psipw7thaward.org · 11/22/2013 · author's personal copy journal of membrane science383...

This article appeared in a journal published by Elsevier. The attachedcopy is furnished to the author for internal non-commercial researchand education use, including for instruction at the authors institution

and sharing with colleagues.

Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies areencouraged to visit:

http://www.elsevier.com/copyright


Author's personal copy

Journal of Membrane Science 383 (2011) 214– 223

Contents lists available at SciVerse ScienceDirect

Journal of Membrane Science

j ourna l ho me pag e: www.elsev ier .com/ locate /memsci

The role of sulphonated polymer and macrovoid-free structure in the supportlayer for thin-film composite (TFC) forward osmosis (FO) membranes

Natalia Widjojoa, Tai-Shung Chunga,∗, Martin Weberb, Christian Maletzkoc, Volker Warzelhanb

a Department of Chemical and Biomolecular Engineering, National University of Singapore, 10 Kent Ridge Crescent, Singapore 117602, Singaporeb Polymer Research Engineering Plastics, BASF SE, GKT/B-B1, 67056 Ludwigshafen, Germanyc Engineering Plastics, BASF SE, E-KTE/NE-F206, 67056 Ludwigshafen, Germany

a r t i c l e i n f o

Article history:Received 19 May 2011Received in revised form 13 July 2011Accepted 21 August 2011Available online 26 August 2011

Keywords:Thin film composite (TFC) membranesForward osmosisSulphonated polymerInterfacial polymerizationSponge-like structure

a b s t r a c t

A new approach to fabricate thin film composite (TFC) membranes via interfacial polymerization forforward osmosis (FO) applications has revealed that it is possible to design TFC-FO membranes withfully sponge-like structure and likely anti-fouling characteristics while maintaining a high water flux.Not only does the sulphonated material in the substrate of TFC-FO membranes play the key role to createmacrovoid-free structure but also induces hydrophilic properties with enhanced water fluxes. It is foundthat the TFC-FO membranes containing a 50 wt% sulphonated material in the membrane substrate exhibita fully sponge-like structure, while those with lower or without sulphonated content show finger-likestructures. In terms of FO performance, the TFC-FO membranes with 50 wt% sulphonated material contentcan achieve the highest water flux of 33.0 LMH against DI water and 15 LMH against the 3.5 wt% NaClmodel solution using 2 M NaCl as the draw solution tested under the pressure retarded osmosis (PRO)mode. The value of 15 LMH for seawater desalination is the highest reported so far. Despite the debateson whether TFC-FO membranes should possess a finger-like or sponge-like structure, it is proven that thedegree of hydrophilicity of membrane substrates is a much stronger factor enhancing the water flux in FOtests. Meanwhile, a fully sponge-like structure with expected anti-fouling property is preferred for long-term membrane stability. Furthermore, the structural parameter indicating the internal concentrationpolarization (ICP) can be remarkably decreased with an increase in sulphonated material content inmembrane substrates.

© 2011 Elsevier B.V. All rights reserved.

1. Introduction

The forward osmosis (FO) process is an emerging technologyfor water purification and seawater desalination. Compared to thetraditional pressure driven process such as reverse osmosis (RO),FO membranes take the inherent advantages of osmotically drivenprocesses. In particular, there are several aspects of FO processeswhich can go beyond the current RO technology [1–3]; namely: (1)no or low hydraulic pressure is required in the FO process, thus itmay lower energy consumption and equipment costs [1]; (2) waterrecovery in the FO process may be potentially higher than that ofRO process [1]; (3) much reversible fouling has been observed inthe FO process [4]; (4) FO can be applicable to wider applicationsother than seawater desalination and waste water treatment suchas power generation [5–7], juice or food concentration [8,9], andprotein and pharmaceutical enrichment [10,11].

∗ Corresponding author. Tel.: +65 65166645; fax: +65 67791936.E-mail address: [email protected] (T.-S. Chung).

Principally, FO processes employ the chemical potential differ-ence between the feed and the concentrated draw solution across asemi-permeable membrane, so-called the osmotic pressure gradi-ent, as the driving force to induce a net flow of clean water throughthe membrane into the draw solution. However, the major chal-lenges to fully explore FO potential as a new generation waterproduction technology are: (1) the inadequate number of com-mercially available high performance FO membranes [1,2]; (2) thelimitation of non-toxic draw solutions which can be recycled effec-tively and easily with low energy expenses and minimal reverseflux [12–14].

The desired FO membranes consist of the following charac-teristics: (1) an ultrathin defect-free semi-permeable active layerwith a high water flux and high solute rejection; (2) a thin sup-porting layer with high porosity and low internal concentrationpolarization (ICP); (3) hydrophilic nature to enhance water flux andreduce membrane fouling; and (4) sufficient mechanical strengthand robust to sustain backwash, cleaning and vibration in industrialoperations.

To the present, flat sheet FO membranes made of cellulosetriacetate (CTA) from Hydration Technologies Inc. are dominant

0376-7388/$ – see front matter © 2011 Elsevier B.V. All rights reserved.doi:10.1016/j.memsci.2011.08.041

chencts

Highlight

chencts

Highlight


N. Widjojo et al. / Journal of Membrane Science 383 (2011) 214– 223 215

Fig. 1. The chemical structure of (a) PESU; (b) PESU-co-sPPSU 11.

in commercial applications such as water reuse and purifica-tion for industries, military, space exploration, emergency relief,and recreational purposes. However, CTA has a relatively lowwater permeability and salt rejection and can be easily degradedby draw solution such as ammonium bicarbonate [1]. Other FOmembranes in the form of flat asymmetric and hollow fiber con-figurations were also developed via phase inversion technique[15–21]. Similar to CTA, most of these membranes exhibit rela-tively low water fluxes especially under tests for osmotic seawaterdesalination.

For decades, thin film composite (TFC) membranes with highsalt rejections have been used in RO processes [22–24]. The com-mercially available TFC-RO membranes consist of three layers;namely, a thin polyamide layer as a selective layer, a porouspolysulfone layer as the intermediate layer for easy thin-filmpolymerization on top of it and better interactions with under-neath non-woven fabrics, and a thick layer of non-woven fabricsas the mechanical support layer to withstand high pressures.Except the polyamide layer, other layers are hydrophobic innature. McCutcheon and Elimelech [25] have explored conven-tional TFC-RO membranes in FO applications and found that TFC-ROmembranes intensified ICP and significantly decreased flux in FOprocesses since hydrophobic sub-layers tend to hinder the osmot-ically driven water diffusion cross the membrane. In addition, thethick non-woven fabric provides an additional resistance towardswater transport in FO processes.

The preferred support layer for FO membranes has to be asthin as possible [25,26] or non-existent [26] with a minimalresistance for osmotically driven water transport across the mem-brane. By eliminating non-woven fabrics used in traditional TFC-ROmembranes, Yip et al. [27] and Wang [28,29] are the pioneersin fabricating TFC-FO flat-sheet and hollow fiber membranes,respectively, on the polysulfone support for FO applications. Theformer showed a water flux up to 18 LMH under FO tests usingwater as the feed and 1.5 M NaCl as the draw solute, and a saltrejection greater than 97% at 400 psi (27.57 bar) under RO tests[27], while the latter showed a water flux up to 32.2–42.6 LMHunder PRO tests using water as the feed and 0.5 M NaCl as thedraw solution with reverse salt flux about 4 gMH, and a saltrejection of 91% at 1.01 bar under RO tests [28,29]. They allclaimed that an ideal support substrate for TFC-FO membranesshould comprise a thin layer of sponge-like structure near the topedge of membrane cross-section with a fully finger-like structureunderneath.

Recently, Wang et al. [30] in our group modified physicochem-ical properties of the substrate material by blending a hydrophilicsulphonated polysulfone (SPSf) with conventional hydrophobicpolysulfone (PSf), and then carried out the thin-film polymeriza-tion on top of it. The resultant TFC-FO flat-sheet membranes showfluxes up to 47.5 LMH in the PRO mode with salt leakages up to12.4 gMH using 2 M NaCl as a draw solution. A comparison of thesetwo flat-sheet membranes made by Yip et al. [27] and Wang et al.

[30], the latter demonstrates a higher water flux than the former.Therefore, one hypothesizes that the water flux in FO processes canbe significantly enhanced by an increase in substrate hydrophilic-ity and finger-like structure in the membrane substrate of PSf/SPSfTFC-FO membranes.

Since it is hard to make a fair comparison between Yip et al. [27]and Wang et al. [30] works because they used different substratethicknesses and monomers for thin-film polymerizations, the pur-poses of this work are to (1) investigate the effect of sulphonatedmaterials on TFC-FO membranes by using a sulphonated copoly-mer made of polyethersulfone (PES) and polyphenylsulfone (PPSU)and its blends with PESU in substrates; and (2) to examine if thefinger-like structure is essential for the fabrication of high fluxFO membranes. To the best of our knowledge, no PES and thissulphonated copolymer have been reported in the literature forTFC-FO membranes. It is assumed that the hydrophilic nature ofthis polymer can not only enhance water flux of the resultant FOmembranes, but also change morphology from a fully finger-like tosponge-like structure. If so, this study will have significant impactto future development of next generation FO membranes.

2. Experimental

2.1. Materials

The PESU-co-sPPSU 11 hydrophilic polymer with 11 wt%sulphonation degree and polyethersulfone (PESU E6020P) used asmaterials for membrane substrates were kindly supplied by BASFSE Company, Germany. Fig. 1 describes the chemical structuresof PESU-co-sPPSU 11 and PESU E6020P materials. PESU-co-sPPSUis a sulphonated copolymer made of polyethersulfone (PESU) andpolyphenylsulfone (PPSU). This material cannot form a free stand-ing asymmetric membrane substrate due to its highly hydrophilicin nature and slow precipitation rate. Consequently, polymerblends consisting of polyethersulfone (PESU E6020P) and differ-ent concentrations of PESU-co-sPPSU 11 hydrophilic polymer wereused to prepare the substrates for thin-film polymerization appli-cations.

N-methyl-2-pyrrolidone (NMP) from Merck and ethylene gly-col (EG) from Sigma–Aldrich were employed as the solventand additive, respectively, in the fabrication of membrane sub-strates for TFC-FO membranes. M-phenylenediamine (MPD) fromSigma–Aldrich with >99% purity and trimesoyl chloride (TMC) fromSigma–Aldrich with 98% purity were used as received for interfacialpolymerization process. n-Hexane from Merck with >99.0% puritywas utilized as the solvent for TMC.

For FO tests, sodium chloride (NaCl) supplied by Sigma–Aldrichwas dissolved in deionized (DI) water at 0.5, 1, 2, 3, 4, and 5 Mconcentrations and used as the draw solution. In osmotic seawaterdesalination tests, seawater as a model solution (3.5 wt% NaCl in DIwater) was used as the feed solution.

chencts

Highlight

chencts

Highlight


216 N. Widjojo et al. / Journal of Membrane Science 383 (2011) 214– 223

Table 1Compositions of casting solutions with different concentrations of sulphonated polymer.

Composition of solution casting (in wt%) Casting solution 1 (0 wt%sulphonated polymer)

Casting solution 2 (25 wt%sulphonated polymer)

Casting solution 3 (50 wt%sulphonated polymer)

PESU E6020P 16 12 8PESU-co-sPPSU 11 sulphonated polymer 0 4 8EG 16 16 16NMP 68 68 68

2.2. Fabrication of membrane substrates for interfacialpolymerization applications

To fabricate membrane substrates for interfacial polymeriza-tion with different concentrations of sulphonated polymer, threepolymer solutions were prepared as displayed in Table 1. The cast-ing solutions were allowed to degas overnight prior casting ontoa glass plate with a casting knife of 100 m in thickness. The as-cast membranes were then immersed into a water coagulation bathimmediately at room temperature and kept for 1 day to ensure com-plete precipitation. In addition, the membranes were washed withDI water before interfacial polymerization was carried out.

2.3. Fabrication of TFC-FO membranes via interfacialpolymerization

The formation of a polyamide layer in the membrane substrateswas based on the interfacial polymerization process [22–24]. Themembrane substrate was first immersed in 2 wt% of MPD in DIwater for 1 min. Thereafter, a filter paper was used to remove thewater droplets in the membrane surface. Subsequently, the topsurface of membranes was brought into contact with the 0.05 wt%TMC solution in n-hexane for 15 s. The resultant TFC-FO membraneswere dried at 60 C for 1 min, followed by drying in the air for 2 min.The TFC-FO membranes were then washed in DI water before FOtests.

2.4. Characterizations of membrane substrates

2.4.1. Morphology and topologyThe morphology of different membrane substrates was

observed by using a field emission scanning electron microscope(FESEM JEOL JSM-6700LV). The samples were prepared by fractur-ing the membrane in liquid nitrogen and then coated with platinumusing a sputtering coater (JEOL LFC-1300).

The surface topology of membrane substrates was analyzed byusing an Agilent Technologies 5500 Scanning Probe Microscope(AFM). A tapping mode (Acoustic AC) was operated on the sam-ples in air at room temperature with the image scanning size of2 m × 2 m. The mean surface roughness (Ra) was calculated toqualitatively compare the surface roughness of the membranes.

2.4.2. Pore size and pore size distributionThe fabricated flat sheet membrane substrate was first tested to

measure its pure water permeability (PWP) (in L m−2 bar−1 h−1) byan ultrafiltration membrane permeation cell with a sample diam-eter of 5 cm [31,32]. Subsequently, the membrane was subjectedto neutral solute (polyethylene glycol (PEG) or polyethylene oxide(PEO)) separation tests by flowing them through the membrane’stop surface under a pressure of 25 psi (1.72 bar) on the liquid side.The concentrations of the neutral solutes were measured by a totalorganic carbon analyzer (TOC ASI-5000A, Shimadzu, Japan). Themeasured feed (Cf) and permeate (Cp) concentrations were usedfor the calculation of the effective solute rejection coefficient R (%):

R =(

1 − Cp

Cf

)× 100% (1)

In this work, solutions containing 200 ppm of different molecularweights of PEG or PEO were used as the neutral solutes for the char-acterizations of membrane pore size and pore size distribution. Therelationship between Stokes radius (rs, nm) and molecular weight(Mw, g mol−1) of these neutral solutes can be expressed as:

for PEG r = 16.73 × 10−12 × M0.557 (2)

for PEO r = 10.44 × 10−12 × M0.587 (3)

From Eqs. (2) and (3), the radius (r) of a hypothetical solute at a givenMw can be calculated. The mean effective pore size and the poresize distribution were then obtained according to the traditionalsolute transport approach by ignoring influences of the steric andhydrodynamic interaction between solute and membrane pores,the mean effective pore radius (p) and the geometric standarddeviation (p) can be assumed to be the same as s (the geometricmean radius of solute at R = 50%) and g (the geometric standarddeviation defined as the ratio of the rs at R = 84.13% over that atR = 50%). Therefore, based on p and p, the pore size distribution ofa membrane can be expressed as the following probability densityfunction:

dR(dp)ddp

= 1

dp ln p√

2exp

[− (ln dp − ln p)2

2(ln p)2

](4)

2.4.3. Membrane porosity (ε)To measure the porosity of membrane substrates, wet mem-

branes were taken out from the water bath followed by careful andquick removal of excess water on the surface by tissue paper. Thewet membrane were then weighed (m1, g), freeze dried overnight,and re-weighed (m2, g). Subsequently, the absorbed water is calcu-lated as m1 − m2, and the dry weight of the membrane is m2. Sincethe densities of both water (w, 1.00 g/cm3), and polymers: PESUE6020P (p1, 1.37 g/cm3), and PESU-co-sPPSU 11 (p2, 1.43 g/cm3)are known, the overall porosity ε was then obtained as follows:

ε = (m1 − m2)/w

(m1 − m2)/w + m2/p(5)

2.4.4. Mechanical strengthsThe mechanical properties of membrane substrates were mea-

sured by using an Instron 5542 tensile testing equipment. Theflat sheet membranes were cut into stripes with 5 mm width andclamped at the both ends with an initial gauge length of 25 mm anda testing rate of 10 mm/min. At least three stripes were tested foreach casting condition to obtain the average values of tensile stress,extension at break and Young’s modulus of the membranes.

2.5. Mass transport characteristics of TFC-FO membranes

The water permeability and salt permeability of TFC-FO mem-branes were determined by testing the membranes under the ROmode following the method used by Wang et al. [30]. The waterpermeability coefficient (A) was obtained from the pure water per-meation flux under the applied trans-membrane pressure of 50 psi(3.45 bar). The salt rejection (Rs) was determined from the mea-sured conductivities of permeate and feed by using feed water



Fig. 2. Schematic diagram of the laboratory-scale FO set-up (co-current crossflowof feed and draw solutions are used).

containing 400 ppm NaCl at 50 psi (3.45 bar). The salt permeabil-ity coefficient (B), which is the intrinsic property of a membrane,was determined based on the solution-diffusion theory [5,33,34]:

1 − Rs

Rs= B

A(P − )(6)

2.6. Water reclamation through forward osmosis tests of TFC-FOmembranes

FO experiments were conducted on a lab-scale circulating fil-tration unit, as shown in Fig. 2 [17,30]. The cross-flow permeationcell was a plate and frame design with a rectangular channel oneach side of the membrane. The solution flow velocities duringFO tests were kept at 8.33 cm/s for both feed and draw solutionswhich co-currently flowed through the cell channels. The tempera-tures of the feed and draw solutions were maintained at 22 ± 0.5 C.The membranes were tested under two different modes: (1) pres-sure retarded osmosis (PRO mode) where the draw solution facesagainst the dense selective layer and (2) FO mode where the feedwater side faces against the dense selective layer.

The draw solutions were prepared from NaCl solutions with dif-ferent concentrations. The change of draw solution concentrationwas ignored because the ratio of water permeation flux to the vol-ume of the draw solution was less than 2% during the FO testing.When using the deionized water as the feed, the salt leakage can becalculated by measuring the conductivity in the feed solution at theend of experiment. A balance (EK-4100i, A&D Company Ltd., Japan)connected to a computer recorded down the mass of water per-meating into the draw solution over a selected period of time. Thewater permeation flux was then calculated according to the weightchange of feed water. The water permeation flux (Jv, L m−2 h−1,abbreviated as LMH) is calculated from the volume change of thefeed or draw solution.

Jv = V

Sm t(7)

where V (L) is the permeation water collected over a predeter-mined time t (h) in the FO process duration; Sm is the effectivemembrane surface area (m2).

The salt concentration in the feed water was determined fromthe conductivity measurement using a calibration curve for the sin-gle salt solution. The salt leakage, salt reverse-diffusion from thedraw solution to the feed, Js in g m−2 h−1 (abbreviated as gMH), isthereafter determined from the increase of the feed conductivity:

Js = (CtVt)Sm t

(8)

where Ct and Vt are the salt concentration and the volume of thefeed at the end of FO tests, respectively.

The water flux in FO processes can be modeled by the followingequations [5,35]. For the PRO mode (selective layer against the drawsolution):

Jw = 1Km

InAD,m − Jw + B

AF,b + B(9)

For the FO mode (selective layer against the feed solution):

Jw = 1Km

InAD,b + B

AF,m + Jw + B(10)

where D,b and F,b refer to the osmotic pressures in the respectivebulk draw solution and feed, D,m and F,m are the correspond-ing osmotic pressures on membrane surfaces facing the draw andfeed solutions after considering the external concentration polar-ization effect (ECP) [36,37] as shown in Fig. 3. The relationshipamong solute diffusion resistivity within the porous layer Km, diffu-sivity Ds, membrane structural parameter S, membrane turtuosity, membrane thickness l and membrane porosity, ε can be repre-sented as follows:

Km = S

Ds= l

εDs(11)

3. Results and discussion

3.1. Characteristics and performance of membrane substrates

It is well-known that blending a hydrophilic material into apolymeric casting solution can transform the phase inversion pathand alter membrane properties, such as morphology, mechani-cal strength, hydrophilicity, pore size and its distribution. Fig. 4exemplifies the SEM morphology of membrane substrates castfrom different concentrations of sulphonated polymer as tabulatedin Table 1. The cross-section of membranes cast from pure PESUE6020P (0 wt% sulphonated material) results in membranes with alarge number of finger-like macrovoids similar to those literaturereports made from PSf [27–30]. On the other hand, the membranesubstrate containing 50 wt% sulphonated polymer exhibits a fullysponge-like structure with no observed macrovoids. This is due tothe delayed demixing induced by the sulphonated material whichwould reduce macrovoid formation at higher sulphonated materialcontent. In summary, traces of macrovoids can be easily observed atthe bottom surface of membranes (500× magnification) cast fromdopes comprising 0 and 25 wt% sulphonated material, while nomacrovoids can be found in those membranes containing 50 wt%sulphonated material. The thicknesses of these membranes are inthe range of 40–55 m.

Under SEM observation at a magnification of 15,000, allcast membranes show almost the same top surface morphol-ogy even though they were cast from different concentrations ofsulphonated polymer. However, they have different surface rough-ness as examined by AFM and displayed in Fig. 5. The top surfaceroughness decreases with an increase in sulphonated polymerconcentration. This phenomenon might be due to a less suddencontraction of membrane surface during a delayed demixing pro-cess as compared to that of an instantaneous demixing. The mildsolvent exchange and slow phase inversion help form a smooth



Fig. 3. Schematic diagram of the effect of concentration polarization under PRO and FO modes.

Fig. 4. Typical morphology of membrane substrates with different concentrations of sulphonated polymer for TFC fabrication: (a) no sulphonated polymer; (b) 25 wt%sulphonated material; (c) 50 wt% sulphonated material.



Fig. 5. SFM images and mean surface roughness of membranes cast from: (a) 0 wt% sulphonated material; (b) 25 wt% sulphonated material; (c) 50 wt% sulphonated material.

surface. Similar phenomena have been reported by Peng et al. [38]and Sukitpaneenit and Chung [39].

Table 2 and Fig. 6 represent the PWP and pore size characteris-tics of membrane substrates cast with different concentrations ofsulphonated polymer. It is interesting to take note that the PWP ofmembrane substrates follows the order: 25 wt% sulphonated poly-mer >0 wt% sulphonated polymer >50 wt% sulphonated polymer.The membrane substrate comprising 25 wt% sulphonated polymerhas a higher PWP than that made of pure PESU due to the incrementin hydrophilicity. However, the membrane substrate containing50 wt% sulphonated polymer results in a lower PWP than that con-taining 25 wt% sulphonated polymer. This interesting phenomenonis due to the fact that the former has a higher degree of water-induced swelling than the latter. In addition, the MWCO of as-castmembranes is in the following order: 50 wt% sulphonated polymer>25 wt% sulphonated polymer >0 wt% sulphonated polymer. Thissequence is consistent with our hypothesis that the sulphonatedpolymer results in a relatively higher porosity and larger pore sizesince it induces delayed demixing during phase inversion, as con-firmed in Table 2.

Further characterization on contact angle indicates that PESUmembrane substrates without any sulphonation material have acontact angle of 77.3. On the other hand, the substrates compris-ing 25 or 50 wt% sulphonation material have relatively low contactangles in the range of 15–20 or lower because the measured waterdroplet spread immediately on membrane surfaces due to theincreased hydrophilicity. Table 3 summarizes mechanical strengthsof these substrates. Young’s modulus decreases, while elongation atbreak increases with an increase in sulphonated polymer content.When the content of sulphonated materials goes beyond 50 wt%,the mechanical strengths of membrane substrates are not sufficientto be tested under FO processes.

3.2. Characteristics and performance of TFC-FO membranes

TFC-FO membranes were conducted on the aforementionedmembrane substrates consisting of different concentrations ofsulphonated polymer via interfacial polymerization for FO stud-ies. Fig. 7 displays a typical image of top surface and selective

layer morphology of TFC membrane on a substrate comprising50 wt% sulphonated polymer. The calculated thickness of selectivepolyamide layer is 180.5 nm.

Table 4 compares the PRO and FO performance of TFC-FOmembranes using 2 M NaCl as a draw solution. The thin-film poly-merization conducted on a sponge-like substrate made of 50 wt%sulphonated polymer has the best FO performance. It has the high-est water flux and the lowest salt leakages tested under both PROand FO modes. Water fluxes of 33 and 21 LMH can be achievedfor PRO and FO modes respectively using a 2 M NaCl as the drawsolution.

Based on the membrane substrate morphology, our resultsshow the opposite trend as compared to the work reported byYip et al. [27], Tiraferri et al. [40] and Wang et al. [28,29]. It washypothesized in their papers that an ideal substrate for TFC-FOmembranes should possess a large portion of finger-like struc-ture with a thin layer of sponge-like structure near the membranetop surface to enhance water transport through the TFC-FO mem-branes. On the contrary, in our work, the TFC-FO membraneswith fully sponge-like structure exhibits the highest water flux inFO tests as compared to those with part of finger-like structure,i.e. 25 wt% sulphonated material, or fully finger-like structure, i.e.0 wt% sulphonated material. The sponge-like structure is preferablesince it can provide better performance stability for the TFC-FOmembrane in the long term. Clearly, the opposite trend betweenour and their works is mainly due to the higher hydrophilic andporous nature of our membrane substrates. A membrane substratewith a highly finger-like structure is not the essential requirementto form a high flux TFC-FO membrane.

In comparison of PRO and FO data in Table 4, the water fluxes inthe PRO mode for all TFC-FO membranes are much higher thanthose of FO mode. In the PRO mode, the internal concentrationpolarization can be eliminated significantly since the selective layeris located on the membrane’s top surface. However, in the FO mode,the water fluxes are reduced tremendously due to the severe ICPeffect within the porous bottom layer.

Table 5 summarizes the basic transport properties. The waterand salt permeability coefficients of TFC-FO membranes have rela-tively similar values regardless the concentration of sulphonated

Table 2Summary of mean effective pore size (p), PWP and MWCO of membrane substrates cast with different concentrations of sulphonated polymer.

Composition of solution casting 0 wt% sulphonated content 25 wt% sulphonated content 50 wt% sulphonated content

p (nm) 12.5 15.4 16.8p 2.17 2.03 2.00PWP (L m−2 bar−1 h−1) 370.01 464.33 188.63MWCO (Da) 296,116 345,469 414,829Porosity (%) 80.33 84.29 84.94



Fig. 6. (a) Solute separation curves, (b) probability density function curves and (c) cumulative pore size distribution curve.

Table 3Mechanical properties of membrane substrates cast with different concentrations of sulphonated polymer.

Membrane ID Young’s modulus (MPa) Tensile strength (MPa) Elongation at break (%)

Membrane substrate with casting solution 1 (0 wt% sulphonated polymer) 151.7 ± 7.9 5.5 ± 0.2 55.9 ± 6.7Membrane substrate with casting solution 2 (25 wt% sulphonated polymer) 123.4 ± 2.9 3.5 ± 0.03 26.2 ± 3.2Membrane substrate with casting solution 3 (50 wt% sulphonated polymer) 80.1 ± 4.8 3.7 ± 0.3 64.4 ± 6.2

Fig. 7. Typical FESEM images of TFC-FO membrane with 50 wt% sulphonated material in the support layer.

Table 4PRO and FO performance of TFC-FO membranes with different concentrations of sulphonated polymer in the membrane support layer.

Sample ID Substrate PRO mode FO mode

Water flux (LMH) Salt leakage (gMH) Water flux (LMH) Salt leakage (gMH)

1 TFC-FO with casting solution 1 (0 wt% sulphonated polymer) 13.5 3.7 10.5 3.12 TFC-FO with casting solution 2 (25 wt% sulphonated polymer) 21.0 3.7 16.5 3.13 TFC-FO with casting solution 3 (50 wt% sulphonated polymer) 33.0 2.8 21.0 2.2



Table 5Transport properties and structural parameters of TFC-FO membranes with different concentrations of sulphonated materials in the membrane substrates.

Membrane Waterpermeability, A(L m−2 h−1 bar−1)

Salt rejectiona (%) Salt permeability, B(L m−2 h−1 (m s−1))

Km (s m−1) Sb (m)

TFC-FO with casting solution 1 (0 wt% sulphonated polymer) 0.68 91.4 0.22 6.36 × 105 9.63 × 10−4

TFC-FO with casting solution 2 (25 wt% sulphonated polymer) 0.72 91.5 0.23 3.38 × 105 5.12 × 10−4

TFC-FO with casting solution 3 (50 wt% sulphonated polymer)* 0.73 91 0.25 2.14 × 105 3.24 × 10−4

a Tested at 50 psi (3.45 bar) with 400 ppm NaCl solution.b Structural parameters were calculated based on experiments under the FO mode using 2 M NaCl as the draw solution and DI water as the feed.

0 1 2 3 4 50

10

20

30

40

50

PROFO

Rev

erse

sal

t flu

x (g

MH

)

Dra w solution conce ntra tion, NaCl (M)

0 1 2 3 4 510

20

30

40

50

60

Wat

er fl

ux (L

MH

)

Dra w solution conce ntra tion, NaCl (M)

PROFO

Fig. 8. The water fluxes and salt leakages of TFC-FO membranes (50 wt% sulphonated material in the membrane substrate) in the FO and PRO tests with varying draw solutionconcentrations (NaCl) using DI water as feed.

material in membrane substrates. Interestingly, the calculatedstructural parameter (S) decreases with an increase in sulphonatedcontent in membrane substrates. This implies that a lower ICP effectcan be achieved when the TFC-FO membrane was developed on amembrane substrates consisting of higher hydrophilic or porousnature for FO applications.

Furthermore, since the TFC-FO membrane comprising 50 wt%sulphonated content shows the best FO performances, it wasfurther tested for both PRO and FO modes using different NaCl con-centrations as draw solutions as shown in Fig. 8. The water fluxesincrease linearly in the PRO and FO modes at low draw solutionconcentrations, while it seems to be leveled off at higher concen-trations. This phenomenon is most likely attributed by the dilutiveECP within the boundary layer at the membrane surface and ICPwithin the support layer which considerably reduce an efficiencyof osmotic driving force due to higher salt leakage at higher drawsolution concentration.

3.3. Osmotic seawater desalination

To be applicable for sea desalination, a model seawater solu-tion (3.5 wt% NaCl) was used to test the best performance TFC-FOmembranes. Compared to previous tests using DI water as a feedsolution, Fig. 9 indicates that water fluxes decrease in both PROand FO modes when seawater is used as the feed. This is due to thereduction of the overall osmotic pressure difference between thefeed and the draw solution. The use of feed seawater also furtherreduces the water flux difference between PRO and FO modes ascompared to those using DI water as the feed. Since both ICP andECP are functions of water flux and effective mass transfer coef-ficient, the reduced water flux when using seawater in the testsreduces the concentration polarization which in turn diminishesthe water flux difference between PRO and FO modes under thesame draw solution.

2 3 4 512

14

16

18

20

22

24

26

28

PRO FO

Wat

er fl

ux (L

MH

)

Draw solution concentration, NaCl (M)

PRO~15 LMH

Fig. 9. The water flux in the FO and PRO tests with varying draw solution concen-trations (NaCl) using seawater concentration (3.5 wt% NaCl) as feed.

In this work, it is found that the water flux can achieve 15 LMHfor the PRO mode and 13.5 LMH for the FO mode, respectively,using 2 M NaCl as a draw solution. To the best of our knowledge,this is the best data for TFC-FO membranes tested with seawaterconcentration using 2 M NaCl as the draw solution.

4. Conclusions

This work introduces the effects of different amounts ofsulphonated material in the TFC-FO membrane substrates on theoverall FO performance and membrane morphology. It reveals acontradictory finding that the TFC-FO membranes derived fromsubstrates of hydrophobic nature and finger-like structures do not



necessarily facilitate a higher water flux in FO processes than thoseof hydrophilic characteristics and sponge-like structures. The fol-lowing conclusions can be further drawn from this work:

1. The TFC-FO membranes comprising the most hydrophilic char-acteristics of membrane substrate, i.e. 50 wt% sulphonatedcontent, exhibit fully sponge-like structure morphology and thehighest water flux of 33.0 and 21 LMH tested under PRO and FOmodes, respectively, against DI water with 2 M NaCl as the drawsolution. Using 3.5 wt% NaCl as the mode feed seawater and 2 MNaCl as the draw solution, this membrane shows the highestwater flux of 15 LMH under the PRO mode, which is the best sofar among available literatures.

2. The TFC-FO membranes without comprising sulphonated mate-rial in the membrane substrates exhibit a fully finger likestructure with relatively low water fluxes of 13.5 and 10.5 LMHtested under PRO and FO modes, respectively, against DI waterwith 2 M NaCl as the draw solution.

3. The structural parameter, an indicator of potential internal con-centration polarization (ICP), can be decreased with an increasein sulphonated material content in membrane substrates.

Acknowledgements

The author would like to thank BASF SE, Germany for fundingthis research project with a grant number of R-279-000-283-597.The author would like to thank the Singapore National ResearchFoundation (NRF) for support through the Competetive ResearchProgram for the project entitled, “New Advanced FO membranesand membrane systems for wastewater treatment, water reuseand seawater desalination” (grant number R-279-000-339-281).Thanks are also due to Dr. K.Y. Wang, Ms. S. Zhang, and Ms. X. Lifor their help and suggestions on this work. We are also appreci-ated Ms. X.R. Teh for her help on the experimental works. Specialthanks are due to Ms. M.L. Chua for her help on the polymer densitymeasurements.

Nomenclature

A water permeability (L m−2 h−1 bar−1)B salt permeability (L m−2 h−1)Cf feed concentration (mol L−1)Cp permeate concentration (mol L−1)Ct salt concentration (mol L−1)dp pore diameter (nm)ds solute diameter (nm)Ds diffusion coefficient of salt in the membrane sub-

strate (m2 s−1)ECP external concentration polarizationFO forward osmosisICP internal concentration polarizationJs reverse salt flux (g m−2 h−1)Jw water flux (L m−2 h−1)k water transport coefficient (m s−1)kb mass transfer coefficient (m s−1)kD mass transfer coefficient (m s−1)Km solute diffusion resistivity within the porous layer

(s m−1)l thickness (m)m membrane weight (g)M molecular weight (g mol−1)

MWCO molecular weight cut-off (kDa)P pressure (bar)PRO pressure-retarded osmosisr Stokes radius (nm)R solute rejectionS membrane structural parameter (m)Sm effective membrane surface area (m2)TFC thin-film-compositet operation time interval (h)V water permeation volume (L)Vt volume of the feed at a time interval of t (L)ε membrane porosity osmotic pressure (bar) material density (g/cm3) geometric mean radius geometric standard deviation turtuosity

Subscriptsb bulk solutionD draw solution sideF feed solution sidei inside of the active layer within the porous supportm membranes solute

References

[1] M.A. Shannon, P.W. Bohn, M. Elimelech, J.G. Georgiadis, B.J. Marinas, A.M.Mayes, Science and technology for water purification in the coming decades,Nature 452 (2008) 301.

[2] T.S. Chung, S. Zhang, K.Y. Wang, J.C. Su, M.M. Ling, Forward osmosis processes:yesterday, today and tomorrow, Desalination, in press.

[3] L. Liu, M. Wang, D. Wang, C. Gao, Current patents of forward osmosis membraneprocess, Recent. Pat. Chem. Eng. 2 (2009) 76.

[4] B. Mi, M. Elimelech, Chemical and physical aspects of organic fouling of forwardosmosis membranes, J. Membr. Sci. 320 (2008) 292.

[5] K.L. Lee, R.W. Baker, H.K. Londsdale, Membranes for power generation bypressure-retarded osmosis, J. Membr. Sci. 8 (1981) 141.

[6] K. Gerstandt, K.V. Peinemann, S.E. Skilhagen, T. Thorsen, T. Holt, Membraneprocesses in energy supply for an osmotic power plant, Desalination 224 (2008)64.

[7] S. Loeb, Production of energy from concentrated brines by pressure-retardedosmosis. I. Preliminary technical and economic correlations, J. Membr. Sci. 1(1976) 49.

[8] B. Jiao, A. Cassano, E. Drioli, Recent advances on membrane processes for theconcentration of fruit juices: a review, J. Food Eng. 63 (2004) 303.

[9] K.B. Petrotos, H.N. Lazarides, Osmotic concentration of liquid foods, J. Food Eng.49 (2001) 201.

[10] Q. Yang, K.Y. Wang, T.S. Chung, A novel dual-layer forward osmosis membranefor protein enrichment and concentration, Sep. Purif. Technol. 69 (2009) 269.

[11] M.M. Ling, T.S. Chung, Novel dual-stage FO system for sustainable proteinenrichment using nanoparticles as intermediate draw solutes, J. Membr. Sci.372 (2011) 201.

[12] J.R. McChtcheon, R.L. McGinnis, M. Elimelech, A novel ammonia–carbon dioxideforward (direct) osmosis desalination process, Desalination 174 (2005) 1.

[13] M.M. Ling, K.Y. Wang, T.S. Chung, Highly water soluble magnetic nanoparticlesas novel draw solutes in forward osmosis for water reuse, Ind. Eng. Chem. Res.49 (2010) 5869.

[14] Q.C. Ge, J.C. Su, T.S. Chung, G. Amy, Hydrophilic superparamagnetic nanoparti-cles: synthesis, characterization and application in forward osmosis processes,Ind. Eng. Chem. Res. 50 (2011) 382.

[15] K.Y. Wang, T.S. Chung, J.J. Qin, Polybenzimidazole (PBI) nanofiltration hollowfiber membranes applied in forward osmosis process, J. Membr. Sci. 300 (2007)6.

[16] K.Y. Wang, Q. Yang, T.S. Chung, Enhanced forward osmosis from chemicallymodified polybenzimidazole (PBI) nanofiltration hollow fiber membranes witha thin wall, Chem. Eng. Sci. 64 (2009) 1577.

[17] K.Y. Wang, R.C. Ong, T.S. Chung, Double-selective layer forward osmosis mem-branes for the elimination of internal concentration polarization within theporous sublayer, Ind. Chem. Eng. Res. 49 (2010) 4824.

[18] Q. Yang, K.Y. Wang, T.S. Chung, Dual-layer hollow fibers with enhanced flux asnovel forward osmosis membranes for water production, Environ. Sci. Technol.43 (2009) 2800.



[19] J. Su, Q. Yang, J.F. Teo, T.S. Chung, Cellulose acetate nanofiltration hollow fibermembranes for forward osmosis processes, J. Membr. Sci. 355 (2010) 36.

[20] S. Zhang, K.Y. Wang, T.S. Chung, H.M. Chen, Y.C. Jean, G. Amy, Well-constructedcellulose acetate membranes for forward osmosis: minimized internal concen-tration polarization with an ultra-thin selective layer, J. Membr. Sci. 360 (2010)522.

[21] R. Hausman, B. Digman, I.C. Escobar, M. Coleman, T.S. Chung, Functionalizationof polybenzimidazole membranes to impart negative charge and hydrophilic-ity, J. Membr. Sci. 363 (2010) 195.

[22] J.E. Cadotte, Interfacially synthesized reverse osmosis membrane, US Patent4,277,344 (1981).

[23] R.J. Petersen, Composite reverse-osmosis and nanofiltration membranes, J.Membr. Sci. 83 (1993) 81.

[24] S. Verissimo, K.V. Peinemann, J. Bordado, Thin-film composite hollow fibermembranes: an optimized manufacturing method, J. Membr. Sci. 264 (2005)48.

[25] J.R. McCutcheon, M. Elimelech, Influence of membrane support layer hydropho-bicity on water flux in osmotically driven membrane processes, J. Membr. Sci.318 (2008) 458.

[26] J.C. Su, T.S. Chung, Experimental and theoretical study of sublayer structureand its effect on concentration polarization and membrane performance in FOprocesses, J. Membr. Sci. 376 (2011) 214.

