than one cation is simulta-

eously transported across the membrane with one of these ionseing preferentially transported.

The validity of the model outlined above (i.e. Eqs. (20), (28)nd (29)) was successfully verified using experimental transportata [68]. This model for the case of constant anion concentra-ion (Eq. (29)) was subsequently applied by the same authors
30] to characterize the performance of PIMs in the processing ofomplex acidic nuclear wastes when different calixarene-crownther carriers were used and the composition of the source phaseas varied.

3 Membrane Science 281 (2006) 7–41

owccct

dwainmPiaobpsiapiotedrcfittss(sdptptp

{

wc

[cF

V

Fo

tt

t((

[(

V

[

wc{

d

[

na


The steady-state models outlined above are generally validnly for the initial stages of most membrane transport processeshen the ratio between the free carrier concentration and the

oncentration of the metal–carrier complex in the PIM is suffi-iently high. These models are, however, not applicable in theases of slow interfacial kinetics or when metal accumulation inhe membrane takes place.

A more general modeling approach overcoming most of theisadvantages of the steady-state PIM models mentioned aboveas proposed by Kolev et al. [60]. The models based on this

pproach and developed by the same research group [6,47,60,87]nvolve more complex mathematics requiring the application ofumerical techniques. They were successfully used to describeathematically the extraction and transport of Au(III) [60],d(II) [47], Cd(II) [6] and Cu(II) [6] across PVC membranes

ncorporating Aliquat 336 chloride as the carrier. These modelsre based on assumptions (iv)–(vi) regarding the Fickian naturef the membrane mass transfer and the rapid mass transport inoth the bulk and the stagnant layers of the source and receivinghases. The latter assumption is justified in the case of efficienttirring when the bulk and interface concentrations are almostdentical. These models, unlike the steady-state models outlinedbove, can be used to describe membrane extraction and trans-ort involving metal accumulation in the membrane and slownterfacial complexation reactions. In those cases the diffusionf both the metal–carrier complex and the free complex withinhe membrane must be taken into account. To simplify the math-matical description in such cases, it has been assumed that theiffusion coefficients of the complexed and uncomplexed car-ier are equal. This means that the total concentration of thearrier in the membrane is uniform and constant and it is suf-cient to describe the diffusion of either the uncomplexed or

he complexed carrier only. These models have been appliedo PIM systems operating in ‘extraction mode’ only, i.e. bothides of the membrane are in contact with the same sourceolution only. If the two compartments of the transport cellFig. 5) are identical, the transport cell can be considered asymmetrical with respect to the membrane. This simplifies theevelopment of the mathematical model as only half the trans-ort cell (Fig. 11), consisting of one compartment and half ofhe membrane adjacent to it, must be considered. If the sourcehase contains a negatively charged metal chloride complexhe following interfacial ion-pair formation process can takelace:

MClmn−}aq + nRClmem

k+1�k−1

{MClmRn}mem + nClaq− (30)

here k+1 and k−1 are the forward and the backward kinetic rateonstants and RCl represents the Aliquat 336 chloride.

Due to steric and viscosity related restrictions n is usually 16,47,60]. If the chloride ion is in excess, which is usually thease, the transient interfacial concentration of MClmn− (x = δ,ig. 11) can be described by Eq. (31):

d[MClmn−]aq

dt= −A{k+1[RCl]nmem[MClm

n−]aq

− k−1[MClmRn]mem} (31)

m(

i

ig. 11. Schematic diagram of the membrane extraction system where only halff the transport cell is considered (δ is half the membrane thickness).

This concentration is identical to the bulk concentration ofhis chemical species under the assumption of ideal mixing inhe aqueous phases.

The mass transfer of the free Aliquat 336 chloride withinhe membrane can be described by the Fick’s second law (Eq.32)) with initial and boundary conditions expressed by Eqs.32a)–(32d), respectively:

∂[RCl]mem

∂t= D

∂2[RCl]mem

∂x2 (32)

RCl]mem(0, x) = [RCl]0mem (32a)

∂[RCl]mem

∂x

)x=0

= 0 (32b)

d[MClmn−]

dt= −AD

n

(∂[RCl]mem

∂x

)x=δ

(32c)

MClmn−]x=0 = [MClm

n−]0

(32d)

here x is the axial distance and D is the diffusion coeffi-ient of RClmem which was assumed to be equal to that ofMClmRn}mem.

The concentration of the ion-pair {MClmRn}mem can beetermined as

MClmRn]mem = [RCl]0mem − [RCl]mem

n(33)

Eqs. (31)–(33) cannot be solved analytically because of theon-linearity of Eq. (31) with respect to the dependent vari-bles [RCl]mem and [MClmn−]aq. The implicit finite-difference
ethod [60] was used for the simultaneous solution of Eqs.
31)–(33).After converting Eqs. (31) and (32) at n = 1 into equations

n dimensionless quantities and variables it was shown that the

Memb

ot

α

β

wp

dkpvc

ttaPttT1stccdlcc

‘

Ft1

cwtatec3

D

w3

9

iiopfwca(actm


verall membrane extraction process can be characterized bywo dimensionless groups (Eqs. (34) and (35)) [60]:

= k−1δ

D

δ

L

[RCl]0mem

[MClm−]0aq

(34)

= Kex[MClm−]

0aq (35)

here L = V/A is the characteristic length of the aqueous (source)hase.