[27] N.Y. Yip, A. Tiraferri, W.A. Phillip, J.D. Schiffman, M. Elimelech, High perfor-mance thin-film composite forward osmosis membrane, Environ. Sci. Technol.44 (2010) 3812.

[28] R. Wang, L. Shi, C.Y. Tang, S. Chou, C. Qiu, A.G. Fane, Characterization of novelforward osmosis hollow fiber membranes, J. Membr. Sci. 355 (2010) 158.

[29] S. Chou, L. Shi, R. Wang, C.Y. Tang, C. Qiu, A.G. Fane, Characteristics and potentialapplications of a novel forward osmosis hollow fiber membrane, Desalination261 (2010) 365.

[30] K.Y. Wang, T.S. Chung, G. Amy, Developing thin-film-composite forwardosmosis membranes based on the PES/SPSf substrate through interfacial poly-merization, AIChE J., in press.

[31] S. Singh, K. Khulbe, T. Matsuura, P. Ramamurthy, Membrane characteri-zation by solute transport and atomic force microscopy, J. Membr. Sci.142 (1998) 111.

[32] K.Y. Wang, T. Matsuura, T.S. Chung, W.F. Guo, The effects of flow angle and shearrate within the spinneret on the separation performance of poly(ethersulfone)(PES) ultrafiltration hollow fiber membranes, J. Membr. Sci. 240 (2004) 67.

[33] H.K. Londsdale, U. Merten, R.L. Riley, Transport properties of cellulose acetateosmosis membranes, J. Appl. Polym. Sci. 9 (1965) 1341.

[34] S. Loeb, G.D. Mehta, A two-coefficient water transport equation for pressure-retarded osmosis, J. Membr. Sci. 4 (1979) 351.

[35] G.D. Mehta, S. Loeb, Internal polarization in the porous substructure of asemipermeable membrane under pressure-retarded osmosis, J. Membr. Sci. 4(1978) 261.

[36] J.R. McCuthcheon, M. Elimelech, Modeling water flux in forward osmosis:implications for improved membrane design, AIChE J. 53 (2007) 1736.

[37] J.R. McCuthcheon, M. Elimelech, Influence of concentrative and dilutive internalconcentration polarization on flux behavior in forward osmosis, J. Membr. Sci.284 (2006) 237.

[38] M. Peng, H. Li, L. Wu, Q. Zheng, Y. Chen, W. Gu, Porous poly(vinylidene fluo-ride) membrane with highly hydrophobic surface, J. Appl. Polym. Sci. 98 (2005)1358.

[39] P. Sukitpaneenit, T.S. Chung, Molecular design of the morphology and pore sizeof PVDF hollow fiber membranes for ethanol–water separation employing themodified pore-flow concept, J. Membr. Sci. 374 (2011) 67.

[40] A. Tiraferri, N.Y. Yip, W.A. Phillip, J.D. Schiffman, M. Elimelech, Relating perfor-mance of thin-film composite forward osmosis membranes to support layerformation and structure, J. Membr. Sci. 367 (2011) 340.


Thin-film composite forward osmosis membranes with novel hydrophilicsupports for desalination

Gang Han a, Tai-Shung Chung a,n, Masahiro Toriida b, Shoji Tamai b

a Department of Chemical and Biomolecular Engineering, National University of Singapore, 4 Engineering Drive 4, Singapore 117602, Singaporeb New Materials Development Center, Mitsui Chemicals, Inc., 580-32 Nagaura Sodegaura-City, Chiba 299-0265, Japan


Article history:

Received 5 June 2012

Received in revised form

20 August 2012

Accepted 2 September 2012Available online 10 September 2012

Keywords:

Composite membrane

Interfacial polymerization

Sulphonated poly(ether ketone)

Hydrophilic substrates

Forward osmosis

Desalination

a b s t r a c t

In this work, a novel sulphonated poly(ether ketone) (SPEK) polymer with super-hydrophilic nature

was designed as the substrate material to fabricate high performance thin-film composite (TFC)

membranes for desalination via forward osmosis (FO). m-Phenylenediamine (MPD) and 1,3,5-trime-

soylchloride (TMC) were employed as the monomers for the interfacial polymerization reaction to form

a thin aromatic polyamide selective layer. It has been demonstrated that blending a certain SPEK

material into the polysulfone (PSU) substrate of TFC-FO membranes not only plays the key role to form

a fully sponge-like structure, but also enhances membrane hydrophilicity and reduces structure

parameter. The TFC-FO membrane comprising 50 wt% SPEK in the substrate shows the highest water

flux of 50 LMH against deionized water and 22 LMH against the 3.5 wt% NaCl model solution,

respectively, when using 2 M NaCl as the draw solution tested under the pressure retarded osmosis

(PRO) mode (draw solution flows against the selective layer). It is found that the hydrophilicity and

thickness of the substrates for TFC-FO membranes play much stronger roles in facilitating high water

flux in FO for desalination compared to those made from hydrophobic substrates full of finger-like

structures. Moreover, the reduced membrane structural parameter indicates that the internal

concentration polarization (ICP) can be remarkably reduced via blending a hydrophilic material into

the membrane substrates. Thermal treatment of TFC-FO membranes with optimized conditions can

also improve the membrane performance and mechanical strength.

& 2012 Elsevier B.V. All rights reserved.

1. Introduction

Global water scarcity is a significant problem that continues togrow worse with the rapid economic development and popula-tion growth. Development of alternative water sources by apply-ing innovative membrane technologies with low costs, such asseawater desalination and wastewater reclamation, has becomeone of the most promising approaches to meet this criticalchallenge [1–5]. Over the past few decades, reverse osmosis(RO) technology has established as the industry benchmark formembrane-based water reuse and desalination because of itssuperior product water quality and competitive cost. However, itsefficiency and sustainable operation are hampered by consider-able energy consumption and membrane fouling [1,2,4]. Recently,forward (direct) osmosis (FO) has gained much attention as anemerging alternative technology to conventional pressure-drivenmembrane processes for seawater desalination [5–9], wastewater

treatment [10], food processing [11], protein, pharmaceuticalenrichment [12,13], and power generation [14–17].

In lieu of the applied hydraulic pressure, FO employs the osmoticpressure gradient as the driving force to induce osmotic flow from alower solute/particle (correspondingly higher solvent) concentra-tion feed solution to a higher solute/particle (correspondingly lowersolvent) concentration draw solution [5,18]. The absence of externalhydraulic pressure in FO is expected to reduce system energyconsumption, equipment costs and the associated deleteriouseffects to the quality of the product water [1,2,19]. Moreover, FOmay offer the advantages of higher rejections to a wide range ofcontaminants and lower membrane fouling propensities and higherwater recovery compared to pressure-driven processes [2,20–22].However, the inadequate number of commercially available highperformance FO membranes [5,18,22–24] and the lack of drawsolutes with friendly characteristics of low cost, non-toxicity,minimal reverse salt flux and easy of recycle [25–27] deter thefull implementation of FO technology. In order to develop poly-meric FO membranes with appropriate separation performance,two fabrication techniques have been often adopted: (1) asym-metric membranes made by non-solvent induced phase inversion[9,28–32], and (2) thin-film-composite (TFC) membranes made by


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

Journal of Membrane Science

0376-7388/$ - see front matter & 2012 Elsevier B.V. All rights reserved.

http://dx.doi.org/10.1016/j.memsci.2012.09.005

n Corresponding author. Tel.: þ65 65166645; fax: þ65 67791936.

E-mail address: [email protected] (T.-S. Chung).

Journal of Membrane Science 423–424 (2012) 543–555


interfacial polymerization on porous support layers [6,23,24,33–39].Compared to the conventional asymmetric membranes (i.e., celluloseacetate-based membranes), TFC membranes have advantages suchas a higher water permeability, greater solute rejection, and non-biodegradability [23,33,40,41].

Aromatic polyamide TFC membranes were invented in 1970s.Since then, they have dominated the production of modernRO membranes [36–39,42–44]. Generally, the polyamide TFCmembranes have a highly anisotropic structure consisting of a poroussupport layer and a thin rejection layer prepared via interfacialpolymerization between an aqueous polyfunctional amine solutionand a polyfunctional acyl chloride dissolved in an apolar organicsolvent [6,7,17,23,33–39]. The support layer and active rejection layercan be independently tailored. Hence it permits easy optimization ofthe overall composite membrane with desirable separation perfor-mance (i.e., permeability, selectivity, and stability). These advantagesinspire membrane scientists to explore polyamide TFC membranesfor FO applications [6,7,23,24,33–35].

The design scheme for TFC-FO membranes is different fromthat for RO membranes. One major difference between RO and FOTFC membranes is their requirements for the membrane support.Usually, in order to withstand the high trans-membrane pressureof up to 60–100 bar, a thick and mechanically strong hydrophobicsupport should be required for TFC-RO membranes. However, FOis generally conducted under low or no applied pressures, a thinsupport is desired. In addition, a thin support layer can effectivelyminimize the effect of internal concentration polarization (ICP) inthe FO process. This ICP generally cannot be eliminated by increasingthe cross-flow velocity and turbulence along the surface because ithappens inside the porous support layer which is different from theexternal concentration polarization (ECP) encountered in pressuredriven membrane processes. Moreover, in FO processes both the topand bottom membrane surfaces have to simultaneously contact withtwo solutions, and water transport through the semi-permeablemembrane is based on the solution-diffusion mechanism, thus thesubstrate physicochemical properties (hydrophilicity, porosity, poresize, pore-size distribution, and substructure resistance) also playvery important roles in the overall FO performance [23].

The desired TFC-FO membranes should be: (1) an ultrathinsemi-permeable active layer with a high solute rejection andwater permeation flux; (2) a thin supporting layer with low ICP;(3) hydrophilic nature to enhance FO performance and reducemembrane fouling; and (4) sufficient mechanical strength androbust to sustain backwash, cleaning and vibration in industrialoperations [5,6,23,45]. Studies have shown that the physicochem-ical properties of the substrates play key roles in determining theformation and morphology of the polyamide layer as well as theseparation performance of the resultant TFC-FO membranes. Forexample, the rate and amount of m-phenylenediamine (MPD)diffusion into the reaction interface may alter the breadth of thereaction zone, the extent of interfacial polymerization reaction,and the polyamide layer thickness [23,46]. Kim and Kim [47]reported that the modification of the polysulfone support byplasma with hydrophilic materials prior to interfacial polymer-ization led to an increase in not only the water flux and rejectionbut also chlorine resistance. Singh et al. [48] showed that poly-sulfone support with a larger pore size resulted in the TFCmembrane with a high water flux but a low rejection. Ghoshand Hoek [49] reported that the hydrophobicity of the supportlayer influenced the meniscus shape that formed between theorganic and aqueous phases during interfacial polymerization andtherefore determined the properties of the polyamide layer andthe overall separation performance.

Recently, high performance TFC-FO membranes have been fab-ricated by introducing a certain amount of hydrophilic sulphonatedpolymers, like sulphonated polysulfone [23] and PESU-co-sPPSU 11

hydrophilic polymer [6], and sulphonated poly(arylene ether sul-fone)s [50] into the substrate membrane matrix. The blendedsulphonated materials not only adjust the hydrophilicity and themorphology of the membrane substrates but also significantlyenhance the FO performance of the result TFC membranes. Sulpho-nated poly(ether ketone) (SPEK) polymers are one of the mostimportant hydrophilic sulphonated aromatic polymers [51–53]. Inaddition to the unique chemical and physical properties such as highhydrolytic, thermal and oxidation stabilities, adjustable hydrophili-city and excellent mechanical properties, they are characteristic oflow costs and easy processability [54]. According to the aforemen-tioned requirements for TFC-FO membranes, we believe the additionof SPEK into the support substrate may significantly enhance theperformance of the TFC-FO membranes.

Therefore, the objectives of this study are (1) to fabricate highperformance TFC-FO membranes for desalination using a MitsuiChemicals’ hydrophilic SPEK as part of substrate materials; (2) toexamine the relationship between substrate morphology andresultant TFC-FO membranes’ performance as a function of SPEKcontent; and (3) to reveal the fundamental science bridging SPEKchemistry, SPEK loadings, membrane substrates properties, TFCformation, thermal treatment and membrane separation perfor-mance. To our best knowledge, this is the first study reporting thedevelopment and application of the hydrophilic SPEK material forTFC-FO membranes.

2. Experimental

2.1. Materials

Udels polysulfone (PSU, UDEL P-3500) was used as thepolymer material for the fabrication of membrane substrates.m-Phenylenediamine (MPD) with 499% purity and trimesoylchloride (TMC) with 98% purity supplied by Sigma-Aldrich wereused as the monomers for the interfacial polymerization reaction.N-hexane from Merck with 499.0% purity was utilized as thesolvent for the TMC monomer. Bis(4-hydroxy-3,5-dimethyl-phenyl)methane (TMBPF) and 4,40-difluorobenzophenone (DFBP) fromTokyo Kasei Kogyo Co. Ltd. and the 50% fuming sulfuric acid,dimethylsulfoxide (DMSO) from Wako Junyaku Kogyo Co. Ltd. wereused as received for the synthesis of SPEK polymer. N-methyl-2-pyrrolidone (NMP) and diethylene glycol (DEG) purchased fromMerck were employed as the solvent and pore former, respectively.For FO desalination tests, sodium chloride (NaCl) purchased fromMerck was dissolved in deionized (DI) water and used as the drawsolution. Polyethylene oxide (PEO) and polyethylene glycol (PEG)from Sigma-Aldrich with different molecular weights were used tomeasure the mean pore size and its distribution of membranesubstrates through the solute transport method. The deionizedwater was produced by a Milli-Q unit (Millipore) with a resistivityof 15 MO cm.

2.2. Synthesis of SPEK polymer

The sulphonated poly (ether ketone) (SPEK) polymer wassynthesized by Mitsui Chemicals using a method similar to thatdescribed by Zhou and Xiao et al. [52,55]. Typically, a mixturecontaining 772 g of DMSO, 257 g of toluene, 16.47 g (0.039 mol)of 5,50-carbonylbis(2-fluorobenzene sulfonic acid) sodium salt,76.59 g (0.351 mol) of DFBP, 99.98 g (0.390 mol) of TMBPF and67.38 g (0.488 mol) of potassium carbonate was charged into afive-necked reactor equipped with a nitrogen-introducing tube, areflux condenser, a stirrer and a thermometer. The reactionsystem was then heated to, and maintained at 140 1C with stirringfor 12 h under a nitrogen atmosphere to remove water generated

G. Han et al. / Journal of Membrane Science 423–424 (2012) 543–555544


by the system. The reaction mixture was then distilled for 2 h anddiluted by 585 g of toluene at 100 1C. The polymer powder wasprecipitated in 2500 g of methanol, then filtered and washed withmethanol and deionized water. 161.5 g (yield of 91%) of thepolymer product was collected after drying at 50 1C for 8 h and150 1C for 4 h under nitrogen atmosphere. Fig. 1 describes thechemical structures of commercial PSU and the synthesized SPEKpolymer with a degree of sulphonation (DS) of 10 wt%, deter-mined by titration results. The inherent viscosity of the polymer ina solvent (NMP/DMSO¼1/1 (wt/wt)) was 96 cm3 g1, measuredat a concentration of 0.005 g cm3 at 35 1C.

2.3. Preparation of membrane substrates

The membrane substrates were prepared by the conventionalLoeb–Sourirajan wet phase inversion method [23]. The polymersolutions with various SPEK content were prepared at roomtemperature (about 23 1C) and the solution compositions aresummarized in Table 1. The solutions were degassed for morethan 24 h after completely dissolution. Afterwards, the homo-geneous solution was cast using a knife blade at a height of100 mm over a clean glass plate at ambient temperature. Then theentire assembly was immediately immersed into a water bath atroom temperature. The phase inversed substrate membraneswere removed from the water bath and washed thoroughly withdeionized water and then stored in deionized water at roomtemperature.

2.4. Fabrication of polyamide TFC-FO membranes

The polyamide TFC-FO membranes were fabricated on the topsurface of the cast membrane substrates via the interfacialpolymerization reaction between MPD and TMC, as shown inFig. 2. The membrane substrate was firstly immersed in a 2 wt%MPD aqueous solution for 120 s. Then, the excess water dropletson the membrane surface were removed by a filter paper.Subsequently, the top surface of the substrate membrane wasbrought into contact with a 0.2 wt% TMC/hexane solution for120 s immediately. In the whole process, membranes were fixedin a frame so that only the top surface of the substrate wasexposed to the reactants. The resultant TFC-FO membranes werethen thermally treated in hot water and the treatment condi-tions will be discussed later. All the TFC-FO membranes werefinally kept in deionized water at room temperature for furthercharacterization.

2.5. Characterizations

2.5.1. Morphology, contact angle, and membrane porosity

Membrane morphology was observed via a field-emission scan-ning electron microscope (FESEM, JEOL JSM-6700F). The membranesamples dried by a freeze dryer (ModulyoD, Thermo ElectronCorporation, USA) were frozen and then fractured in liquid nitrogen

before sputtering with platinum using a sputtering coater (JEOLLFC-1300).

The surface hydrophilicity of membrane substrates was mea-sured by a Contact Angle Geniometer (Rame Hart, USA) usingMilli-Q deionized water as the probe liquid at room temperature(about 23 1C). To minimize the experimental error, the contactangle was randomly measured at more than 10 different locationsfor each sample and the average value was reported.

To measure the membrane substrate porosity, wet membraneswere carefully blotted using tissue paper to remove the excesswater on the surface. Then the wet membranes were immediatelyweighed (m1, g), freeze dried overnight, and re-weighed (m2, g).The water content was calculated as m1m2, and the dry weightof the membrane was m2. Since the densities of both water (rw,1.00 g cm3), PSU (rp1, 1.24 g cm3) and SPEK (rp2, 1.18 g cm3)are known, the overall porosity e (%) of the membrane was thenobtained as follows [6,23]:

e¼ ðm1m2Þ=rw

ðm1m2Þ=rwþm2=rp

ð1Þ

2.5.2. Pore size and pore-size distribution of membrane substrates

The mean effective pore size and pore size distribution of eachmembrane substrate were examined by solute rejection experi-ments and the detailed procedures were reported elsewhere[6,23,56]. The membrane substrate was first tested to measureits pure water permeability (PWP) (in L/(m2 bar h)) by using anultrafiltration membrane permeation cell with a sample diameterof 3.5 cm. Subsequently, the membrane was subjected to theneutral solutes (polyethylene glycol (PEG) or polyethylene oxide(PEO)) separation tests with a concentration of 200 ppm under apressure of 1 bar on the liquid solution side. The concentrations ofthe neutral solutes in the feed and permeate were measured via atotal organic carbon analyzer (TOC ASI-5000A, Shimadzu, Japan).The measured feed (Cf) and permeate (Cp) concentrations wereused for the calculation of the effective solute rejection coefficientR (%):

R¼ 1Cp

Cf

100% ð2Þ

Fig. 1. The chemical structure of: (a) PSU and (b) SPEK with a DS¼10 wt%.

Table 1Compositions of casting solutions for the fabrication of membrane substrates.

Composition of

casting

solutions (wt%)

Casting

solution 1

(50 wt% SPEK)

Casting

solution 2

(50 wt% SPEK)

Casting

solution 3

(25 wt% SPEK)

Casting

solution 4

(0 wt%

SPEK)

PSU 7.5 7.5 11.25 15

SPEK 7.5 7.5 3.75 –

DEG – 17 17 17

NMP 85 68 68 68

G. Han et al. / Journal of Membrane Science 423–424 (2012) 543–555 545


The relationship between Stokes radius (rs, nm) and molecularweight (Mw, g mol1) of the neutral solutes can be expressed as:

for PEG : r¼ 16:73 1012M0:557

ð3Þ

for PEO : r¼ 10:44 1012M0:587

ð4Þ

The mean effective pore size and the pore size distribution wereobtained based on the traditional solute transport approach byignoring the influence of the steric and hydrodynamic interactionbetween solute and membrane pores, the mean effective poreradius (mp) and the geometric standard deviation (sp) can beassumed to be the same as ms (the geometric mean radius of soluteat R¼50%) and sg (the geometric standard deviation defined as theratio of the rs at R¼84.13% over that at R¼50%). Therefore, based ondp and sp, the pore size distribution of a membrane can beexpressed as the following probability density function:

dRðdpÞ

ddp¼

1

dp lnsp

ffiffiffiffiffiffi2pp exp

ðlndplnmpÞ2

2ðlnspÞ2

" #ð5Þ

2.5.3. Mass transport characteristics of TFC-FO membranes

The water and salt permeability of the TFC-FO membraneswere characterized by testing the membranes under the ROmode [6]. The RO filtration apparatus applied is a stainless steelcell with an effective membrane area of 19.5 cm2. The waterpermeability coefficient (A) was calculated from the pure waterpermeation fluxes under a trans-membrane pressure varyingfrom 1 to 5 bar. The salt rejection (Rs) was tested by one feedwater containing 200 ppm NaCl and determined according to theconductivity measurements of permeate and feed solutions. Thesalt permeability coefficient (B) was determined according tothe solution-diffusion theory as follows [16,23,57]:

1Rs

Rs¼

B

AðDPDpÞ ð6Þ

where DP and Dp are the hydraulic pressure difference andosmotic pressure difference across the membrane, respectively.

2.5.4. FO performance of TFC-FO membranes

FO performance of the fabricated TFC membranes was eval-uated via a lab-scale cross-flow filtration unit as reported byWidjojo and Zhang et al. [6,9]. The cross-flow permeation cell wasa plate-and-frame design with a spacer free rectangular channel(2.0 cm in length, 2.0 cm in width and 0.28 cm in height) on eachside of the membrane. The solution flow velocities during the FOtests were kept at 0.2 L min1 for both the feed and drawsolutions which flowed counter-currently along the membranes.The temperatures of the feed and draw solutions were maintainedat room temperature of about 23 1C. Each membrane was eval-uated under two different modes: (1) pressure retarded osmosis

(PRO mode) where the draw solution flows against the polyamideselective layer; and (2) FO mode where the draw solution flowsagainst the porous support layer.

NaCl solutions with different concentrations (wt%) and deio-nized water were used as the draw solutions and feed solution,respectively. During FO tests, the dilution of the draw solutionwas ignored, because the ratio of water permeation flux tothe volume of the draw solution was less than 2%. A balance(EK-4100i, A&D Company Ltd., Japan) connected to a computerrecorded down the mass of water permeating into the drawsolution over a selected period of time. The water permeation flux(Jv, L m2 h1, abbreviated as LMH) is calculated from the volumechange of the feed or draw solution:

Jv ¼DV

Dt Smð7Þ

where DV (L) is the permeation water collected over a predeter-mined time Dt (h) in the FO process; Sm is the effective membranesurface area (m2).

The salt concentration in the feed water solution was deter-mined from the conductivity measurement based on a standardconcentration–conductivity curve. The salt leakage rate or saltreverse diffusion flux Js in g m2 h1 (abbreviated as gMH) fromthe draw solution to the feed side was determined from theincrease of the feed conductivity:

Js ¼DðCtVtÞ

Dt Smð8Þ

where Ct and Vt are the salt concentration and the volume of thefeed solution at the end of FO tests, respectively. To minimizeerrors, every experiment was carried out more than twice and theaverage value was reported. The water flux in FO processes wasalso modeled by the following equations [16,58]. For the PROmode (selective layer against the draw solution):

Jw ¼1

Kmln

ApD,mJwþB

ApF,bþBð9Þ


Jw ¼1

Kmln

ApD,bþB

ApF,mþ JwþBð10Þ

where pD,b and pF,b refer to the osmotic pressures in therespective bulk draw solution and feed solution, pD,m and pF,mare the corresponding osmotic pressures on membrane surfacesin the draw and feed solutions. The relationship among solutediffusion resistivity within the porous layer Km, diffusivity Ds,membrane structural parameter S, membrane tortuosity t, mem-brane thickness l and membrane porosity e can be represented asfollows:

Km ¼S

Ds¼

lteDs

ð11Þ

Fig. 2. Scheme of the interfacial polymerization reaction to form the polyamide layer.



2.5.5. Membrane mechanical strengths

The mechanical properties of membrane substrates were mea-sured by using an Instron 5542 tensile testing equipment. The flatsheet membranes were cut into stripes with 5 mm width andclamped at the both ends with an initial gauge length of 30 mm anda testing rate of 10 mm/min. At least five stripes were tested foreach casting condition to obtain the average values of tensile stress,Young’s modulus and elongation at break of the membranes.

2.5.6. Other characterizations

A FTIR (Fourier transform infrared spectroscopy) spectroscope(Bio-Rad FTS 135) over the range of 700–4000 cm1 in theattenuated total reflectance (ATR) mode was used to characterizethe membranes. Thermogravimetric analysis (TGA) data was col-lected on a TGA 2050 Thermo-gravimetric Analyzer (SDT 2960, TAInstruments, New Castle, DE) with a heating rate of 10 1C/min in anitrogen flow of 100 mL/min.

The ion exchange capacity (IEC) and the degree of sulphona-tion (DS) of the synthesized SPEK material were obtained fromthe titration results and the detail procedures were reportedelsewhere [52]. The DS and IEC values were calculated with thefollowing equations, modified from Guan et al. [59]:

DSð%Þ ¼0:434MV

W0:08MV 100 ð12Þ

IEC ¼1000DS

434þ81DSð13Þ

where W, M and V are the mass of SPEK (g), concentration ofNaOH (mol/l) and volume of NaOH (ml) reacted with the SPEK,respectively; the 434 and 81 are the molecular weights of the PEKrepeat unit and the –SO3H, respectively.


3.1. Characterization of the synthesized SPEK polymer

The DS and IEC values of the synthesized SPEK polymer arecalculated to be 10 wt% and 0.49 meq/g, respectively, suggestingthe successful preparation of the sulphonated material, which isfurther proven by the TGA measurement as shown in Fig. 3(a).Although the PSU and SPEK polymer samples were dried at 80 1C

overnight, they had to be exposed to air at atmospheric condi-tions for around 20 min when preparing for the TGA tests. It canbe seen that for PSU polymer, there is only one major weight lossstarting from around 500 1C which is ascribed to the decomposi-tion of the polymer backbone. However, three weight-loss stagesare observed for the synthesized SPEK polymer. The initial massloss below 100 1C is due to the evaporation of water bounded tothe sulphonic groups, indicating that the hydrophilic nature of theSPEK polymer which can easily adsorb certain amount of moist-ure during the short time when exposed to air. The mass losscentered at around 450 1C is attributed to the degradation of thesulphonic groups which can be observed from the FTIR spectrum(Fig. 3(b)). The final decomposition of the polymer is observed atabove 530 1C, which is in agreement with typical fully aromaticpolymers [54,55,60]. The FTIR spectra of the SPEK and PSUpolymers are shown in Fig. 3(b). The peaks at 1651, 1500 and1240 cm1 are attributed to the carbonyl stretching of phenylketone group (–Ar–C(QO)–Ar–), CQC of benzene and phenylester group (–Ar–O–), respectively. The SQO stretching vibrationof PhSO2Ph is present at 1165 cm1. The characteristic absorptionat 1192 cm1 is due to the asymmetric stretching vibration of thearomatic SO3H [51]. Compared with PSU, the synthesized SPEKpolymer is more hydrophilic because of the sulphonic groups andis expected to be a good substrate material for the fabrication ofhigh performance TFC FO membranes [6,23].


The synthesized SPEK material (15 wt% in NMP) cannot form afree standing asymmetric membrane substrate when pure water isused as the non-solvent to induce the phase inversion maybe due toits highly hydrophilic nature and slow precipitation rate. Conse-quently, polymer blends consisting of PSU and different concentra-tions of SPEK hydrophilic polymer were used to prepare thesubstrates for TFC-FO membranes. Table 1 summarizes the polymercomposition in casting solutions for substrate preparation.

3.2.1. Role of the substrate membrane structures

It is well accepted that membrane substrates for TFC-FOmembranes should have a relatively high membrane porosity andproper pore size for the purpose of enhancing water permeability,

600 800 1000 1200 1400 1600 1800 2000

Tran

smitt

ance

, %

Wavenumber, cm-10 100 200 300 400 500 600 700 800 900

0

20

40

60

80

100

120 PSU

SPEK

Wei

ght (

%)

Temperature (°C)

Fig. 3. TGA curves (a), and FTIR spectra (b) of PSU and synthesized SPEK polymer.

Table 2Summary of the mean effective pore size (mp), PWP and MWCO of the PSU/SPEK (50 wt% SPEK) membrane substrates cast from solutions with/without DEG additive.

Membrane mp (nm) sP PWP [L/(m2 bar h)] Porosity (%) MWCO (kDa)

Casting solution 1 (50 wt% SPEK without DEG) 8.3 1.22 95.5 62.6 31.1

Casting solution 2 (50 wt% SPEK with DEG) 10.7 1.25 152.7 77.2 66



while maintaining good mechanical properties. This can be partiallyachieved by the addition of some pore forming agents in the polymersolution. Table 2 and Fig. 4 display the pore size characteristics andthe PWP values of membrane substrates cast from solutions 1 and2 as tabulated in Table 1. The PWP can be significantly increased from95.5 L/(m2 bar h) to 152.7 L/(m2 bar h) after adding the pore formerDEG. Meanwhile, the MWCO of the as-cast membranes also shows anincrement form 31 kDa to 66 kDa. This is mainly because of theenlarged membrane mean pore size from 8.3 nm to 10.3 nm andmembrane porosity from 62.6% to 77.2%. In addition, as shown inFig. 4, the addition of DEG can slightly broaden the pore sizedistribution of the substrate membranes.

Fig. 5 shows the morphology of aforementioned membranesubstrates cast from solutions 1 and 2. Both substrates exhibit a

fully sponge-like structure with no observed macrovoids. This isdue to the delayed demixing induced by the hydrophilic sulpho-nated material which can suppress the formation of marovoids[6] and reduced membrane thickness [61,62]. Under SEM obser-vation at a high magnification of 50,000, the membrane substratecontaining DEG exhibits bigger pore sizes at the top and bottomsurface and more porous cross-section morphology than thosecast from solution 2. These observations are consistent with theresults shown in Table 2 and Fig. 4.

Table 3 presents the FO performance of the TFC membranesformed on the substrates cast from solutions 1 and 2 in terms ofwater flux and salt reverse flux using 2 M NaCl as the drawsolution and deionized water as the feed solution. The TFCmembranes constructed on the substrates cast from solution2 containing DEG exhibit a higher water flux and a comparablesalt reverse flux. These indicate that the addition of the poreformer DEG in the casting solution for substrates is favorable forthe fabrication of TFC-FO membranes with better performancedue to the reduced substrate resistance.

0 10 20 30 40 50

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

Pore radius, dp(nm)

Prob

abili

ty d

ensi

ty fu

nctio

n, (n

m-1

)

Casting solution 2 (50 wt% SPEK with DEG) Casting solution 1 (50 wt% SPEK without DEG)

Fig. 4. Pore-size distribution of the PSU/SPEK (50 wt% SPEK) membrane substrates.

with/without DEG additive.

Top X10,000

Top X50,000

Top X10,000

Top X50,000

1µm 1µm 100 nm 100 nm

Bottom X10,000

Bottom X50,000

Bottom X10,000

Bottom X50,000

1µm 1µm 100 nm 100 nm

Cross section X3,000




1µm 1µm

100 nm 100 nm

Fig. 5. Typical morphology of membrane substrates cast from solutions 1 and 2.

Table 3PRO and FO performance of the TFC-FO membranes formed on the PSU/SPEK

(50 wt% SPEK) membrane substrates cast from solutions with/without DEG

additive.

Sample Substrate PRO mode FO mode

Water

flux

(LMH)

Salt

leakage

(gMH)

Water

flux

(LMH)

Salt

leakage

(gMH)

TFC-FO Casting solution 1

(50 wt% SPEK) without

DEG

40 8 30 7

TFC-FO Casting solution 2

(50 wt% SPEK) with DEG

50 9 35 7

All TFC-FO membranes are thermally treated in 95 1C water for 1 min. Draw

solution: 2 M, NaCl; Feed solution: DI water.



3.2.2. Role of the sulphonated material blending concentration

It has been known that blending a hydrophilic material into acasting solution made of hydrophobic polymers can alter originalmembrane properties such as morphology, hydrophilicity, poresize and its distribution, as well as mechanical strength becauseof different phase inversion paths [6,23]. A similar phenomenon isobserved in this study. Fig. 6 shows the SEM morphology ofmembrane substrates cast from solutions comprising differentSPEK concentrations as tabulated in Table 1. The membrane castfrom pure PSU (solution 4) has a cross-section morphology full offinger-like macrovoids. However, the cross-sections of the mem-branes cast from the solutions containing 25 wt% (solution 3) and50 wt% (solution 2) SPEK exhibit a fully sponge-like structurewithout macrovoids. This is due to the fact that the sulphonatedmaterial induces the delayed demixing and suppresses themacrovoid formation. Traces of macrovoids can be observed atthe bottom surface of the membranes (10,000 magnification)cast from pure PSU (solution 4), while no macrovoids can befound in those membranes containing 25 wt% and 50 wt% SPEK.In addition, the pore size at the bottom surface, the membranethickness and the porous structure across the membrane signifi-cantly decrease with an increase in SPEK content. All the castmembrane substrates show a similar top surface morphologywith no visual pores observed at a magnification of 10,000.

Fig. 7 and Table 4 display and summarize the pore sizecharacteristics and PWP values of the membrane substrates as afunction of SPEK content, respectively. With increasing the SPEKcontent from 0 wt% to 25 wt%, the mean pore size slightlyincreases from 19.8 nm to 21.5 nm. However, a higher SPEKcontent of 50 wt% corresponds to a much smaller mean pore sizeof 10.7 nm. The pore size distribution also narrows down with anincrease in SPEK content, as indicated by the decreasing trend ofthe geometric standard deviation from 1.54 nm, 1.45 nm to1.25 nm for substrates containing 0 wt%, 25 wt% and 50 wt% SPEK,respectively. In addition, the PWP values follow the order of poresizes. In other words, the PWP value of the substrate containing

25 wt% SPEK is the highest, followed by that containing 0 wt%SPEK and then 50 wt% SPEK. The highest PWP for the substrateblended with 25 wt% SPEK arises from the effects of increasedhydrophilicity, pore size and porosity. However, a further increasein SPEK content to 50 wt% does not help PWP. On the contrary, itsPWP value is even lower than that of the PSU substrate (i.e.,comprising 0 wt% SPEK). This is due to the fact that the former hasa much higher degree of water induced chain swelling than thelatter, thus reduces water transport across the membrane [6].

Contact angle measurements indicate that the PSU membranewithout blending any SPEK has a contact angle of 81.01, while themembrane substrates blended with 25 wt% and 50 wt% SPEK haverelatively low contact angles of 76.11 and 59.41, respectively. Thedecreased water contact angles are mainly because of the additionof hydrophilic SPEK. Table 5 shows the mechanical strengths ofthese membrane substrates. With an increment in SPEK content,

Top X10,000

Bottom X10,000



Top X10,000

Bottom X10,000



Top X10,000

Bottom X10,000



1 µm 1 µm 10 µm

10 µm

10 µm

100 nm

Fig. 6. Typical substrate morphology as a function of SPEK concentration in substrates for TFC membrane fabrication: (a) casting solution 4 (0 wt% SPEK); (b) casting

solution 3 (25 wt% SPEK); and (c) casting solution 2 (50 wt% SPEK).

0 10 20 30 40 50 60 70 80

0.00

0.04

0.08

0.12

0.16

0.20

Casting solution 3 (25 wt% SPEK polymer)



Prob

abili

ty d

ensi

ty fu

nctio

n, (n

m-1

)

Pore radius, dp(nm)

Fig. 7. Pore size distribution of membrane substrates with different SPEK

concentrations.



Young’s modulus decreases, while the elongation at break increases.However, a further increase in SPEK content beyond 50 wt%, themembranes become too weak to be characterized under FOprocesses.


3.3.1. Effects of the thermal treatment

The thin polyamide layer formed via interfacial polymerizationbetween TMC and MPD (Fig. 2) is the selective layer whichdetermines the separation performance of the TFC-FO mem-branes. Curing temperature, reaction duration and posttreatmentare three important parameters controlling the quality andseparation performance of the polyamide layer. The post thermaltreatment in hot water helps to elute un-reacted monomers suchas MPD out of the TFC membranes. Table 6 summaries the FOperformance of the TFC membranes constructed on the PSU andPSU/SPEK (50 wt% SPEK) substrates after being treated withdifferent temperatures and durations. For the TFC-FO membranesconsisting of a hydrophilic PSU/SPEK (50 wt% SPEK) substrate,although a high water flux can be achieved without any thermaltreatment, a high salt revere flux is accompanied. When the TFC-FO membrane was treated at relative moderate conditions, suchas at 80 1C for 1 min, a decreased water flux and a comparablehigh reverse salt flux were observed which may be due to thethermal induced reduction in membrane permeability. However,when the thermal treatment conditions became too strong,for example 100 1C for 1 min or 95 1C for 2 min, the TFC-FOmembranes shrank and exhibited a significant reductionin performance because of the defect formation. Similar

phenomenon was observed for TFC-FO membranes with a PSUsubstrate (i.e., sample ID 7 in Table 6). In comparison, the TFC-FOmembranes treated with 95 1C water for 1 min show the bestperformance indicated by the highest Jw/Js values. In addition,different substrates show different responses to thermal treat-ment, indicating that the effects of thermal treatment on TFC-FOmembrane are restrained by the properties of the substrate.

In fact, thermal treatment will also simultaneously alter thesubstrate properties, particularly for the SPEK material used in this

Table 4Summary of mean effective pore size (mp), PWP and MWCO of membrane substrates cast from solutions with different SPEK concentrations.

Membrane mp (nm) sP PWP [L/(m bar h)] MWCO (kDa) Porosity (%) Contact angle (deg.)

Casting solution 4 (0 wt% SPEK) 19.8 1.54 385.4 305.1 77.4 81.071.8



Table 5Mechanical properties of membrane substrates cast from solutions with different SPEK concentrations and thermal treatments.


Membrane substrate from casting solution 4 (0 wt% SPEK) 276.8735.4 7.370.8 20.375.2



Membrane substrate from casting solution 2 (50 wt% SPEK)—95 1C water for 1 min 69.174.8 4.670.1 50.375.8

Table 6Effects of thermal treatment on FO performance of TFC membranes made from different substrates.

Sample ID Temperature (1C) Time (min) PRO mode FO mode

Water flux (Jw, LMH) Salt leakage (Js, gMH) Jw/Js (L/g) Water flux (Jw, LMH) Salt leakage (Js, gMH) Jw/Js (L/g)

1 – – 55 16 3.4 37 11 3.4

2 80 1 45 16 2.8 30 12 2.5

3 95 1 50 9 5.5 35 7 5.0

4 95 2 25 10 2.5 18 8 2.3

5 100 1 Membrane shrinkage (defects)

6 95 1 38 6 6.3 23 4 5.8

7 100 1 25 8 3.12 17 6 2.8

Substrate 1–5: TFC-FO membranes with substrates cast from casting solution 2 (50 wt% SPEK);

6 and 7: TFC-FO membranes with substrates cast from casting solution 4 (0 wt% SPEK)

Experimental conditions: Feed: deionized water; draw solution: 2 M NaCl; crossflow velocity and temperature of both feed and draw solutions of 0.2 L min1 and 23 1C,

respectively.