The numerical values of these two dimensionless groupsetermine whether the extraction process is under diffusion,inetic or mixed diffusion-kinetic control (Fig. 12) [60]. It isossible to estimate whether for a particular extraction system,ariations in the value of the diffusion coefficient and the kineticonstants will influence its performance.

The model outlined above was fitted to experimental extrac-ion data using a simplex optimization algorithm to determinehe values of the kinetic rate constants, the extraction constantnd the diffusion coefficient in the extraction of Au(III) [60],d(II) [47] and Cd(II) [6] from their hydrochloric acid solu-

ions into PVC/Aliquat 336 chloride membranes of differenthickness and containing 20–50% (w/w) Aliquat 336 chloride.he diffusion coefficient values obtained were of the order of0−13 m2 s−1, similar to the values obtained by fitting the steady-tate transport models outlined above to the initial stages of PIMransport experiments [41,43,66,92]. As expected, the diffusionoefficients increased with increasing Aliquat 336 chloride con-entration in the PIMs studied. An order of magnitude higheriffusion coefficient values for 50% membranes compared toower concentration membranes indicated possible structural
hanges in the membrane occurring at Aliquat 336 chloride con-entrations higher than 40%.
The modeling approach outlined above was also applied to thenon-facilitated’ transport of thiourea across PVC/Aliquat 336

ig. 12. Influence of the kinetic and diffusion parameters on the character ofhe extraction process. (Reproduced with permission from Ref. [60]. Copyright997 Elsevier Science.)

pOirtrespi

vgttcowcchtirbs

rane Science 281 (2006) 7–41 37

hloride PIMs separating acidic source and receiving phaseshere thiourea was partially protonated [87]. It was assumed

hat molecular thiourea was only transported across the PIMsnd that its interfacial PIM concentration was determined byhe corresponding partition constant. By fitting the model to thexperimental PIM extraction data an empirical expression foralculating the diffusion coefficient of thiourea in PVC/Aliquat36 chloride membranes was derived (Eq. (36)) [87]:

= 2.43 × 10−18C3.627 (36)

here C is the percentage membrane concentration of Aliquat36 chloride. Eq. (36) is valid for C between 30 and 50% (w/w).

. The future of PIM research

One of the main goals of this review is to provide deepernsight into the factors that control the transport rate, selectiv-ty and stability of PIMs by collating the transport phenomenabserved by various authors and relating these to the membraneroperties. As discussed in the various sections, a number ofactors have been found to influence the performance of PIMsith the most important amongst them being: (1) the membrane

omposition, (2) the properties of the base polymers, the carriersnd the plasticizers, (3) the morphology of the membrane and4) the chemistry of the aqueous solutions making up the sourcend receiving phases. It has been observed that there is an intri-ate relationship between these factors and so it is essential toake them into account in any holistic approach to the transport

echanisms in PIMs.As we have documented, there is a respectable number of

ublished papers on PIMs and this number is increasing steadily.ne factor stands out in the majority of these papers and this

s the need to achieve a balance in membrane composition withespect to the three major constituents of the membrane, namely,he base polymer, the carrier and the plasticizer. While the car-ier is essential for the transport of the target solute in PIMs,xcessive amounts of carrier can result in carrier aggregation inome cases and exudation in others. Also, excessive amount oflasticizer can lead to “bleeding” from the membrane, whereasnsufficient plasticizer can lead to very low permeabilities.

The base polymers have often been thought to merely pro-ide mechanical support for the other constituents in order toive stability to the membrane, however, research has indicatedhat the bulk properties of the polymers can be important fac-ors governing the transport of target solutes. Thus the optimumomposition for a PIM depends quite strongly on the physic-chemical properties of each constituent of the membrane asell as on their compatibilities. At the present time, often a PIM

omposition appears to be 40% (w/w) base polymer, 40% (w/w)arrier and 20% (w/w) plasticizer although not all researchersave used this composition. There is also some evidence thathis composition may not result in the best PIM performance
n all cases. Considerable effort can be expected in future PIMsesearch to study these effects and to develop a model that cane used in the prediction of the optimum composition for a givenet of constituents. Although a large number of carriers has been

38 L.D. Nghiem et al. / Journal of Mem

Fig. 13. Historical and future development of liquid membranes and PIMs. Thew[

smtlibcaCiIop

vaftfPcaoPaaeeoc

A

fP

Bs

R

idth of the bars represents the amount of effort merited (adapted from Cussler78]).

tudied in PIMs so far, there is an apparent lack of focus on com-ercially available carriers. As we have shown in this review,

he number of PIM investigations using such carriers remainsimited. Furthermore, only a handful of plasticizers have beennvestigated to date. Most of these plasticizers seem to haveeen chosen not because of their commercial availability at lowost or clear industrial application but rather because of theirpplication in ISE membranes. It is also interesting to note thatTA and PVC have been used as the base polymers in most

f not all of the PIMs studied so far with CTA predominating.t can be expected that future research will expand the numberf commercially available carriers, plasticizers as well as baseolymers that can be used.

Given the demonstrated superior stability of PIMs over thearious other types of liquid membranes (e.g. SLMs) and thedequate permeability in practical sense and selectivity of PIMsor industrial applications, Cussler [78] has predicted (with cau-ion about speculation with regard the timeline) a decline inundamental research on SLMs and an increased interest inIMs research in the near future and as a result of this, practi-al applications will emerge (Fig. 13). In fact, this prediction islready becoming reality if one considers the increasing numberf papers being published on PIMs. However, we do not believeIM systems will replace traditional solvent extraction systems,nd certainly not in the near future, but will find a role in nichereas such as amino acid separation in biotechnology, fructosenrichment in food processing technology, precious metal recov-ry from electronic scrap and catalytic converters, the treatmentf radioactive waste streams and in environmental clean up ofontaminated waters.

cknowledgements

The authors are grateful to the Australian Research Councilor financial support under its Discovery Project scheme and torofessor Geoff Stevens from the Department of Chemical and

brane Science 281 (2006) 7–41

iomolecular Engineering of The University of Melbourne fortimulating discussions.

eferences

[1] A.M. Sastre, A. Kumar, J.P. Shukla, R.K. Singh, Improved techniquesin liquid membrane separations: an overview, Sep. Purif. Meth. 27 (2)(1998) 213–298.

[2] J. de Gyves, E.R. de San Miguel, Metal ion separations by supportedliquid membranes, Ind. Eng. Chem. Res. 38 (6) (1999) 2182–2202.

[3] X.J. Yang, A.G. Fane, Effect of membrane preparation on the life-time of supported liquid membranes, J. Membr. Sci. 133 (2) (1997)269–273.

[4] X.J. Yang, A.G. Fane, J. Bi, H.J. Griesser, Stabilization of supportedliquid membranes by plasma polymerization surface coating, J. Membr.Sci. 168 (1/2) (2000) 29–37.

[5] A.J. Schow, R.T. Peterson, J.D. Lamb, Polymer inclusion membranescontaining macrocyclic carriers for use in cation separations, J. Membr.Sci. 111 (2) (1996) 291–295.

[6] L. Wang, R. Paimin, R.W. Cattrall, S. Wei, S.D. Kolev, The extrac-tion of cadmium(II) and copper(II) from hydrochloric acid solutionsusing an Aliquat 336/PVC membranes, J. Membr. Sci. 176 (1) (2000)105–111.

[7] G. Argiropoulos, R.W. Cattrall, I.C. Hamilton, S.D. Kolev, R. Paimin,The study of a membrane for extracting gold(III) from hydrochloricacid solutions, J. Membr. Sci. 138 (2) (1998) 279–285.

[8] L. Bromberg, G. Levin, O. Kedem, Transport of metals through gelledsupported liquid membranes containing carrier, J. Membr. Sci. 71 (1/2)(1992) 41–50.

[9] C. Fontas, R. Tayeb, S. Tingry, M. Hidalgo, P. Seta, Transport of plat-inum(IV) through supported liquid membrane (SLM) and polymericplasticized membrane (PPM), J. Membr. Sci. 263 (1–2) (2005) 96–102.

[10] A. Gherrou, H. Kerdjoudj, R. Molinari, P. Seta, Preparation and char-acterization of polymeric plasticized membranes (PPM) embedding acrown ether carrier application to copper ions transport, Mater. Sci.Eng. C 25 (4) (2005) 436–443.

[11] R. Tayeb, C. Fontas, M. Dhahbi, S. Tingry, P. Seta, Cd(II) transportacross supported liquid membranes (SLM) and polymeric plasticizedmembranes (PPM) mediated by Lasalocid A, Sep. Purif. Technol. 42(2) (2005) 189–193.

[12] K.L. Thunhorst, R.D. Noble, C.N. Bowman, Properties of the transportof alkali metal salts through polymeric membranes containing benzo-18-crown-6 crown ether functional groups, J. Membr. Sci. 156 (2)(1999) 293–302.

[13] A. Gherrou, H. Kerdjoudj, R. Molinari, P. Seta, E. Drioli, Fixed sitesplasticized cellulose triacetate membranes containing crown ethers forsilver(I), copper(II) and gold(III) ions transport, J. Membr. Sci. 228(2) (2004) 149–157.

[14] O. Arous, H. Kerdjoudj, P. Seta, Comparison of carrier-facilitatedsilver (i) and copper (ii) ions transport mechanisms in a supportedliquid membrane and in a plasticized cellulose triacetate membrane, J.Membr. Sci. 241 (2) (2004) 177–185.

[15] M. Sugiura, Coupled-ion transport through a solvent polymeric mem-brane, J. Colloid Interf. Sci. 81 (2) (1981) 385–389.

[16] A. Legin, S. Makarychev-Mikhailov, D. Kirsanov, J. Mortensen, Y.Vlasov, Solvent polymeric membranes based on tridodecylmethy-lammonium chloride studied by potentiometry and electrochemicalimpedance spectroscopy, Anal. Chim. Acta 514 (1) (2004) 107–113.

[17] A.M. Neplenbroek, D. Bargeman, C.A. Smolders, Supported liquidmembranes: stabilization by gelation, J. Membr. Sci. 67 (2/3) (1992)149–165.

[18] A.M. Neplenbroek, D. Bargeman, C.A. Smolders, The stability of sup-
ported liquid membranes, Desalination 79 (2/3) (1990) 303–312.
[19] J.S. Kim, S.K. Kim, J.W. Ko, E.T. Kim, S.H. Yu, M.H. Cho, S.G.Kwon, E.H. Lee, Selective transport of cesium ion in polymericCTA membrane containing calixcrown ethers, Talanta 52 (6) (2000)1143–1148.