0 10 20 30 40 50 60 70 80

0.00

0.05

0.10

0.15

0.20

0.25 Casting solution 2 (50 wt% SPEK with DEG) Casting solution 2 (50 wt% SPEK with DEG) (95 °C for 1 min)

Pore radius, dp(nm)

Prob

abili

ty d

ensi

ty fu

nctio

n, (n

m-1

)

Fig. 8. Pore-size distribution of PSU/SPEK (50 wt% SPEK) membrane substrates

with/without thermal treatment.



study. Fig. 8 and Table 7 present the pore size characteristics, PWPand MWCO of the membrane substrates cast from solution 2(Table 1) with or without the thermal treatment in 95 1C water.It is interesting to take note that after thermal treatment the poresize distribution of the substrate membrane significantly becomesbroader with a much increased MWCO, while keeping a relativelysimilar membrane mean pore size. Moreover, the porosity andPWP of the membrane substrates decrease remarkably afterthermal treatment. These results suggest that the porous structure

has been changed, which may be due to the pseudo-cross-linkingreaction happened during the thermal treatment. Fig. 9 illustratesthe possible pathways for the formation of a sulphone linkagebetween polymer chains. The formation of the bridges occurs viaan electrophilic aromatic substitution (SEAr) with a Friedel–Craftstype acylation mechanism [54,63].

FTIR spectroscopy and TGA analysis were employed to furthercharacterize the pseudo-cross-linking reaction during the thermaltreatment. Fig. 10 shows the typical FTIR spectrum of a PSU/SPEK

Table 7Summary of mean effective pore size (mp), PWP and MWCO of the PSU/SPEK (50 wt% SPEK) membrane substrates with/without thermal treatment.

Membrane mp (nm) sP PWP [L/(m2 bar h)] Porosity (%) MWCO (kDa) Contact angle (deg.)

Casting solution 2 (50 wt% SPEK with DEG) 10.7 1.25 152.7 77.2 66 59.471.5

Casting solution 2 (50 wt% SPEK with DEG) (95 1C water for 1 min) 13.1 2.03 84.1 50.4 248.8 6872.4

Fig. 9. Scheme of the possible pathways for the formation of a sulfone linkage.

600 800 1000 1200 1400 1600 1800 2000

PSU/SPEK(50 wt%)-95 °C for 1min

PSU/SPEK(50 wt%)

Tran

smitt

ance

, %

Wavenumber, cm-1

16511500

12401080 1165

Fig. 10. FTIR spectra of the membrane substrates cast from solution 2 (50 wt%

SPEK) with/without thermal treatment.

0 100 200 300 400 500 600 700 800 90020

30

40

50

60

70

80

90

100

110

120

130 PSU&SPEK(50 wt%) PSU&SPEK(50 wt%)-95 °C for 1min

Wei

ght (

%)

Temperature (°C)

Fig. 11. TGA curves of the membrane substrates cast from solution 2 (50 wt%

SPEK) with/without thermal treatment.



(50 wt%) membrane treated at 95 1C for 1 min and compares withthat of an untreated PSU/SPEK(50 wt%) membrane. The char-acteristic peaks of the SPEK and PSU materials can be observedin the SPEK/PSU hybrid membrane when comparing with theirindividual spectra shown in Fig. 3. However, the peaks at1080 cm1, 1165 cm1, 1192 cm1 and 1240 cm1 due to theabsorption of 1:2:4-substituted phenyl rings, the SQO stretchingvibration of PhSO2Ph, the asymmetric stretching vibration of thearomatic SO3H, and the phenyl ester group (–Ar–O–), respec-tively, show significantly shifts toward low frequency. In addition,the characteristic absorption at 1500 cm1 and 1651 cm1,which are attributed to the CQC of benzene and carbonylstretching of phenyl ketone group (–Ar–C(QO)–Ar–), respec-tively, also slightly move to the low frequency direction [51,64].Fig. 11 compares the thermal decomposition behavior of thethermal treated PSU/SPEK (50 wt%) membrane substrate with anuntreated PSU/SPEK (50 wt%) membrane. As aforementioned, themass loss in Fig. 3 occurs at around 450 1C because of thedecomposition of sulfonic groups. Interestingly, this mass loss isclearly reduced for the thermal-treated membrane, indicating aloss of some sulfonic groups during the post thermal treatment.This phenomenon is consistent with the mechanism proposed inFig. 9 that the bonds formation between adjacent polymericchains consumes some SO2 moieties.

Furthermore, Table 7 shows that the contact angle of themembrane substrate decreases after thermal treatment. Thisreduction in membrane hydrophilicity also confirms the con-sumption of some sulphonate groups. On the other hand, if theabove hypothesis is occurred, the mechanical strength of themembrane substrates will increase after thermal treatment due tothe enhanced interaction between polymer backbones. As dis-played in Table 5, Young’s modulus (MPa), tensile strength (MPa)and elongation at break (%) are all significantly increased. Insummary, all the results suggest that the interaction betweenmacromolecular chains during the thermal treatment happensthrough the proposed pathways.

3.3.2. Effects of SPEK concentration on the TFC layer and

FO performance

TFC-FO membranes were also fabricated on membrane sub-strates consisting of different SPEK concentrations. For easy com-parison, DEG was used as the pore-former for the substrates and allthe TFC-FO membranes were treated in 95 1C water for 1 min.

Fig. 12 displays the SEM morphology of TFC-FO membranes onsubstrates cast from solutions 2, 3 and 4, respectively. A typical‘‘ridge-and-valley’’ morphology is observed for the TFC polyamidelayer. However, the average thickness of the polyamide layervaries with SPEK content in the substrates and follows the orderof 25 wt% SPEK (solution 3)40 wt% SPEK (solution 4)450 wt%SPEK (solution 2). The thickness of the polyamide layer is anaverage of at least three separate samples. Since the conditionsfor interfacial polymerization and post-treatment are the sameand these three substrates also show a similar top surface porestructure (Fig. 6), the difference in substrate hydrophilicity maybe mainly responsible for the deviation in their TFC layerthicknesses. Since interfacial polymerization between MPD andTMC occurs predominantly in the organic phase due to therelatively low solubility of TMC in water. The MPD moleculesmust diffuse from the water phase into the organic phase andreact with the TMC molecules during the initial stage of inter-facial polymerization that results in the formation of the nascentpolyamide film consisting of many pendant acid chlorides. Theadsorbed MPD monomers within the porous substrate can furtherdiffuse out and react with these acid chlorides [65,66]. Comparedwith the hydrophobic PSU substrate, the substrates containing25 wt% SPEK may absorb more MPD molecules in water and reactwith TMC, thus produces a thicker polyamide layer. On the otherhand, a further increase in SPEK content to 50 wt% may not onlyenhance water content in the porous substrate but also limit thediffusion of MPD into the reaction zone due to the hydrogenbonding between MPD and sulfonic groups. As a result, thethickness of the TFC layer on highly hydrophilic substratesbecomes smaller.

Top X10,000 Cross section X10,000 Top X50,000 Cross section X50,000

410255244322

241270167247

285210205185145

1µm 100 nm

50 wt% SPEK

25 wt% SPEK

0 wt% SPEK

a

b

c

Fig. 12. Typical morphology of TFC-FO membranes with different SPEK concentrations in the substrates: (a) casting solution 4 (0 wt% SPEK); (b) casting solution 3 (25 wt%

SPEK); and (c) casting solution 2 (50 wt% SPEK).



Table 8 compares the performance of the TFC-FO membranesunder both PRO and FO modes using 2 M NaCl as the drawsolution and deionized water as the feed. The TFC-FO membranescomprising a fully sponge-like structure substrate made of 50 wt%SPEK show the highest water flux and slightly increased saltleakages for both PRO and FO modes. Water fluxes of 50 and 35LMH can be achieved for PRO and FO modes, respectively, whenusing a 2 M NaCl as the draw solution. A similar sponge-likestructure and high performance FO membranes were observed byWidjojo et al. [6] using other type hydrophilic material. Therefore,a highly finger-like structure is not an essential requirement toform a high water flux TFC-FO membrane as claimed by others[33]. In fact, a sponge-like structure with relatively hydrophiliccharacteristics is preferable since it can provide better perfor-mance stability for the TFC-FO membranes with enhanced anti-fouling properties. In comparison of PRO and FO data in Table 8,the water fluxes under PRO are much higher than those under FOdue to severe internal concentration polarization within theporous substrate under the FO mode.

Table 9 summarizes the basic transport properties of theTFC-FO membranes. The water permeability coefficient rises with

an increase in SPEK content in the substrates, while, the saltpermeability coefficient also slightly increases. Interestingly, thecalculated structure parameter (S) significantly decreases with anincrease in SPEK content in the membrane substrates. Thisimplies that the ICP effect can be significantly mitigated whenthe TFC-FO membrane is fabricated on a membrane substratewith higher hydrophilic and porous nature.

Furthermore, since the TFC-FO membrane constructed on thesubstrate containing 50 wt% SPEK (solution 2 as shown in Table 1)shows the best FO performance with the lowest S value, it wasfurther tested under both PRO and FO modes using different NaClconcentrations as draw solutions and deionized water as the feedsolution. As shown in Fig. 13, the water flux linearly increases inboth PRO and FO modes at low draw solution concentrations,while it seems to be leveled off at higher NaCl concentrations.This is most likely owing to the dilutive external concentrationpolarization (ECP) near the membrane surface and ICP within theporous substrate. In addition, a higher draw solution concentra-tion may induce a higher salt leakage and thus reduces the overallosmotic driving force across the membrane [6].


Actual membrane performance for seawater desalination isimportant for industrial applications. Fig. 14 shows the watertransport performance as a function of NaCl concentration in drawsolutions under both PRO and FO modes when using a modelseawater solution (3.5 wt% NaCl) as the feed. Our best TFC-FOmembrane shows a water flux of 22 LMH under the PRO mode or17 LMH under the FO mode using 2 M NaCl as a draw solution. Tothe best of our knowledge, these water fluxes outperform other flatsheet TFC-FO membranes in the application of seawater desalina-tion. A comparison of Figs. 13 and 14 also indicates that waterfluxes decrease in both PRO and FO modes when seawater is usedas the feed. This is due to a reduction in overall osmotic pressuredifference between the feed and the draw solution. In addition, thedifference in water flux between PRO and FO modes is decreasedduring seawater desalination. This is resulted from a smallerosmotic driving force during seawater desalination and a smaller

Table 8Performance of TFC-FO membranes with different SPEK content in the membrane

substrates.

Sample Substrate PRO mode FO mode

Water

flux

(LMH)

Salt

leakage

(gMH)

Water

flux

(LMH)

Salt

leakage

(gMH)

TFC-FO

0

Casting solution 4

(0 wt% SPEK)

38 6 23 4

TFC-FO

25

Casting solution 3

(25 wt% SPEK)

42 7 29 5

TFC-FO

50

Casting solution 2

(50 wt% SPEK)

50 9 35 7

All TFC-FO membranes are treated in 95 1C water for 1 min.

Experimental conditions: Feed: deionized water; draw solution: 2 M NaCl;

crossflow velocity and temperature of both feed and draw solutions of 0.2 L min1

and 23 1C, respectively.

Table 9Transport properties and structural parameters of TFC-FO membranes with different SPEK content in membrane substrates.

Membranea Water permeability, A (L/(m2 bar h)) Salt rejection (%)b Salt permeability, B (L m2 h1) Km (s m1) S (m)c

TFC-FO 0 0.50 91 0.041 0.12105 1.82104

TFC-FO 25 0.67 90 0.062 0.10105 1.56104

TFC-FO 50 0.75 89.5 0.068 0.071105 1.07104

a TFC-FO 0 represents the TFC-FO membrane with a substrate cast from solution 4 (0 wt% SPEK).b Tested at 1 bar (14.5 psi) with 200 ppm NaCl solution.c Structural parameters were calculated based on experiments under the FO mode using 2 M NaCl as the draw solution and deionized water as the feed.

0.0 0.5 1.0 1.5 2.0 2.50

10

20

30

40

50

60 PRO FO

Wat

er fl

ux (L

MH

)

Draw solution concentration, NaCl (M)0.0 0.5 1.0 1.5 2.0 2.5

0

3

6

9

12

15 PRO FO

Rev

erse

sal

t flu

x (g

MH

)


Fig. 13. The water fluxes and salt leakages of TFC-FO membranes (50 wt% SPEK polymer in the membrane substrates) in the PRO and FO tests with different draw solution

concentrations using deionized water as feed.



water flux tends to reduce ICP and ECP effects. Therefore thedifference in water flux between PRO and FO modes is diminished.

4. Conclusions

In this work, we have demonstrated hydrophilic sulphonatedpoly(ether ketone) (SPEK) is a good substrate material for thefabrication of high performance flat sheet TFC-FO membranes. Thefollowing conclusions can be further drawn from the current work:

1. Introduction of SPEK into the membrane substrates increasesmembrane hydrophilicity, reduces membrane thickness andchanges membrane morphology from a finger-like to a sponge-like structure. Membrane ductility can also be enhanced viablending SPEK material.

2. The TFC-FO membranes with the most hydrophilic nature ofsubstrates, i.e., 50 wt% SPEK exhibit the lowest membranethickness, fully sponge-like structure morphology and the high-est water flux of 50 LMH and 35 LMH tested under PRO and FOmodes, respectively, when using deionized water as the feed and2 M NaCl as the draw solution. These membranes also show thehighest water flux of 22 LMH under the PRO mode when amodel seawater solution (3.5 wt% NaCl) used as the feed and2 M NaCl as the draw solution. To the best of our knowledge,this seawater desalination water flux outperforms other flatsheet TFC-FO membranes among available published literatures.

3. The degree of membrane hydrophilicity and thickness of thesubstrates for TFC-FO membranes may play much stronger rolesin enhancing the water transport in FO processes compared tomembrane morphology (sponge-like or finger-like). TFC-FO mem-branes derived from substrates of hydrophobic nature and finger-like structure do not necessarily facilitate a higher water flux inFO processes than those of hydrophilic characteristics and fullysponge-like structures. Post-treatment procedures of TFC-FOmembranes can significantly alter the membrane performance.

4. It has been demonstrated that the membrane structure para-meter, an indicator of potential internal concentration polar-ization (ICP), can be significantly decreased by blending acertain amount of hydrophilic sulphonated materials into themembrane substrates.

Acknowledgments

This research was funded by the Singapore National ResearchFoundation under its Competitive Research Program for the project

entitled, ‘‘Advanced FO Membranes and Membrane Systems forWastewater Treatment, Water Reuse and Seawater Desalination’’(Grant No. R-279-000-336-281). Special thanks are due to Dr. JincaiSu and Mitsui Chemicals for their valuable suggestions.

Nomenclature

A water permeability (L/m2 h bar)B salt permeability (L m2 h1)Cf feed concentration (mol L1)Cp permeate concentration (mol L1)Ct salt concentration (mol L1)Ds diffusion coefficient of salt in the membrane sub-

strate (m2 s1)Js reverse salt flux (g m2 h1)Jw water flux (L m2 h1)Km solute diffusion resistivity within the porous layer

(s m1)l thickness (m)m membrane weight (g)P pressure (bar)R solute rejectionS membrane structural parameter (m)Sm effective membrane surface area (m2)Dt operation time interval (h)DV water permeation volume (L)Vt volume of the feed at a time interval of Dt (L)e porosityp osmotic pressure (bar)r material density (g cm3)s geometric standard deviationt tortuositym geometric mean radiusdp pore diameter (nm)ds solute diameter (nm)

Subscripts

s solutem membraneb bulk solutionD draw solution sideF feed solution side

References

[1] R.F. Service, Desalination freshens up, Science 313 (2006) 1088–1090.[2] M.A. Shannon, P.W. Bohn, M. Elimelech, J.G. Georgiadis, B.J. Marinas,

A.M. Mayes, Science and technology for water purification in the comingdecades, Nature 452 (2008) 301–310.

[3] G.M. Geise, H.S. Lee, D.J. Miller, B.D. Freeman, J.E. McGrath, D.R. Paul, Waterpurification by membranes: The role of polymer science,, J. Polym. Sci.: Part BPolym. Phys. 48 (2010) 1685–1718.

[4] B. Van der Bruggen, K. Borghgraef, C. Vinckier, Causes of water supplyproblems in urbanised regions in developing countries, Water ResourcesManage. 24 (2010) 1885–1902.

[5] T.S. Chung, S. Zhang, K.Y. Wang, J.C. Su, M.M. Ling, Forward osmosisprocesses: yesterday, today and tomorrow, Desalination 287 (2012) 78–81.

[6] N. Widjojo, T.S. Chung, M. Weber, C. Maletzko, V. Warzelhan, The role ofsulphonated polymer and macrovoid-free structure in the support layer forthin-film composite (TFC) forward osmosis (FO) membranes, J. Membr. Sci.383 (2011) 214–223.

[7] S. Chou, L. Shi, R. Wang, C.Y. Tang, C.Q. Qiu, A.G. Fane, Characteristics,potential applications of a novel forward osmosis hollow fiber membrane,Desalination 261 (2010) 365–372.

0.5 1.0 1.5 2.0 2.55

10

15

20

25

30 PRO FO

Wat

er fl

ux (L

MH

)

Draw solution concentration, M(NaCl)

Feed: 3.5 wt% NaCl

Fig. 14. The water flux in the PRO and FO tests with varying draw solution

concentrations (NaCl) using seawater (3.5 wt% NaCl) as feed.



[8] S. Zhao, L. Zou, Relating solution physicochemical properties to internal concen-tration polarization in forward osmosis, J. Membr. Sci. 379 (2011) 459–467.

[9] S. Zhang, K.Y. Wang, T.S. Chung, H.M. Chen, Y.C. Jean, G. Amy, Well-constructed cellulose acetate membranes for forward osmosis: minimizedinternal concentration polarization with an ultra-thin selective layer, J.Membr. Sci. 360 (2010) 522–535.

[10] J.C. Su, T.S. Chung, B.J. Helmer, J.S. Wit, Enhanced double-skinned FO mem-branes with inner dense layer for wastewater treatment and macromoleculerecycle using sucrose as draw solute, J. Membr. Sci. 396 (2012) 92–100.

[11] K.B. Petrotos, H.N. Lazarides, Osmotic concentration of liquid foods, J. FoodEng. 49 (2001) 201–206.

[12] X. Jin, J. Shan, C. Wang, J. Wei, C.Y. Tang, Rejection of pharmaceuticals byforward osmosis membranes, J. Hazard. Mater. 227–228 (2012) 55–61.

[13] M.M. Ling, T.S. Chung, Novel dual-stage FO system for sustainable proteinenrichment using nanoparticles as intermediate draw solutes, J. Membr. Sci.372 (2011) 201–209.

[14] S. Loeb, Osmotic power-plants, Science 189 (1975) 654–655.[15] S. Loeb, Production of energy from concentrated brines by pressure-retarded

osmosis. I. Preliminary technical and economic correlations, J. Membr. Sci. 1(1976) 49–63.

[16] K.L. Lee, R.W. Baker, H.K. Londsdale, Membranes for power generation bypressure-retarded osmosis, J. Membr. Sci. 8 (1981) 141–171.

[17] S. Chou, R. Wang, L. Shi, Q. She, C.Y. Tang, A.G. Fane, Thin-film compositehollow fiber membranes for pressure retarded (PRO) process with highpower density, J. Membr. Sci. 389 (2012) 25–33.

[18] S. Zhao, L. Zou, C.Y. Tang, D. Mulcahy, Recent developments in forwardosmosis: opportunities and challenges, J. Membr. Sci. 396 (2012) 1–21.

[19] K. Ravilious, Add salt as required: the recipe for fresh water, New Sci. 207(2773) (2010) 39–41.

[20] Z.Y. Li, V.Y. Quintanilla, R.V. Linares, Q.Y. Li, T. Zhan, G. Amy, Flux patternsand membrane fouling propensity during desalination of seawater byforward osmosis, Water Res. 46 (2012) 195–204.

[21] B.X. Mi, M. Elimelech, Organic fouling of forward osmosis membranes:fouling reversibility and cleaning without chemical reagents, J. Membr. Sci.348 (2010) 337–345.

[22] L. Liu, M. Wang, D. Wang, C. Gao, Current patents of forward osmosismembrane process, Recent Pat. Chem. Eng. 2 (2009) 76–82.

[23] K.Y. Wang, T.S. Chung, G. Amy, Developing thin-film-composite forwardosmosis membranes based on the PES/SPSf substrate through interfacialpolymerization, AIChE J. 58 (2012) 770–781.

[24] G. Han, S. Zhang, X. Li, N. Widjojo, T.S. Chung, Thin film composite forwardosmosis membranes based on polydopamine modified polysulfone sub-strates with enhancements in both water flux and salt rejection, Chem.Eng. Sci. 80 (2012) 219–231.

[25] M.M. Ling, K.Y. Wang, T.S. Chung, Highly water soluble magnetic nanopar-ticles as novel draw solutes in forward osmosis for water reuse, Ind. Eng.Chem. Res. 49 (2010) 5869–5876.

[26] S.K. Yen, F.M. Haja N., M. Su, K.Y. Wang, T.S. Chung, Study of draw solutesusing 2-methylimidazole-based compounds in forward osmosis, J. Membr.Sci. 364 (2010) 242–252.

[27] Q.C. Ge, J.C. Su, T.S. Chung, G. Amy, Hydrophilic superparamagnetic nano-particles: synthesis, characterization and application in forward osmosisprocesses, Ind. Eng. Chem. Res. 50 (2011) 382–388.

[28] R. Hausman, B. Digman, I.C. Escobar, M. Coleman, T.S. Chung, Functionaliza-tion of polybenzimidizole membranes to impart negative charge and hydro-philicity, J. Membr. Sci. 363 (2010) 195–203.

[29] Q. Yang, K.Y. Wang, T.S. Chung, Dual-layer hollow fibers with enhanced fluxas novel forward osmosis membranes for water production, Environ. Sci.Technol. 43 (2009) 2800–2805.

[30] J.C. Su, Q. Yang, J.F. Teo, T.S. Chung, Cellulose acetate nanofiltration hollow fibermembranes for forward osmosis processes, J. Membr. Sci. 355 (2010) 36–44.

[31] K.Y. Wang, R.C. Ong, T.S. Chung, Double-skinned forward osmosis mem-branes for reducing internal concentration polarization within the poroussublayer, Ind. Eng. Chem. Res. 49 (2010) 4824–4831.

[32] J. Herron, Asymmetric forward osmosis membranes, US Patent 7445712,Hydration Technologies Inc., Albany, OR, US, 2005.

[33] N.Y. Yip, A. Tiraferri, W.A. Phillip, J.D. Schiffman, M. Elimelech, Highperformance thin-film composite forward osmosis membrane, Environ. Sci.Technol. 44 (2010) 3812–3818.

[34] N.N. Bui, M.L. Lind, E.M.V. Hoek, J.R. McCutcheon, Electrospun nanofibersupported thin film composite membranes for engineered osmosis, J. Membr.Sci. 385 (2011) 10–19.

[35] R. Wang, L. Shi, C.Y. Tang, S.R. Chou, C.Q. Qiu, A.G. Fane, Characterization of novelforward osmosis hollow fiber membranes, J. Membr. Sci. 355 (2010) 158–167.

[36] J.E. Cadotte, Interfacially synthesized reverse osmosis membrane, US patent4277344, 1981.

[37] R.J. Petersen, Composite reverse-osmosis and nanofiltration membranes,J. Membr. Sci. 83 (1993) 81–150.

[38] S. Verissimo, K.V. Peinemann, J. Bordado, Thin-film composite hollow fibermembranes: an optimized manufacturing method, J. Membr. Sci. 264 (2005)48–55.

[39] A.K. Ghosh, B.H. Jeong, X.F. Huang, E.M.V. Hoek, Impacts of reaction andcuring conditions on polyamide composite reverse osmosis membraneproperties, J. Membr. Sci. 311 (2008) 34–45.

[40] A. Sagle, B. Freeman, Fundamentals of Membranes for Water Treatment, TheFuture of Desalination in Texas, vol. 2, Report No. 363, Texas WaterDevelopment Board, Austin, TX, 2004, p. 137.

[41] A.P. Rao, S.V. Joshi, J.J. Trivedi, C.V. Devmurari, V.J. Shah, Structure–performancecorrelation of polyamide thin film composite membranes: effect of coatingconditions on film formation, J. Membr. Sci. 211 (2003) 13–24.

[42] P.W. Morgan, Condensation polymers: by interfacial and solution methods,Polymer Reviews, vol. 10, Wiley, New York, 1965, pp. 19–64.

[43] J.E. Cadotte, K.E. Cobian, R.H. Forester, R.J. Petersen, Continued Evaluation ofInsitu-formed Condensation Polymers for Reverse Osmosis Membranes, NTISReport no. PB-253193, Wiley, New York, 1976 April.

[44] I.J. Roh, S.Y. Park, J.J. Kim, C.K. Kim, Effects of the polyamide molecularstructure on the performance of reverse osmosis membranes, J. Polym. Sci.:Part B: Polym. Phys. 36 (1998) 1821–1830.

[45] X. Li, K.Y. Wang, B.J. Helmer, T.S. Chung, Thin-film composite membranes andformation mechanism of thin-film layer on hydrophilic cellulose acetatepropionate substrates for forward osmosis processes, Ind. Eng. Chem. Res. 51(2012) 10039–10050.

[46] P.W. Morgan, S.L. Kwolek, Interfacial polycondensation 2. Fundamentals ofpolymer formation at liquid interfaces, J. Polym. Sci. A1 (34) (1996) 531–559.

[47] H.I. Kim, S.S. Kim, Plasma treatment of polypropylene and polysulfonesupports for thin film composite reverse osmosis membrane, J. Membr. Sci.286 (2006) 193–201.

[48] P.S. Singh, S.V. Joshi, J.J. Trivedi, C.V. Devmurari, A. PrakashRao, P.K. Ghosh,Probing the structural variations of thin film composite RO membranesobtained by coating polyamide over polysulfone membranes of different poredimensions, J. Membr. Sci. 278 (2006) 19–25.

[49] A.K. Ghosh, E.M.V. Hoek, Impacts of support membrane structure andchemistry on polyamide–polysulfone interfacial composite membranes, J.Membr. Sci. 336 (2009) 140–148.

[50] W. Xie, H. Ju, G.M. Geise, B.D. Freeman, J.I. Mardel, A.J. Hill, J.E. McGrath,Effect of free volume on water and salt transport properties in directlycopolymerized disulfonated poly(arylene ether sulfone) random copolymers,Macromolecules 44 (2011) 4428–4438.

[51] J.H. Chen, Q.L. Liu, A.M. Zhu, Q.G. Zhang, Dehydration of acetic acid bypervaporation using SPEK-C/PVA blend membranes, J. Membr. Sci. 320(2008) 416–422.

[52] Z.Z. Zhou, J.H. Cheng, T.S. Chung, T.A. Hattona, M.T. Katsunori Nishiura,S. Tamai, Enantiomeric resolution of tryptophan via stereoselective bindingin an ion-exchange membrane partitioned free flow isoelectric focusingsystem, Chem. Eng. J. 174 (2011) 522–529.

[53] L.K. Chen, C.S. Wu, M.C. Chen, K.L. Hsu, H.C. Li, C.H. Hsieh, M.H. Hsiao,C.L. Chang, P.P. Jen Chu, Cross-linked norbornene sulfonated poly(etherketone)s for proton exchange membrane, J. Membr. Sci. 361 (2010) 143–153.

[54] M.L. Di Vona, E. Sgreccia, S. Licoccia, G. Alberti, L. Tortet, P. Knauth, Analysisof temperature-promoted and solvent-assisted cross-linking in sulfonatedpoly(ether ether ketone) (SPEEK) proton-conducting membranes, J. Phys.Chem. B 113 (2009) 7505–7512.

[55] Y.C. Xiao, M.L. Chng, T.S. Chung, M. Toriida, S. Tamai, H. Chen, Y.C.J. Jean,Asymmetric structure and enhanced gas separation performance induced byin situ growth of silver nanoparticles in carbon membranes, Carbon 48(2010) 409–416.

[56] S. Singh, K.C. Khulbe, T. Matsuura, P. Ramamurthy, Membrane characteriza-tion by solute transport and atomic force microscopy, J. Membr. Sci. 142(1998) 111–127.

[57] S. Loeb, G.D. Mehta, A two-coefficient water transport equation for pressure-retarded osmosis, J. Membr. Sci. 4 (1979) 351–362.

[58] G.D. Mehta, S. Loeb, Internal polarization in the porous substructure of asemipermeable membrane under pressure-retarded osmosis, J. Membr. Sci. 4(1978–1979) 261–265.

[59] R. Guan, H. Zou, D. Lu, C. Gong, Y. Liu, Polyethersulfone sulfonated bychlorosulfonic acid and its membrane characteristics, Eur. Polym. J. 41(2005) 1554–1560.

[60] Y. Li, T.S. Chung, Highly selective sulfonated polyethersulfone (SPES)-basedmembranes with transition metal counterions for hydrogen recovery andnatural gas separation, J. Membr. Sci. 308 (2008) 128–135.

[61] N. Vogrin, C. Strpnik, V. Musil, M. Brumen, The wet phase separation: theeffect of cast solution thickness on the appearance of macrovoids in themembrane forming ternary cellulose acetate/acetone/water system, J.Membr. Sci. 207 (2002) 139–141.

[62] D.F. Li, T.S. Chung, J. Ren, R. Wang, Thickness dependence of macrovoidevolution in wet phase-inversion asymmetric membranes, Ind. Eng. Chem.Res. 43 (2004) 1553–1556.

[63] G.A. Olah, T. Mathew, G.K. Surya Prakash, Nafion-H catalysed sulfonylation ofaromatics with arene/alkenesulfonic acids for the preparation of sulfones,Chem. Commun. (2001) 1696–1697.

[64] M.L. Di Vona, D. Marani, C. D’Ottavi, M. Trombetta, E. Traversa, I. Beurroies,P. Knauth, S. Licoccia, A simple new route to covalent organic/inorganic hybridproton exchange polymeric membranes, Chem. Mater. 18 (2006) 69–75.

[65] Y. Jin, Z. Su, Effects of polymerization conditions on hydrophilic groups inaromatic polyamide thin films, J. Membr. Sci. 330 (2009) 175–179.

[66] S.M. Aharoni, Rigid aromatic fractal polyamides, Polym. Adv. Technol. 6(1995) 373–382.


(This is a sample cover image for this issue. The actual cover is not yet available at this time.)










A sulfonated polyphenylenesulfone (sPPSU) as the supporting substratein thin film composite (TFC) membranes with enhanced performance forforward osmosis (FO)

Natalia Widjojo a,b, Tai-Shung Chung a,⇑, Martin Weber c, Christian Maletzko d, Volker Warzelhan c

a Department of Chemical and Biomolecular Engineering, National University of Singapore, 10 Kent Ridge Crescent, Singapore 117602, Singaporeb BASF South East Asia Pte Ltd., A-GMM/F, 61 Science Park Road #03-01, Singapore 117525, Singaporec Advanced Materials & Systems Research, BASF SE, GMV/W-B001, 67056 Ludwigshafen, Germanyd Engineering Plastics, BASF SE, E-KTE/NE-F206, 67056 Ludwigshafen, Germany

h i g h l i g h t s

" Sulfonated polyphenylenesulfone (sPPSU) used for forward osmosis (FO) membranes." Sponge-like structure and hydrophilic FO membranes formed by sPPSU support." 4.4-Fold increment on water flux using sPPSU for FO instead of non-sulfonated." Significant improvement on the water flux for seawater desalination application.


Article history:Received 10 October 2012Received in revised form 28 December 2012Accepted 4 January 2013Available online 11 January 2013

Keywords:Thin film composite (TFC) membranesDirectly sulfonated polymerForward osmosisInterfacial polymerizationSponge-like structureHydrophilic substrate

a b s t r a c t

The new sulfonated polyphenylenesulfone (sPPSU) materials synthesized via direct route with variouscontent of sulfonated units, i.e., 2.5 and 5 mol% 3,30-di-sodiumdisulfate-4,40-dichlorodiphenyl sulfone(sDCDPS) monomer, have been effectively implemented as supporting layers of the thin film composite(TFC) membranes for forward osmosis (FO) applications. Not only does the hydrophilic nature of mem-brane substrates essentially facilitate the water transport across the membrane during the FO process,but also possibly provide anti-fouling characteristics as well as induce the formation of fully sponge-likestructures. Compared to TFC-FO membranes made of hydrophobic non-sulfonated PPSU supporting lay-ers, those made of hydrophilic sPPSU supporting layers comprising 2.5 mol% sDCDPS can achieve a 4.4-fold increment on water flux up to 54 LMH with 8.8 gMH salt reverse flux under the pressure retardedosmosis (PRO) mode using 2 M NaCl as draw solution. Surprisingly, the newly developed TFC-FO mem-branes show a much smaller difference in water flux between PRO and FO modes compared to previousworks, indicating much lower ICP, particularly at low draw solution concentrations, i.e. 0.5–2 M NaCl.When tested for seawater desalination using 3.5 wt% NaCl as the feed and 2 M NaCl as the draw solution,the aforementioned membrane show a water flux up to 22 LMH under the PRO mode, which is the high-est ever reported. Furthermore, the structural parameter indicating the internal concentration polariza-tion (ICP) can be remarkably decreased with an increase in sulphonated material contents in membranesubstrates.

2013 Elsevier B.V. All rights reserved.

1. Introduction

The forward osmosis (FO) process is an emerging technology fornext-generation water purification and seawater desalination [1–3]. FO employs the osmotic pressure difference between the feedand the draw solution across a semi-permeable membrane as thedriving force to induce clean water flow through the membrane

into the draw solution. Compared to the traditional pressure drivenprocess such as reverse osmosis (RO), the FO process exhibits sev-eral features that can potentially surpass the RO technology [1–3]:(1) lower energy consumption and equipment costs [1]; (2) higherwater recovery [1]; (3) more reversible fouling behavior [4]; (4)more extensive applications, such as power generation [5–7], juiceor food concentration [8,9], and protein and pharmaceuticalenrichment [10,11].

To date, the major challenges on FO processes remain on (1)improving the productivity of current FO membranes to compete

1385-8947/$ - see front matter 2013 Elsevier B.V. All rights reserved.http://dx.doi.org/10.1016/j.cej.2013.01.007

⇑ Corresponding author. Tel.: +65 6516 6645; fax: +65 6779 1936.E-mail address: [email protected] (T.-S. Chung).

Chemical Engineering Journal 220 (2013) 15–23


Chemical Engineering Journal

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


with that of RO technology and (2) developing the draw solutionswhich can be recycled effectively with low energy consumptionsand minimal reverse solute fluxes [1,2,12–14]. The preferred FOmembranes should possess the following characteristics [2,15–21]: (1) an ultrathin selective layer with both high water fluxand solute rejection; (2) a thin supporting layer with high porosityand low internal concentration polarization (ICP); (3) hydrophilicproperty to boost water transport and lessen fouling tendency;and (4) adequate mechanical strength, chemical resistance androbustness to sustain backwash, cleaning and vibration in indus-trial operations.

Vast developments of FO membranes have been in the form ofasymmetric flat sheet or hollow fiber membranes via phase inver-sion technique [15–21]. The current flat asymmetric FO mem-branes available in the market are made of cellulose triacetate(CTA) produced by Hydration Technologies Inc. Although thesemembranes have been used for broad applications, they have rela-tively low water permeability and salt rejection particularly for os-motic seawater desalination.

The next development of FO membranes was inspired by thethin film composite (TFC) approach originally developed for ROmembranes which consists of three layers: (1) a thin polyamide(PA) selective layer; (2) a porous polysulfone (PSU) layer as theintermediate layer; and (3) a relatively thick layer of non-wovenfabrics to withstand high pressures in the RO process [22–24].The feasibility of applying a typical TFC-RO membrane for the FOprocess has been initially explored [25]. However, the thick andhydrophobic nature of supporting layers in TFC-RO membraneslessen the water productivity with intensified internal concentra-tion polarization (ICP) due to the additional resistance towardswater transport through the membranes [25,26].

By eliminating non-woven fabrics used in the traditional TFC-RO membranes, Yip et al. [27] and Wang et al. [28,29] have beenthe pioneers in developing TFC-FO flat-sheet and hollow fibermembranes, respectively, on the PSU or polyethersulfone (PESU)support for FO applications. The flat sheet membranes demon-strated a water flux up to 18 LMH under FO tests using water as

the feed and 1.5 M NaCl as the draw solute, and a salt rejectiongreater than 97% at 400 psi (27.57 bar) under RO tests [27], whilethe hollow fiber ones exhibited a water flux up to 32.2–42.6LMH under the pressure retarded osmosis (PRO) mode using wateras the feed and 0.5 M NaCl as the draw solution with a reverse saltflux about 4 gMH, and a salt rejection of 91% at 1.01 bar under ROtests [28,29]. Based on the aforementioned FO performance, it wasclaimed that an ideal support substrate for TFC-FO membranesshould comprise of a thin layer of sponge-like structure near thetop edge of membrane cross-section with a fully finger-like struc-ture underneath.

Subsequently, the FO performance of TFC-FO membranes wasfound to be further enhanced if the hydrophilicity of membranesubstrates was increased as reported by Wang et al. [30]. By blend-ing 3 wt% of hydrophilic sulfonated PSU (sPSU) and the hydropho-bic PESU materials into the membrane substrates, the resultantTFC-FO flat-sheet membranes showed a water flux up to 47.5LMH in the PRO mode with a salt leakage up to 12.4 gMH using2 M NaCl as a draw solution. Hence, it is postulated that theenhancement of FO performance can be realized by: (1) increasingthe hydrophilicity of membrane substrates; (2) fabricating mem-brane substrates with finger-like structures to enhance watertransport through the membranes.

Widjojo et al. [31] disclosed a new approach to design high per-formance TFC-FO membranes with fully sponge-like structure andlikely anti-fouling characteristics. The substrate of their TFC-FOmembranes contained a 50 wt% sulfonated material made by apost-sulfonation method. Not only did their membranes exhibit afully sponge-like structure but also attain a water flux of 33.0LMH against DI water and 15 LMH against the 3.5 wt% NaCl modelsolution using 2 M NaCl as the draw solution tested under the PROmode. The value of 15 LMH for seawater desalination is the highestreported so far.

However, the post sulfonation polymers have several drawbacksas membrane materials: (1) the difficulty of controlling the sulfona-tion process, i.e. the position of sulfonated groups, the degree of sul-fonation and the unclear reaction chemistry, which become critical

Nomenclature

A water permeability, L m2 h1 bar1

B salt permeability, L m2 h1

Cf feed concentration, mol L1

Cp permeate concentration, mol L1

Ct salt concentration, mol L1

dp pore diameter, nmds solute diameter, nmDs diffusion coefficient of salt in the membrane substrate,

m2 s1

ECP external concentration polarizationFO forward osmosisICP internal concentration polarizationJs reverse salt flux, g m2 h1

Jw water flux, L m2 h1

k water transport coefficient, m s1

kb mass transfer coefficient, m s1

kD mass transfer coefficient, m s1

Km solute diffusion resistivity within the porous layer,s m1

l thickness, mm membrane weight, gM molecular weight, g mol1

MWCO molecular weight cut-off, kDaP pressure, bar

PRO pressure-retarded osmosisr Stokes radius, nmR solute rejectionS membrane structural parameter, mSm effective membrane surface area, m2

TFC thin-film-compositeDt operation time interval, hDV water permeation volume, LVt volume of the feed at a time interval of Dt, Le membrane porosityp osmotic pressure, barq material density, g cm3

l geometric mean radiusr geometric standard deviations turtuosity

Subscriptsb bulk solutionD draw solution sideF feed solution sidei inside of the active layer within the porous supportm membranes solute

16 N. Widjojo et al. / Chemical Engineering Journal 220 (2013) 15–23


issues for scale up and reproducibility in membrane performance[32]; (2) the post sulfonated materials may not be able to form con-tinuous asymmetric membranes unless blending with PESU due toslow precipitation [31]. To overcome the aforementioned problems,BASF SE Germany has synthesized sulfonated polyphenylenesulf-one (sPPSU) with various concentrations of sulfonated monomer(3,30-di-sodiumdisulfate-4,40-dichlorodiphenyl sulfone or sDCDPS),i.e., 2.5 (referred to as sPPSU-2,5) and 5 mol% (referred to as sPPSU-5), following the direct sulfonation synthesis route developed byMcGrath et al. [32]. Therefore, the objectives of this work are: (1)to study the membrane morphology, formation and mechanicalstrength of asymmetric flat sheets made fully of directly sulfonatedmaterials with different degrees of sulfonation; (2) to investigatethe performance of directly sulfonated materials as membrane sub-strates for FO applications via TFC method. As a benchmark, a non-sulfonated polyphenylenesulfone (PPSU) material was also investi-gated and compared. To the best of our knowledge, this is the firsttime of applying directly sulfonated materials as membrane sub-strates in the TFC-FO membranes.