Mem
[20] J.S. Kim, S.K. Kim, M.H. Cho, S.H. Lee, J.Y. Kim, S.-G. Kwon, E.-H. Lee, Permeation of silver ion through polymeric CTA membranecontaining acyclic polyether bearing amide and amine end-group, Bull.Kor. Chem. Soc. 22 (10) (2001) 1076–1080.

[21] R. Bloch, A. Finkelstein, O. Kedem, D. Vofsi, Metal-ion separation bydialysis through solvent membranes, Ind. Eng. Chem. Prod. Res. Dev.6 (2) (1967) 231–237.

[22] R.W. Cattrall, Chemical sensors, in: R.G. Compton (Ed.), OxfordChemistry Primers, vol. 52, Oxford University Press, New York, 1997.

[23] G.J. Moody, R.B. Oke, J.D.R. Thomas, Calcium-sensitive electrodebased on a liquid ion exchanger in a poly(vinyl-chloride) matrix, Ana-lyst 95 (1136) (1970) 910–918.

[24] J.S. Gardner, J.O. Walker, J.D. Lamb, Permeability and durabilityeffects of cellulose polymer variation in polymer inclusion membranes,J. Membr. Sci. 229 (1/2) (2004) 87–93.

[25] F.W. Billmeyer, Textbook of Polymer Science, 3rd ed., Wiley, NewYork, 1984.

[26] R.P. Wool, Polymer entanglements, Macromolecules 26 (7) (1993)1564–1569.

[27] J. Rais, C.V. Mason, K.D. Abney, Use of PVC plasticized membranesfor uptake of radioactive cesium and strontium, Sep. Sci. Technol. 32(5) (1997) 951–969.

[28] P.J. Flory, Principles of Polymer Chemistry, Cornell University Press,Ithaca, 1953.

[29] D.A. Lannon, E.J. Hoskins, Effect of plasticizers, fillers, and otheradditives on physical properties, in: P.D. Ritchie (Ed.), Physics of Plas-tics, Iliffe Books Ltd., London, 1965.

[30] T.G. Levitskaia, J.D. Lamb, K.L. Fox, B.A. Moyer, Selective carrier-mediated cesium transport through polymer inclusion membranesby calix[4]arene-crown-6 carriers from complex aqueous mixtures,Radiochim. Acta 90 (1) (2002) 43–52.

[31] A.Y. Nazarenko, J.D. Lamb, Selective transport of lead(II) and stron-tium(II) through a crown ether-based polymer inclusion membranecontaining dialkylnaphthalenesulfonic acid, J. Inclusion Phenom. 29(3/4) (1997) 247–258.

[32] J.C. Aguilar, E.R.D.S. Miguel, J.D. Gyves, R.A. Bartsch, M. Kim,Design, synthesis and evaluation of diazadibenzocrown ethers as Pb2+

extractants and carriers in plasticized cellulose triacetate membranes,Talanta 54 (6) (2001) 1195–1204.

[33] W. Walkowiak, R.A. Bartsch, C. Kozlowski, J. Gega, W.A. Charewicz,B. Amiri-Eliasi, Separation and removal of metal ionic species by poly-mer inclusion membranes, J. Radioanal. Nucl. Chem. 246 (3) (2000)643–650.

[34] C. Kozlowski, W. Apostoluk, W. Walkowiak, A. Kita, Removal ofCr(VI), Zn(II) and Cd(II) ions by transport across polymer inclusionmembranes with basic ion carriers, Fizykochemiczne Problemy Min-eralurgii 36 (2002) 115–122.

[35] M. Cox, Solvent extraction in hydrometallurgy, in: J. Rydberg, et al.(Eds.), Solvent Extraction Principles and Practice, Mercel Dekker, Inc.,New York, 2004.

[36] G.W. Stevens, J.M. Perera, F. Grieser, Interfacial aspects of metal ionextraction in liquid–liquid systems, Rev. Chem. Eng. 17 (2) (2001)87–110.

[37] G.W. Stevens, J.M. Perera, F. Grieser, Metal ion extraction, Curr. Opin.Colloid Interf. Sci. 2 (6) (1997) 629–634.

[38] W. Walkowiak, M. Ulewicz, C.A. Kozlowski, Application of macrocy-cle compounds for metal ions removal and separation—a review, ARSSep. Acta 1 (2002) 87–98.

[39] C.A. Kozlowski, W. Walkowiak, Transport of Cr(VI), Zn(II), and Cd(II)ions across polymer inclusion membranes with tridecyl(pyridine) oxideand tri-n-octylamine, Sep. Sci. Technol. 39 (13) (2004) 3127–3141.

[40] J. Rydberg, G.R. Choppin, C. Musikas, T. Sekine, Solvent extractionequilibria, in: J. Rydberg, et al. (Eds.), Solvent Extraction Principles
and Practice, Marcel Dekker, Inc., New York, 2004.
[41] J.C. Aguilar, M. Sanchez-Castellanos, E. Rodriguez de San Miguel, J.de Gyves, Cd(II) and Pb(II) extraction and transport modeling in SLMand PIM systems using Kelex 100 as carrier, J. Membr. Sci. 190 (1)(2001) 107–118.

brane Science 281 (2006) 7–41 39

[42] M. Ulewicz, W. Walkowiak, J. Gega, B. Pospiech, Zinc(II) selectiveremoval from other transition metal ions by solvent extraction andtransport through polymer inclusion membranes with D2EHPA, ARSSep. Acta 2 (2003) 47–55.

[43] J. de Gyves, A.M. Hernandez-Andaluz, E.R.D.S. Miguel, LIX(R)-loaded polymer inclusion membrane for copper(II) transport. 2. Opti-mization of the efficiency factors (permeability, selectivity, and stabil-ity) for LIX(R) 84-I, J. Membr. Sci. 268 (2) (2006) 142–149.

[44] J.K. Sears, J.R. Darby, Technology of Plasticizers, John Wiley & Sons,New York, 1982, p. 1174.

[45] M. Sugiura, Effect of polyoxyethylene n-alkyl ethers on carrier-mediated transport of lanthanide ions through cellulose triacetate mem-branes, Sep. Sci. Technol. 27 (2) (1992) 269–276.

[46] H. Matsuoka, M. Aizawa, S. Suzuki, Uphill transport of uranium acrossa liquid membrane, J. Membr. Sci. 7 (1) (1980) 11–19.

[47] S.D. Kolev, R.W. Cattrall, R. Paimin, I.D. Potter, Y. Sakai, Theoret-ical and experimental study of palladium(II) extraction into Aliquat336/PVC membranes, Anal. Chim. Acta 413 (2000) 241–246.

[48] R.S. Barshtein, G.A. Kotlyarevskii, The mechanism of plasticizationof poly(vinyl chloride), Plasticheskie Massy (7) (1965) 13–14.

[49] R.S. Barshtein, G.A. Kotlyarevskii, Plasticizers for poly(vinyl chloride)and its copolymers, Plasticheskie Massy (2) (1961) 57–60.

[50] W.S. Gibbons, H.M. Patel, R.P. Kusy, Effects of plasticizers on themechanical properties of poly(vinyl chloride) membranes for electrodesand biosensors, Polymer 38 (11) (1997) 2633–2642.

[51] W.S. Gibbons, R.P. Kusy, Influence of plasticizer configurationalchanges on the mechanical properties of highly plasticized poly(vinylchloride), Polymer 39 (26) (1998) 6755–6765.

[52] W.S. Gibbons, R.P. Kusy, Influence of plasticizer configurationalchanges on the dielectric characteristics of highly plasticized poly(vinylchloride), Polymer 39 (14) (1998) 3167–3178.

[53] S.P. Kusumocahyo, T. Kanamori, K. Sumaru, S. Aomatsu, H. Mat-suyama, M. Teramoto, T. Shinbo, Development of polymer inclusionmembranes based on cellulose triacetate: carrier-mediated transport ofcerium(III), J. Membr. Sci. 244 (1/2) (2004) 251–257.

[54] P.K. Mohapatra, P.N. Pathak, A. Kelkar, V.K. Manchanda, Novel poly-mer inclusion membrane containing a macrocyclic ionophore for selec-tive removal of strontium from nuclear waste solution, New J. Chem.28 (8) (2004) 1004–1009.

[55] Y.M. Scindia, A.K. Pandey, A.V.R. Reddy, Coupled-diffusion transportof Cr(VI) across anion-exchange membranes prepared by physical andchemical immobilization methods, J. Membr. Sci. 249 (1/2) (2005)143–152.

[56] M. Sugiura, M. Kikkawa, Effect of plasticizer on carrier-mediatedtransport of zinc ion through cellulose triacetate membranes, Sep. Sci.Technol. 22 (11) (1987) 2263–2268.

[57] M.E. Duffey, D.F. Evans, E.L. Cussler, Simultaneous diffusion of ionsand ion pairs across liquid membranes, J. Membr. Sci. 3 (1) (1978)1–14.

[58] C.A. Kozlowski, W. Walkowiak, Applicability of liquid membranes inchromium(VI) transport with amines as ion carriers, J. Membr. Sci.226 (1/2) (2005) 143–150.

[59] J.Y. Xu, L.J. Wang, W. Shen, R. Paimin, X.G. Wang, The influence ofthe interior structure of Aliquat 336/PVC membranes to their extractionbehavior, Sep. Sci. Technol. 39 (15) (2004) 3527–3539.

[60] S.D. Kolev, G. Argiropoulos, R.W. Cattrall, I.C. Hamilton, R. Paimin,Mathematical modelling of membrane extraction of gold(III) fromhydrochloric acid solutions, J. Membr. Sci. 137 (1/2) (1997) 261–269.

[61] L. Wang, W. Shen, Chemical and morphological stability of Aliquat336/PVC membranes in membrane extraction: a preliminary study,Sep. Purif. Technol. 46 (1/2) (2005) 51–62.

[62] R. Tripathi, A.K. Pandey, S. Sodaye, B.S. Tomar, S.B. Manohar, S.
Santra, K. Mahata, P. Singh, S. Kailas, Backscattering spectrometrystudies on metal ion distribution in polymer inclusion membranes,Nucl. Instrum. Meth. Phys. Res., Sect. B 211 (1) (2003) 138–144.
[63] S.C. Lee, J.D. Lamb, M. Cai, J.T. Davis, Cs+ and Ba2+ selectivetransport by a novel self-assembled isoguanosine ionophore through

4 Mem
polymer inclusion and bulk liquid membranes, J. Inclusion Phenom.Macrocyclic Chem. 40 (1/2) (2001) 51–57.

[64] C.A. Kozlowski, W. Walkowiak, Removal of chromium(VI) from aque-ous solutions by polymer inclusion membranes, Water Res. 36 (19)(2002) 4870–4876.