2. Experimental

2.1. Materials

PPSU, sPPSU-2,5 and sPPSU-5 used as materials for membranesubstrates were synthesized by BASF SE Company, Germany. Thechemical structures of PPSU and sPPSU are illustrated in Fig. 1.N-methyl-2-pyrrolidone (NMP) from Merck and ethylene glycol(EG) from Sigma Aldrich were employed as the solvent and addi-tive, respectively, in the fabrication of membrane substrates forTFC-FO membranes. M-phenylenediamine (MPD) from Sigma Al-drich with >99% purity and trimesoyl chloride (TMC) from SigmaAldrich with 98% purity were used as received for the interfacialpolymerization process. N-heptane from Merck with >99.0% puritywas utilized as the solvent for TMC.

For FO tests, sodium chloride (NaCl) supplied by Sigma Aldrichwas dissolved in deionized (DI) water at 0.5, 1, 2, 3 and 4 M con-centrations and used as draw solutions. In the osmotic seawaterdesalination test, a model solution of seawater (3.5 wt% NaCl inDI water) was used as the feed solution.

2.2. Fabrication of membrane substrates for TFC-FO membranes

Membrane substrates were cast from polymer solutions com-prising three different degrees of sulfonation by varying the DCDPScontent in polymer materials. The composition of each dope solu-tion was the same with the formulation of polymer/EG/NMP

(wt%) = 13/16/71, while the polymers used were PPSU, sPPSU-2,5,and sPPSU-5, respectively. The casting solutions were allowed todegas overnight prior casting onto a glass plate with a 100 lmcasting knife. The as-cast membranes were then immersed into awater coagulation bath immediately at room temperature and keptfor 1 day to ensure complete precipitation. In addition, the mem-branes were washed with DI water before the interfacial polymer-ization was carried out.

The procedure of depositing a PA layer via interfacial polymer-ization on membrane substrates was adopted from our previouswork with some modifications [31]. The membrane substratewas first immersed in 2 wt% of MPD in DI water for 2 min. There-after, a filter paper was used to remove the water droplets on themembrane surface. Subsequently, the top surface of membraneswas brought into contact with the 0.05 wt% TMC solution in n-hep-tane for 15 s. The resultant TFC-FO membranes were dried in theair for 10 min. The TFC-FO membranes were then washed in DIwater before FO tests.

2.3. Characterizations of membrane substrates

2.3.1. Morphology and mechanical strength of membrane substratesThe morphology of different membrane substrates was ob-

served by using a field emission scanning electron microscope(FESEM JEOL JSM-6700LV). The samples were prepared by fractur-ing the membrane in liquid nitrogen and then coated with plati-num using a sputtering coater (JEOL LFC-1300).

The mechanical properties of membrane substrates were mea-sured by an Instron 5542 tensile testing equipment. The flat sheetmembranes were cut into stripes with 5 mm width and clamped atthe both ends with an initial gauge length of 25 mm and a testingrate of 10 mm/min. At least three stripes were tested for each cast-ing condition to obtain the average values of tensile stress, exten-sion at break and Young’s modulus of the membranes.

2.3.2. Pore size and pore size distributionThe fabricated flat sheet membrane substrate was first tested to

measure its pure water permeability (PWP) (in L/m2 bar h) by anultrafiltration membrane permeation cell with a sample diameterof 5 cm [33,34]. Subsequently, the membrane was subjected toneutral solute (polyethylene glycol (PEG) or polyethylene oxide(PEO)) separation tests by flowing them through the membrane’stop surface under a pressure of 25 psi (1.72 bar) on the liquid side.The concentrations of the neutral solutes were measured by a totalorganic carbon analyzer (TOC ASI-5000A, Shimadzu, Japan). Themeasured feed (Cf) and permeate (Cp) concentrations were usedfor the calculation of the effective solute rejection coefficient R (%):

* SO2 O

m

SO2 O *

n

HO3S SO3H

* SO2 O

m

(a)

(b) Fig. 1. Chemical structures of: (a) PPSU (non-sulfonated); (b) sPPSU-2,5 (m = 97.5 mol%; n = 2.5 mol%) and sPPSU-5 (m = 95 mol%; n = 5 mol%).

N. Widjojo et al. / Chemical Engineering Journal 220 (2013) 15–23 17


R ¼ 1 Cp

Cf

100% ð1Þ

In this work, solutions containing 200 ppm of different molecu-lar weights of PEG or PEO were used as the neutral solutes forthe characterizations of membrane pore size and pore sizedistribution. The relationship between Stokes radius (rs, nm) andmolecular weight (Mw, g mol1) of these neutral solutes can beexpressed as:

For PEG r ¼ 16:73 1012 M0:557 ð2Þ

For PEO r ¼ 10:44 1012 M0:587 ð3Þ

From Eqs. (2) and (3), the radius (r) of a hypothetical solute at a gi-ven Mw can be calculated. The mean effective pore size and the poresize distribution were then obtained according to the traditionalsolute transport approach by ignoring influences of the steric andhydrodynamic interaction between solute and membrane pores,the mean effective pore radius (lp) and the geometric standarddeviation (rp) can be assumed to be the same as ls (the geometricmean radius of solute at R = 50%) and rg (the geometric standarddeviation defined as the ratio of the rs at R = 84.13% over that atR = 50%). Therefore, based on lp and rp, the pore size distributionof a membrane can be expressed as the following probability den-sity function:

dRðdpÞddp

¼ 1dp ln rp

ffiffiffiffiffiffiffi2pp exp

ðln dp ln lpÞ2

2ðlnrpÞ2

" #ð4Þ

2.3.3. Membrane porosity (e)To measure the porosity of membrane substrates, wet mem-

branes were taken out from the water bath followed by carefuland quick removal of excess water on the surface by tissue paper.The wet membrane were then weighed (m1, g), freeze dried over-night, and re-weighed (m2, g). Subsequently, the absorbed wateris calculated as m1–m2, and the dry weight of the membrane ism2. Since the densities of both water (qw, 1.00 g/cm3), and poly-mers: PPSU (qp1, 1.41 g/cm3), sPPSU-2,5 (qp2, 1.35 g/cm3), andsPPSU-5 (qp3, 1.42 g/cm3), are known, the overall porosity e wasthen obtained as follows:

e ¼ m1 m2=qw

ðm1 m2Þ=qw þm2=qpð5Þ

2.4. Mass transport characteristics of TFC-FO membranes

The water permeability and salt permeability of TFC-FO mem-branes were determined by testing the membranes using a RO testsetup following Wang et al. [30]. The water permeability coeffi-cient (A) was acquired from the pure water permeation flux underthe applied trans-membrane pressure of 25 psi (1.72 bar). The saltrejection (Rs) was determined from the measured conductivities ofpermeate and feed by using feed water containing 400 ppm NaCl at25 psi (1.72 bar). The salt permeability coefficient (B), which is theintrinsic property of a membrane, was calculated based on thesolution-diffusion theory [5,35,36]:

1 Rs

Rs¼ B

AðDP DpÞ ð6Þ

2.5. FO testing procedure of TFC-FO membranes

A lab-scale circulating filtration unit was utilized for testing theFO performance of TFC-FO membranes [17,30]. The cross-flowpermeation cell was a plate and frame design with a rectangular

channel on each side of the membrane. The solution flow velocitiesduring FO tests were kept at 8.33 cm/s for both feed and draw solu-tions which co-currently flowed through the cell channels. Thetemperatures of the feed and draw solutions were maintained at22 ± 0.5 C. The membranes were tested under two differentmodes: (1) PRO mode where the draw solution faces against thedense selective layer and (2) FO mode where the feed water sidefaces against the dense selective layer.

The draw solutions were prepared from NaCl solutions with dif-ferent concentrations. The change of draw solution concentrationwas ignored because the ratio of water permeation flux to the vol-ume of the draw solution was less than 2% during the FO testing.When using the deionized water as the feed, the salt leakage canbe calculated by measuring the conductivity in the feed solutionat the end of experiment. A balance (EK-4100i, A&D CompanyLtd., Japan) connected to a computer recorded down the mass ofwater permeating into the draw solution over a selected periodof time. The water permeation flux was then calculated accordingto the weight change of feed water. The water permeation flux (Jv,L m2 h1, abbreviated as LMH) is calculated from the volumechange of the feed or draw solution.

Jv ¼DV

SmDtð7Þ

where DV (L) is the permeation water collected over a predeter-mined time Dt (h) in the FO process duration; Sm is the effectivemembrane surface area (m2).

The salt concentration in the feed water was determined fromthe conductivity measurement using a calibration curve for thesingle salt solution. The salt leakage, salt reverse diffusion fromthe draw solution to the feed, Js in g m2 h1 (abbreviated asgMH), is thereafter determined from the increase of the feedconductivity:

Js ¼DðCtVtÞ

SmDtð8Þ

where Ct and Vt are the salt concentration and the volume of thefeed at the end of FO tests, respectively.

The water flux in FO processes can be modeled by the followingequations [5,37]. For the PRO mode (selective layer against thedraw solution):

Jw ¼1

Kmln

ApD;m Jw þ BApF;b þ B

ð9Þ


Jw ¼1

Kmln

ApD;b þ BApF;m þ Jw þ B

ð9Þ

where pD,b and pF,b refer to the osmotic pressures in the respec-tive bulk draw solution and feed, pD;m and pF;m are the corre-sponding osmotic pressures on membrane surfaces facing thedraw and feed solutions after considering the external concentra-tion polarization effect (ECP) [38,39]. The relationship amongsolute diffusion resistivity within the porous layer Km, diffusivityDs, membrane structural parameter S, membrane turtuosity s,membrane thickness l and membrane porosity, e can be repre-sented as follows:

Km ¼S

Ds¼ ls

eDsð10Þ

To apply the TFC-FO membranes for seawater desalination, aseawater model solution (3.5 wt% NaCl) was used to test waterfluxes under FO and PRO modes.





During the fabrication of membrane substrates, it can be ob-served that membrane substrates cast from directly sulfonatedmaterials with 2.5 and 5 mol% sDCDPS monomer (sPPSU-2,5 andsPPSU-5) exhibit slower precipitation rates as compared to thatof non-sulfonated PPSU materials. This is a common phenomenonsince the sulfonated materials tend to facilitate delay demixing ascompared to that of non-sulfonated ones. However, compared tothe previous sulfonated materials synthesized via post sulfonationmethod which require blending with other polymer to form asym-metric membranes [30,31], the directly sulfonated polymers havethe ability to form free standing asymmetric membranes withoutblending with other polymer.

Fig. 2 illustrates the SEM images of membrane substrates castfrom non-sulfonated and sulfonated materials. The thicknesses ofthese membranes are in the range of 35–50 lm. As expected, themembrane substrate from non-sulfonated PPSU (Fig. 2a) exhibitsnumbers of macrovoids due to the instantaneous demixing, whilethe membrane substrates from sPPSU (Fig. 2b and c) display a fullysponge-like structure without formation of macrovoids. Thesponge-like structure is preferable since it can provide better per-formance stability for the TFC-FO membrane in the long term. Incontrast to the relatively dense bottom surface of membrane sub-strate from non-sulfonated materials, those membranes from sulfo-nated materials are fully porous. The membrane made of sPPSU-5with the highest sulfonation content possesses the most porousbottom surface (Fig. 2c).

The typical membrane morphology as shown in Fig. 2b and c isnot only suitable for membrane substrates for TFC-FO membranesbut also for other water applications, such as microfiltration/ultra-filtration (MF/UF) membranes, because it offers limited resistanceto the water transport, long-term stability as well as higherhydrophilicity. Table 1 clearly shows the contact angles of topand bottom surfaces of membrane substrates from sPPSU are sig-nificantly lower than those of non-sulfonated ones.

Table 2 and Fig. 3 represent the PWP and pore size characteris-tics of membrane substrates from non-sulfonated and directly sul-fonated PPSU materials. It is interesting to take a note that the PWPof membrane substrates follows the order: non-sulfonatedPPSU > sPPSU-2,5 > sPPSU-5. Although the non-sulfonated PPSUmembranes are hydrophobic, it possesses a large number of mac-rovoids as well as relatively large pores on the top surface(Fig. 2a). Hence, this maybe the main reason why it has the highestPWP among all membrane substrates. Meanwhile, the membranesubstrate from sPPSU-5 polymer (5 mol% sDCDPS) results in a low-er PWP than that containing sPPSU-2,5 (2.5 mol% sDCDPS). Thisphenomenon is due to the fact that the former has a higher degreeof water-induced swelling than the latter.

The MWCO of as-cast membranes is in the following order:sPPSU-2,5 (2.5 mol% sDCDPS) > non-sulfonated PPSU > sPPSU-5(5 mol% sDCDPS) as displayed in Table 2. The MWCO in thesPPSU-2,5 is higher than non-sulfonated PPSU because its sulfonicgroup induces delay demixing and results in larger pore sizes.However, the MWCO of sPPSU-5 membranes is smaller than othersdue to the effect of larger swelling behavior in the highly sulfo-nated materials.

Fig. 2. Typical morphology of membrane substrates from: (a) PPSU; (b) sPPSU-2,5; and (c) sPPSU-5.



Table 3 summarizes mechanical strengths of the membranesubstrates. In general, Young’s modulus decreases, while elonga-tion at break increases with an increase in sulfonation degree ofmembrane substrates. sPPSU-5 shows the weakest mechanicalstrength among all polymers.


To fabricate TFC-FO membranes, a thin film was deposited oneach aforementioned membrane substrate via interfacial polymer-ization. Fig. 4 presents a typical SEM image of the ridge-valley topsurface and selective layer morphology of the TFC membrane on asubstrate made of the sPPSU-2,5 material. The calculated thicknessof the selective PA layer on each membrane substrate is within therange of 115–125 nm.

Table 4 further compares the PRO and FO performance of TFC-FO membranes made of substrates with different sulfonation con-tent using 2 M NaCl as a draw solution. The results are in agree-ment with our hypothesis that the TFC-FO membranes withmore hydrophilic and fully sponge-like structure, i.e. sPPSU mate-rials, exhibit much higher water fluxes for both PRO and FO tests ascompared to those with hydrophobic and macrovoid structure sub-strates, i.e. PSU [27], blends of post sulfonated PSU and PESU mate-rials [30], PESU material [31]. The thin-film polymerizationperformed on a sponge-like substrate made from sPPSU-5 with5 mol% sDCDPS monomer show the highest water flux for FO appli-cation. However, the reproducibility of these TFC-FO membranes isnot good which is indicated by the large salt leakage variations dueto the lower mechanical strength of materials with a higher degreeof sulfonation. Therefore, we subsequently focus on the TFC-FOmembranes derived from sPPSU-2,5 membrane substrates sincethey provide reasonably high water fluxes and sufficient mechan-ical strength for FO applications. The acquired water flux can beup to 54 LMH under the PRO mode using 2 M NaCl as a draw solu-tion which is about 5.4 times compared to that of TFC-FO mem-branes derived from non-sulfonated PPSU substrates, as shown inTable 4. Meanwhile, the respective salt reverse flux can be con-trolled below 10 gMH.

Fig. 5 displays the FO performance of TFC-FO membranes usingmembrane substrates made from a directly sulfonated materialcontaining 2.5 mol% sDCDPS monomer for both PRO and FO modesas a function of draw solute (NaCl) concentration. The water fluxesincrease linearly in both PRO and FO modes at low draw solutionconcentrations. Fluxes measured under the FO mode seem to beleveled off at higher concentrations. This phenomenon is mostlikely attributed by the dilutive external concentration polarization(ECP) within the boundary layer at the membrane surface and ICPwithin the support layer which considerably reduce an efficiencyof osmotic driving force due to a higher salt leakage at a higherdraw solution concentration. Surprisingly, the newly developedTFC-FO membranes show a much smaller difference in water fluxbetween PRO and FO modes compared to previous works, indicat-ing much lower ICP, particularly at low draw solution concentra-tions, i.e. 0.5–2 M NaCl [28–31]. Fig. 6 illustrates a comparison ofpercentage change of water flux difference between PRO and FO

Table 1Contact angles of top and bottom surfaces of membrane substrates from non-sulfonated and sulfonated PPSU.

Membrane ID Contact angle (top surface) Contact angle (bottom surface)

PPSU (non-sulfonated) 86.2 ± 1.05 90.9 ± 1.43sPPSU-2,5 (2.5 mol% sDCDPS) 76.5 ± 1.82 68.3 ± 1.19sPPSU-5 (5 mol% sDCDPS) 62.1 ± 1.71 55.3 ± 1.09

Table 2Summary of PWP and pore size characteristics of membrane substrates from non-sulfonated and direct sulfonated PPSU materials.

Membrane ID PPSU (non-sulfonated) sPPSU-2,5 (2.5 mol% DCDPS) sPPSU-5 (5 mol% DCDPS)

lp (nm) 8.38 10.72 8.29rp 2.37 1.99 2.27PWP (L/m2 bar h) 6529 846.4 241.3MWCO (Da) 194,674 195,423 169,913Porosity, e (%) 65.0 83.41 84.18

0 20 40 600.00

0.02

0.04

0.06

0.08

Pore size, dp (nm)

Prob

abilit

y de

nsity

func

tion,

nm-1

PPSU sPPSU-GK39 sPPSU-GK15

Fig. 3. Probability density curve of membrane substrates with different sulfonationcontents: (a) PPSU; (b) sPPSU-2,5; (c) sPPSU-5.

Table 3Mechanical properties of membrane substrates from non-sulfonated and sulfonated materials.


PPSU (non-sulfonated) 241.0 ± 16.2 7.89 ± 0.29 18.4 ± 3.77sPPSU-2,5 (2.5 mol% sDCDPS) 75.7 ± 2.72 3.67 ± 0.10 37.2 ± 4.45sPPSU-5 (5 mol% sDCDPS) 14.2 ± 0.77 0.97 ± 0.05 43.9 ± 2.34



modes against the water flux obtained under the PRO mode as afunction of draw solution concentration for the current and previ-ous works [28,30]. This phenomenon indicates that the TFC-FOmembranes using direct sulfonation material substrate withhydrophilic and sponge-like morphology could significantly reducethe ICP effects in the substrate layer.

In comparison to the non-sulfonated PPSU, the use of directlysulfonated polymers as membrane substrates can significantlyeliminate the wetting problem in the FO process. From the PROtests on the TFC-FO membranes with membrane substrates fromnon-sulfonated materials, the water flux only increases slightlyfrom 8 to 12 LMH with the increment of NaCl draw solutionconcentrations from 0.5 to 4 M as shown in Fig. 7. This stronglyindicates that the wetting problem in the membrane substratesleading to more severe ICP due to the hydrophobic nature ofPPSU.

Table 5 summarizes the basic transport properties. The waterand salt permeability coefficients of TFC-FO membranes fromnon-sulfonated materials shows higher values compared to thosefrom sulfonated materials. It is hypothesized that the macrovoidsstructure in non-sulfonated membrane substrates facilitates thewater transport under low pressure testing. Interestingly, the cal-culated structural parameter (S) decreases with an increase in sul-fonated content in membrane substrates. This implies that a lowerICP effect can be achieved when the TFC-FO membrane was devel-oped on a membrane substrates consisting of higher hydrophilicityor porous nature for FO applications.


To be applied in the seawater desalination, a model seawatersolution (3.5 wt% NaCl) was used to test the best performance

Fig. 4. Typical FESEM images of TFC-FO membranes with sPPSU-2,5 (2.5 mol% sDCDPS) as the membrane substrate.

Table 4PRO and FO performance of TFC-FO membranes with non-sulfonated and sulfonated polymer materials in the membrane support layers (feed: DI water; draw solution: 2 M NaCl).

ID Membrane substrate PRO mode FO mode

Water flux (LMH) Salt leakage (gMH) Water flux (LMH) Salt leakage (gMH)

1 PPSU (non-sulfonated) 10 2.1 10 2.32 sPPSU-2,5 (2.5 mol% sDCDPS) 54 8.8 48 7.63 sPPSU-5 (5 mol% sDCDPS) 78.75–85 13.5–35 62.8 14.9–35

0 1 2 3 420

30

40

50

60

70

80

90

PRO mode FO mode

Wat

er fl

ux (L

MH

)

NaCl draw solution concentration (M)

0 1 2 3 4

5

10

15

20

PRO mode FO mode

Salt

reve

rse

flux

(gM

H)


Ideal Jw=AπD,b

Fig. 5. The water fluxes and reverse salt fluxes of TFC-FO membranes (membrane substrates from sPPSU-2,5) in the FO and PRO tests with varying draw solutionconcentrations (NaCl) using DI water as feed.



TFC-FO membranes. Compared to previous tests using DI water asa feed solution, Fig. 8 indicates that water fluxes decrease in bothPRO and FO modes when seawater is used as the feed. This isdue to the reduction of the overall osmotic pressure difference be-tween the feed and the draw solution. In this work, it is found thatthe water flux can achieve up to 22 LMH for both PRO and FOmodes, using 2 M NaCl as a draw solution. It shows that the useof fully sulfonated membranes can improve the water flux up to50% compared to the previous result in the TFC-FO membranes

using 50/50 (wt%) PESU/post sulfonated polymer blends as mem-brane substrates (PRO 15 LMH) [31] or sPSU/PESU polymerblends as membrane substrates (PRO 12.7 LMH) [30].

4. Conclusions

The newly synthesized direct sulfonated PPSU materials withdifferent sulfonation contents have been introduced for the mem-brane substrates of TFC-FO membranes with hydrophilic andsponge-like properties. Compared to the previous works on TFC-FO membranes [28–31], the new membranes made of fully sulfo-nated materials in the support layer demonstrate its uniquenessto significantly lower the water flux differences between PROand FO modes against the initial PRO water flux, indicating muchlower ICP in the overall FO process. The following conclusionscan be drawn from this work:

1. The directly sulfonated materials alone can form continuousmembranes with good mechanical strength. Due to their hydro-philic characteristics, the membrane substrates from directlysulfonated materials exhibit fully sponge-like structures withporous bottom surfaces which can facilitate water transportthrough the membranes.

2. Compared to the TFC-FO membranes made of non-sulfonatedsupport layers, those with membrane substrates from sPPSU-2,5 can achieve much higher water flux up to 54 LMH (5times improvement) with 8.8 gMH salt reverse flux underPRO mode when tested using 2 M NaCl as draw solution.Using 3.5 wt% NaCl as the mode feed seawater and 2 M NaClas the draw solution, this membrane shows the highest waterflux of 22 LMH under the PRO mode, which is the best in theliteratures.

Fig. 6. A comparison of percentage change of water flux difference between PROand FO modes against PRO water flux at different draw solution concentrations.

0 1 2 3 4

10

15

20

25

30

35

40


Wat

er fl

ux (L

MH

)

PROFO

Fig. 7. The water flux and reverse salt flux of TFC-FO membranes made ofsubstrates from PPSU as a function of draw solution concentration (NaCl) using DIwater as feed.

Table 5Transport properties and structural parameters of TFC-FO membranes with PPSU and sPPSU-2,5 as membrane substrates.

Membrane substrates Water permeability, A (L m2 h1 bar1) Salt rejectiona (%) Salt permeability, B (L m2 h-1 (m s1)) Km (s m1) Sb (m)

PPSU (non-sulfonated) 12.53 81.71 5.78 1.94 106 2.94 103

sPPSU-2,5 (2.5 mol% sDCDPS) 3.23 84.10 1.05 4.31 105 6.52 104

a Tested at 25 psi (1.72 bar) with 400 ppm NaCl solution.b Structural parameters were calculated based on experiments under the FO mode using 2 M NaCl as the draw solution and DI water as the feed.

2 3 4 520

25

30

35

40

45

Wat

er fl

ux (L

MH

)


PROFO

Fig. 8. The water flux in the FO and PRO tests of TFC-FO membranes (membranesubstrates from sPPSU-2,5) with varying draw solution concentrations usingseawater model solution (3.5 wt% NaCl) as the feed.



3. It is interesting to note that the water fluxes of TFC-FO mem-branes made of fully sulfonated substrates increase linearlywith the increment of draw solution concentrations (i.e. up to4 M NaCl), while those of TFC-FO from non-sulfonated supportlayers reaches a plateau at very low draw solution concentra-tions. This indicates that the application of directly sulfonatedmaterials as membrane substrates not only significantly reducethe wetting problem but also facilitate water transport in theTFC-FO membranes.

4. The structural parameter, an indicator of potential internal con-centration polarization (ICP), can be decreased with an increasein sulfonated material content in membrane substrates.

Acknowledgements

The authors would like to thank BASF SE, Germany for fundingthis research project with a Grant Number of R-279-000-283-597and the Singapore National Research Foundation under its Compet-itive Research Program for the project entitled ‘‘Advanced FOMembranes and Membrane Systems for Wastewater Treatment,Water Reuse and Seawater Desalination: Module designs and inte-grated systems for sustainable processes’’ (Grant Number: R-279-000-339-281). Thanks are also due to Ms. S. Zhang and Ms. X. Lifor their suggestions on this work. We are also appreciated Ms. J.Gao for her help on the experimental works. Special thanks aredue to Ms. M.L. Chua for her help on the polymer densitymeasurements.

References

[1] S. Zhao, L. Zou, C.Y. Tang, D. Mulcahy, Recent developments in forwardosmosis: opportunities and challenges, J. Membr. Sci. 396 (2012) 1.

[2] T.S. Chung, S. Zhang, K.Y. Wang, J. Su, M.M. Ling, Forward osmosis processes:yesterday, today and tomorrow, Desalination 287 (2011) 78.

[3] L. Liu, M. Wang, D. Wang, C. Gao, Current patents of forward osmosismembrane process, Recent. Patent. Chem. Eng. 2 (2009) 76.

[4] B. Mi, M. Elimelech, Chemical and physical aspects of organic fouling offorward osmosis membranes, J. Membr. Sci. 320 (2008) 292.

[5] K.L. Lee, R.W. Baker, H.K. Londsdale, Membranes for power generation bypressure-retarded osmosis, J. Membr. Sci. 8 (1981) 141.

[6] K. Gerstandt, K.V. Peinemann, S.E. Skilhagen, T. Thorsen, T. Holt, Membraneprocesses in energy supply for an osmotic power plant, Desalination 224(2008) 64.

[7] S. Loeb, Production of energy from concentrated brines by pressure-retardedosmosis. I. Preliminary technical and economic correlations, J. Membr. Sci. 1(1976) 49.

[8] B. Jiao, A. Cassano, E. Drioli, Recent advances on membrane processes for theconcentration of fruit juices: a review, J. Food Eng. 63 (2004) 303.

[9] K.B. Petrotos, H.N. Lazarides, Osmotic concentration of liquid foods, J. Food Eng.49 (2001) 201.

[10] Q. Yang, K.Y. Wang, T.S. Chung, A novel dual-layer forward osmosis membranefor protein enrichment and concentration, Sep. Purif. Technol. 69 (2009) 269.

[11] M.M. Ling, T.S. Chung, Novel dual-stage FO system for sustainable proteinenrichment using nanoparticles as intermediate draw solutes, J. Membr. Sci.372 (2011) 201.

[12] J.R. McCutcheon, R.L. McGinnis, M. Elimelech, A novel ammonia–carbondioxide forward (direct) osmosis desalination process, Desalination 174 (2005)1.

[13] M.M. Ling, K.Y. Wang, T.S. Chung, Highly water soluble magnetic nanoparticlesas novel draw solutes in forward osmosis for water reuse, Ind. Eng. Chem. Res.49 (2010) 5869.

[14] S. Phuntshoa, H.K. Shon, S. Hong, S. Lee, S. Vigneswaran, A novel low energyfertilizer driven forward osmosis desalination for direct fertigation: evaluatingthe performance of fertilizer draw solutions, J. Membr. Sci. 375 (2011) 172.

[15] K.Y. Wang, T.S. Chung, J.J. Qin, Polybenzimidazole (PBI) nanofiltration hollowfiber membranes applied in forward osmosis process, J. Membr. Sci. 300 (2007)6.

[16] K.Y. Wang, Q. Yang, T.S. Chung, Enhanced forward osmosis from chemicallymodified polybenzimidazole (PBI) nanofiltration hollow fiber membranes witha thin wall, Chem. Eng. Sci. 64 (2009) 1577.

[17] K.Y. Wang, R.C. Ong, T.S. Chung, Double-skinned forward osmosis membranesfor reducing internal concentration polarization within the porous sublayer,Ind. Chem. Eng. Res. 49 (2010) 4824.

[18] Q. Yang, K.Y. Wang, T.S. Chung, Dual-layer hollow fibers with enhanced flux asnovel forward osmosis membranes for water production, Environ. Sci. Technol.43 (2009) 2800.

[19] J. Su, Q. Yang, J.F. Teo, T.S. Chung, Cellulose acetate nanofiltration hollow fibermembranes for forward osmosis processes, J. Membr. Sci. 355 (2010) 36.

[20] S. Zhang, K.Y. Wang, T.S. Chung, H.M. Chen, Y.C. Jean, G. Amy, Well-constructedcellulose acetate membranes for forward osmosis: minimized internalconcentration polarization with an ultra-thin selective layer, J. Membr. Sci.360 (2010) 522.

[21] R.C. Ong, T.S. Chung, Fabrication and positron annihilation spectroscopy (PAS)characterization of cellulose triacetate membranes for forward osmosis, J.Membr. Sci. 394–395 (2012) 1.

[22] J.E. Cadotte, Interfacially Synthesized Reverse Osmosis Membrane, US Patent4277344, 1981.

[23] R.J. Petersen, Composite reverse-osmosis and nanofiltration membranes, J.Membr. Sci. 83 (1993) 81.

[24] Y.H. La, R. Sooriyakumaran, D.C. Miller, M. Fujiwara, Y. Terui, K. Yamanaka, B.D.McCloskey, B.D. Freeman, R.D. Allen, Novel thin film composite membranecontaining ionizable hydrophobes: pH-dependent reverse osmosis behaviorand improved chlorine resistance, J. Mater. Chem. 20 (2010) 4615.

[25] J.R. McCutcheon, M. Elimelech, Influence of membrane support layerhydrophobicity on water flux in osmotically driven membrane processes, J.Membr. Sci. 318 (2008) 458.

[26] J. Su, T.S. Chung, Experimental and theoretical study of sublayer structure andits effect on concentration polarization and membrane performance in FOprocesses, J. Membr. Sci. 376 (2011) 214.

[27] N.Y. Yip, A. Tiraferri, W.A. Phillip, J.D. Schiffman, M. Elimelech, Highperformance thin-film composite forward osmosis membrane, Environ. Sci.Technol. 44 (2010) 3812.

[28] R. Wang, L. Shi, C.Y. Tang, S. Chou, C. Qiu, A.G. Fane, Characterization of novelforward osmosis hollow fiber membranes, J. Membr. Sci. 355 (2010) 158.

[29] S. Chou, L. Shi, R. Wang, C.Y. Tang, C. Qiu, A.G. Fane, Characteristics andpotential applications of a novel forward osmosis hollow fiber membrane,Desalination 261 (2010) 365.

[30] K.Y. Wang, T.S. Chung, G. Amy, Developing thin-film-composite forwardosmosis membranes based on the pes/spsf substrate through interfacialpolymerization, AIChE J. 58 (2011) 770.

[31] N. Widjojo, T.S. Chung, M. Weber, C. Maletzko, V. Warzelhan, The role ofsulphonated polymer and macrovoid-free structure in the support layer forthin-film composite (TFC) forward osmosis (FO) membranes, J. Membr. Sci.383 (2011) 214.

[32] G.M. Geise, H.S. Lee, D.J. Miller, B.D. Freeman, J.E. McGrath, D.R. Paul, Waterpurification by membrane: the role of polymer science, J. Polym. Sci. Part B:Polym. Phys. 48 (2010) 1685.

[33] S. Singh, K. Khulbe, T. Matsuura, P. Ramamurthy, Membrane characterizationby solute transport and atomic force microscopy, J. Membr. Sci. 142 (1998)111.

[34] K.Y. Wang, T. Matsuura, T.S. Chung, W.F. Guo, The effects of flow angle andshear rate within the spinneret on the separation performance ofpoly(ethersulfone) (PES) ultrafiltration hollow fiber membranes, J. Membr.Sci. 240 (2004) 67.

[35] H.K. Londsdale, U. Merten, R.L. Riley, Transport properties of cellulose acetateosmosis membranes, J. Appl. Polym. Sci. 9 (1965) 1341.

[36] S. Loeb, G.D. Mehta, A two-coefficient water transport equation for pressure-retarded osmosis, J. Membr. Sci. 4 (1979) 351.

[37] G.D. Mehta, S. Loeb, Internal polarization in the porous substructure of asemipermeable membrane under pressure-retarded osmosis, J. Membr. Sci. 4(1978) 261.

[38] J.R. McCutcheon, M. Elimelech, Modeling water flux in forward osmosis:implications for improved membrane design, AIChE J. 53 (2007) 1736.

[39] J.R. McCutcheon, M. Elimelech, Influence of concentrative and dilutive internalconcentration polarization on flux behavior in forward osmosis, J. Membr. Sci.284 (2006) 237.


High Performance Thin-Film Composite Forward Osmosis HollowFiber Membranes with Macrovoid-Free and Highly Porous Structurefor Sustainable Water ProductionPanu Sukitpaneenit and Tai-Shung Chung*

Department of Chemical and Biomolecular Engineering, National University of Singapore, 10 Kent Ridge Crescent 4 EngineeringDrive 4, Singapore 117576, Singapore

*S Supporting Information

ABSTRACT: The development of high-performance andwell-constructed thin-film composite (TFC) hollow fibermembranes for forward osmosis (FO) applications ispresented in this study. The newly developed membranesconsist of a functional selective polyamide layer formed byhighly reproducible interfacial polymerization on a poly-ethersulfone (PES) hollow fiber support. Using dual-layercoextrusion technology to design and effectively control thephase inversion during membrane formation, the support wasdesigned to possess desirable macrovoid-free and fully sponge-like morphology. Such morphology not only provides excellentmembrane strength, but it has been proven to minimizeinternal concentration polarization in a FO process, thusleading to the water flux enhancement. The fabricatedmembranes exhibited relatively high water fluxes of 32−34LMH and up to 57−65 LMH against a pure water feed using 2 M NaCl as the draw solution tested under the FO and pressureretarded osmosis (PRO) modes, respectively, while consistently maintaining relatively low salt leakages below 13 gMH for allcases. With model seawater solution as the feed, the membranes could display a high water flux up to 15−18 LMH, which iscomparable to the best value reported for seawater desalination applications.

INTRODUCTIONTo address or alleviate the water crisis, fresh water productionthrough desalination and wastewater reclamation processes hasreceived worldwide attention.1 Currently, reverse osmosis(RO) is a dominant technology in seawater desalination andwater treatment markets. However, it is energy intensive innature. Furthermore, there have been questions concerning thenegative environmental impact of current RO desalinationtechnologies.1 With these concerns, there is an urgent need fora novel sustainable technology that consumes less energy andchemicals, and has minimal impact on the environment for thefuture water-energy nexus.Forward osmosis (FO) is an emerging, low-energy, and

green membrane process in comparison to conventionaldesalination and wastewater treatment processes. FO utilizesthe osmotic pressure difference between two solutionsseparated by a semipermeable membrane to induce sponta-neous water transport from the feed solution of low osmoticpressure to the draw solution of high osmotic pressure whilemost solutes are rejected by the FO membrane.2−4 In additionto this low hydraulic operating pressure, FO may offeradvantages of higher rejections to a wide range of contaminantsand lower membrane-fouling propensities compared to theconventional RO process.5−8 These superior FO properties

have attracted considerable attention in many fields of scienceand engineering including liquid food processing,9 pharma-ceutical and protein enrichment,10 and power generation,4,11,12

in addition to its great potential for seawater desalination,fertigation, water treatment, and reclamation processes.5,13−16

However, one of the major challenges to the current FOtechnology is the lack of effective membranes, which is theheart of the system. The ideal FO membranes need to providehigh water permeability, low solute permeability, sufficient highmechanical strength, good chemical stability, and low internalconcentration polarization (ICP).3,4 To meet the abovecharacteristics, most FO membranes are designed to have anultrathin active layer supported by a relatively thin and fullyporous substrate. One promising approach is the fabrication ofthin-film composite (TFC) hollow fiber membranes made byinterfacial polymerization because they possess higher surfaceareas and a better flow pattern for the FO process.17−19 Whileseveral studies have reported the preparation of flat sheet TFCmembranes in the recent FO development,20−22 few studies

Received: April 19, 2012Revised: May 28, 2012Accepted: June 4, 2012Published: June 4, 2012

Article

pubs.acs.org/est

© 2012 American Chemical Society 7358 dx.doi.org/10.1021/es301559z | Environ. Sci. Technol. 2012, 46, 7358−7365

pubs.acs.org/est

http://pubs.acs.org/action/showImage?doi=10.1021/es301559z&iName=master.img-000.jpg&w=238&h=148

have been devoted to the fabrication and characterization ofTFC hollow fiber membranes but without revealing the detailsof interfacial polymerization.18,19 Moreover, most investigatedTFC hollow fiber membranes consist of a top thin polyamide(PA) rejection layer and large finger-like macrovoids in aporous membrane support. These macrovoids may beundesirable because they are mechanically weak points, whichmay cause membrane failure under continuous vibration andbackwashing operations.23 On the other hand, hollow fibermembranes with a favorable macrovoid-free structure could bethe more preferential design in the industrial applications.In this work, we have demonstrated novel TFC-FO hollow

fiber membranes with a high flux and a favorable macrovoid-free substructure for water production through FO processes.The feasibility of fabricating well-defined FO hollow fibermembrane supports has been demonstrated by employing theadvanced coextrusion technologies to effectively control thephase inversion and create appropriate membrane supports forinterfacial polymerization. A simple and highly reproducibleinterfacial polymerization method, using the m-phenylenedi-amine (MPD) and trimesoyl chloride (TMC) monomers, wasdeveloped. The simple and cost-effective fabrication techniqueusing a dual-layer spinneret demonstrated through this studycould provide another platform for future FO membranedevelopment.

MATERIALS AND METHODS

Fabrication of Macrovoid-Free PES Hollow FiberMembrane Supports. The hollow fiber membranes wereprepared by a dry-jet wet spinning process employing theadvanced coextrusion technology through a dual-layer spinner-et. A detailed description of the hollow fiber spinning process isdocumented elsewhere.24 The dual-layer spinneret and itsdimension used in this study are described in Figure S1 of the

Supporting Information (SI). The spinning conditions for thePES hollow fiber membrane supports are listed in Table S1 ofthe SI. The detailed procedures for the polymer dopepreparation, spinning process, membrane post-treatment, andmodule fabrication are described in the SI.