[65] P. Lacan, C. Guizard, P. Le Gall, D. Wettling, L. Cot, Facilitated trans-port of ions through fixed-site carrier membranes derived from hybridorganic–inorganic materials, J. Membr. Sci. 100 (2) (1995) 99–109.

[66] J.D. Lamb, A.Y. Nazarenko, Lead(II) ion sorption and transport usingpolymer inclusion membranes containing tri-octylphosphine oxide, J.Membr. Sci. 134 (2) (1997) 255–259.

[67] M. Sugiura, Transport of lanthanide ions through cellulose triacetatemembranes containing hinokitiol and flavonol as carriers, Sep. Sci.Technol. 25 (11/12) (1990) 1189–1199.

[68] T.G. Levitskaia, D.M. Macdonald, J.D. Lamb, B.A. Moyer, Predictionof the carrier-mediated cation flux through polymer inclusion mem-branes via fundamental thermodynamic quantities: complexation studyof bis(dodecyloxy)calix[4]arene-crown-6 with alkali metal cations,Phys. Chem. Chem. Phys. 2 (7) (2000) 1481–1491.

[69] R. Bloch, Hydrometallurgical separations by solvent membranes, in:J.E. Flinn (Ed.), Membrane Science and Technology, Plenum Press,New York, 1970, pp. 171–187.

[70] B. Wionczyk, W. Apostoluk, K. Prochaska, C. Kozlowski, Properties of4-(1′-n-tridecyl)pyridine N-oxide in the extraction and polymer inclu-sion membrane transport of Cr(VI), Anal. Chim. Acta 428 (1) (2001)89–101.

[71] K.C. Sole, A.M. Feather, P.M. Cole, Solvent extraction in southernAfrica: an update of some recent hydrometallurgical developments,Hydrometallurgy 78 (1/2) (2005) 52–78.

[72] I. Ivanov, Increased current efficiency of zinc electrowinning inthe presence of metal impurities by addition of organic inhibitors,Hydrometallurgy 72 (1/2) (2004) 73–78.

[73] P.J.D. Lloyd, Principles of industrial solvent extraction, in: J. Ryd-berg, et al. (Eds.), Solvent Extraction Principles and Practices, MarcelDekker, Inc., New York, 2004.

[74] C.A. Kozlowski, J. Jablonska, Removal and separation of zinc and cad-mium ions by amine solvent extraction and liquid membranes, Przem.Chem. 82 (1) (2003) 27–34.

[75] C.A. Kozlowski, W. Walkowiak, W. Pellowski, J. Koziol, Competi-tive transport of toxic metal ions by polymer inclusion membranes, J.Radioanal. Nucl. Chem. 253 (3) (2002) 389–394.

[76] A.J.B. Kemperman, D. Bargeman, T. vandenBoomgaard, H. Strath-mann, Stability of supported liquid membranes: state of the art, Sep.Sci. Technol. 31 (20) (1996) 2733–2762.

[77] P.R. Danesi, L. Reichley-Yinger, P.G. Rickert, Lifetime of supportedliquid membranes: the influence of interfacial properties, chemicalcomposition and water transport on the long-term stability of the mem-branes, J. Membr. Sci. 31 (2/3) (1987) 117–145.

[78] E.L. Cussler, Facilitated transport, in: R.W. Baker, et al. (Eds.), Mem-brane Separation Systems: Recent Developments and Future Direc-tions, Noyes Data Corporation, New Jersey, 1991, pp. 242–275.

[79] M. Sugiura, M. Kikkawa, S. Urita, Carrier-mediated transport of rareearth ions through cellulose triacetate membranes, J. Membr. Sci. 42(1/2) (1989) 47–55.

[80] C. Bourget, B. Jakovljevic, D. Nucciarone, CYANEX(R) 301 binaryextractant systems in cobalt/nickel recovery from acidic sulphate solu-tions, Hydrometallurgy 77 (3/4) (2005) 203.

[81] R.-S. Juang, H.-C. Kao, Extraction separation of Co(II)/Ni(II) fromconcentrated HCl solutions in rotating disc and hollow-fiber membranecontactors, Sep. Purif. Technol. 42 (1) (2005) 65.

[82] M.S. El Sayed, Uranium extraction from gattar sulfate leach liquorusing Aliquat-336 in a liquid emulsion membrane process, Hydromet-allurgy 68 (1–3) (2003) 51.

[83] K.M. White, B.D. Smith, P.J. Duggan, S.L. Sheahan, E.M. Tyndall,Mechanism of facilitated saccharide transport through plasticized cel-lulose triacetate membranes, J. Membr. Sci. 194 (2) (2001) 165–175.

[84] J.A. Riggs, B.D. Smith, Facilitated transport of small carbohydratesthrough plasticized cellulose triacetate membranes. Evidence for fixed-

brane Science 281 (2006) 7–41

site jumping transport mechanism, J. Am. Chem. Soc. 119 (11) (1997)2765–2766.

[85] T.A. Munro, B.D. Smith, Facilitated transport of amino acids by fixed-site jumping, Chem. Commun. 22 (1997) 2167–2168.