Interfacial Polymerization of Thin-Film-Composite(TFC) Forward Osmosis Hollow Fiber Membranes. Theformation of a polyamide thin layer on the inner surface of PEShollow fiber supports was achieved by an interfacial polymer-ization (IP) between MPD monomers in the aqueous phaseand TMC monomers in the organic phase. The detailedspecification of the experiment setup and preparation steps isdisclosed in the SI.

Characterizations of PES Hollow Fiber MembraneSupports and TFC-FO Hollow Fiber Membranes. Thefabricated hollow fiber membrane supports were first tested tomeasure pure water permeability (PWP) (L m2− h−1 bar−1 orLMHbar−1) using a lab-scale circulating filtration unit, asdescribed in SI or eslewhere.10,25 The membranes were thencharacterized by solute rejection measurements. Feed solutionscontaining 200 ppm of different molecular weights of PEG orPEO were utilized as the neutral solutes to estimate pore size,pore size distribution, and molecular weight cutoff (MWCO)according to the solute transport method described in the SI orelsewhere.17,26 The water permeability A, salt rejection Rs, andsalt permeability B of TFC-FO hollow fiber membranes weredetermined by testing the membranes under the RO modefollowing the method described elsewhere.19,25

Water Reclamation through Forward Osmosis Testsof TFC-FO Hollow Fiber Membranes and Evaluation ofStructural Parameters of Supporting Layers. Forwardosmosis (FO) experiments were conducted on a lab-scalecirculating filtration unit.25 The detailed experimental setup,operating conditions, and the determination of structural

Figure 1. SEM micrographs of different bulk and surface morphologies of PES hollow fiber membrane supports (spinning code: PESwater).

Environmental Science & Technology Article

dx.doi.org/10.1021/es301559z | Environ. Sci. Technol. 2012, 46, 7358−73657359


parameters are described in the SI. A schematic diagram of FOsetup is shown in Figure S2 of the SI.

RESULTS AND DISCUSSIONMorphology, Microstructure, and Separation Charac-

teristics of PES Hollow Fiber Supports. A typicalmorphology of PES hollow fiber membrane supports spunusing water as a bore-fluid (PESwater) is illustrated in Figure 1.The hollow fibers have a macrovoid-free and fully sponge-likestructure with a high degree of concentricity. The inner surfaceof as-spun membranes exhibits relatively small pores andsmooth surface, compared to the outer surface which possessesa highly porous structure and relatively large pore sizes of 1 μm(estimated from the FESEM observation, Figure 1). Similarbulk and surface morphologies as discussed above are obtainedfor the hollow fibers spun with other conditions (70/30 wt %water/NMP and 40/30/30 wt % water/NMP/PEG asnonsolvent bore-fluids), as illustrated in Figure S3 of the SI.These morphological features are consistent with our strategy,employing dual-layer spinneret to facilitate membrane phaseinversion during fiber spinning, as depicted in Figure 2. The

pure NMP solvent was fed at the outer channel to induce adelayed demixing23 (slow phase inversion and mild solvent

exchange) in the air-gap region prior to entering the watercoagulant, attempting to form a macrovoid-free cross-sectionand a porous structure at the membrane’s outer surface. Incontrast, an instantaneous demixing (rapid phase inversion andfast solvent exchange) was developed by introducing anonsolvent at the bore-fluid channel to create a relativelydense inner surface. Moreover, based on our observation duringexperiments, the fabricated hollow fiber membranes spun fromthe dual-layer coextrusion technique tend to have a uniformdimension and no irregular shape, which reflects the uniformsolvent exchange rate during the phase inversion process.Another important parameter that plays an important role ineliminating larger finger-like macrovoids and meanwhilecreating highly porous and open-cell structure is the additionof PEG and water nonsolvent additives with optimumconcentrations into dope solutions. PEG, a highly hydrophilicpolymer and a weak nonsolvent of a PES/NMP system, isknown as an effective additive to enhance pore formation,improve pore interconnectivity, and prevent macrovoidformation. Water, a strong nonsolvent of the system, is added(with a relative small amount) to enhance dope viscosity and tobring the dope formulation close to the binodal compositionwhich could favor the macrovoid suppression.27

The inner surface roughness of hollow fibers was examinedby AFM and is displayed in Figure S4. All as-spun fibers have asimilar surface roughness of 2−3 nm regardless of bore-fluidspinning conditions. It can be seen that the membranes spunwith pure water exhibited slightly higher surface roughnesscompared to others. Similar phenomenon has been reported byWidjojo et al.28 and Sukitpaneenit and Chung29 where theinstantaneous demixing induced by strong nonsolvent, e.g.water, tends to result in membranes with a rougher surface (Ra= 2.85 nm). In other words, the delayed demixing could ratherresult in a smoother surface and a smaller surface roughness asshown in the case of membranes spun with water/NMP (Ra =2.25 nm) and water/NMP/PEG (Ra = 2.69 nm). Nevertheless,one may observe a slight increase in surface roughness of themembranes PESwater/NMP/PEG, compared to the membranePESwater/NMP. This may be due to the fact that (1) addingPEG, known as a pore forming additive, in the bore-fluid mayenhance the pore evolution at the contacting inner surfaceduring membrane formation, and (2) the highly hydrophilicPEG polymer in the bore-fluid may complicate the precip-itation rate and preferentially create a relatively higher surfaceroughness. In other words, introducing PEG into a bore-fluidsolution could enhance the delayed demixing and slow theprecipitation rate to some certain extent, leading to formationof a membrane with relatively higher surface porosity/

Figure 2. Strategies to control the phase inversion process with the aidof coextrusion technology employing a dual-layer spinneret.

Table 1. Summary of Mean Effective Pore Size (μp), PWP, MWCO, Porosity, Water Contact Angle, and Mechanical Propertiesof PES Hollow Fiber Membrane Supports

membrane ID PESwater PESwater/NMP PESwater/NMP/PEG

mean pore size, μp (nm) (at the inner surface) 15.01 16.50 17.05geometric standard variation, σp 1.30 1.33 1.34PWP (L m2− h−1 bar−1) (at 1 bar) 835 856 1021MWCO (Da) 106 967 132 349 142 672porosity (%) 80.0 ± 1.4 82.0 ± 1.0 80.9 ± 1.6water contact angle, θ (°) (at the inner surface) 64.6 ± 2.0 56.5 ± 2.1 54.8 ± 2.8tensile strength (MPa) 5.88 ± 0.27 6.97 ± 0.23 5.81 ± 0.09elongation at break (%) 48.13 ± 5.78 49.82 ± 8.34 50.90 ± 4.65Young’s modulus (MPa) 273.80 ± 21.58 228.30 ± 49.56 196.27 ± 27.31




roughness. Besides, the highly hydrophilic characteristic of PEGmay further hinder the solvent/nonsolvent exchange rateduring phase inversion at the inner surface due to high affinitywith added PEG in the polymer dope, and thus enhance thesurface roughness.27

Basic characteristics of PES hollow fiber membrane supportsbefore interfacial polymerization, which include the mean poresize, pure water permeability (PWP), molecular-weight cutoff(MWCO), porosity, and water contact angle, are listed in Table1. All spun PES hollow fiber membranes possess a small poresize in a range of 15−17 nm and a narrow pore size distributionprofile (small geometric standard variation of 1.30−1.34),which is essential to produce a continuous and homogeneouspolyamide layer through interfacial polymerization.28,30 A clearcomparison of pore size and pore size distribution of theresultant membranes is depicted in Figure S5.As shown in Table 1, the PESwater, PESwater/NMP, and

PESwater/NMP/PEG supports display high PWP values of 835,856, and 1021 L m2− h−1 bar−1 at 1 bar, respectively. Thistendency is apparently attributed to their corresponding poresize, pore size distribution, and high overall porosity (80−82%)for each respective membrane. Additionally, the membranespossess moderately low water contact angles of 55−65° (<90°)at the inner surface that not only reflect their reasonably goodhydrophilicity, but also facilitate water transport.28,30 Themechanical strength of these hollow fiber supports wasinvestigated and is also summarized in Table 1. In mostcases, they have comparable tensile strength of 5.8−7 MPa,elongation at break of 48−50%, and Young’s modulus of 200−270 MPa within the error ranges. As anticipated, the hollowfiber membranes in this present work display a bettermechanical strength (5.8−7 MPa) compared to the previousstudies on TFC-FO hollow fiber (4.1−5.4 MPa)18 and otherPES ultrafiltration membranes (0.94−1.65 MPa)31 reported inthe literature. This is significantly contributed to the presenceof macrovoid-free sponge-like structure in the hollow fibers.Overall, the membrane supports fabricated in this study have

possessed desirable characteristics of being supports forinterfacial polymerization. These include the mean pore size

in nanorange, high porosity, and high hydrophilicity, macro-void-free with enhanced mechanical properties.

Characteristics and Performance of TFC-FO HollowFiber Membranes. The active polyamide layer of the TFC-FO hollow fiber membranes after interfacial polymerization onthe aforementioned membrane supports is depicted in Figure 3.The selective layer has a uniform ridge-and-valley morphology,which is a typical characteristic of polyamide membranesformed using an interfacial polymerization.32 The estimatedthickness of the selective polyamide layer is 380 ± 25 nmregardless of different membrane supports. There is nosignificant change observed on the outer surface morphologyafter IP, indicating no IP solutions penetrating across themembrane bulk.The fabricated TFC-FO hollow fiber membranes with

different membrane supports were evaluated for their FOperformance in both FO and PRO operating modes using DIwater feed and 2 M NaCl as a draw solution. The membraneperformance was expressed in terms of water flux and saltreverse flux as summarized in Table 2. Interestingly, allmembranes demonstrate remarkable high water fluxes andreasonably low salt leakages. Based on the experimental results,

Figure 3. Typical morphology of TFC-FO hollow fiber membranes with PESwater/NMP supports layer after interfacial polymerization.

Table 2. PRO and FO Performance of Developed TFC-FOHollow Fiber Membranes with Different Bore-FluidSpinning Conditionsa

PRO mode FO mode

membrane IDwater flux(LMH)

saltleakage(gMH)

water flux(LMH)

saltleakage(gMH)

TFC-FO with PESwatersupports

57.1 6.93 32.1 6.15

TFC-FO withPESwater/NMP supports

62.7 10.30 34.1 7.10

TFC-FO withPESwater/NMP/PEGsupports

65.1 12.34 34.5 9.87

aExperimental conditions: DI water as the feed solution, 2.0 M NaClas the draw solution.




TFC-FO-PESwater hollow fiber membranes exhibit the waterfluxes of 32.1 and 57.1 LMH with relatively low reverse fluxesof 6.15 and 6.93 gMH for FO and PRO modes, respectively. Inthe case of the TFC-FO-PESwater/NMP/PEG membranes, thewater flux moves up to 34.5 and 65.1 LMH but with slightlyincreased reverse salt fluxes of 9.87 and 12.34 gMH, forrespective FO and PRO modes, under the same 2.0 M NaCldraw solution. The increase in water flux and reverse salt flux ofthe resultant TFC-FO membranes in both FO and PRO modesis in the following order: TFC-FO-PESwater < TFC-FO-PESwater/NMP < TFC-FO-PESwater/NMP/PEG.A comparison of FO performance of the resultant TFC-FO

hollow fiber membranes with other thin-film compositemembranes and commercial membranes reported in theliterature for FO applications is presented in Table3.18,19,28,30,33−36 Overall, the fabricated TFC-FO membranesshow relatively high water fluxes in both FO and PRO modes;the water fluxes as high as 32−34.5 LMH under the FO modeand exceeding up to 57−65 LMH under the PRO mode withacceptable reverse salt fluxes of 6−12 gMH could be achievedusing a 2 M NaCl draw solution and pure water as the feed.Compared to other TFC and commercially available HTI flatsheet membranes, the developed TFC hollow fiber membranesnot only offer high water fluxes and less salt leakages, whichcould be attributed to the inherent characteristics of the hollowfiber membrane configuration, but also possess a well-structured macrovoid-free morphology with enhanced mechan-ical properties to sustain the membrane stability in the realapplication.

Another interesting finding is that there is a close relationshipbetween the pore characteristics of membrane supports derivedfrom different bore-fluid conditions and the FO performance oftheir corresponding TFC-FO membranes. As all membranesupports possess almost equivalent overall porosity andhydrophilicity, the membrane support with a smaller poresize and a narrower pore size distribution could yield the TFC-FO membrane (i.e., TFC-FO-PESwater) with an impressive highwater flux but less salt leakage, compared to those with a largerpore size and a wider pore size distribution,22 i.e., TFC-FO withPESwater/NMP and TFC-FO with PESwater/NMP/PEG membranes.The significant contribution of TFC-FO support layers on

the internal concentration polarization (ICP) was investigated,which is typically expressed in terms of structural parameter, S.The structural parameters of different hollow fiber supportswere determined by fitting the FO mode experiment data inTable 2 with the equations S11 and S12 in the SI for eachmembrane condition. The water permeability (A), saltpermeability (B) coefficients, solute diffusion resistivity withinthe porous layer (Km) involved in the calculation, and thecorresponding structural parameters of TFC-FO hollow fibermembranes with the supports derived from different bore-fluidcompositions are tabulated in Table 4. From the results, thewater and salt permeabilities obtained from RO tests are inagreement with our FO experimental data. Membranespossessing a higher intrinsic water permeability have tendencyto achieve a higher water flux in actual FO operation. In termsof structural parameter, all membranes have structuralparameters in a range of 219−261 μm (2.19−2.61 × 10−4

Table 3. Comparison of FO Membrane Performancea

membranewater flux, Jv (FO/PRO) (L m‑2

h‑1)reverse salt flux, Js (FO/PRO) (g m‑2

h‑1 )draw

solution ref

TFC-FO with PESwater supports 22.5/25.6 2.8/3.2 0.5 M NaCl this workTFC-PES hollow fiber (#B-FO dope formula) 13.5/32.2 1.8/3.5 0.5 M NaCl 18TFC-PES hollow fiber (#C-FO dope formula) 18.5/42.6 1.5/4.0 0.5 M NaCl 19TFC-PESU E6020P and PESU-co-sPPSU 11 flat sheetmembrane

14.8/15.5 2.2/2.5 0.5 M NaCl 28

TFC-PES/SPSf flat sheet membrane 13.0/24.1 3.6/4.5 0.5 M NaCl 30TFC-Cellulose acetate propionate (CAP) membrane 8.0/15.0 0.8/0.8 0.5 M NaCl 33HTI flat sheet membrane 18.6 (PRO) 0.5 M NaCl 34TFC-FO with PESwater supports 32.1/57.1 6.15/6.93 2.0 M NaCl this workTFC-FO with PESwater/NMP supports 34.1/62.7 7.10/10.30 2.0 M NaCl this workTFC-FO with PESwater/NMP/PEG supports 34.5/65.1 9.87/12.34 2.0 M NaCl this workTFC-PES hollow fiber (#B-FO dope formula) 22.5/45.0 2.5/4.4 2.0 M NaCl 18TFC-PES hollow fiber (#C-FO dope formula) 29.5/68.0 2.6/5.8 2.0 M NaCl 19TFC-PESU E6020P and PESU-co-sPPSU 11 flat sheetmembrane

21.0/33.0 2.2/2.8 2.0 M NaCl 28

TFC-PES/SPSf flat sheet membrane 26.0/47.5 8.3/12.4 2.0 M NaCl 30TFC-cellulose acetate propionate (CAP) membrane 17/35 (PRO) 1.85/1.9 (FO) 2.0 M NaCl 33HTI flat sheet membrane 13.0 (FO) 10.5 2.0 M NaCl 35HTI flat sheet membrane 11.2 (PRO) 2.0 M NaCl 36aExperiment conducted at 20−25 °C; FO/PRO in bracket refers to FO and PRO testing modes in FO experiment.

Table 4. Intrinsic Separation Properties and Structural Parameters of TFC-FO Hollow Fiber Membranes

membrane IDwater permeability, A (Lm‑2 h‑1

bar‑1)salt rejectiona

(%)salt permeability, B (Lm‑2

h‑1) Km (sm‑1) S (m)b

TFC-FO with PESwater supports 1.18 87.95 0.135 1.45 × 105 2.19 × 10−4

TFC-FO with PESwater/NMP supports 1.64 83.04 0.281 1.65 × 105 2.52 × 10−4

TFC-FO with PESwater/NMP/PEG supports 1.83 81.52 0.348 1.73 × 105 2.61 × 10−4

a200 ppm NaCl as the feed solution in the RO test under an applied pressure of 1 bar. bCalculated based on the experiments under the FO modeusing 2 M NaCl as the draw solution and DI water as the feed solution.



m), which is relatively low compared to the values of mostmembranes investigated in the literature. The S values obtainedare slightly smaller than those reported for commercial HTI FOmembranes (481 μm)35 and TFC hollow fiber membranes(550−595 μm),18,21 but nearly 2 orders of magnitude smallerthan the ones for conventional RO membranes (∼ 40 mm forBW30 membranes).21 The lower the structural parameter, theless severe the ICP. Hence, much less ICP would beencountered with our membranes during FO processes, andthe effective driving force would be preserved to the highestextent.It has been hypothesized in previous studies18,19 that an ideal

membrane support for TFC-FO hollow fiber membranesshould possesses an array of large finger-like structures with athin layer of sponge-like structures near the functional selectivelayer. Our results reveal that the TFC-FO macrovoid-freehollow fiber supports with a highly sponge-like structure couldbe an alternative favorable membrane morphology to reduceICP, and thus enhance water permeation flux. In other words, amembrane substrate with a highly finger-like structure is not theessential requirement to form a high flux TFC-FO hollow fibermembrane.In addition, the TFC-FO hollow fiber membranes with

PESwater supports and the lowest structural parameter werefurther evaluated for the membrane performance under variousdraw solution concentrations (0.5−2 M NaCl), as illustrated inFigure 4. Regardless of the FO and PRO operating modes, thewater flux increases with an increase in draw solutionconcentration because a larger effective osmotic pressuredifference provides a greater driving force. The water flux inPRO mode is comparable with the water flux in FO mode atrelatively low draw solution concentrations, e.g. 0.5 M NaCl;but there seems to be more deviation at increased drawsolution concentrations yielding a much higher water flux forthe PRO mode. For all cases, membranes show higher saltleakages at higher draw solution concentrations whereby onlyslight larger reverse salt fluxes were observed for the PRO case.In addition, one may observe that there is a rapid proportionalincrement of water fluxes along the increase of draw solutionconcentrations in the PRO mode whereas the water flux in FOmode shows relative plateau tendency at relatively high drawsolution concentrations. This implies the evolution and impact

of ICP within the support layer as a function of draw solutionstrengths. Moreover, this phenomenon also signifies that waterflux tested under the FO mode may experience severer ICPrelative to that under the PRO mode especially at elevated drawsolution concentrations. A comparison between experimentalwater flux and the model prediction is also shown in Figure 4. Itis observed that the model curve fits well the experimental flux.

Potentials of Developed TFC-FO Hollow FiberMembranes for Water Production through ForwardOsmosis Seawater Desalination. The newly developedTFC-FO hollow fiber membranes were characterized toevaluate their potential for seawater desalination. A modelseawater solution (3.5 wt % NaCl) was employed to test themembrane performance. Using 2 M NaCl as a draw solutionand under PRO mode, the TFC-FO hollow fiber membranesexhibit water production of 15, 16, and 18 LMH with differentsupports of PESwater, PESwater/NMP, and PESwater/NMP/PEG,respectively. Compared to previous tests using DI water as afeed solution, the water fluxes attained from all TFC-FOmembranes decrease when seawater is used as the feed. This isdue to the reduction of the overall osmotic pressure differencebetween the feed and the draw solution.A performance comparison of the resultant membranes with

the literature for FO seawater desalination under the sameoperating conditions mentioned above is illustrated in Figure 5.Comparing with various membrane materials includingcellulose acetate (CA) membranes,37 commercial HTImembranes,36 thin-film composite flat sheet, and hollow fibermembranes,19,28,30,33 the TFC-FO hollow fiber membranesdeveloped in this work displays a higher water flux, andcomparable to the best in the literature tested under the sameconditions using seawater as the feed. With such encouragingmembrane performance, and together with the benefits of cost-effective membrane synthesis and fabrication through a dual-layer coextrusion approach which requires a relatively smallamount of solvent to facilitate phase inversion during themembrane fabrication, the newly developed TFC-FO hollowfiber membrane can be a promising candidate for osmoticseawater desalination applications.

Figure 4. Water flux (a) and reverse salt flux (b) for TFC-FO membranes (PESwater supports) under FO and PRO operating modes using water asthe feed solution against different draw solution concentrations.




ASSOCIATED CONTENT

*S Supporting InformationMaterials; procedures for hollow fiber spinning, membranepost-treatment, and module fabrication; protocols for physicaland mass transport characterizations of membranes; forwardosmosis setup and operating conditions; additional SEM andAFM images of the membrane supports; pore size distributionprofiles. This material is available free of charge via the Internetat http://pubs.acs.org.

AUTHOR INFORMATION

Corresponding Author*Tel.: 65 6516 6645; fax: 65 6779 1936; e-mail: [email protected].

NotesThe authors declare no competing financial interest.

ACKNOWLEDGMENTS

We thank the Singapore National Research Foundation (NRF)support through the Competitive Research Program for theproject entitled “New Advanced FO membranes andmembrane systems for wastewater treatment, water reuse andseawater desalination” with grant R-279-000-311-281. Specialthanks are also given to Dr. Jincai Su, Ms. Sui Zhang, Ms. XueLi, and Ms. Fitri Juniwati for their valuable assistance. Dr. PanuSukitpaneenit also acknowledges the World Future Foundation(WFF) for awarding his Ph.D. Prize in Environmental andSustainability Research, 2012.

REFERENCES(1) Elimelech, M.; Phillip, W. A. The future of seawater desalination:Energy, technology and the environment. Science 2011, 333, 712−717.

(2) Cath, T. Y.; Childress, A. E.; Elimelech, M. Forward osmosis:Principles, applications, and recent developments. J. Membr. Sci. 2006,281, 70−87.(3) Zhao, S.; Zou, L.; Tang, C. Y.; Mulcahy, D. Recent developmentsin forward osmosis: Opportunities and challenges. J. Membr. Sci. 2012,396, 1−21.(4) Chung, T. S.; Zhang, S.; Wang, K. Y.; Su, J.; Ling, M. M. Forwardosmosis processes: Yesterday, today and tomorrow. Desalination 2012,287, 78−81.(5) Cornelissen, E. R.; Harmsen, D.; de Korte, K. F.; Ruiken, C. J.;Qin, J.-J.; Oo, H.; Wessels, L. P. Membrane fouling and processperformance of forward osmosis membranes on activated sludge. J.Membr. Sci. 2008, 319, 158−168.(6) Mi, B.; Elimelech, M. Organic fouling of forward osmosismembranes: Fouling reversibility and cleaning without chemicalreagents. J. Membr. Sci. 2010, 348, 337−345.(7) Zhao, S.; Zou, L. Effects of working temperature on separationperformance, membrane scaling and cleaning in forward osmosisdesalination. Desalination 2011, 278, 157−164.(8) Mi, B.; Elimelech, M. Gypsum scaling and cleaning in forwardosmosis: Measurement and mechanisms. Environ. Sci. Technol. 2010,44, 2022−2028.(9) Jiao, B.; Cassano, A.; Drioli, E. Recent advances on membraneprocesses for the concentration of fruit juices: A review. J. Food Eng.2004, 63, 303−324.(10) Yang, Q.; Wang, K. Y.; Chung, T. S. A novel dual-layer forwardosmosis membrane for protein enrichment and concentration. Sep.Purif. Technol. 2009, 69, 269−274.(11) Lee, K. L.; Baker, R. W.; Lonsdale, H. K. Membranes for powergeneration by pressure-retarded osmosis. J. Membr. Sci. 1981, 8, 141−171.(12) Gerstandt, K.; Peinemann, K. -V.; Skilhagen, S. E.; Thorsen, T.;Holt, T. Membrane processes in energy supply for an osmotic powerplant. Desalination 2008, 224, 64−70.(13) McCutcheon, J. R.; McGinnis, R. L.; Elimelech, M. A novelammonia-carbon dioxide forward (direct) osmosis desalinationprocess. Desalination 2005, 174, 1−11.(14) Phuntsho, S.; Shon, H. K.; Hong, S.; Lee, S.; Vigneswaran, S. Anovel low energy fertilizer driven forward osmosis desalination fordirect fertigation: Evaluating the performance of fertilizer drawsolutions. J. Membr. Sci. 2011, 375, 172−181.(15) Cartinella, J. L.; Cath, T. Y.; Flynn, M. T.; Miller, G. C.; Hunter,K. W.; Childress, A. E. Removal of natural steroid hormones fromwastewater using membrane contactor processes. Environ. Sci. Technol.2006, 40, 7381−7386.(16) Hoover, L. A.; Phillip, W. A.; Tiraferri, A.; Yip, N. Y.; Elimelech,M. Forward with osmosis: Emerging applications for greatersustainability. Environ. Sci. Technol. 2011, 45, 9824−9830.(17) Yang, Q.; Wang, K. Y.; Chung, T. S. Dual-layer hollow fiberswith enhanced flux as novel forward osmosis membranes for waterproduction. Environ. Sci. Technol. 2009, 43, 2800−2805.(18) Wang, R.; Shi, L.; Tang, C. Y.; Chou, S.; Qiu, C.; Fane, A. G.Characterization of novel forward osmosis hollow fiber membranes. J.Membr. Sci. 2010, 355, 158−167.(19) Chou, S.; Shi, L.; Wang, R.; Tang, C. Y.; Qiu, C.; Fane, A. G.Characteristics and potential applications of a novel forward osmosishollow fiber membrane. Desalination 2010, 261, 365−372.(20) Yip, N. Y.; Tiraferri, A.; Phillip, W. A.; Schiffman, J. D.;Elimelech, M. High performance thin-film composite forward osmosismembrane. Environ. Sci. Technol. 2010, 44, 3812−3818.(21) Wei, J.; Qiu, C.; Tang, C. Y.; Wang, R.; Fane, A. G. Synthesisand characterization of flat-sheet thin-film composite forward osmosismembranes. J. Membr. Sci. 2011, 372, 292−302.(22) Tiraferri, A.; Yip, N. Y.; Phillip, W. A.; Schiffman, J. D.;Elimelech, M. Relating performance of thin-film composite forwardosmosis membranes to support layer formation and structure. J.Membr. Sci. 2011, 367, 340−352.(23) Peng, N.; Widjojo, N.; Sukitpaneenit, P.; Teoh, M. M.;Lipscomb, G. G.; Chung, T. S.; Lai, J.-Y. Evolution of polymeric

Figure 5. Current status of potential water flux reported in theliterature for FO seawater desalination applications. Note: The modelseawater (3.5 wt % NaCl) solution as the feed solution, PRO mode,2.0 M NaCl draw solution. The membrane materials include celluloseacetate,37 commercial HTI,36 TFC cellulose acetate propionate,33 TFChollow fiber (finger-like macrovoids),19 TFC PES/sulfonated PSF,30

and TFC sulfonated PSF.28



http://pubs.acs.org

mailto:[email protected]



hollow fibers as sustainable technologies: Past, present, and future.Prog. Polym. Sci. 2012, No. http://dx.doi.org/10.1016/j.progpolyms-ci.2012.01.001.(24) Sukitpaneenit, P.; Chung, T. S. Molecular elucidation ofmorphology and mechanical properties of PVDF hollow fibermembranes from aspects of phase inversion, crystallization andrheology. J. Membr. Sci. 2009, 340, 192−205.(25) Su, J.; Yang, Q.; Teo, J. F.; Chung, T. S. Cellulose acetatenanofiltration hollow fiber membranes for forward osmosis processes.J. Membr. Sci. 2010, 355, 36−44.(26) Wang, K. Y.; Matsuura, T.; Chung, T. S.; Guo, W. F. The effectsof flow angle and shear rate within the spinneret on the separationperformance of poly(ethersulfone) (PES) ultrafiltration hollow fibermembranes. J. Membr. Sci. 2004, 240, 67−79.(27) Liu, Y.; Koops, G. H.; Strathman, H. Characterization ofmorphology controlled polyethersulfone hollow fiber membranes bythe addition of polyethylene glycol to the dope and bore liquidsolution. J. Membr. Sci. 2003, 223, 187−199.(28) Widjojo, N.; Chung, T. S.; Weber, M.; Maletzko, C.; Warzelhan,V. The role of sulphonated polymer and macrovoid-free structure inthe support layer for thin-film composite (TFC) forward osmosis(FO) membranes. J. Membr. Sci. 2011, 383, 214−223.(29) Sukitpaneenit, P.; Chung, T. S. Molecular design of themorphology and pore size of PVDF hollow fiber membranes forethanol-water separation employing the modified pore-flow concept. J.Membr. Sci. 2011, 374, 67−82.(30) Wang, K. Y.; Chung, T. S.; Amy, G. Developing thin-film-composite forward osmosis membranes on the PES/SPSf substratethrough interfacial polymerization. AIChE J. 2012, 58, 770−781.(31) Chung, T. S.; Qin, J.-J.; Gu, J. Effect of shear rate within thespinneret on morphology, separation performance and mechanicalproperties of ultrafiltration polyethersulfone hollow fiber membranes.Chem. Eng. Sci. 2000, 55, 1077−1091.(32) Ghosh, A. K.; Jeong, B.-H.; Huang, X. F.; Hoek, E. M. V.Impacts of reaction and curing conditions on polyamide compositereverse osmosis membrane properties. J. Membr. Sci. 2008, 311, 34−35.(33) Li, X.; Wang, K. Y.; Helmer, B.; Chung, T. S. Thin-filmcomposite membranes and formation mechanism of thin-film layers onhydrophilic cellulose acetate propionate substrates for forward osmosisprocesses. Ind. Eng. Chem. Res. 2012, No. 10.1021/ie2027052.(34) Gray, G. T.; McCutcheon, J. R.; Elimelech, M. Internalconcentration polarization in forward osmosis: Role of membraneorientation. Desalination 2006, 197, 1−8.(35) Phillip, W. A.; Yong, J. S.; Elimelech, M. Reverse draw solutepermeation in forward osmosis: Modeling and experiments. Environ.Sci. Technol. 2010, 44, 5170−5176.(36) McCutcheon, J. R.; McGinnis, R. L.; Elimelech, M. Desalinationby ammonia-carbon dioxide forward osmosis: Influence of draw andfeed solution concentrations on process performance. J. Membr. Sci.2006, 278, 114−123.(37) Zhang, S.; Wang, K. Y.; Chung, T. S.; Chen, H.; Jean, Y. C.;Amy, G. Well-constructed cellulose acetate membranes for forwardosmosis: Minimized internal concentration polarization with an ultra-thin selective layer. J. Membr. Sci. 2010, 360, 522−535.



High Performance Thin-Film Composite Forward Osmosis HollowFiber Membranes with Macrovoid-Free and Highly Porous Structurefor Sustainable Water ProductionPanu Sukitpaneenit and Tai-Shung Chung*

Department of Chemical and Biomolecular Engineering, National University of Singapore, 10 Kent Ridge Crescent 4 EngineeringDrive 4, Singapore 117576, Singapore


ABSTRACT: The development of high-performance andwell-constructed thin-film composite (TFC) hollow fibermembranes for forward osmosis (FO) applications ispresented in this study. The newly developed membranesconsist of a functional selective polyamide layer formed byhighly reproducible interfacial polymerization on a poly-ethersulfone (PES) hollow fiber support. Using dual-layercoextrusion technology to design and effectively control thephase inversion during membrane formation, the support wasdesigned to possess desirable macrovoid-free and fully sponge-like morphology. Such morphology not only provides excellentmembrane strength, but it has been proven to minimizeinternal concentration polarization in a FO process, thusleading to the water flux enhancement. The fabricatedmembranes exhibited relatively high water fluxes of 32−34LMH and up to 57−65 LMH against a pure water feed using 2 M NaCl as the draw solution tested under the FO and pressureretarded osmosis (PRO) modes, respectively, while consistently maintaining relatively low salt leakages below 13 gMH for allcases. With model seawater solution as the feed, the membranes could display a high water flux up to 15−18 LMH, which iscomparable to the best value reported for seawater desalination applications.

INTRODUCTIONTo address or alleviate the water crisis, fresh water productionthrough desalination and wastewater reclamation processes hasreceived worldwide attention.1 Currently, reverse osmosis(RO) is a dominant technology in seawater desalination andwater treatment markets. However, it is energy intensive innature. Furthermore, there have been questions concerning thenegative environmental impact of current RO desalinationtechnologies.1 With these concerns, there is an urgent need fora novel sustainable technology that consumes less energy andchemicals, and has minimal impact on the environment for thefuture water-energy nexus.Forward osmosis (FO) is an emerging, low-energy, and

green membrane process in comparison to conventionaldesalination and wastewater treatment processes. FO utilizesthe osmotic pressure difference between two solutionsseparated by a semipermeable membrane to induce sponta-neous water transport from the feed solution of low osmoticpressure to the draw solution of high osmotic pressure whilemost solutes are rejected by the FO membrane.2−4 In additionto this low hydraulic operating pressure, FO may offeradvantages of higher rejections to a wide range of contaminantsand lower membrane-fouling propensities compared to theconventional RO process.5−8 These superior FO properties

have attracted considerable attention in many fields of scienceand engineering including liquid food processing,9 pharma-ceutical and protein enrichment,10 and power generation,4,11,12

in addition to its great potential for seawater desalination,fertigation, water treatment, and reclamation processes.5,13−16

However, one of the major challenges to the current FOtechnology is the lack of effective membranes, which is theheart of the system. The ideal FO membranes need to providehigh water permeability, low solute permeability, sufficient highmechanical strength, good chemical stability, and low internalconcentration polarization (ICP).3,4 To meet the abovecharacteristics, most FO membranes are designed to have anultrathin active layer supported by a relatively thin and fullyporous substrate. One promising approach is the fabrication ofthin-film composite (TFC) hollow fiber membranes made byinterfacial polymerization because they possess higher surfaceareas and a better flow pattern for the FO process.17−19 Whileseveral studies have reported the preparation of flat sheet TFCmembranes in the recent FO development,20−22 few studies

Received: April 19, 2012Revised: May 28, 2012Accepted: June 4, 2012Published: June 4, 2012

Article

pubs.acs.org/est

© 2012 American Chemical Society 7358 dx.doi.org/10.1021/es301559z | Environ. Sci. Technol. 2012, 46, 7358−7365

pubs.acs.org/est


have been devoted to the fabrication and characterization ofTFC hollow fiber membranes but without revealing the detailsof interfacial polymerization.18,19 Moreover, most investigatedTFC hollow fiber membranes consist of a top thin polyamide(PA) rejection layer and large finger-like macrovoids in aporous membrane support. These macrovoids may beundesirable because they are mechanically weak points, whichmay cause membrane failure under continuous vibration andbackwashing operations.23 On the other hand, hollow fibermembranes with a favorable macrovoid-free structure could bethe more preferential design in the industrial applications.In this work, we have demonstrated novel TFC-FO hollow

fiber membranes with a high flux and a favorable macrovoid-free substructure for water production through FO processes.The feasibility of fabricating well-defined FO hollow fibermembrane supports has been demonstrated by employing theadvanced coextrusion technologies to effectively control thephase inversion and create appropriate membrane supports forinterfacial polymerization. A simple and highly reproducibleinterfacial polymerization method, using the m-phenylenedi-amine (MPD) and trimesoyl chloride (TMC) monomers, wasdeveloped. The simple and cost-effective fabrication techniqueusing a dual-layer spinneret demonstrated through this studycould provide another platform for future FO membranedevelopment.

MATERIALS AND METHODS

Fabrication of Macrovoid-Free PES Hollow FiberMembrane Supports. The hollow fiber membranes wereprepared by a dry-jet wet spinning process employing theadvanced coextrusion technology through a dual-layer spinner-et. A detailed description of the hollow fiber spinning process isdocumented elsewhere.24 The dual-layer spinneret and itsdimension used in this study are described in Figure S1 of the

Supporting Information (SI). The spinning conditions for thePES hollow fiber membrane supports are listed in Table S1 ofthe SI. The detailed procedures for the polymer dopepreparation, spinning process, membrane post-treatment, andmodule fabrication are described in the SI.

Interfacial Polymerization of Thin-Film-Composite(TFC) Forward Osmosis Hollow Fiber Membranes. Theformation of a polyamide thin layer on the inner surface of PEShollow fiber supports was achieved by an interfacial polymer-ization (IP) between MPD monomers in the aqueous phaseand TMC monomers in the organic phase. The detailedspecification of the experiment setup and preparation steps isdisclosed in the SI.

Characterizations of PES Hollow Fiber MembraneSupports and TFC-FO Hollow Fiber Membranes. Thefabricated hollow fiber membrane supports were first tested tomeasure pure water permeability (PWP) (L m2− h−1 bar−1 orLMHbar−1) using a lab-scale circulating filtration unit, asdescribed in SI or eslewhere.10,25 The membranes were thencharacterized by solute rejection measurements. Feed solutionscontaining 200 ppm of different molecular weights of PEG orPEO were utilized as the neutral solutes to estimate pore size,pore size distribution, and molecular weight cutoff (MWCO)according to the solute transport method described in the SI orelsewhere.17,26 The water permeability A, salt rejection Rs, andsalt permeability B of TFC-FO hollow fiber membranes weredetermined by testing the membranes under the RO modefollowing the method described elsewhere.19,25

Water Reclamation through Forward Osmosis Testsof TFC-FO Hollow Fiber Membranes and Evaluation ofStructural Parameters of Supporting Layers. Forwardosmosis (FO) experiments were conducted on a lab-scalecirculating filtration unit.25 The detailed experimental setup,operating conditions, and the determination of structural

Figure 1. SEM micrographs of different bulk and surface morphologies of PES hollow fiber membrane supports (spinning code: PESwater).




parameters are described in the SI. A schematic diagram of FOsetup is shown in Figure S2 of the SI.

RESULTS AND DISCUSSIONMorphology, Microstructure, and Separation Charac-

teristics of PES Hollow Fiber Supports. A typicalmorphology of PES hollow fiber membrane supports spunusing water as a bore-fluid (PESwater) is illustrated in Figure 1.The hollow fibers have a macrovoid-free and fully sponge-likestructure with a high degree of concentricity. The inner surfaceof as-spun membranes exhibits relatively small pores andsmooth surface, compared to the outer surface which possessesa highly porous structure and relatively large pore sizes of 1 μm(estimated from the FESEM observation, Figure 1). Similarbulk and surface morphologies as discussed above are obtainedfor the hollow fibers spun with other conditions (70/30 wt %water/NMP and 40/30/30 wt % water/NMP/PEG asnonsolvent bore-fluids), as illustrated in Figure S3 of the SI.These morphological features are consistent with our strategy,employing dual-layer spinneret to facilitate membrane phaseinversion during fiber spinning, as depicted in Figure 2. The

pure NMP solvent was fed at the outer channel to induce adelayed demixing23 (slow phase inversion and mild solvent

exchange) in the air-gap region prior to entering the watercoagulant, attempting to form a macrovoid-free cross-sectionand a porous structure at the membrane’s outer surface. Incontrast, an instantaneous demixing (rapid phase inversion andfast solvent exchange) was developed by introducing anonsolvent at the bore-fluid channel to create a relativelydense inner surface. Moreover, based on our observation duringexperiments, the fabricated hollow fiber membranes spun fromthe dual-layer coextrusion technique tend to have a uniformdimension and no irregular shape, which reflects the uniformsolvent exchange rate during the phase inversion process.Another important parameter that plays an important role ineliminating larger finger-like macrovoids and meanwhilecreating highly porous and open-cell structure is the additionof PEG and water nonsolvent additives with optimumconcentrations into dope solutions. PEG, a highly hydrophilicpolymer and a weak nonsolvent of a PES/NMP system, isknown as an effective additive to enhance pore formation,improve pore interconnectivity, and prevent macrovoidformation. Water, a strong nonsolvent of the system, is added(with a relative small amount) to enhance dope viscosity and tobring the dope formulation close to the binodal compositionwhich could favor the macrovoid suppression.27

The inner surface roughness of hollow fibers was examinedby AFM and is displayed in Figure S4. All as-spun fibers have asimilar surface roughness of 2−3 nm regardless of bore-fluidspinning conditions. It can be seen that the membranes spunwith pure water exhibited slightly higher surface roughnesscompared to others. Similar phenomenon has been reported byWidjojo et al.28 and Sukitpaneenit and Chung29 where theinstantaneous demixing induced by strong nonsolvent, e.g.water, tends to result in membranes with a rougher surface (Ra= 2.85 nm). In other words, the delayed demixing could ratherresult in a smoother surface and a smaller surface roughness asshown in the case of membranes spun with water/NMP (Ra =2.25 nm) and water/NMP/PEG (Ra = 2.69 nm). Nevertheless,one may observe a slight increase in surface roughness of themembranes PESwater/NMP/PEG, compared to the membranePESwater/NMP. This may be due to the fact that (1) addingPEG, known as a pore forming additive, in the bore-fluid mayenhance the pore evolution at the contacting inner surfaceduring membrane formation, and (2) the highly hydrophilicPEG polymer in the bore-fluid may complicate the precip-itation rate and preferentially create a relatively higher surfaceroughness. In other words, introducing PEG into a bore-fluidsolution could enhance the delayed demixing and slow theprecipitation rate to some certain extent, leading to formationof a membrane with relatively higher surface porosity/

Figure 2. Strategies to control the phase inversion process with the aidof coextrusion technology employing a dual-layer spinneret.

Table 1. Summary of Mean Effective Pore Size (μp), PWP, MWCO, Porosity, Water Contact Angle, and Mechanical Propertiesof PES Hollow Fiber Membrane Supports

membrane ID PESwater PESwater/NMP PESwater/NMP/PEG

mean pore size, μp (nm) (at the inner surface) 15.01 16.50 17.05geometric standard variation, σp 1.30 1.33 1.34PWP (L m2− h−1 bar−1) (at 1 bar) 835 856 1021MWCO (Da) 106 967 132 349 142 672porosity (%) 80.0 ± 1.4 82.0 ± 1.0 80.9 ± 1.6water contact angle, θ (°) (at the inner surface) 64.6 ± 2.0 56.5 ± 2.1 54.8 ± 2.8tensile strength (MPa) 5.88 ± 0.27 6.97 ± 0.23 5.81 ± 0.09elongation at break (%) 48.13 ± 5.78 49.82 ± 8.34 50.90 ± 4.65Young’s modulus (MPa) 273.80 ± 21.58 228.30 ± 49.56 196.27 ± 27.31




roughness. Besides, the highly hydrophilic characteristic of PEGmay further hinder the solvent/nonsolvent exchange rateduring phase inversion at the inner surface due to high affinitywith added PEG in the polymer dope, and thus enhance thesurface roughness.27

Basic characteristics of PES hollow fiber membrane supportsbefore interfacial polymerization, which include the mean poresize, pure water permeability (PWP), molecular-weight cutoff(MWCO), porosity, and water contact angle, are listed in Table1. All spun PES hollow fiber membranes possess a small poresize in a range of 15−17 nm and a narrow pore size distributionprofile (small geometric standard variation of 1.30−1.34),which is essential to produce a continuous and homogeneouspolyamide layer through interfacial polymerization.28,30 A clearcomparison of pore size and pore size distribution of theresultant membranes is depicted in Figure S5.As shown in Table 1, the PESwater, PESwater/NMP, and

PESwater/NMP/PEG supports display high PWP values of 835,856, and 1021 L m2− h−1 bar−1 at 1 bar, respectively. Thistendency is apparently attributed to their corresponding poresize, pore size distribution, and high overall porosity (80−82%)for each respective membrane. Additionally, the membranespossess moderately low water contact angles of 55−65° (<90°)at the inner surface that not only reflect their reasonably goodhydrophilicity, but also facilitate water transport.28,30 Themechanical strength of these hollow fiber supports wasinvestigated and is also summarized in Table 1. In mostcases, they have comparable tensile strength of 5.8−7 MPa,elongation at break of 48−50%, and Young’s modulus of 200−270 MPa within the error ranges. As anticipated, the hollowfiber membranes in this present work display a bettermechanical strength (5.8−7 MPa) compared to the previousstudies on TFC-FO hollow fiber (4.1−5.4 MPa)18 and otherPES ultrafiltration membranes (0.94−1.65 MPa)31 reported inthe literature. This is significantly contributed to the presenceof macrovoid-free sponge-like structure in the hollow fibers.Overall, the membrane supports fabricated in this study have

possessed desirable characteristics of being supports forinterfacial polymerization. These include the mean pore size

in nanorange, high porosity, and high hydrophilicity, macro-void-free with enhanced mechanical properties.

Characteristics and Performance of TFC-FO HollowFiber Membranes. The active polyamide layer of the TFC-FO hollow fiber membranes after interfacial polymerization onthe aforementioned membrane supports is depicted in Figure 3.The selective layer has a uniform ridge-and-valley morphology,which is a typical characteristic of polyamide membranesformed using an interfacial polymerization.32 The estimatedthickness of the selective polyamide layer is 380 ± 25 nmregardless of different membrane supports. There is nosignificant change observed on the outer surface morphologyafter IP, indicating no IP solutions penetrating across themembrane bulk.The fabricated TFC-FO hollow fiber membranes with

different membrane supports were evaluated for their FOperformance in both FO and PRO operating modes using DIwater feed and 2 M NaCl as a draw solution. The membraneperformance was expressed in terms of water flux and saltreverse flux as summarized in Table 2. Interestingly, allmembranes demonstrate remarkable high water fluxes andreasonably low salt leakages. Based on the experimental results,

Figure 3. Typical morphology of TFC-FO hollow fiber membranes with PESwater/NMP supports layer after interfacial polymerization.

Table 2. PRO and FO Performance of Developed TFC-FOHollow Fiber Membranes with Different Bore-FluidSpinning Conditionsa

PRO mode FO mode

membrane IDwater flux(LMH)

saltleakage(gMH)

water flux(LMH)

saltleakage(gMH)

TFC-FO with PESwatersupports

57.1 6.93 32.1 6.15

TFC-FO withPESwater/NMP supports

62.7 10.30 34.1 7.10

TFC-FO withPESwater/NMP/PEGsupports

65.1 12.34 34.5 9.87

aExperimental conditions: DI water as the feed solution, 2.0 M NaClas the draw solution.




TFC-FO-PESwater hollow fiber membranes exhibit the waterfluxes of 32.1 and 57.1 LMH with relatively low reverse fluxesof 6.15 and 6.93 gMH for FO and PRO modes, respectively. Inthe case of the TFC-FO-PESwater/NMP/PEG membranes, thewater flux moves up to 34.5 and 65.1 LMH but with slightlyincreased reverse salt fluxes of 9.87 and 12.34 gMH, forrespective FO and PRO modes, under the same 2.0 M NaCldraw solution. The increase in water flux and reverse salt flux ofthe resultant TFC-FO membranes in both FO and PRO modesis in the following order: TFC-FO-PESwater < TFC-FO-PESwater/NMP < TFC-FO-PESwater/NMP/PEG.A comparison of FO performance of the resultant TFC-FO

hollow fiber membranes with other thin-film compositemembranes and commercial membranes reported in theliterature for FO applications is presented in Table3.18,19,28,30,33−36 Overall, the fabricated TFC-FO membranesshow relatively high water fluxes in both FO and PRO modes;the water fluxes as high as 32−34.5 LMH under the FO modeand exceeding up to 57−65 LMH under the PRO mode withacceptable reverse salt fluxes of 6−12 gMH could be achievedusing a 2 M NaCl draw solution and pure water as the feed.Compared to other TFC and commercially available HTI flatsheet membranes, the developed TFC hollow fiber membranesnot only offer high water fluxes and less salt leakages, whichcould be attributed to the inherent characteristics of the hollowfiber membrane configuration, but also possess a well-structured macrovoid-free morphology with enhanced mechan-ical properties to sustain the membrane stability in the realapplication.

Another interesting finding is that there is a close relationshipbetween the pore characteristics of membrane supports derivedfrom different bore-fluid conditions and the FO performance oftheir corresponding TFC-FO membranes. As all membranesupports possess almost equivalent overall porosity andhydrophilicity, the membrane support with a smaller poresize and a narrower pore size distribution could yield the TFC-FO membrane (i.e., TFC-FO-PESwater) with an impressive highwater flux but less salt leakage, compared to those with a largerpore size and a wider pore size distribution,22 i.e., TFC-FO withPESwater/NMP and TFC-FO with PESwater/NMP/PEG membranes.The significant contribution of TFC-FO support layers on

the internal concentration polarization (ICP) was investigated,which is typically expressed in terms of structural parameter, S.The structural parameters of different hollow fiber supportswere determined by fitting the FO mode experiment data inTable 2 with the equations S11 and S12 in the SI for eachmembrane condition. The water permeability (A), saltpermeability (B) coefficients, solute diffusion resistivity withinthe porous layer (Km) involved in the calculation, and thecorresponding structural parameters of TFC-FO hollow fibermembranes with the supports derived from different bore-fluidcompositions are tabulated in Table 4. From the results, thewater and salt permeabilities obtained from RO tests are inagreement with our FO experimental data. Membranespossessing a higher intrinsic water permeability have tendencyto achieve a higher water flux in actual FO operation. In termsof structural parameter, all membranes have structuralparameters in a range of 219−261 μm (2.19−2.61 × 10−4

Table 3. Comparison of FO Membrane Performancea

membranewater flux, Jv (FO/PRO) (L m‑2

h‑1)reverse salt flux, Js (FO/PRO) (g m‑2

h‑1 )draw

solution ref

TFC-FO with PESwater supports 22.5/25.6 2.8/3.2 0.5 M NaCl this workTFC-PES hollow fiber (#B-FO dope formula) 13.5/32.2 1.8/3.5 0.5 M NaCl 18TFC-PES hollow fiber (#C-FO dope formula) 18.5/42.6 1.5/4.0 0.5 M NaCl 19TFC-PESU E6020P and PESU-co-sPPSU 11 flat sheetmembrane

14.8/15.5 2.2/2.5 0.5 M NaCl 28

TFC-PES/SPSf flat sheet membrane 13.0/24.1 3.6/4.5 0.5 M NaCl 30TFC-Cellulose acetate propionate (CAP) membrane 8.0/15.0 0.8/0.8 0.5 M NaCl 33HTI flat sheet membrane 18.6 (PRO) 0.5 M NaCl 34TFC-FO with PESwater supports 32.1/57.1 6.15/6.93 2.0 M NaCl this workTFC-FO with PESwater/NMP supports 34.1/62.7 7.10/10.30 2.0 M NaCl this workTFC-FO with PESwater/NMP/PEG supports 34.5/65.1 9.87/12.34 2.0 M NaCl this workTFC-PES hollow fiber (#B-FO dope formula) 22.5/45.0 2.5/4.4 2.0 M NaCl 18TFC-PES hollow fiber (#C-FO dope formula) 29.5/68.0 2.6/5.8 2.0 M NaCl 19TFC-PESU E6020P and PESU-co-sPPSU 11 flat sheetmembrane

21.0/33.0 2.2/2.8 2.0 M NaCl 28

TFC-PES/SPSf flat sheet membrane 26.0/47.5 8.3/12.4 2.0 M NaCl 30TFC-cellulose acetate propionate (CAP) membrane 17/35 (PRO) 1.85/1.9 (FO) 2.0 M NaCl 33HTI flat sheet membrane 13.0 (FO) 10.5 2.0 M NaCl 35HTI flat sheet membrane 11.2 (PRO) 2.0 M NaCl 36aExperiment conducted at 20−25 °C; FO/PRO in bracket refers to FO and PRO testing modes in FO experiment.

Table 4. Intrinsic Separation Properties and Structural Parameters of TFC-FO Hollow Fiber Membranes

membrane IDwater permeability, A (Lm‑2 h‑1

bar‑1)salt rejectiona

(%)salt permeability, B (Lm‑2

h‑1) Km (sm‑1) S (m)b

TFC-FO with PESwater supports 1.18 87.95 0.135 1.45 × 105 2.19 × 10−4

TFC-FO with PESwater/NMP supports 1.64 83.04 0.281 1.65 × 105 2.52 × 10−4

TFC-FO with PESwater/NMP/PEG supports 1.83 81.52 0.348 1.73 × 105 2.61 × 10−4

a200 ppm NaCl as the feed solution in the RO test under an applied pressure of 1 bar. bCalculated based on the experiments under the FO modeusing 2 M NaCl as the draw solution and DI water as the feed solution.



m), which is relatively low compared to the values of mostmembranes investigated in the literature. The S values obtainedare slightly smaller than those reported for commercial HTI FOmembranes (481 μm)35 and TFC hollow fiber membranes(550−595 μm),18,21 but nearly 2 orders of magnitude smallerthan the ones for conventional RO membranes (∼ 40 mm forBW30 membranes).21 The lower the structural parameter, theless severe the ICP. Hence, much less ICP would beencountered with our membranes during FO processes, andthe effective driving force would be preserved to the highestextent.It has been hypothesized in previous studies18,19 that an ideal

membrane support for TFC-FO hollow fiber membranesshould possesses an array of large finger-like structures with athin layer of sponge-like structures near the functional selectivelayer. Our results reveal that the TFC-FO macrovoid-freehollow fiber supports with a highly sponge-like structure couldbe an alternative favorable membrane morphology to reduceICP, and thus enhance water permeation flux. In other words, amembrane substrate with a highly finger-like structure is not theessential requirement to form a high flux TFC-FO hollow fibermembrane.In addition, the TFC-FO hollow fiber membranes with

PESwater supports and the lowest structural parameter werefurther evaluated for the membrane performance under variousdraw solution concentrations (0.5−2 M NaCl), as illustrated inFigure 4. Regardless of the FO and PRO operating modes, thewater flux increases with an increase in draw solutionconcentration because a larger effective osmotic pressuredifference provides a greater driving force. The water flux inPRO mode is comparable with the water flux in FO mode atrelatively low draw solution concentrations, e.g. 0.5 M NaCl;but there seems to be more deviation at increased drawsolution concentrations yielding a much higher water flux forthe PRO mode. For all cases, membranes show higher saltleakages at higher draw solution concentrations whereby onlyslight larger reverse salt fluxes were observed for the PRO case.In addition, one may observe that there is a rapid proportionalincrement of water fluxes along the increase of draw solutionconcentrations in the PRO mode whereas the water flux in FOmode shows relative plateau tendency at relatively high drawsolution concentrations. This implies the evolution and impact

of ICP within the support layer as a function of draw solutionstrengths. Moreover, this phenomenon also signifies that waterflux tested under the FO mode may experience severer ICPrelative to that under the PRO mode especially at elevated drawsolution concentrations. A comparison between experimentalwater flux and the model prediction is also shown in Figure 4. Itis observed that the model curve fits well the experimental flux.

Potentials of Developed TFC-FO Hollow FiberMembranes for Water Production through ForwardOsmosis Seawater Desalination. The newly developedTFC-FO hollow fiber membranes were characterized toevaluate their potential for seawater desalination. A modelseawater solution (3.5 wt % NaCl) was employed to test themembrane performance. Using 2 M NaCl as a draw solutionand under PRO mode, the TFC-FO hollow fiber membranesexhibit water production of 15, 16, and 18 LMH with differentsupports of PESwater, PESwater/NMP, and PESwater/NMP/PEG,respectively. Compared to previous tests using DI water as afeed solution, the water fluxes attained from all TFC-FOmembranes decrease when seawater is used as the feed. This isdue to the reduction of the overall osmotic pressure differencebetween the feed and the draw solution.A performance comparison of the resultant membranes with

the literature for FO seawater desalination under the sameoperating conditions mentioned above is illustrated in Figure 5.Comparing with various membrane materials includingcellulose acetate (CA) membranes,37 commercial HTImembranes,36 thin-film composite flat sheet, and hollow fibermembranes,19,28,30,33 the TFC-FO hollow fiber membranesdeveloped in this work displays a higher water flux, andcomparable to the best in the literature tested under the sameconditions using seawater as the feed. With such encouragingmembrane performance, and together with the benefits of cost-effective membrane synthesis and fabrication through a dual-layer coextrusion approach which requires a relatively smallamount of solvent to facilitate phase inversion during themembrane fabrication, the newly developed TFC-FO hollowfiber membrane can be a promising candidate for osmoticseawater desalination applications.

Figure 4. Water flux (a) and reverse salt flux (b) for TFC-FO membranes (PESwater supports) under FO and PRO operating modes using water asthe feed solution against different draw solution concentrations.




ASSOCIATED CONTENT

*S Supporting InformationMaterials; procedures for hollow fiber spinning, membranepost-treatment, and module fabrication; protocols for physicaland mass transport characterizations of membranes; forwardosmosis setup and operating conditions; additional SEM andAFM images of the membrane supports; pore size distributionprofiles. This material is available free of charge via the Internetat http://pubs.acs.org.

AUTHOR INFORMATION

Corresponding Author*Tel.: 65 6516 6645; fax: 65 6779 1936; e-mail: [email protected].


ACKNOWLEDGMENTS

We thank the Singapore National Research Foundation (NRF)support through the Competitive Research Program for theproject entitled “New Advanced FO membranes andmembrane systems for wastewater treatment, water reuse andseawater desalination” with grant R-279-000-311-281. Specialthanks are also given to Dr. Jincai Su, Ms. Sui Zhang, Ms. XueLi, and Ms. Fitri Juniwati for their valuable assistance. Dr. PanuSukitpaneenit also acknowledges the World Future Foundation(WFF) for awarding his Ph.D. Prize in Environmental andSustainability Research, 2012.

REFERENCES(1) Elimelech, M.; Phillip, W. A. The future of seawater desalination:Energy, technology and the environment. Science 2011, 333, 712−717.

(2) Cath, T. Y.; Childress, A. E.; Elimelech, M. Forward osmosis:Principles, applications, and recent developments. J. Membr. Sci. 2006,281, 70−87.(3) Zhao, S.; Zou, L.; Tang, C. Y.; Mulcahy, D. Recent developmentsin forward osmosis: Opportunities and challenges. J. Membr. Sci. 2012,396, 1−21.(4) Chung, T. S.; Zhang, S.; Wang, K. Y.; Su, J.; Ling, M. M. Forwardosmosis processes: Yesterday, today and tomorrow. Desalination 2012,287, 78−81.(5) Cornelissen, E. R.; Harmsen, D.; de Korte, K. F.; Ruiken, C. J.;Qin, J.-J.; Oo, H.; Wessels, L. P. Membrane fouling and processperformance of forward osmosis membranes on activated sludge. J.Membr. Sci. 2008, 319, 158−168.(6) Mi, B.; Elimelech, M. Organic fouling of forward osmosismembranes: Fouling reversibility and cleaning without chemicalreagents. J. Membr. Sci. 2010, 348, 337−345.(7) Zhao, S.; Zou, L. Effects of working temperature on separationperformance, membrane scaling and cleaning in forward osmosisdesalination. Desalination 2011, 278, 157−164.(8) Mi, B.; Elimelech, M. Gypsum scaling and cleaning in forwardosmosis: Measurement and mechanisms. Environ. Sci. Technol. 2010,44, 2022−2028.(9) Jiao, B.; Cassano, A.; Drioli, E. Recent advances on membraneprocesses for the concentration of fruit juices: A review. J. Food Eng.2004, 63, 303−324.(10) Yang, Q.; Wang, K. Y.; Chung, T. S. A novel dual-layer forwardosmosis membrane for protein enrichment and concentration. Sep.Purif. Technol. 2009, 69, 269−274.(11) Lee, K. L.; Baker, R. W.; Lonsdale, H. K. Membranes for powergeneration by pressure-retarded osmosis. J. Membr. Sci. 1981, 8, 141−171.(12) Gerstandt, K.; Peinemann, K. -V.; Skilhagen, S. E.; Thorsen, T.;Holt, T. Membrane processes in energy supply for an osmotic powerplant. Desalination 2008, 224, 64−70.(13) McCutcheon, J. R.; McGinnis, R. L.; Elimelech, M. A novelammonia-carbon dioxide forward (direct) osmosis desalinationprocess. Desalination 2005, 174, 1−11.(14) Phuntsho, S.; Shon, H. K.; Hong, S.; Lee, S.; Vigneswaran, S. Anovel low energy fertilizer driven forward osmosis desalination fordirect fertigation: Evaluating the performance of fertilizer drawsolutions. J. Membr. Sci. 2011, 375, 172−181.(15) Cartinella, J. L.; Cath, T. Y.; Flynn, M. T.; Miller, G. C.; Hunter,K. W.; Childress, A. E. Removal of natural steroid hormones fromwastewater using membrane contactor processes. Environ. Sci. Technol.2006, 40, 7381−7386.(16) Hoover, L. A.; Phillip, W. A.; Tiraferri, A.; Yip, N. Y.; Elimelech,M. Forward with osmosis: Emerging applications for greatersustainability. Environ. Sci. Technol. 2011, 45, 9824−9830.(17) Yang, Q.; Wang, K. Y.; Chung, T. S. Dual-layer hollow fiberswith enhanced flux as novel forward osmosis membranes for waterproduction. Environ. Sci. Technol. 2009, 43, 2800−2805.(18) Wang, R.; Shi, L.; Tang, C. Y.; Chou, S.; Qiu, C.; Fane, A. G.Characterization of novel forward osmosis hollow fiber membranes. J.Membr. Sci. 2010, 355, 158−167.(19) Chou, S.; Shi, L.; Wang, R.; Tang, C. Y.; Qiu, C.; Fane, A. G.Characteristics and potential applications of a novel forward osmosishollow fiber membrane. Desalination 2010, 261, 365−372.(20) Yip, N. Y.; Tiraferri, A.; Phillip, W. A.; Schiffman, J. D.;Elimelech, M. High performance thin-film composite forward osmosismembrane. Environ. Sci. Technol. 2010, 44, 3812−3818.(21) Wei, J.; Qiu, C.; Tang, C. Y.; Wang, R.; Fane, A. G. Synthesisand characterization of flat-sheet thin-film composite forward osmosismembranes. J. Membr. Sci. 2011, 372, 292−302.(22) Tiraferri, A.; Yip, N. Y.; Phillip, W. A.; Schiffman, J. D.;Elimelech, M. Relating performance of thin-film composite forwardosmosis membranes to support layer formation and structure. J.Membr. Sci. 2011, 367, 340−352.(23) Peng, N.; Widjojo, N.; Sukitpaneenit, P.; Teoh, M. M.;Lipscomb, G. G.; Chung, T. S.; Lai, J.-Y. Evolution of polymeric

Figure 5. Current status of potential water flux reported in theliterature for FO seawater desalination applications. Note: The modelseawater (3.5 wt % NaCl) solution as the feed solution, PRO mode,2.0 M NaCl draw solution. The membrane materials include celluloseacetate,37 commercial HTI,36 TFC cellulose acetate propionate,33 TFChollow fiber (finger-like macrovoids),19 TFC PES/sulfonated PSF,30

and TFC sulfonated PSF.28



http://pubs.acs.org




hollow fibers as sustainable technologies: Past, present, and future.Prog. Polym. Sci. 2012, No. http://dx.doi.org/10.1016/j.progpolyms-ci.2012.01.001.(24) Sukitpaneenit, P.; Chung, T. S. Molecular elucidation ofmorphology and mechanical properties of PVDF hollow fibermembranes from aspects of phase inversion, crystallization andrheology. J. Membr. Sci. 2009, 340, 192−205.(25) Su, J.; Yang, Q.; Teo, J. F.; Chung, T. S. Cellulose acetatenanofiltration hollow fiber membranes for forward osmosis processes.J. Membr. Sci. 2010, 355, 36−44.(26) Wang, K. Y.; Matsuura, T.; Chung, T. S.; Guo, W. F. The effectsof flow angle and shear rate within the spinneret on the separationperformance of poly(ethersulfone) (PES) ultrafiltration hollow fibermembranes. J. Membr. Sci. 2004, 240, 67−79.(27) Liu, Y.; Koops, G. H.; Strathman, H. Characterization ofmorphology controlled polyethersulfone hollow fiber membranes bythe addition of polyethylene glycol to the dope and bore liquidsolution. J. Membr. Sci. 2003, 223, 187−199.(28) Widjojo, N.; Chung, T. S.; Weber, M.; Maletzko, C.; Warzelhan,V. The role of sulphonated polymer and macrovoid-free structure inthe support layer for thin-film composite (TFC) forward osmosis(FO) membranes. J. Membr. Sci. 2011, 383, 214−223.(29) Sukitpaneenit, P.; Chung, T. S. Molecular design of themorphology and pore size of PVDF hollow fiber membranes forethanol-water separation employing the modified pore-flow concept. J.Membr. Sci. 2011, 374, 67−82.(30) Wang, K. Y.; Chung, T. S.; Amy, G. Developing thin-film-composite forward osmosis membranes on the PES/SPSf substratethrough interfacial polymerization. AIChE J. 2012, 58, 770−781.(31) Chung, T. S.; Qin, J.-J.; Gu, J. Effect of shear rate within thespinneret on morphology, separation performance and mechanicalproperties of ultrafiltration polyethersulfone hollow fiber membranes.Chem. Eng. Sci. 2000, 55, 1077−1091.(32) Ghosh, A. K.; Jeong, B.-H.; Huang, X. F.; Hoek, E. M. V.Impacts of reaction and curing conditions on polyamide compositereverse osmosis membrane properties. J. Membr. Sci. 2008, 311, 34−35.(33) Li, X.; Wang, K. Y.; Helmer, B.; Chung, T. S. Thin-filmcomposite membranes and formation mechanism of thin-film layers onhydrophilic cellulose acetate propionate substrates for forward osmosisprocesses. Ind. Eng. Chem. Res. 2012, No. 10.1021/ie2027052.(34) Gray, G. T.; McCutcheon, J. R.; Elimelech, M. Internalconcentration polarization in forward osmosis: Role of membraneorientation. Desalination 2006, 197, 1−8.(35) Phillip, W. A.; Yong, J. S.; Elimelech, M. Reverse draw solutepermeation in forward osmosis: Modeling and experiments. Environ.Sci. Technol. 2010, 44, 5170−5176.(36) McCutcheon, J. R.; McGinnis, R. L.; Elimelech, M. Desalinationby ammonia-carbon dioxide forward osmosis: Influence of draw andfeed solution concentrations on process performance. J. Membr. Sci.2006, 278, 114−123.(37) Zhang, S.; Wang, K. Y.; Chung, T. S.; Chen, H.; Jean, Y. C.;Amy, G. Well-constructed cellulose acetate membranes for forwardosmosis: Minimized internal concentration polarization with an ultra-thin selective layer. J. Membr. Sci. 2010, 360, 522−535.



Development of Thin-Film Composite forward Osmosis Hollow FiberMembranes Using Direct Sulfonated Polyphenylenesulfone (sPPSU)as Membrane SubstratesPeishan Zhong,†,‡ Xiuzhu Fu,‡ Tai-Shung Chung,‡,* Martin Weber,§ and Christian Maletzko∥

†Functionalized Materials and Nanostructures, Global Research Center Singapore (A-GMM/F), BASF South East Asia Pte Ltd 61Science Park Road, Singapore 117525‡Department of Chemical & Biomolecular Engineering, National University of Singapore, 10 Kent Ridge Crescent, Singapore 119260§Advanced Materials & Systems Research, Performance Materials, BASF SE GMV/W-B1, 67056 Ludwigshafen, Germany∥Performance Materials, BASF SE G-PMS/PS-F206, 67056 Ludwigshafen, Germany


ABSTRACT: This study investigates a new approach to fabricate thin-filmcomposite (TFC) hollow fiber membranes via interfacial polymerization forforward osmosis (FO) applications. Different degrees of sulfonation ofpolyphenylenesulfone (PPSU) were adopted as membrane substrates toinvestigate their impact on water flux. It has been established that the degreeof sulfonation plays a role in both creating a macrovoid-free structure andinducing hydrophilicity to bring about higher water fluxes. The fabricatedmembranes exhibit extremely high water fluxes of 30.6 and 82.0 LMHagainst a pure water feed using 2.0 M NaCl as the draw solution testedunder FO and pressure retarded osmosis (PRO) modes, respectively, whilemaintaining low salt reverse fluxes below 12.7 gMH. The structuralparameter (S) displays remarkable decreases of up to 4.5 times as themembrane substrate is switched from a nonsulfonated to sulfonated one. Inaddition, the newly developed TFC-FO membranes containing 1.5 mol %sPPSU in the substrate achieves a water flux of 22 LMH in seawater desalination using a 3.5 wt % NaCl model solution and 2.0M NaCl as the draw solution under the PRO mode. To the best of our knowledge, this value is the highest ever reported forseawater desalination using flat and hollow fiber FO membranes. The use of sulfonated materials in the FO process opens up afrontier for sustainable and efficient production of potable water.

INTRODUCTION

Water scarcity in the 21st century is a pressing global issue1,2

and forward osmosis has attracted much attention as anemerging technology3,4 to alleviate this problem. It also bringsabout potential applications including power generation,5,6

wastewater treatment, food processing,7,8 and protein enrich-ment.9 There are many advantages of the forward osmosisprocess which include requiring much less energy to induce anet flow of water in comparison with the traditional watertreatment process such as reverse osmosis (RO) and lowmembrane fouling.10−12 The use of an osmotic pressuregradient also helps reduce the use of depleting fuel sourceswhich has brought about detrimental environmental problemsto our earth.The current dominant FO membrane in the commercial

application comes from Hydration Technology Inc. (HTI).However, these membranes are cellulose-based and faceproblems such as degradation by microorganisms, narrow pHworking range and low salt rejection. In contrast, the polyamidethin film composite (TFC) membrane fabricated frominterfacial polymerization upon a suitable substrate is an

attractive technique to mitigate the mentioned problems. Thistechnique has been widely applied in the reverse osmosis (RO)process13,14 and its concept has been extended to the FOprocess. Many studies have worked on the flat sheet15−17

configuration while few studies have investigated TFCmembranes in the hollow fiber configuration.18−20

The key prerequisites for the development of desirable TFCmembranes for FO include (1) ultrathin semipermeable activelayer with low water transport resistance and high salt rejection;(2) highly porous and thin supporting layer to minimizeinternal concentration polarization (ICP); (3) low foulingpropensity and (4) sufficiently high mechanical strength.10

These fundamental requirements have to be met in order tomeet and achieve the attractive qualities, offered by the forwardosmosis process. However, to date, the development of forwardosmosis encounters the challenge of ineffective membranes that

Received: March 26, 2013Revised: May 31, 2013Accepted: June 3, 2013Published: June 3, 2013

Article

pubs.acs.org/est

© 2013 American Chemical Society 7430 dx.doi.org/10.1021/es4013273 | Environ. Sci. Technol. 2013, 47, 7430−7436

pubs.acs.org/est

play the critical role in most FO processes.11 Hence, there is aneed to search for suitable materials. These materials arepreferred to be hydrophilic.21

In this work, direct sulfonated polyphenylenesulfone(sPPSU) with different concentrations of sulfonated monomer(sDCDPS), that is, 1.5 and 2.5 mol %, were applied assubstrates for FO membranes. The objectives of this study is to(1) investigate the effect of sulfonation degree on themorphology of hollow fiber substrates spun from differentspinning parameters; (2) explore the effect of sulfonationdegree of the TFC membrane substrate on FO performance;and (3) explore their potential for seawater desalination. It hasbeen postulated that an increase in hydrophilicity of membranesubstrates would increase the FO water flux of thesemembranes and provide a solution to the current issues facedin the FO process.A study by Shi et al has investigated the effect of substrate

structure on the formation of the TFC layer upon it andconcluded that the substrate should possess an MWCO of lessthan 300 kDa in order to achieve a desirable semipermeableskin.22 Therefore, the MWCO of the substrates fabricated inthis work will be controlled within the desirable range. Theconcentration of the PEG400 which acts as a pore former wasvaried in this work and the optimal condition was used.In this work, hollow fibers with an inner selective layer were

designed due to advantages which include (1) feed being ableto be delivered into the lumen equally, (2) easy protection ofthe defect-free separation layer, and (3) low transportresistance in the permeate side.23 The formulation of thedope to yield a macrovoid-free structure was also explored asmacrovoids are viewed as mechanically weaker points on themembrane. The effect of PEG400 and the addition of water tothe dope solution for macrovoid suppression were studiedalong with the spinning parameters.

MATERIALS AND METHODSMaterials. Polyphenylenesulfone (PPSU), 1.5 mol % and

2.5 mol % sulfonated polyphenylenesulfone (sPPSU) were usedas the materials for the fabrication of various membranesubstrates. The sPPSU materials were synthesized via the directsulfonation synthesis route developed by McGrath et al.24 The1.5 mol % and 2.5 mol % sPPSU have IEC values of 5.1 and 8.2meq/100g polymer. N-methyl-2-pyrrolidone (NMP) andpolyethylene glycol 400 (PEG400) from Merck were used asthe solvent and pore former in the dope solution preparation,respectively. More information on the chemicals used forinterfacial polymerization can be found in the SupportingInformation (SI).Cloud Point Measurements. To determine the binodal

composition of the PPSU/PEG400/NMP/water system, cloudpoint data were obtained by means of a titration method. It isknown that a dope formulation near to the binodal curve willresult in membranes with macrovoid-free structures upon phaseinversion.25,26 Distilled water was slowly added to the system ofpolymer, additive, and solvent. It was observed that the additionof water caused some local coagulation of the dope which wasleft to stir till it became homogeneous again. This procedurewas repeated till the dope became permanently turbid. 93 wt %of the amount of water required to cause phase separation wasadded to each dope (coagulation value).Hollow Fiber Spinning of Hollow Fiber Membrane

Substrates. The hollow fiber membrane substrates wereprepared via a dry-jet wet spinning process employing an

advanced coextrusion technology using a dual-layer spinneret.Polymer solutions were prepared according to the concen-trations as described in Table S1 of Supporting Information(SI). The spinning parameters as well as the membrane post-treatment and module fabrication are described in the SI.

Interfacial Polymerization for Fabrication of TFC-FOHollow Fiber Membranes. The M-phenylenediamine(MPD) solution in aqueous phase and trimesoyl chloride(TMC) solution in the organic phase were brought into contactleading to the formation of a polyamide layer throughinterfacial polymerization. Interfacial polymerization wasperformed on the lumen side of hollow fibers. The reactantsolutions were pumped into the lumen side by a peristalticpump with a flow rate of 3.5 mL/min from the bottom to topof the module held in a vertical hmmand 0.1 wt % sodiumdodecyl sulfate (SDS) was fed to the lumen side of hollowfibers for 5 min.After that, the excess MPD residual solutions/droplets were

removed by purging a sweeping air for 5 min using acompressed air gun. Then a 0.15 wt % TMC-solution inhexane was brought into contact with the MPD saturated onthe membrane’s inner surface, leading to form a thin film ofpolyamide (PA) as a selective inner layer of hollow fibers.Subsequently, a purged sweeping air was applied for 1 min withan attempt to remove the residual solvents/droplets after the IPreaction. The resultant TFC membranes were heat-cured at 65°C in an oven for 15 min and subsequently rinsed with DIwater several times and stored in DI water before furthercharacterizations.

Characterizations of Sulfonated PPSU Hollow FiberSubstrates and TFC-FO Membranes. The fabricated hollowfiber membrane substrate was first tested to measure its purewater permeability (PWP) (in L/m2·bar.hr) by an ultrafiltrationmembrane permeation setup. Subsequently, the membrane wassubjected to neutral solutes of progressively higher molecularweights (polyethylene glycol (PEG) or polyethylene oxide(PEO)) at 200 ppm to estimate its pore size, pore sizedistribution and molecular weight cutoff (MWCO) based onthe solute transport method as elaborated more in the SI. Themass transport characteristics: water permeability, A, saltrejection R, and salt permeability of the TFC-FO hollowfiber membranes were also evaluated using the RO mode.

Pore Size and Pore Size Distribution. The fabricated hollowfiber membrane substrate was first tested to measure its purewater permeability (PWP) (in L/m2·bar.hr) by an ultrafiltrationmembrane permeation setup. Subsequently, the membrane wassubjected to neutral solutes of progressively higher molecularweights (polyethylene glycol (PEG) or polyethylene oxide(PEO)) separation tests by flowing them through themembrane’s top surface under a pressure of 1 bar on thelumen side. Between runs of different solutes, the membranewas flushed thoroughly with DI water. The concentrations ofthe neutral solutes were measured by a total organic carbonanalyzer (TOC ASI-5000A, Shimadzu, Japan). The measuredfeed (Cf) and permeate (Cp) concentrations were used for thecalculation of the effective solute rejection coefficient R (%):

= − ×⎛⎝⎜

⎞⎠⎟R

C

C1 100%p

f (1)

In this work, solutions containing 200 ppm of differentmolecular weights of PEG or PEO were used as the neutralsolutes for the characterizations of membrane pore size and


dx.doi.org/10.1021/es4013273 | Environ. Sci. Technol. 2013, 47, 7430−74367431

pore size distribution. The relationship between Stokes radius(rs, nm) and molecular weight (Mw, gmol−1) of these neutralsolutes can be expressed as follows:

= × ×−r Mfor PEG: 16.73 10 12 0.557 (2)

= × ×−r Mfor PEO: 10.44 10 12 0.587 (3)

From eqs 2 and 3, the radius (r) of a hypothetical solute at agiven Mw can be calculated. The mean effective pore size andthe pore size distribution were then obtained according to thetraditional solute transport approach by ignoring influences ofthe steric and hydrodynamic interaction between solute andmembrane pores, the mean effective pore radius (μp) and thegeometric standard deviation (σp) can be assumed to be thesame as μs (the geometric mean radius of solute at R = 50%)and σg (the geometric standard deviation defined as the ratio ofthe rs at R = 84.13% over that at R = 50%). Therefore, based onμp and σp, the pore size distribution of a membrane can beexpressed as the following probability density function:

σ π

μ

σ= −

−⎡⎣⎢⎢

⎤⎦⎥⎥

R d

d d

dd ( )

d1

ln 2exp

(ln ln )

2(ln )p

p p p

p p2

p2

(4)

Mass Transport Characteristics of TFC-FO Membranes.The water permeability and salt permeability of TFC-FOmembranes were determined by testing the membranes underthe RO mode. The water permeability coefficient (A) wasobtained from the pure water permeation flux under the appliedtrans-membrane pressure of 1 bar. The salt rejection (Rs) wasdetermined from the measured conductivities of permeate andfeed by using feedwater containing 1000 ppm NaCl at 1 bar.The salt permeability coefficient (B), which is the intrinsicproperty of a membrane, was determined based on thesolution-diffusion theory:27

π−

=Δ − Δ

RR

BA P

1( )

s

s (5)

Forward Osmosis Performance Tests. The forwardosmosis experiments were carried out in a lab-scale cross-flowfiltration unit. The feed and draw solutions were kept at roomtemperature of about 23 °C and fed cocurrently into themodule. The flow rate at the lumen side was 0.1 L/min whilethat at the shell side was 0.2 L/min. Each membrane was testedunder the PRO mode (active layer facing draw solution) andFO mode (active layer facing feed solution). The change in

Figure 1. Morphology of membrane substrates fabricated from (a) PPSU; (b) 1.5 mol % sPPSU; and (c) 2.5 mol % sPPSU.



feed solution weight was monitored by a computer connectedto a balance (EK-4100i, A&D Company Ltd., Japan).The water flux (Lm2−h−1, abbreviated as LMH) is calculated

as follows:

= ΔΔ

JV

A t.v (6)

Where ΔV (L) is the permeation water collected over apredetermined time Δt (h) in the FO process duration; A is theeffective membrane surface area (m2).The salt concentration in the feedwater was determined from

the conductivity measurement using a calibration curve for thesingle salt solution. The salt leakage or salt reverse-diffusionfrom the draw solution to the feed, Js in g m−2 h−1 (abbreviatedas gMH), is thereafter determined from the increase of the feedconductivity:

=Δ

·ΔJ

V C VA t

( )s

t t(7)

where Ct and Vt are the salt concentration and the volume ofthe feed at the end of FO tests, respectively.The water flux in FO processes can be modeled by the

following equations28

For the PRO mode:

ππ

=− +

+J

K

A J B

A B1

ln D

Fw

m

,m w

,b (8)

For the FO mode:

ππ

=+

+ +J

K

A B

A J B1

ln D

Fw

m

,b

,m w (9)

Where πD,b and πF,b refer to the osmotic pressures in therespective bulk draw and feed solutions, respectively, while πD,mand πF,m are the corresponding osmotic pressures on themembrane surfaces facing the draw and feed solutions,respectively. The relationship among solute diffusion resistivityKm, within the porous layer, diffusivity Ds, membrane structuralparameter S, membrane tortuosity τ, membrane thickness l, andmembrane porosity ε can be represented as follows:

τε

= =KSD

lDm

s s (10)

The membrane structural parameter S is an intrinsicmembrane property indicative of the degree of internalconcentration polarization and a decrease in S positivelyinfluences the FO performance of membranes.