[86] I.L. Jenkins, Solvent extraction chemistry in the atomic energyindustry—a review, Hydrometallurgy 4 (1) (1979) 1–20.

[87] Y. Sakai, R.W. Cattrall, I.D. Potter, S.D. Kolev, R. Paimin, Transportof thiourea through an Aliquat 336/polyvinyl chloride membrane, Sep.Sci. Technol. 35 (13) (2000) 1979–1990.

[88] C. Kozlowski, M. Ulewicz, W. Walkowiak, Separation of zinc andcadmium ions from aqueous chloride solutions by ion flotation andliquid membranes, Fizykochemiczne Problemy Mineralurgii 34 (2000)141–151.

[89] T. Hayashita, Heavy metal ion separation by functional polymericmembranes, in: A. Bartsch Richard, J.D. Way (Eds.), Chemical Sep-arations with Liquid Membranes, American Chemical Society, Wash-ington, DC, 1996.

[90] R.C. Weast, M.J. Astle, CRC Handbook of Chemistry and Physics,60th ed., CRC Press, Boca Raton, FL, 1980.

[91] M. Matsumoto, T. Takagi, K. Kondo, Separation of lactic acid usingpolymeric membrane containing a mobile carrier, J. Ferment. Bioeng.85 (5) (1998) 483–487.

[92] G. Salazar-Alvarez, A.N. Bautista-Flores, E.R. de San Miguel, M.Muhammed, J. de Gyves, Transport characterisation of a PIM systemused for the extraction of Pb(II) using D2EHPA as carrier, J. Membr.Sci. 250 (1/2) (2005) 247–257.

[93] S.C. Lee, J.D. Lamb, M.H. Cho, C.H. Rhee, J.S. Kim, A lipophilicacyclic polyether dicarboxylic acid as Pb2+ carrier in polymer inclu-sion and bulk liquid membranes, Sep. Sci. Technol. 35 (5) (2000)767–778.

[94] M.-F. Paugam, J. Buffle, Comparison of carrier-facilitated copper(II)ion transport mechanisms in a supported liquid membrane and ina plasticized cellulose triacetate membrane, J. Membr. Sci. 147 (2)(1998) 207–215.

[95] D. Filippou, Innovative hydrometallurgical processes for the primaryprocessing of zinc, Miner. Process. Extr. Metall. Rev. 25 (2004)205–252.

[96] Y. Baba, K. Hoaki, J.M. Perera, G.W. Stevens, T.J. Cardwell, R.W.Cattrall, S.D. Kolev, Separation of palladium(II) from copper(II) acidicsolutions using PVC membranes containing D2EHPA, Chim. Anal. 54(10) (2001) 69–72.

[97] M. Sugiura, Effect of quaternary ammonium salts on carrier-mediatedtransport of lanthanide ions through cellulose triacetate membranes,Sep. Sci. Technol. 28 (7) (1993) 1453–1463.

[98] M. Resina, J. Macanas, J. de Gyves, M. Munoz, Zn(II), Cd(II) andCu(II) separation through organic–inorganic hybrid membranes con-taining di-(2-ethylhexyl) phosphoric acid or di-(2-ethylhexyl) dithio-phosphoric acid as a carrier, J. Membr. Sci. 268 (1) (2006)57–64.

[99] M.D. Ballinas, E.R. De San Miguel, M.T.D. Rodriguez, O. Silva, M.Munoz, J. de Gyves, Arsenic(V) removal with polymer inclusion mem-branes from sulfuric acid media using DBBP as carrier, Environ. Sci.Technol. 38 (3) (2004) 886–891.

[100] J.D. Lamb, A.Y. Nazarenko, J.C. Uenishi, H. Tsukube, Silver(I) ion-selective transport across polymer inclusion membranes containing newpyridino- and bipyridino-podands, Anal. Chim. Acta 373 (2/3) (1998)167–173.

[101] J.S. Kim, S.H. Yu, M.H. Cho, O.J. Shon, J.A. Rim, S.H. Yang, J.K.Lee, S.J. Lee, Calix[4]azacrown ethers in polymeric CTA membrane,Bull. Kor. Chem. Soc. 22 (5) (2001) 519–522.

[102] J. Kim, T.-H. Ahn, M. Lee, A.J. Leong, L.F. Lindoy, B.R. Rumbel,B.W. Skelton, T. Strixner, G. Wei, A.H. White, Metal ion recog-nition. The interaction of cobalt(II), nickel(II), copper(II), zinc(II),
cadmium(II), silver(I) and lead(II) with N-benzylated macrocyclesincorporating O2N2-, O3N2- and O2N3-donor sets, J. Chem. Soc.,Dalton Trans. (21) (2002) 3993–3998.
[103] M. Ulewicz, C. Kozlowski, W. Walkowiak, Removal of Zn(II), Cd(II)and Cu(II) ions by polymer inclusion membrane with side-armed

Mem

G

22

AABBtCCCCDDDDDDDDDEFFIIHKLMNPPPRSSTTTTTTTTT


diphosphaza-16-crown-6 ethers, Fizykochemiczne Problemy Miner-alurgii 38 (2004) 131–138.

[104] C.A. Kozlowski, T. Girek, W. Walkowiak, J.J. Koziol, Application ofhydrophobic [beta]-cyclodextrin polymer in separation of metal ionsby plasticized membranes, Sep. Purif. Technol. 46 (3) (2005) 136–144.