RESULTS AND DISCUSSION

Phase Separation Properties. It can be observed inFigure S1 of SI that as the degree of sulfonation of the PPSUpolymer increases, a higher amount of water is needed toinduce phase separation. In other words, it can be inferred thatan increase in the degree of sulfonation in the dope systemresults in a slower phase inversion rate. Though the increase insulfonation degree requires a larger amount of water to lead tophase separation, this increment in water addition is relativelysmall.

Characteristics and Performance of Membrane Sub-strates before Interfacial Polymerization. It is known thatthe degree of sulfonation of polymers used for membranefabrication can influence the morphology, pore size, and poresize distribution as well as water permeability. This influence onmorphology is well-exemplified as observed in Figure 1.The hollow fibers fabricated from nonsulfonated polypheny-

lenesulfone display numerous finger-like macrovoids through-out the entire wall thickness due to instantaneous demixing.Strathmann29 found a close correlation between membranestructure and precipitation rate. In general, systems with fastprecipitation rates tend to form finger-like structure, whereassystems with slow precipitation rates result in sponge-likestructure. On the other hand, the membrane substratefabricated from 2.5 mol % sPPSU exhibits a fully sponge-likestructure with no macrovoids and this can be attributed to thedelayed demixing phenomenon induced by the sulfonatedmaterial. In summary, the degree of sulfonation has a directimpact on the amount of macrovoids formed, following theorder of PPSU > 1.5 mol % sPPSU > 2.5 mol % sPPSU. Theoccurrence of macrovoids can also be linked to the viscosity ofthe dope solutions. The PPSU dope, with the lowest viscosityat 3−4 folds smaller than the sPPSU dopes as listed in Table S3

Table 1. Summary of Mean Effective Pore Size (μp), PWP, MWCO, and Contact Angles of TFC Membrane Substrates

membrane ID μp (nm) σp MWCO (Da) PWP (LMHbar−1) contact angle (°)

PPSU 9.17 ± 0.52 1.52 ± 0.09 62319 ± 3205 150.1 ± 20.2 90.0 ± 4.71.5 mol % sPPSU 6.31 ± 0.35 1.67 ± 0.08 39731 ± 1753 213.4 ± 9.6 69.3 ± 1.82.5 mol % sPPSU 9.11 ± 0.41 1.65 ± 0.11 73877 ± 5139 238.7 ± 8.1 63.4 ± 4.2

Figure 2. Morphology of 2.5% mol sPPSU TFC membranes after interfacial polymerization (a) inner surface; (b) edge of TFC layer; (c) outersurface.



of SI, facilities a higher degree of nonsolvent intrusion duringphase inversion and tends to form the most macrovoids. Inaddition, it can be observed that the reduction in phaseinversion kinetics induced by the sulfonation leads to anincrease hollow fiber wall thickness. Comparing PPSU andsPPSU hollow fibers, the sPPSU fibers have thicker walls aslonger durations for phase inversion are allowed for the nascentmembranes to adjust their thicknesses and more NMP solventcan flow out before the membrane contour is fixed.30 Thesurface on the lumen side of hollow fibers exhibit relativelysmaller pores as compared to the surface on the shell sidewhich possesses a highly porous morphology with numerouslarge pores.Table 1 and Figure S2 of SI show the PWP and pore size

characteristics of the TFC membrane substrates fabricated frompolymers of different degrees of sulfonation. As observed,increasing the degree of sulfonation of the membrane substrateincreases the PWP. This is consistent with the values of contactangles which verify the increase in hydrophilicity along withsulfonation degree. It has been shown that better hydrophilicityfacilities water transport.31,32 Comparing 1.5 mol % and 2.5 mol% sPPSU membrane substrates, the latter has larger pores of9.11 nm as compared to 6.31 nm of the former. At identicalspinning conditions, the 2.5 mol % sPPSU polymer givesmembrane substrates with a relatively larger pore size due tothe delayed mixing phenomenon as discussed previously.Characteristics and Performance of TFC-FO Hollow

Fiber Membranes. Figure 2 depicts the typical polyamidelayer of the TFC-FO hollow fiber membranes after interfacialpolymerization was performed on the membrane substrates. Auniform layer of about 300 nm comprising ridges and valleyscovers the inner selective side of the hollow fiber.The fabricated TFC-FO membranes were evaluated for their

performance under the PRO and FO mode using DI water asthe feed and 0.5 M NaCl as the draw solution. The results aresummarized in Table 2. The TFC-FO membranes fabricatedfrom sPPSU substrates show relatively higher water fluxes thanthat fabricated from PPSU substrate, with a slight increase insalt reverse flux displayed by TFC 1.5 mol % sPPSU. Thisincrease in water flux can be attributed to the increase inhydrophilicity of the substrate which aids in water transportthrough the substrate.Among the results, according to the Js/Jw values, the TFC 2.5

mol % sPPSU membrane exhibits the best performance withwater fluxes of 37.7 and 18.0 LMH with relatively low reversesalt fluxes of 7.0 and 2.6 gMH for PRO and FO modes,

respectively. Referring back to Table 1 as discussed previously,the PPSU substrate possesses almost similar pore size andMWCO as the 2.5 mol % sPPSU substrate. Therefore, whilemaintaining similar substrate characteristics, the effect ofsulfonation can be more clearly demonstrated. ComparingTFC 2.5 mol % sPPSU with PPSU membranes, the water fluxincreases by more than 66%, whereas the reverse salt fluxremains almost constant. It is evident that sulfonation plays animportant role in enhancing water fluxes of TFC-FOmembranes. The TFC membrane synthesized on the 1.5 mol% sPPSU substrate displays a higher water flux than that on 2.5mol % sPPSU. This may be attributed to the higher degree ofswelling of the 2.5 mol % sPPSU substrate, hence leading to alower water flux. The degree of swelling for PPSU, 1.5 mol %sPPSU and 2.5 mol % sPPSU are 254.1 ± 8.3%, 269.5 ± 10.4%,and 298.5 ± 4.5%, respectively.The water and salt permeabilites obtained from the

pressurized tests in Table 3 are in good correlation with theFO data in Table 2. Table 3 summarizes the basic transportproperties of the three various TFC-FO hollow fibermembranes. The water permeability of the TFC PPSU iscomparatively higher that than of the TFC sPPSU hollow fibermembranes due to the presence of macrovoids and shows arelatively lower salt rejection rate. The values for both thesulfonated TFC membranes are relatively similar. Thecalculated structural parameter (S) significantly decreaseswhen a sulfonated substrate replaces a nonsulfonated substrate.This implies that a lower internal concentration polarizationeffect can be achieved when sulfonated materials are used forthe membrane substrates. This diminishing effect of internalconcentration polarization is evident from the increase in waterfluxes in both the PRO and FO modes. In short, a decrease instructural parameter (S) has been substantiated to contribute toa higher water flux from the comparison between the TFCPPSU and TFC 1.5 mol % and 2.5 mol % sPPSUs.The TFC-FO membranes fabricated on the 1.5 mol %

sPPSU substrate possesses the lowest structural parameter (S)and hence was chosen for further evaluation of the effect ofincreasing draw solution concentration on water flux. As seen inSI Figure S3, the increase in draw solution concentration from0.5 to 2.0 M NaCl leads to an increase in water flux in both thePRO and FO modes. This is attributed to the increase ineffective osmotic pressure difference. Another interestingobservation is that the water flux in the PRO mode increasessteadily with the increase in draw solution concentration whilethat of the FO mode increases at a much slower rate. This

Table 2. PRO and FO Performance of TFC-FO Hollow Fiber Membranes using a 0.5M NaCl Draw Solution

PRO mode FO mode

membrane ID water flux (LMH) salt reverse flux (gMH) Js/Jw (g/L) water flux (LMH) salt reverse flux (gMH) Js/Jw (g/L)

TFC PPSU 22.64 ± 2.5 7.73 ± 0.51 0.34 12.37 ± 1.2 2.69 ± 0.21 0.22TFC 1.5 mol % sPPSU 49.39 ± 6.2 11.00 ± 1.36 0.22 22.51 ± 2.3 5.49 ± 0.35 0.24TFC 2.5 mol % sPPSU 37.71 ± 1.76 6.98 ± 0.71 0.19 17.98 ± 0.17 2.63 ± 0.32 0.15

Table 3. Transport Properties and Structural Parameters of TFC-FO Hollow Fiber Membranes

membrane ID water permeability, A (Lm−2h−1bar−1) salt rejectiona (%) salt permeability, B (Lm−2h−1) Km (sm−1) Sb (m)

TFC PPSU 3.15 ± 0.07 86.8 ± 0.7 0.0952 ± 0.003 5.07 × 105 7.46 × 10−4

TFC 1.5 mol % sPPSU 1.99 ± 0.02 90.9 ± 0.3 0.0399 ± 0.002 1.11 × 105 1.63 × 10−4

TFC 2.5 mol % sPPSU 1.80 ± 0.11 87.9 ± 0.9 0.0490 ± 0.011 1.63 × 105 2.40 × 10−4

aTested at 1 bar with 1000 ppm NaCl solution. bStructural parameters were calculated based on experiments under the FO mode using 0.5 M NaClas draw solution and DI water as feed.



phenomenon is indicative of the onset of ICP within the poroussupport layer which is more severe in the case of FO mode.Osmotic Seawater Desalination. The newly developed

TFC 1.5 mol % sPPSU FO hollow fiber membranes werechosen for characterization for its potential for seawaterdesalination. 3.5 wt % NaCl was used as a model for seawatersolution to test the membrane performance, along with 2.0 MNaCl solution as the draw solution under the PRO mode.Table 4 summarizes the water flux as a function of module

length. Interestingly, the water flux increases with a decrease in

module length. This is due to the effects of ICP and externalconcentration polarization (ECP) on the effective osmoticdriving force across the membrane in the hollow fiberconfiguration. In other words, the 2.0 M NaCl draw solutionis quickly diluted inside the membrane due to high waterinflow, while the seawater concentration near the TFC layer isquickly built up. As a result, the osmotic driving force across theTFC layer is severely reduced. However, both ICP and ECPeffects can be effectively mitigated by reducing the fiber length.The TFC 1.5 mol % sPPSU with a short fiber length exhibitswater production of 22 LMH which is among the best inliterature to date. In addition, a quick comparison tocommercial HTI membranes33 and other data found fromliterature as shown in Figure 3 shows that the TFC-FO hollowfiber membranes fabricated in this work surpasses that of HTI’s

and many others. Clearly, this work establishes the potentialprospect of using sulfonated polyphenylenesulfone as a newmaterial for substrates for TFC-FO hollow fibers fabricatedfrom interfacial polymerization and provides useful insights inapplying such membranes for real industrial applications.

ASSOCIATED CONTENT*S Supporting InformationMaterials; procedures for hollow fiber spinning, spinningparameters, membrane post-treatment and module fabrication;characterization techniques such as FESEM, viscosity. Thismaterial is available free of charge via the Internet at http://pubs.acs.org.

AUTHOR INFORMATIONCorresponding Author*Phone: (65) 6516 6645; fax: (65) 6779 1936; e-mail:[email protected].


ACKNOWLEDGMENTSWe thank BASF SE, Germany for funding this work with agrant number of R-279-000-363-597. Thanks are due to Ms. LiXue, Dr. Panu Sukitpaneenit and their help and suggestions tothis work. This research was also partially funded by theSingapore National Research Foundation under its Competitiveresearch Program for the project entitled, “Advanced FOMembranes and Membrane Systems for Wastewater Treat-ment, Water Reuse and Seawater Desalination” (grant number:R-279-000-336-281).

REFERENCES(1) Shannon, M. A.; Bohn, P. W.; Elimelech, M.; Georgiadis, J. G.;Marinas, B. J.; Mayes, A. M. Science and technology for waterpurification in the coming decades. Nature 2008, 452 (7185), 301−310.(2) Semiat, R. Energy issues in desalination processes. Environ. Sci.Technol. 2008, 42 (22), 8193−8201.(3) Zhao, S.; Zou, L.; Tang, C. Y.; Mulcahy, D. Recent developmentsin forward osmosis: Opportunities and challenges. J. Membr. Sci. 2012,396 (0), 1−21.(4) Zhao, S.; Zou, L.; Mulcahy, D. Effects of membrane orientationon process performance in forward osmosis applications. J. Membr. Sci.2011, 382 (1−2), 308−315.(5) Lee, K. L.; Baker, R. W.; Lonsdale, H. K. Membranes for powergeneration by pressure-retarded osmosis. J. Membr. Sci. 1981, 8 (2),141−171.(6) Achilli, A.; Cath, T. Y.; Childress, A. E. Power generation withpressure retarded osmosis: An experimental and theoretical inves-tigation. J. Membr. Sci. 2009, 343 (1−2), 42−52.(7) Garcia-Castello, E. M.; McCutcheon, J. R.; Elimelech, M.Performance evaluation of sucrose concentration using forwardosmosis. J. Membr. Sci. 2009, 338 (1−2), 61−66.(8) Jiao, B.; Cassano, A.; Drioli, E. Recent advances on membraneprocesses for the concentration of fruit juices: A review. J. Food Eng.2004, 63 (3), 303−324.(9) Yang, Q.; Wang, K. Y.; Chung, T. S. A novel dual-layer forwardosmosis membrane for protein enrichment and concentration. Sep.Purif. Technol. 2009, 69 (3), 269−274.(10) Chung, T. S.; Li, X.; Ong, R. C.; Ge, Q.; Wang, H.; Han, G.Emerging forward osmosis (FO) technologies and challenges aheadfor clean water and clean energy applications. Curr. Opin. Chem. Eng.2012, 1 (3), 246−257.

Table 4. TFC Hollow Fiber Membranes Synthesized on 1.5mol % sPPSU for Seawater Desalinationa

membrane ID fiber length (cm) flux (LMH)

1 13.5 14.01 ± 0.052 10.3 16.23 ± 0.123 8.5 22.13 ± 1.21

aUnder the PRO mode using a model seawater of 3.5 wt % NaCl and a2.0 M NaCl solution as the draw solution.

Figure 3. A comparison of water fluxes reported in literature for FOseawater desalination applications. Model seawater (3.5 wt % NaCl)solution was used as the feed solution, whereas 2.0 M NaCl as thedraw solution. Commercial HTI,33,34 cellulose acetate propionate,16

TFC hollow fiber (macrovoids),18 TFC PES/sPsf,15 TFC sPsf,21 TFCPES (macrovoid-free).20.



http://pubs.acs.org

http://pubs.acs.org


(11) Li, Z. Y.; Yangali-Quintanilla, V.; Valladares-Linares, R.; Li, Q.;Zhan, T.; Amy, G. Flux patterns and membrane fouling propensityduring desalination of seawater by forward osmosis. Water Res. 2012,46 (1), 195−204.(12) Mi, B.; Elimelech, M. Organic fouling of forward osmosismembranes: Fouling reversibility and cleaning without chemicalreagents. J. Membr. Sci. 2010, 348 (1−2), 337−345.(13) Veríssimo, S.; Peinemann, K. V.; Bordado, J. Thin-filmcomposite hollow fiber membranes: An optimized manufacturingmethod. J. Membr. Sci. 2005, 264 (1−2), 48−55.(14) Shintani, T.; Matsuyama, H.; Kurata, N. Development of achlorine-resistant polyamide reverse osmosis membrane. Desalination2007, 207 (1−3), 340−348.(15) Wang, K. Y.; Chung, T. S.; Amy, G. Developing thin-film-composite forward osmosis membranes on the PES/SPSf substratethrough interfacial polymerization. AIChE J. 2012, 58 (3), 770−781.(16) Li, X.; Wang, K. Y.; Helmer, B.; Chung, T. S. Thin-filmcomposite membranes and formation mechanism of thin-film layers onhydrophilic cellulose acetate propionate substrates for forward osmosisprocesses. Ind. Eng. Chem. Res. 2012, 51 (30), 10039−10050.(17) Yip, N. Y.; Tiraferri, A.; Phillip, W. A.; Schiffman, J. D.;Elimelech, M. High performance thin-film composite forward osmosismembrane. Environ. Sci. Technol. 2010, 44 (10), 3812−3818.(18) Chou, S.; Shi, L.; Wang, R.; Tang, C. Y.; Qiu, C.; Fane, A. G.Characteristics and potential applications of a novel forward osmosishollow fiber membrane. Desalination 2010, 261 (3), 365−372.(19) Wang, R.; Shi, L.; Tang, C. Y.; Chou, S.; Qiu, C.; Fane, A. G.Characterization of novel forward osmosis hollow fiber membranes. J.Membr. Sci. 2010, 355 (1−2), 158−167.(20) Sukitpaneenit, P.; Chung, T. S. High Performance thin-filmcomposite forward osmosis hollow fiber membranes with macrovoid-free and highly porous structure for sustainable water production.Environ. Sci. Technol. 2012, 46 (13), 7358−7365.(21) Widjojo, N.; Chung, T. S.; Weber, M.; Maletzko, C.; Warzelhan,V. The role of sulphonated polymer and macrovoid-free structure inthe support layer for thin-film composite (TFC) forward osmosis(FO) membranes. J. Membr. Sci. 2011, 383 (1−2), 214−223.(22) Shi, L.; Chou, S. R.; Wang, R.; Fang, W. X.; Tang, C. Y.; Fane,A. G. Effect of substrate structure on the performance of thin-filmcomposite forward osmosis hollow fiber membranes. J. Membr. Sci.2011, 382 (1−2), 116−123.(23) Zhang, G.; Song, X.; Ji, S.; Wang, N.; Liu, Z. Self-assembly ofinner skin hollow fiber polyelectrolyte multilayer membranes by adynamic negative pressure layer-by-layer technique. J. Membr. Sci.2008, 325 (1), 109−116.(24) Geise, G. M.; Lee, H. S.; Miller, D. J.; Freeman, B. D.; McGrath,J. E.; Paul, D. R. Water purification by membrane: The role of polymerscience. J. Polym. Sci., Part B: Polym. Phys. 2010, 48 (15), 1685.(25) Doi, S.; Hamanaka, K. Pore size control technique in thespinning of polysulfone hollow fiber ultrafiltration membranes.Desalination 1991, 80 (2−3), 167−180.(26) Liu, Y.; Koops, G. H.; Strathmann, H. Characterization ofmorphology controlled polyethersulfone hollow fiber membranes bythe addition of polyethylene glycol to the dope and bore liquidsolution. J. Membr. Sci. 2003, 223 (1−2), 187−199.(27) Loeb, S.; Mehta, G. D. A two-coefficient water transportequation for pressure-retarded osmosis. J. Membr. Sci. 1978, 4 (0),351−362.(28) Mehta, G. D.; Loeb, S. Internal polarization in the poroussubstructure of a semipermeable membrane under pressure-retardedosmosis. J. Membr. Sci. 1978, 4 (0), 261−265.(29) Strathmann, H., Production of microporous media by phaseinversion processes. In Materials Science of Synthetic Membranes;American Chemical Society: Washington, DC, 1985; Vol. 269, pp165−195.(30) Xing, D. Y.; Peng, N.; Chung, T. S. Formation of celluloseacetate membranes via phase inversion using ionic liquid, [BMIM]-SCN, as the solvent. Ind. Eng. Chem. Res. 2010, 49 (18), 8761−8769.

(31) Li, Y.; Chung, T. S.; Chan, S. Y. High-affinity sulfonatedmaterials with transition metal counterions for enhanced proteinseparation in dual-layer hollow fiber membrane chromatography. J.Chromatogr., A 2008, 1187 (1−2), 285−288.(32) Rahimpour, A.; Madaeni, S. S.; Ghorbani, S.; Shockravi, A.;Mansourpanah, Y. The influence of sulfonated polyethersulfone(SPES) on surface nano-morphology and performance of poly-ethersulfone (PES) membrane. Appl. Surf. Sci. 2010, 256 (6), 1825−1831.(33) McCutcheon, J. R.; McGinnis, R. L.; Elimelech, M. Desalinationby ammonia−carbon dioxide forward osmosis: Influence of draw andfeed solution concentrations on process performance. J. Membr. Sci.2006, 278 (1−2), 114−123.(34) Phillip, W. A.; Yong, J. S.; Elimelech, M. Reverse draw solutepermeation in forward osmosis: Modeling and experiments. Environ.Sci. Technol. 2010, 44 (13), 5170−5176.



Development of Thin-Film Composite forward Osmosis Hollow FiberMembranes Using Direct Sulfonated Polyphenylenesulfone (sPPSU)as Membrane SubstratesPeishan Zhong,†,‡ Xiuzhu Fu,‡ Tai-Shung Chung,‡,* Martin Weber,§ and Christian Maletzko∥

†Functionalized Materials and Nanostructures, Global Research Center Singapore (A-GMM/F), BASF South East Asia Pte Ltd 61Science Park Road, Singapore 117525‡Department of Chemical & Biomolecular Engineering, National University of Singapore, 10 Kent Ridge Crescent, Singapore 119260§Advanced Materials & Systems Research, Performance Materials, BASF SE GMV/W-B1, 67056 Ludwigshafen, Germany∥Performance Materials, BASF SE G-PMS/PS-F206, 67056 Ludwigshafen, Germany


ABSTRACT: This study investigates a new approach to fabricate thin-filmcomposite (TFC) hollow fiber membranes via interfacial polymerization forforward osmosis (FO) applications. Different degrees of sulfonation ofpolyphenylenesulfone (PPSU) were adopted as membrane substrates toinvestigate their impact on water flux. It has been established that the degreeof sulfonation plays a role in both creating a macrovoid-free structure andinducing hydrophilicity to bring about higher water fluxes. The fabricatedmembranes exhibit extremely high water fluxes of 30.6 and 82.0 LMHagainst a pure water feed using 2.0 M NaCl as the draw solution testedunder FO and pressure retarded osmosis (PRO) modes, respectively, whilemaintaining low salt reverse fluxes below 12.7 gMH. The structuralparameter (S) displays remarkable decreases of up to 4.5 times as themembrane substrate is switched from a nonsulfonated to sulfonated one. Inaddition, the newly developed TFC-FO membranes containing 1.5 mol %sPPSU in the substrate achieves a water flux of 22 LMH in seawater desalination using a 3.5 wt % NaCl model solution and 2.0M NaCl as the draw solution under the PRO mode. To the best of our knowledge, this value is the highest ever reported forseawater desalination using flat and hollow fiber FO membranes. The use of sulfonated materials in the FO process opens up afrontier for sustainable and efficient production of potable water.

INTRODUCTION

Water scarcity in the 21st century is a pressing global issue1,2

and forward osmosis has attracted much attention as anemerging technology3,4 to alleviate this problem. It also bringsabout potential applications including power generation,5,6

wastewater treatment, food processing,7,8 and protein enrich-ment.9 There are many advantages of the forward osmosisprocess which include requiring much less energy to induce anet flow of water in comparison with the traditional watertreatment process such as reverse osmosis (RO) and lowmembrane fouling.10−12 The use of an osmotic pressuregradient also helps reduce the use of depleting fuel sourceswhich has brought about detrimental environmental problemsto our earth.The current dominant FO membrane in the commercial

application comes from Hydration Technology Inc. (HTI).However, these membranes are cellulose-based and faceproblems such as degradation by microorganisms, narrow pHworking range and low salt rejection. In contrast, the polyamidethin film composite (TFC) membrane fabricated frominterfacial polymerization upon a suitable substrate is an

attractive technique to mitigate the mentioned problems. Thistechnique has been widely applied in the reverse osmosis (RO)process13,14 and its concept has been extended to the FOprocess. Many studies have worked on the flat sheet15−17

configuration while few studies have investigated TFCmembranes in the hollow fiber configuration.18−20

The key prerequisites for the development of desirable TFCmembranes for FO include (1) ultrathin semipermeable activelayer with low water transport resistance and high salt rejection;(2) highly porous and thin supporting layer to minimizeinternal concentration polarization (ICP); (3) low foulingpropensity and (4) sufficiently high mechanical strength.10

These fundamental requirements have to be met in order tomeet and achieve the attractive qualities, offered by the forwardosmosis process. However, to date, the development of forwardosmosis encounters the challenge of ineffective membranes that

Received: March 26, 2013Revised: May 31, 2013Accepted: June 3, 2013Published: June 3, 2013

Article

pubs.acs.org/est

© 2013 American Chemical Society 7430 dx.doi.org/10.1021/es4013273 | Environ. Sci. Technol. 2013, 47, 7430−7436

pubs.acs.org/est

play the critical role in most FO processes.11 Hence, there is aneed to search for suitable materials. These materials arepreferred to be hydrophilic.21

In this work, direct sulfonated polyphenylenesulfone(sPPSU) with different concentrations of sulfonated monomer(sDCDPS), that is, 1.5 and 2.5 mol %, were applied assubstrates for FO membranes. The objectives of this study is to(1) investigate the effect of sulfonation degree on themorphology of hollow fiber substrates spun from differentspinning parameters; (2) explore the effect of sulfonationdegree of the TFC membrane substrate on FO performance;and (3) explore their potential for seawater desalination. It hasbeen postulated that an increase in hydrophilicity of membranesubstrates would increase the FO water flux of thesemembranes and provide a solution to the current issues facedin the FO process.A study by Shi et al has investigated the effect of substrate

structure on the formation of the TFC layer upon it andconcluded that the substrate should possess an MWCO of lessthan 300 kDa in order to achieve a desirable semipermeableskin.22 Therefore, the MWCO of the substrates fabricated inthis work will be controlled within the desirable range. Theconcentration of the PEG400 which acts as a pore former wasvaried in this work and the optimal condition was used.In this work, hollow fibers with an inner selective layer were

designed due to advantages which include (1) feed being ableto be delivered into the lumen equally, (2) easy protection ofthe defect-free separation layer, and (3) low transportresistance in the permeate side.23 The formulation of thedope to yield a macrovoid-free structure was also explored asmacrovoids are viewed as mechanically weaker points on themembrane. The effect of PEG400 and the addition of water tothe dope solution for macrovoid suppression were studiedalong with the spinning parameters.

MATERIALS AND METHODSMaterials. Polyphenylenesulfone (PPSU), 1.5 mol % and

2.5 mol % sulfonated polyphenylenesulfone (sPPSU) were usedas the materials for the fabrication of various membranesubstrates. The sPPSU materials were synthesized via the directsulfonation synthesis route developed by McGrath et al.24 The1.5 mol % and 2.5 mol % sPPSU have IEC values of 5.1 and 8.2meq/100g polymer. N-methyl-2-pyrrolidone (NMP) andpolyethylene glycol 400 (PEG400) from Merck were used asthe solvent and pore former in the dope solution preparation,respectively. More information on the chemicals used forinterfacial polymerization can be found in the SupportingInformation (SI).Cloud Point Measurements. To determine the binodal

composition of the PPSU/PEG400/NMP/water system, cloudpoint data were obtained by means of a titration method. It isknown that a dope formulation near to the binodal curve willresult in membranes with macrovoid-free structures upon phaseinversion.25,26 Distilled water was slowly added to the system ofpolymer, additive, and solvent. It was observed that the additionof water caused some local coagulation of the dope which wasleft to stir till it became homogeneous again. This procedurewas repeated till the dope became permanently turbid. 93 wt %of the amount of water required to cause phase separation wasadded to each dope (coagulation value).Hollow Fiber Spinning of Hollow Fiber Membrane

Substrates. The hollow fiber membrane substrates wereprepared via a dry-jet wet spinning process employing an

advanced coextrusion technology using a dual-layer spinneret.Polymer solutions were prepared according to the concen-trations as described in Table S1 of Supporting Information(SI). The spinning parameters as well as the membrane post-treatment and module fabrication are described in the SI.

Interfacial Polymerization for Fabrication of TFC-FOHollow Fiber Membranes. The M-phenylenediamine(MPD) solution in aqueous phase and trimesoyl chloride(TMC) solution in the organic phase were brought into contactleading to the formation of a polyamide layer throughinterfacial polymerization. Interfacial polymerization wasperformed on the lumen side of hollow fibers. The reactantsolutions were pumped into the lumen side by a peristalticpump with a flow rate of 3.5 mL/min from the bottom to topof the module held in a vertical hmmand 0.1 wt % sodiumdodecyl sulfate (SDS) was fed to the lumen side of hollowfibers for 5 min.After that, the excess MPD residual solutions/droplets were

removed by purging a sweeping air for 5 min using acompressed air gun. Then a 0.15 wt % TMC-solution inhexane was brought into contact with the MPD saturated onthe membrane’s inner surface, leading to form a thin film ofpolyamide (PA) as a selective inner layer of hollow fibers.Subsequently, a purged sweeping air was applied for 1 min withan attempt to remove the residual solvents/droplets after the IPreaction. The resultant TFC membranes were heat-cured at 65°C in an oven for 15 min and subsequently rinsed with DIwater several times and stored in DI water before furthercharacterizations.

Characterizations of Sulfonated PPSU Hollow FiberSubstrates and TFC-FO Membranes. The fabricated hollowfiber membrane substrate was first tested to measure its purewater permeability (PWP) (in L/m2·bar.hr) by an ultrafiltrationmembrane permeation setup. Subsequently, the membrane wassubjected to neutral solutes of progressively higher molecularweights (polyethylene glycol (PEG) or polyethylene oxide(PEO)) at 200 ppm to estimate its pore size, pore sizedistribution and molecular weight cutoff (MWCO) based onthe solute transport method as elaborated more in the SI. Themass transport characteristics: water permeability, A, saltrejection R, and salt permeability of the TFC-FO hollowfiber membranes were also evaluated using the RO mode.

Pore Size and Pore Size Distribution. The fabricated hollowfiber membrane substrate was first tested to measure its purewater permeability (PWP) (in L/m2·bar.hr) by an ultrafiltrationmembrane permeation setup. Subsequently, the membrane wassubjected to neutral solutes of progressively higher molecularweights (polyethylene glycol (PEG) or polyethylene oxide(PEO)) separation tests by flowing them through themembrane’s top surface under a pressure of 1 bar on thelumen side. Between runs of different solutes, the membranewas flushed thoroughly with DI water. The concentrations ofthe neutral solutes were measured by a total organic carbonanalyzer (TOC ASI-5000A, Shimadzu, Japan). The measuredfeed (Cf) and permeate (Cp) concentrations were used for thecalculation of the effective solute rejection coefficient R (%):

= − ×⎛⎝⎜

⎞⎠⎟R

C

C1 100%p

f (1)

In this work, solutions containing 200 ppm of differentmolecular weights of PEG or PEO were used as the neutralsolutes for the characterizations of membrane pore size and



pore size distribution. The relationship between Stokes radius(rs, nm) and molecular weight (Mw, gmol−1) of these neutralsolutes can be expressed as follows:

= × ×−r Mfor PEG: 16.73 10 12 0.557 (2)

= × ×−r Mfor PEO: 10.44 10 12 0.587 (3)

From eqs 2 and 3, the radius (r) of a hypothetical solute at agiven Mw can be calculated. The mean effective pore size andthe pore size distribution were then obtained according to thetraditional solute transport approach by ignoring influences ofthe steric and hydrodynamic interaction between solute andmembrane pores, the mean effective pore radius (μp) and thegeometric standard deviation (σp) can be assumed to be thesame as μs (the geometric mean radius of solute at R = 50%)and σg (the geometric standard deviation defined as the ratio ofthe rs at R = 84.13% over that at R = 50%). Therefore, based onμp and σp, the pore size distribution of a membrane can beexpressed as the following probability density function:

σ π

μ

σ= −

−⎡⎣⎢⎢

⎤⎦⎥⎥

R d

d d

dd ( )

d1

ln 2exp

(ln ln )

2(ln )p

p p p

p p2

p2

(4)

Mass Transport Characteristics of TFC-FO Membranes.The water permeability and salt permeability of TFC-FOmembranes were determined by testing the membranes underthe RO mode. The water permeability coefficient (A) wasobtained from the pure water permeation flux under the appliedtrans-membrane pressure of 1 bar. The salt rejection (Rs) wasdetermined from the measured conductivities of permeate andfeed by using feedwater containing 1000 ppm NaCl at 1 bar.The salt permeability coefficient (B), which is the intrinsicproperty of a membrane, was determined based on thesolution-diffusion theory:27

π−

=Δ − Δ

RR

BA P

1( )

s

s (5)

Forward Osmosis Performance Tests. The forwardosmosis experiments were carried out in a lab-scale cross-flowfiltration unit. The feed and draw solutions were kept at roomtemperature of about 23 °C and fed cocurrently into themodule. The flow rate at the lumen side was 0.1 L/min whilethat at the shell side was 0.2 L/min. Each membrane was testedunder the PRO mode (active layer facing draw solution) andFO mode (active layer facing feed solution). The change in

Figure 1. Morphology of membrane substrates fabricated from (a) PPSU; (b) 1.5 mol % sPPSU; and (c) 2.5 mol % sPPSU.



feed solution weight was monitored by a computer connectedto a balance (EK-4100i, A&D Company Ltd., Japan).The water flux (Lm2−h−1, abbreviated as LMH) is calculated

as follows:

= ΔΔ

JV

A t.v (6)

Where ΔV (L) is the permeation water collected over apredetermined time Δt (h) in the FO process duration; A is theeffective membrane surface area (m2).The salt concentration in the feedwater was determined from

the conductivity measurement using a calibration curve for thesingle salt solution. The salt leakage or salt reverse-diffusionfrom the draw solution to the feed, Js in g m−2 h−1 (abbreviatedas gMH), is thereafter determined from the increase of the feedconductivity:

=Δ

·ΔJ

V C VA t

( )s

t t(7)

where Ct and Vt are the salt concentration and the volume ofthe feed at the end of FO tests, respectively.The water flux in FO processes can be modeled by the

following equations28

For the PRO mode:

ππ

=− +

+J

K

A J B

A B1

ln D

Fw

m

,m w

,b (8)

For the FO mode:

ππ

=+

+ +J

K

A B

A J B1

ln D

Fw

m

,b

,m w (9)

Where πD,b and πF,b refer to the osmotic pressures in therespective bulk draw and feed solutions, respectively, while πD,mand πF,m are the corresponding osmotic pressures on themembrane surfaces facing the draw and feed solutions,respectively. The relationship among solute diffusion resistivityKm, within the porous layer, diffusivity Ds, membrane structuralparameter S, membrane tortuosity τ, membrane thickness l, andmembrane porosity ε can be represented as follows:

τε

= =KSD

lDm

s s (10)

The membrane structural parameter S is an intrinsicmembrane property indicative of the degree of internalconcentration polarization and a decrease in S positivelyinfluences the FO performance of membranes.

RESULTS AND DISCUSSION

Phase Separation Properties. It can be observed inFigure S1 of SI that as the degree of sulfonation of the PPSUpolymer increases, a higher amount of water is needed toinduce phase separation. In other words, it can be inferred thatan increase in the degree of sulfonation in the dope systemresults in a slower phase inversion rate. Though the increase insulfonation degree requires a larger amount of water to lead tophase separation, this increment in water addition is relativelysmall.