[105] S. Elshani, S. Chun, B. Amiri-Eliasi, R.A. Bartsch, Highly selectiveBa2+ separations with acyclic, lipophilic di-[N-(X)sulfonyl carbamoyl]polyethers, Chem. Commun. (2) (2005) 279–281.

[106] J.S. Kim, S.H. Lee, S.H. Yu, M.H. Cho, D.W. Kim, S.-G. Kwon,E.-H. Lee, Calix[6]arene bearing carboxylic acid and amide groupsin polymeric CTA membrane, Bull. Kor. Chem. Soc. 23 (8) (2002)1085–1088.

[107] J.S. Gardner, Q.P. Peterson, J.O. Walker, B.D. Jensen, B. Adhikary,R.G. Harrison, J.D. Lamb, Anion transport through polymer inclu-sion membranes facilitated by transition metal containing carriers, J.Membr. Sci. 277 (1/2) (2006) 165–176.

[108] P.R. Danesi, Separation of metal species by supported liquid mem-branes, Sep. Sci. Technol. 19 (11-1) (1984) 857–894.

[109] E.L. Cussler, R. Aris, A. Bhown, On the limits of facilitated diffusion,J. Membr. Sci. 43 (2/3) (1989) 149–164.

[110] N.A. Plate, T.L. Lebedeva, G.A. Shandryuk, L.M. Kardivarenko, V.V.Bagreev, On the mechanism of metal-ions facilitated transport throughpseudo-liquid membranes, J. Membr. Sci. 104 (3) (1995) 197–203.

[111] A.A. Kalachev, L.M. Kardivarenko, N.A. Plate, V.V. Bagreev, Facil-itated diffusion in immobilized liquid membranes: experimental ver-ification of the “jumping” mechanism and percolation threshold inmembrane transport, J. Membr. Sci. 75 (1/2) (1992) 1–5.

[112] S.M. Aharoni, On entanglements of flexible and rodlike polymers,Macromolecules 16 (11) (1983) 1722–1728.

[113] G. Arena, A. Contino, A. Magri, D. Sciotto, J.D. Lamb, Selectivetransport of cesium and strontium ions through polymer inclusionmembranes containing calixarenes as carriers, Supramol. Chem. 10(1) (1998) 5–15.

[114] M. Sugiura, T. Yamaguchi, Coupled transport of picrate anion throughliquid membranes supported by a microporous polymer film, J. ColloidInterf. Sci. 96 (2) (1983) 454–459.

[115] G. Argiropoulos, R.W. Cattrall, I.C. Hamilton, R. Paimin, Determina-tion of the quaternary ammonium ion components of Aliquat 336 byelectrospray mass spectrometry, Anal. Chim. Acta 360 (1–3) (1998)167–169.

lossary

-NPOE: 2-Nitrophenyl octyl ether-NPPE: 2-Nitrophenyl pentyl ether

TTTTT

brane Science 281 (2006) 7–41 41

CMs: Activated composite membranesFM: Atomic force microscopyLMs: Bulk liquid membranesMPP: 4-Benzoyl-3-methyl-1-phenyl-5-pyrazolone

-BuDC18C6: Di-tert-butylcyclohexano-18-crown-6AP: Cellulose acetate propionateMPO: Octyl(phenyl)-N,N-diisobutyl carbamoylmethyl phosphine oxideTA: Cellulose triacetateTB: Cellulose tributyrate2EHDTPA: Di(2-ethylhexyl) dithiophosphoric acid2EHPA: Di(2-ethylhexyl) phosphoric acidBBP: Dibutyl butyl phosphonateC18C6: Dicyclohexano-18-crown-6NNS: Dinonylnaphthalenesulfonic acidOA: Bis(2-ethylhexyl) adipateOP: DioctylphthalateOS: DioctylsebacateOTP: Bis(2-ethylhexyl)terephthalateLMs: Emulsion liquid membranesSMs: Fixed-site membranesTIR: Fourier transform infrared spectrometry

SE: Ion-selective electrodesoG: IsoguanosineIPT: Hinokitiolelex 100: 7-(4-Ethyl-1-methyloctyl)-8-hydroxyquinolineIX® 84-I: 2-Hydroxy-5-nonylacetophenone oximeWc: Critical entanglement molecular weightPHE: p-Nitrophenyl-n-heptyl etherIMs: Polymer inclusion membranesOEs: Polyoxyethylene n-alkyl ethersVC: Poly(vinyl chloride)BS: Rutherford backscattering spectrometryEM: Scanning electron microscopyLMs: Supported liquid membranes2EHP: Tris(2-ethylhexyl)phosphateBEP: Tri(butoxyethyl)phosphateBP: Tri-n-butyl phosphateCP: Tricresyl phosphateDMAC: Tridodecylmethylammonium chlorideDPNO: 4-(1′-n-tridecyl)pyridine N-oxide

g: Glass transition temperatureHF: Tetrahydrofuran

m: Melting temperatureMPP: 4-Trifluoroacetyl-3-methyl-1-phenyl-5-pyrazolone
OA: Tri-n-octyl amineODGA: N,N,N’,N’-Tetraoctyl-3-oxapentanediamideOMAC: Trioctylmethylammonium chlorideOPO: Tri-n-octyl phosphine oxide

review extraction and transport of metal ions and small ...longn/doc/journal papers/extraction...

Documents