Characteristics and Performance of Membrane Sub-strates before Interfacial Polymerization. It is known thatthe degree of sulfonation of polymers used for membranefabrication can influence the morphology, pore size, and poresize distribution as well as water permeability. This influence onmorphology is well-exemplified as observed in Figure 1.The hollow fibers fabricated from nonsulfonated polypheny-

lenesulfone display numerous finger-like macrovoids through-out the entire wall thickness due to instantaneous demixing.Strathmann29 found a close correlation between membranestructure and precipitation rate. In general, systems with fastprecipitation rates tend to form finger-like structure, whereassystems with slow precipitation rates result in sponge-likestructure. On the other hand, the membrane substratefabricated from 2.5 mol % sPPSU exhibits a fully sponge-likestructure with no macrovoids and this can be attributed to thedelayed demixing phenomenon induced by the sulfonatedmaterial. In summary, the degree of sulfonation has a directimpact on the amount of macrovoids formed, following theorder of PPSU > 1.5 mol % sPPSU > 2.5 mol % sPPSU. Theoccurrence of macrovoids can also be linked to the viscosity ofthe dope solutions. The PPSU dope, with the lowest viscosityat 3−4 folds smaller than the sPPSU dopes as listed in Table S3

Table 1. Summary of Mean Effective Pore Size (μp), PWP, MWCO, and Contact Angles of TFC Membrane Substrates

membrane ID μp (nm) σp MWCO (Da) PWP (LMHbar−1) contact angle (°)

PPSU 9.17 ± 0.52 1.52 ± 0.09 62319 ± 3205 150.1 ± 20.2 90.0 ± 4.71.5 mol % sPPSU 6.31 ± 0.35 1.67 ± 0.08 39731 ± 1753 213.4 ± 9.6 69.3 ± 1.82.5 mol % sPPSU 9.11 ± 0.41 1.65 ± 0.11 73877 ± 5139 238.7 ± 8.1 63.4 ± 4.2

Figure 2. Morphology of 2.5% mol sPPSU TFC membranes after interfacial polymerization (a) inner surface; (b) edge of TFC layer; (c) outersurface.



of SI, facilities a higher degree of nonsolvent intrusion duringphase inversion and tends to form the most macrovoids. Inaddition, it can be observed that the reduction in phaseinversion kinetics induced by the sulfonation leads to anincrease hollow fiber wall thickness. Comparing PPSU andsPPSU hollow fibers, the sPPSU fibers have thicker walls aslonger durations for phase inversion are allowed for the nascentmembranes to adjust their thicknesses and more NMP solventcan flow out before the membrane contour is fixed.30 Thesurface on the lumen side of hollow fibers exhibit relativelysmaller pores as compared to the surface on the shell sidewhich possesses a highly porous morphology with numerouslarge pores.Table 1 and Figure S2 of SI show the PWP and pore size

characteristics of the TFC membrane substrates fabricated frompolymers of different degrees of sulfonation. As observed,increasing the degree of sulfonation of the membrane substrateincreases the PWP. This is consistent with the values of contactangles which verify the increase in hydrophilicity along withsulfonation degree. It has been shown that better hydrophilicityfacilities water transport.31,32 Comparing 1.5 mol % and 2.5 mol% sPPSU membrane substrates, the latter has larger pores of9.11 nm as compared to 6.31 nm of the former. At identicalspinning conditions, the 2.5 mol % sPPSU polymer givesmembrane substrates with a relatively larger pore size due tothe delayed mixing phenomenon as discussed previously.Characteristics and Performance of TFC-FO Hollow

Fiber Membranes. Figure 2 depicts the typical polyamidelayer of the TFC-FO hollow fiber membranes after interfacialpolymerization was performed on the membrane substrates. Auniform layer of about 300 nm comprising ridges and valleyscovers the inner selective side of the hollow fiber.The fabricated TFC-FO membranes were evaluated for their

performance under the PRO and FO mode using DI water asthe feed and 0.5 M NaCl as the draw solution. The results aresummarized in Table 2. The TFC-FO membranes fabricatedfrom sPPSU substrates show relatively higher water fluxes thanthat fabricated from PPSU substrate, with a slight increase insalt reverse flux displayed by TFC 1.5 mol % sPPSU. Thisincrease in water flux can be attributed to the increase inhydrophilicity of the substrate which aids in water transportthrough the substrate.Among the results, according to the Js/Jw values, the TFC 2.5

mol % sPPSU membrane exhibits the best performance withwater fluxes of 37.7 and 18.0 LMH with relatively low reversesalt fluxes of 7.0 and 2.6 gMH for PRO and FO modes,

respectively. Referring back to Table 1 as discussed previously,the PPSU substrate possesses almost similar pore size andMWCO as the 2.5 mol % sPPSU substrate. Therefore, whilemaintaining similar substrate characteristics, the effect ofsulfonation can be more clearly demonstrated. ComparingTFC 2.5 mol % sPPSU with PPSU membranes, the water fluxincreases by more than 66%, whereas the reverse salt fluxremains almost constant. It is evident that sulfonation plays animportant role in enhancing water fluxes of TFC-FOmembranes. The TFC membrane synthesized on the 1.5 mol% sPPSU substrate displays a higher water flux than that on 2.5mol % sPPSU. This may be attributed to the higher degree ofswelling of the 2.5 mol % sPPSU substrate, hence leading to alower water flux. The degree of swelling for PPSU, 1.5 mol %sPPSU and 2.5 mol % sPPSU are 254.1 ± 8.3%, 269.5 ± 10.4%,and 298.5 ± 4.5%, respectively.The water and salt permeabilites obtained from the

pressurized tests in Table 3 are in good correlation with theFO data in Table 2. Table 3 summarizes the basic transportproperties of the three various TFC-FO hollow fibermembranes. The water permeability of the TFC PPSU iscomparatively higher that than of the TFC sPPSU hollow fibermembranes due to the presence of macrovoids and shows arelatively lower salt rejection rate. The values for both thesulfonated TFC membranes are relatively similar. Thecalculated structural parameter (S) significantly decreaseswhen a sulfonated substrate replaces a nonsulfonated substrate.This implies that a lower internal concentration polarizationeffect can be achieved when sulfonated materials are used forthe membrane substrates. This diminishing effect of internalconcentration polarization is evident from the increase in waterfluxes in both the PRO and FO modes. In short, a decrease instructural parameter (S) has been substantiated to contribute toa higher water flux from the comparison between the TFCPPSU and TFC 1.5 mol % and 2.5 mol % sPPSUs.The TFC-FO membranes fabricated on the 1.5 mol %

sPPSU substrate possesses the lowest structural parameter (S)and hence was chosen for further evaluation of the effect ofincreasing draw solution concentration on water flux. As seen inSI Figure S3, the increase in draw solution concentration from0.5 to 2.0 M NaCl leads to an increase in water flux in both thePRO and FO modes. This is attributed to the increase ineffective osmotic pressure difference. Another interestingobservation is that the water flux in the PRO mode increasessteadily with the increase in draw solution concentration whilethat of the FO mode increases at a much slower rate. This

Table 2. PRO and FO Performance of TFC-FO Hollow Fiber Membranes using a 0.5M NaCl Draw Solution

PRO mode FO mode

membrane ID water flux (LMH) salt reverse flux (gMH) Js/Jw (g/L) water flux (LMH) salt reverse flux (gMH) Js/Jw (g/L)

TFC PPSU 22.64 ± 2.5 7.73 ± 0.51 0.34 12.37 ± 1.2 2.69 ± 0.21 0.22TFC 1.5 mol % sPPSU 49.39 ± 6.2 11.00 ± 1.36 0.22 22.51 ± 2.3 5.49 ± 0.35 0.24TFC 2.5 mol % sPPSU 37.71 ± 1.76 6.98 ± 0.71 0.19 17.98 ± 0.17 2.63 ± 0.32 0.15

Table 3. Transport Properties and Structural Parameters of TFC-FO Hollow Fiber Membranes

membrane ID water permeability, A (Lm−2h−1bar−1) salt rejectiona (%) salt permeability, B (Lm−2h−1) Km (sm−1) Sb (m)

TFC PPSU 3.15 ± 0.07 86.8 ± 0.7 0.0952 ± 0.003 5.07 × 105 7.46 × 10−4

TFC 1.5 mol % sPPSU 1.99 ± 0.02 90.9 ± 0.3 0.0399 ± 0.002 1.11 × 105 1.63 × 10−4

TFC 2.5 mol % sPPSU 1.80 ± 0.11 87.9 ± 0.9 0.0490 ± 0.011 1.63 × 105 2.40 × 10−4

aTested at 1 bar with 1000 ppm NaCl solution. bStructural parameters were calculated based on experiments under the FO mode using 0.5 M NaClas draw solution and DI water as feed.



phenomenon is indicative of the onset of ICP within the poroussupport layer which is more severe in the case of FO mode.Osmotic Seawater Desalination. The newly developed

TFC 1.5 mol % sPPSU FO hollow fiber membranes werechosen for characterization for its potential for seawaterdesalination. 3.5 wt % NaCl was used as a model for seawatersolution to test the membrane performance, along with 2.0 MNaCl solution as the draw solution under the PRO mode.Table 4 summarizes the water flux as a function of module

length. Interestingly, the water flux increases with a decrease in

module length. This is due to the effects of ICP and externalconcentration polarization (ECP) on the effective osmoticdriving force across the membrane in the hollow fiberconfiguration. In other words, the 2.0 M NaCl draw solutionis quickly diluted inside the membrane due to high waterinflow, while the seawater concentration near the TFC layer isquickly built up. As a result, the osmotic driving force across theTFC layer is severely reduced. However, both ICP and ECPeffects can be effectively mitigated by reducing the fiber length.The TFC 1.5 mol % sPPSU with a short fiber length exhibitswater production of 22 LMH which is among the best inliterature to date. In addition, a quick comparison tocommercial HTI membranes33 and other data found fromliterature as shown in Figure 3 shows that the TFC-FO hollowfiber membranes fabricated in this work surpasses that of HTI’s

and many others. Clearly, this work establishes the potentialprospect of using sulfonated polyphenylenesulfone as a newmaterial for substrates for TFC-FO hollow fibers fabricatedfrom interfacial polymerization and provides useful insights inapplying such membranes for real industrial applications.

ASSOCIATED CONTENT*S Supporting InformationMaterials; procedures for hollow fiber spinning, spinningparameters, membrane post-treatment and module fabrication;characterization techniques such as FESEM, viscosity. Thismaterial is available free of charge via the Internet at http://pubs.acs.org.

AUTHOR INFORMATIONCorresponding Author*Phone: (65) 6516 6645; fax: (65) 6779 1936; e-mail:[email protected].


ACKNOWLEDGMENTSWe thank BASF SE, Germany for funding this work with agrant number of R-279-000-363-597. Thanks are due to Ms. LiXue, Dr. Panu Sukitpaneenit and their help and suggestions tothis work. This research was also partially funded by theSingapore National Research Foundation under its Competitiveresearch Program for the project entitled, “Advanced FOMembranes and Membrane Systems for Wastewater Treat-ment, Water Reuse and Seawater Desalination” (grant number:R-279-000-336-281).

REFERENCES(1) Shannon, M. A.; Bohn, P. W.; Elimelech, M.; Georgiadis, J. G.;Marinas, B. J.; Mayes, A. M. Science and technology for waterpurification in the coming decades. Nature 2008, 452 (7185), 301−310.(2) Semiat, R. Energy issues in desalination processes. Environ. Sci.Technol. 2008, 42 (22), 8193−8201.(3) Zhao, S.; Zou, L.; Tang, C. Y.; Mulcahy, D. Recent developmentsin forward osmosis: Opportunities and challenges. J. Membr. Sci. 2012,396 (0), 1−21.(4) Zhao, S.; Zou, L.; Mulcahy, D. Effects of membrane orientationon process performance in forward osmosis applications. J. Membr. Sci.2011, 382 (1−2), 308−315.(5) Lee, K. L.; Baker, R. W.; Lonsdale, H. K. Membranes for powergeneration by pressure-retarded osmosis. J. Membr. Sci. 1981, 8 (2),141−171.(6) Achilli, A.; Cath, T. Y.; Childress, A. E. Power generation withpressure retarded osmosis: An experimental and theoretical inves-tigation. J. Membr. Sci. 2009, 343 (1−2), 42−52.(7) Garcia-Castello, E. M.; McCutcheon, J. R.; Elimelech, M.Performance evaluation of sucrose concentration using forwardosmosis. J. Membr. Sci. 2009, 338 (1−2), 61−66.(8) Jiao, B.; Cassano, A.; Drioli, E. Recent advances on membraneprocesses for the concentration of fruit juices: A review. J. Food Eng.2004, 63 (3), 303−324.(9) Yang, Q.; Wang, K. Y.; Chung, T. S. A novel dual-layer forwardosmosis membrane for protein enrichment and concentration. Sep.Purif. Technol. 2009, 69 (3), 269−274.(10) Chung, T. S.; Li, X.; Ong, R. C.; Ge, Q.; Wang, H.; Han, G.Emerging forward osmosis (FO) technologies and challenges aheadfor clean water and clean energy applications. Curr. Opin. Chem. Eng.2012, 1 (3), 246−257.

Table 4. TFC Hollow Fiber Membranes Synthesized on 1.5mol % sPPSU for Seawater Desalinationa

membrane ID fiber length (cm) flux (LMH)

1 13.5 14.01 ± 0.052 10.3 16.23 ± 0.123 8.5 22.13 ± 1.21

aUnder the PRO mode using a model seawater of 3.5 wt % NaCl and a2.0 M NaCl solution as the draw solution.

Figure 3. A comparison of water fluxes reported in literature for FOseawater desalination applications. Model seawater (3.5 wt % NaCl)solution was used as the feed solution, whereas 2.0 M NaCl as thedraw solution. Commercial HTI,33,34 cellulose acetate propionate,16

TFC hollow fiber (macrovoids),18 TFC PES/sPsf,15 TFC sPsf,21 TFCPES (macrovoid-free).20.



http://pubs.acs.org

http://pubs.acs.org


(11) Li, Z. Y.; Yangali-Quintanilla, V.; Valladares-Linares, R.; Li, Q.;Zhan, T.; Amy, G. Flux patterns and membrane fouling propensityduring desalination of seawater by forward osmosis. Water Res. 2012,46 (1), 195−204.(12) Mi, B.; Elimelech, M. Organic fouling of forward osmosismembranes: Fouling reversibility and cleaning without chemicalreagents. J. Membr. Sci. 2010, 348 (1−2), 337−345.(13) Veríssimo, S.; Peinemann, K. V.; Bordado, J. Thin-filmcomposite hollow fiber membranes: An optimized manufacturingmethod. J. Membr. Sci. 2005, 264 (1−2), 48−55.(14) Shintani, T.; Matsuyama, H.; Kurata, N. Development of achlorine-resistant polyamide reverse osmosis membrane. Desalination2007, 207 (1−3), 340−348.(15) Wang, K. Y.; Chung, T. S.; Amy, G. Developing thin-film-composite forward osmosis membranes on the PES/SPSf substratethrough interfacial polymerization. AIChE J. 2012, 58 (3), 770−781.(16) Li, X.; Wang, K. Y.; Helmer, B.; Chung, T. S. Thin-filmcomposite membranes and formation mechanism of thin-film layers onhydrophilic cellulose acetate propionate substrates for forward osmosisprocesses. Ind. Eng. Chem. Res. 2012, 51 (30), 10039−10050.(17) Yip, N. Y.; Tiraferri, A.; Phillip, W. A.; Schiffman, J. D.;Elimelech, M. High performance thin-film composite forward osmosismembrane. Environ. Sci. Technol. 2010, 44 (10), 3812−3818.(18) Chou, S.; Shi, L.; Wang, R.; Tang, C. Y.; Qiu, C.; Fane, A. G.Characteristics and potential applications of a novel forward osmosishollow fiber membrane. Desalination 2010, 261 (3), 365−372.(19) Wang, R.; Shi, L.; Tang, C. Y.; Chou, S.; Qiu, C.; Fane, A. G.Characterization of novel forward osmosis hollow fiber membranes. J.Membr. Sci. 2010, 355 (1−2), 158−167.(20) Sukitpaneenit, P.; Chung, T. S. High Performance thin-filmcomposite forward osmosis hollow fiber membranes with macrovoid-free and highly porous structure for sustainable water production.Environ. Sci. Technol. 2012, 46 (13), 7358−7365.(21) Widjojo, N.; Chung, T. S.; Weber, M.; Maletzko, C.; Warzelhan,V. The role of sulphonated polymer and macrovoid-free structure inthe support layer for thin-film composite (TFC) forward osmosis(FO) membranes. J. Membr. Sci. 2011, 383 (1−2), 214−223.(22) Shi, L.; Chou, S. R.; Wang, R.; Fang, W. X.; Tang, C. Y.; Fane,A. G. Effect of substrate structure on the performance of thin-filmcomposite forward osmosis hollow fiber membranes. J. Membr. Sci.2011, 382 (1−2), 116−123.(23) Zhang, G.; Song, X.; Ji, S.; Wang, N.; Liu, Z. Self-assembly ofinner skin hollow fiber polyelectrolyte multilayer membranes by adynamic negative pressure layer-by-layer technique. J. Membr. Sci.2008, 325 (1), 109−116.(24) Geise, G. M.; Lee, H. S.; Miller, D. J.; Freeman, B. D.; McGrath,J. E.; Paul, D. R. Water purification by membrane: The role of polymerscience. J. Polym. Sci., Part B: Polym. Phys. 2010, 48 (15), 1685.(25) Doi, S.; Hamanaka, K. Pore size control technique in thespinning of polysulfone hollow fiber ultrafiltration membranes.Desalination 1991, 80 (2−3), 167−180.(26) Liu, Y.; Koops, G. H.; Strathmann, H. Characterization ofmorphology controlled polyethersulfone hollow fiber membranes bythe addition of polyethylene glycol to the dope and bore liquidsolution. J. Membr. Sci. 2003, 223 (1−2), 187−199.(27) Loeb, S.; Mehta, G. D. A two-coefficient water transportequation for pressure-retarded osmosis. J. Membr. Sci. 1978, 4 (0),351−362.(28) Mehta, G. D.; Loeb, S. Internal polarization in the poroussubstructure of a semipermeable membrane under pressure-retardedosmosis. J. Membr. Sci. 1978, 4 (0), 261−265.(29) Strathmann, H., Production of microporous media by phaseinversion processes. In Materials Science of Synthetic Membranes;American Chemical Society: Washington, DC, 1985; Vol. 269, pp165−195.(30) Xing, D. Y.; Peng, N.; Chung, T. S. Formation of celluloseacetate membranes via phase inversion using ionic liquid, [BMIM]-SCN, as the solvent. Ind. Eng. Chem. Res. 2010, 49 (18), 8761−8769.

(31) Li, Y.; Chung, T. S.; Chan, S. Y. High-affinity sulfonatedmaterials with transition metal counterions for enhanced proteinseparation in dual-layer hollow fiber membrane chromatography. J.Chromatogr., A 2008, 1187 (1−2), 285−288.(32) Rahimpour, A.; Madaeni, S. S.; Ghorbani, S.; Shockravi, A.;Mansourpanah, Y. The influence of sulfonated polyethersulfone(SPES) on surface nano-morphology and performance of poly-ethersulfone (PES) membrane. Appl. Surf. Sci. 2010, 256 (6), 1825−1831.(33) McCutcheon, J. R.; McGinnis, R. L.; Elimelech, M. Desalinationby ammonia−carbon dioxide forward osmosis: Influence of draw andfeed solution concentrations on process performance. J. Membr. Sci.2006, 278 (1−2), 114−123.(34) Phillip, W. A.; Yong, J. S.; Elimelech, M. Reverse draw solutepermeation in forward osmosis: Modeling and experiments. Environ.Sci. Technol. 2010, 44 (13), 5170−5176.



Highly Water-Soluble Magnetic Nanoparticles as Novel Draw Solutes in ForwardOsmosis for Water Reuse

Ming Ming Ling, Kai Yu Wang, and Tai-Shung Chung*

Department of Chemical and Biomolecular Engineering, National UniVersity of Singapore, Singapore 117576

Highly hydrophilic magnetic nanoparticles have been molecularly designed. For the first time, the application ofhighly water-soluble magnetic nanoparticles as novel draw solutes in forward osmosis (FO) was systematicallyinvestigated. Magnetic nanoparticles functionalized by various groups were synthesized to explore the correlationbetween the surface chemistry of magnetic nanoparticles and the achieved osmolality. We verified that magneticnanoparticles capped with polyacrylic acid can yield the highest driving force and subsequently highest water fluxamong others. The used magnetic nanoparticles can be captured by the magnetic field and recycled back into thestream as draw solutes in the FO process. In addition, magnetic nanoparticles of different diameters were alsosynthesized to study the effect of particles size on FO performance. We demonstrate that the engineering of surfacehydrophilicity and magnetic nanoparticle size is crucial in the application of nanoparticles as draw solutes in FO.It is believed that magnetic nanoparticles will soon be extensively used in this area.

Introduction

Water scarcity is gradually emerging as a worldwide problembecause of accelerating population growth and environmentalpollution.1 More and more regions are considered water-poorand face associated problems of food production and publichealth issues.2 Demands for clean water impel an emphasis onresearch in clean water reuse to avert a potential disaster.Previously, multistage flash (MSF), multieffect distillation(MED), and especially reverse osmosis (RO) have been theflagships of seawater desalination.3-5 However, MSF and MEDare energy intensive processes that require high temperaturesfor the evaporation of water from ocean water, while RO is anelectricity intensive process that needs extremely high hydraulicpressures to push water across the RO membrane but reject saltsand other contaminants. MSF, MED, and RO in this centuryare no longer as pragmatic as in the past, especially in light ofthe expected exponential population increase and the limitedenergy resources. Alternative low-energy desalination technolo-gies are urgently needed to facilitate water reuse processes.

Osmosis is a ubiquitous physical phenomenon found in allof biology and involving the great power of molecular transport.Though it has been studied for a long time, only since the 1960shas it been explored for power generation as pressure-retardedosmosis (PRO).6 In forward osmosis (FO), the osmotic pressureacross the semipermeable membrane provides the driving forcethat draws water from a higher-water chemical potential sideto a lower-water chemical potential side without the aid ofhydraulic pressure.7 By comparison, the hydraulic pressure usedin RO must exceed the required threshold osmotic pressure tomove water from a lower to higher water chemical potential.The significant advantages of osmotic pressure requiring noexternal energy source as its drawing force in FO together witha lower membrane fouling propensity than pressure-drivenprocesses explain intensive investigative efforts for furtherexploitation, such as wastewater treatment, drug release, andprotein enrichment.8,9 However, the challenges of FO still liein the fabrication of eligible FO membranes and the readilyseparable draw solutes of high osmotic pressures. The FOmembranes are rigorously delicate, and many researchers have

put forth great effort in studying it.10,11 The selected draw solutesvary across present-day studies that include sugar and am-monium carbon dioxide complex.12,13 Nonetheless, there is along way to go to fulfill the criteria of draw solutes in FO,including (1) high osmolality to generate high osmotic pressuresand (2) easy and efficient separation from water.

According to the criteria of draw solutes in FO, superparamag-netic nanoparticles, which can be easily collected and separatedfrom water by means of a magnetic field, shall be considered asthe draw solute in FO, assuming high osmotic pressures can beinduced by magnetic nanoparticles or some other means. Magneticnanoparticles have assumed an important place in this researchbecause of their high surface-area-to-volume ratio and specialmagnetic behavior.14 So far, they are mainly studied in the fieldof biomedical applications, such as drug delivery and biocatalysis.15,16

However, many researchers in related areas have put forward theidea of using magnetic nanoparticles in FO, concerning the globalwater shortage problem.7,17,18

Herein, we synthesize highly hydrophilic magnetic nanoparticlesand investigate the application of magnetic nanoparticles as drawsolutes in FO. In previously published studies, magnetic nanopar-ticles were found to have high molecular weights and lowsolubilities in water and thus to be of questionable suitability foruse as a draw solute.19 We report fine magnetic nanoparticlessynthesized by a thermal decomposition method in one step withsurfaces capped with super hydrophilic groups to implement a drawsolution of high osmotic pressure. Three different kinds offunctional groups were used to functionalize the magnetic nano-particles, resulting in different osmolalities when they reached thehighest solubility. After the facile separation from water via amagnetic field, magnetic nanoparticles were reused in FO processes.Magnetic nanoparticles of different diameters were also synthesizedand tested for the investigation of possible factors that affect theperformance of magnetic nanoparticles as draw solute in FO. Ourresults demonstrate that functionalized magnetic nanoparticles usedas draw solute are attainable and suggest great potential in a varietyof FO applications.

Experimental Section

Materials. Iron(III) acetylacetonate [Fe(acac)3, 99.9%], tri-ethylene glycol (98%), 2-pyrrolidone (99%), and polyacrylic

* To whom correspondence should be addressed. Tel.: +65-65166645. Fax: +65-67791936. E-mail: [email protected].

Ind. Eng. Chem. Res. 2010, 49, 5869–5876 5869

10.1021/ie100438x 2010 American Chemical SocietyPublished on Web 05/13/2010

acid (PAA) (Mw ) 1800; 98%) were purchased from Sigma-Aldrich. Ethyl acetate (98%) was obtained from Tedia. All thechemicals listed above were used as received. The deionized(DI) water used in the experiments was produced with a Milli-Qunit (Millipore) with a resistivity of 18 MΩ/cm.

Synthesis of Magnetic Nanoparticles. Three kinds of water-soluble magnetic nanoparticles capped with different surfacefunctional groups such as 2-pyrrolidone, triethylene glycol, andpolyacrylic acid were synthesized by the thermal decompositionmethod. Figure 1 illustrates the synthetic routes. 2-Pyrrolidone-magnetic nanoparticles (2-Pyrol-MNPs) and triethylene glycol-magnetic nanoparticles (TREG-MNPs) were prepared in hy-drophilic and high-boiling point organic solvents according tothe literature, and the detail could be found elsewhere.20,21

To investigate and improve the hydrophilicity of nanopar-ticles, polyacrylic acid was selected for functionalization of themagnetic nanoparticle surface because of the copious carboxy-late groups along the polymer chain. These functional groupsmay enhance the interaction between water and magneticnanoparticles because the free carboxylate groups extended inwater may facilitate the dispersibility of polyacrylic acid-magnetic nanoparticles (PAA-MNPs) in the aqueous solution22-24

and increase the driving force for FO. The synthesis of PAA-

MNPs for biomedical applications has been studied using theligand exchange method that transfers the nanoparticle’s surfaceproperty from hydrophobicity to hydrophilicity.22,23 However,compared to that of 2-Pyrol-MNPs and TREG-MNPs, the routeof ligand exchange of polyacrylic acid onto initially hydrophobicmagnetic nanoparticles was very complicated and tedious.

Inspired by Sun’s facile method25 for synthesizing finehydrophobic magnetic nanoparticles using appropriate surfac-tants and high-boiling point solvents, we chose polyacrylic acidas the surfactant and triethylene glycol as the solvent to producehighly hydrophilic PAA-MNPs. To the best of our knowledge,PAA-MNPs prepared by this route have not been reportedpreviously. The synthesis was conducted as follows: 1.0 g ofpolyacrylic acid was first mixed with 25.0 mL of triethyleneglycol under a flow of argon and magnetically stirred; 2.0 mmolof Fe(acac)3 was added when the polyacrylic acid was com-pletely dissolved in triethylene glycol. The mixture was slowlyheated to 190 °C for 30 min and then quickly heated to refluxat 275 °C under a blanket of argon for 30 min. Via removal ofthe heat source, the black homogeneous colloidal suspensionwas cooled down to room temperature. Ethyl acetate (30 mL)was added to the reaction solution and then separated viacentrifugation, resulting in a black precipitation. The blackprecipitation was dissolved in DI water and reprecipitated inethyl acetate three times to remove residuals thoroughly. Viaadjustment of the reaction conditions, 2-Pyrol-MNPs, TREG-MNPs, and PAA-MNPs were prepared with a similar size range(20-30 nm). All three kinds of magnetic nanoparticles exhibitvery hydrophilic properties and are stably dissolved in watereven after being stored for 3 months. The newly synthesizedmagnetic nanoparticle solutions were dialyzed for 36 h tocompletely remove impurities prior to application.

Characterization of Magnetic Nanoparticles. Magneticnanoparticles were imaged using field emission scanningelectronic microscopy (FESEM, JEOL JSM-6700) by drying adispersion of magnetic nanoparticles on amorphous carbon-coated copper grids. The measurements of the size distributionof magnetic nanoparticles were conducted in Nanoparticle SizeAnalyzer (Nano ZS, ZEN3600). The magnetic properties offunctionalized magnetic nanoparticles were recorded in vibratingsample magnetometer (VSM, LakeShore 450-10) with a saturat-ing field of 1 T. Fourier transform infrared spectroscopy (FTIR)

Figure 2. Schematic diagram of a laboratory-scale FO setup combined with a magnetic separator (co-current crossflow of feed and draw solutions).

Figure 1. Routes of synthesis of surface-functionalized magnetic nanoparticles.

5870 Ind. Eng. Chem. Res., Vol. 49, No. 12, 2010

of magnetic nanoparticles, pressed into KBr pellets, wasconducted on a Bio-Rad spectrometer (FTS 135). Thermogravi-metric analysis (TGA) was conducted to determine the weightpercentage of surface-capping groups covalently bonding to themagnetic nanoparticle core, using a Perkin-Elmer TGA 7instrument at a constant heating rate of 5 °C/min from roomtemperature to 600 °C in a nitrogen environment.

Forward Osmosis using Surface-Functionalized MagneticNanoparticles as Draw Solutes. The performance of highlywater-soluble magnetic nanoparticles as draw solutes in the FOsystem was tested on a lab-scale circulating filtration unit, asdepicted in Figure 2. The commercially available HTI FOmembrane (batch 060327-3; Hydration Technologies Inc.,previously Osmotek Inc.) and DI water were employed as theFO membrane and feed solution, respectively. The crossflowpermeation cell was designed in a plate and frame configurationwith a rectangular channel (8.0 cm in length, 1.0 cm in width,and 0.25 cm in height) on either side of the membrane. Thevelocities of both draw and feed solutions, which co-currentlypassed through the permeation cell channel, were maintainedat 6.4 cm/s during the FO testing. The temperatures of the wholeFO system were kept at 22 ( 0.5 °C. The draw solutions of

magnetic nanoparticles were prepared by dissolving certainamount of nanoparticles in water. When magnetic nanoparticleswere used as draw solutes in FO, the magnetic nanoparticlescould be 100% intercepted by FO membranes in the drawsolution side for the diameter of magnetic nanoparticles wasmuch larger than the pore size of FO membranes. Therefore,the mass of magnetic nanoparticles remained constant in drawsolutions. The dilution of the draw solution was negligiblebecause the volumes of draw solutions increased less than 2%during the FO testing.

The water permeation flux (Jv, L ·m-2 ·hr-1, abbreviated asLMH) was calculated from the volume change of the feedsolution.

where ∆V (L) is the permeation water collected over apredetermined time ∆t (hr) during the duration of FO and A isthe effective membrane surface area (m2).

Figure 3. TGA graph of magnetic nanoparticles.

Figure 4. FESEM and size distribution of magnetic nanoparticles.

Figure 5. FTIR spectra of magnetic nanoparticles.

Jv ) ∆V/(A∆t) (1)

Ind. Eng. Chem. Res., Vol. 49, No. 12, 2010 5871

The concentration of surface-capping groups surrounding eachmagnetic nanoparticle was calculated using the formula

where C is the molar concentration of surface-capping groupson magnetic nanoparticles in water (mol/L); F is the density of

the magnetic nanoparticle solution (g/L) while the water densityis assumed to be 1000 g/L; w is the mass percentage of thesurface-capping groups upon magnetic nanoparticles obtainedby TGA via weight percentage change before and after burningmagnetic nanoparticles at high temperatures, as shown in Figure3; V is the volume of the draw solution; and Mw is the molecularweight of the surface-capping groups. Here we assume noadditional volume change in the draw solution after the additionof magnetic nanoparticles except their own individual volumes.

Results and Discussion

Magnetic Nanoparticles. The newly synthesized PAA-MNPsexhibit a narrow size distribution with superparamagnetic andsuper hydrophilic characteristics. These unique properties mayarise from the close matching between the carboxylate groupsof the polyacrylic and magnetite nanocrystal surface.16,22,23 Byincreasing the amount of polyacrylic acid in reaction mixtures,one can effectively control the growth of magnetic nanoparticlecores, and the resultant PAA-MNPs have average diameters of4-30 nm.

Figure 4 shows FESEM images and particle size distributionsof 2-Pyrol-MNPs, TREG-MNPs, and PAA-MNPs. They arealmost monodisperse in water with an average diameter of20-30 nm. It is important to note, even though the particlesare very small, no agglomeration can be observed in thesemagnetic nanoparticles. Figure 5 displays the FTIR spectra ofthe surface chemistry of these magnetic nanoparticles. Thecharacteristic peaks located at 590 cm-1 can be attributed tothe lattice absorption of iron oxide,19-23 while the peaks around1660 cm-1 are caused by the vibrating mode of the C-O bond.Clearly, the O in the C-O bond anchors to Fe on the surfaceof magnetic nanoparticles.19-23 The characteristic bands at2950-2900 cm-1 are typical C-H stretching. For 2-Pyrol-MNPs, the broad band peak at 3350 cm-1 is ascribed to N-Hstretching of 2-pyrrolidone.19,26 In the spectra of TREG-MNPs

Figure 6. Hysteresis loops of magnetic nanoparticles at 300 K.

Figure 7. Water flux and salt reverse flux using a HTI FO membrane (NaClas the draw solution).

Figure 8. Water flux of surface-capping groups and magnetic nanoparticles.

C ) (F - 1000) ·wMw

(2)


and PAA-MNPs, peaks at 1121-1080 cm-1 are referred to asC-O stretching.20-22 A strong absorption band at 1736 cm-1

is found in the spectrum of PAA-MNP, which is a characteristicof the CdO stretching mode for carboxylate groups.21,22 Thepeaks at 1459-1395 and 1621 cm-1 of symmetric and asym-metric C-O stretching modes of carboxylate groups furtherconfirm a large amount of polyacrylic acid attached on thesurface of magnetic nanoparticles.21,22

Magnetic behaviors of 2-Pyrol-MNPs, TREG-MNPs, andPAA-MNPs were investigated using VSM at room temperature.The magnetization curves in Figure 6 show the superparamag-netic properties of the three kinds of magnetic nanoparticles.Coericivity and remanence are not observed by magneticmeasurements. For PAA-MNPs synthesized in triethylene glycolas solvent, its level of saturation magnetization is greatlydecreased compared to that of TREG-MNPs. This phenomenonfurther confirms that polyacrylic acid was successfully bondedto the surface of magnetic nanoparticles in competition withtriethylene glycol in reactions and the multilayer of polyacrylicacid might be formed.27

Forward Osmosis Based on the HTI FO Membrane. TheFO performance of HTI FO membranes was first tested usingsodium chloride as draw solute, and Figure 7 shows the waterflux and salt leakage. The water flux in the FO mode (the drawsolution against the porous layer) is less than that in the PROmode (the draw solution against the dense selective layer) dueto the internal concentration polarization. In other words, moresolutes are rejected by the dense layer in the PRO mode, whilemore solutes could be accumulated inside the membrane poresin the FO mode and retard the flux.28,29

Water Flux Using Versatile Magnetic Nanoparticles asDraw Solutes. To investigate the effect of different surfacechemistries of magnetic nanoparticles as draw solutes on waterflux, 2-Pyrol-MNPs, TREG-MNPs, and PAA-MNPs withdiameters in a similar range of 20-30 nm were first examined.Figures 8 and 9 show the water flux and osmolalities as afunction of surface chemistry and draw solute concentration inPRO and FO modes. For comparison, the surface-cappinggroups were also tested as draw solutes, and their performanceis included in Figures 8 and 9. To elucidate the water fluxdiscrepancy between different draw solutes in water, the

chemical potential energy of surface-capping groups in waterwas calculated with the aid of Material Studio 4.4.30 Figure 10displays the calculated results. There is the smallest potentialenergy gap between water and a 2-pyrrolidone solution, followedby water and a triethylene glycol solution, and then a muchlarger potential energy gap is exemplified between water and apolyacrylic acid solution. A larger potential chemical gapbespeaks a higher osmotic pressure and a stronger driving force.

The theoretical prediction is consistent with the experimentalresults with surface-capping groups. As shown in Figure 8, waterflux at 0.05 mol/L exactly follows this order: polyacrylic acid(PAA) > triethylene glycol (TREG) > 2-pyrrolidone (Pyrol).At high draw solute concentrations, the flux induced bytriethylene glycol is still relatively higher than that of 2-pyr-rolidone. The flux profile induced by polyacrylic acid is differentfrom others. Even though it results in a high flux at 0.05 mol/L, it keeps nearly constant with an increase in polyacrylic acidconcentration. One of the possible reasons is the strong hydrogenbonding between abundant carboxylate groups along polyacrylicacid chains and hydroxyl groups of the HTI FO membranesurface. As a result, the polyacrylic acid may closely adhere to

Figure 9. Osmolality of surface-capping groups and magnetic nanoparticles in water.

Figure 10. Potential energies explaining water flux differences.


the membrane surface or possibly partially block the pore andretard the flux enhancement. In addition, viscosity may play animportant role because the viscosity of polyacrylic acid in wateris found to rapidly increase with an increase in polymerconcentration. Thus, the water transport resistance in the systemincreases and retards any enhancement of the water flux.

Figure 11 tabulates and compares the flux and osmolalitywhen magnetic nanoparticles are employed as draw solutes. Asexpected, PAA-MNPs can yield the highest water flux of10.4-7.7 LMH, followed by TREG-MNPs (8.1-5.9 LMH) andthen 2-Pyrol-MNPs (6.3-4.5 LMH), which is in line with ourpotential energy calculations and the osmolality measurements.

Figure 11. Summary of (a) highest water flux and (b) osmolality of magnetic nanoparticles with different surface chemistries.

Figure 12. Water flux of magnetic nanoparticles before and after recycling.

Figure 13. Size distribution of recycled magnetic nanoparticles (a) 2-Pyrol-MNPs, (b) TREG-MNPs, and (c) PAA-MNPs.


The water flux induced by PAA-MNPs is far superior to thatof pure polyacrylic acid, implying that PAA-MNPs can over-come the hydrogen bonding near the membrane surface andrealize the mobility of polyacrylic acid. This is due to the factthat polyacrylic acid has been evenly dispersed in the solutionand closely attached on the extremely fine and high mass-densitymagnetic nanoparticles. Not only do the nanosize iron oxidecores result in significant contact surfaces between polyacrylicacid and water, but also the covalence bond between polyacrylicacid and magnetic nanoparticles provides sufficient strength forthe integrity of magnetic nanoparticles.

Facile Recovery of Magnetic Nanoparticles by MagneticField Capture. To fulfill the application of highly hydrophilicmagnetic nanoparticles as draw solutes in FO for water reuse,the drawn water and magnetic nanoparticles in the permeateside must be easily separated and the magnetic nanoparticlesmust be easily recycled. A high-gradient magnetic separator(HGMS, model L-1CN, Frantz canister separator, from S. G.Frantz Co., Inc., Trenton, NJ) was employed to provide themagnetic field and test this concept. It was found that themagnetic nanoparticles can be readily captured by the HGMSand the other product is water. Each kind of magnetic nano-particle was then retested in the FO system as draw solute.Figure 12 displays the water flux of magnetic nanoparticlesbefore and after recycling. Flux drops slightly possibly due tothe high strength of the HGMS that causes slight aggregation

of magnetic nanoparticles; the size distribution of recycledmagnetic nanoparticles is shown in Figure 13. However, therecycled magnetic nanoparticles can still be dissolved in waterpromptly, which confirmed surface functional groups have beenfirmly anchored onto iron oxide cores. Future work will aim toovercome the nanoparticle aggregation induced by the HGMS.

Water Flux Using PAA-MNPs of Different Diameters asDraw Solutes. To utilize magnetic nanoparticles more ef-fectively for water reuse, PAA-MNPs, which exhibited thehighest water flux among the three kinds, were chosen for furtherexploration of the effect of magnetic nanoparticle diameter onwater flux. The diameter of magnetic nanoparticles can be tunedby adding 2.5, 2.0, 1.5, and 1.0 g of polyacrylic acid in 25.0mL of solvent to reaction mixtures, where PAA-MNPs withaverage diameters of 4, 5, 7, and 20 nm, respectively, can beobtained. Figure 14 illustrates the diameter distributions of theresultant PAA-MNPs, while Figure 15 shows their correspond-ing water flux and osmolality. The strongest driving force isachieved in the draw solution of 4 nm PAA-MNPs since moremagnetic nanoparticles of smaller diameters can be located perunit volume and it has the highest osmolality. This clearlydemonstrates that water flux with PAA-MNPs as draw soluteincreases with a decrease in MNP diameter. However, the useof smaller PAA-MNPs in FO has some drawbacks. Themagnetic property becomes smaller as a result of a higherpolyacrylic acid content wrapping each particle. In addition, the

Figure 14. Size distribution of PAA-MNPs synthesized by addition of (a) 2.5, (b) 2.0, (c) 1.5, and (d) 1.0 g of polyacrylic acid in 25.0 mL of solvent.

Figure 15. Effect of PAA-MNP particle size on (a) water flux and (b) osmolality.


recovery of smaller PAA-MNPs was unsatisfactory because theirdiameters jump out of the range that the HGSM can capture.Thus, a compromised size of MNPs should be used betweenthe choices of high water flux and ready recovery by themagnetic field.

Conclusion

We have synthesized highly water-soluble magnetic nano-particles and demonstrated that they are novel robust drawsolutes in forward osmosis for the first time. Draw solutions ofmagnetic nanoparticles capped with polyacrylic acid exhibit thehighest water flux among the three different surface-function-alized magnetic nanoparticles. It is believed that water flux canbe further increased via modification of the surface chemistry.Magnetic nanoparticles after being used in the FO process arereadily captured in the magnetic field. The HGMS provides afacile and fast way to facilitate the recovery of magneticnanoparticles in a continuous process. In addition, water fluxcan be enhanced by decreasing the diameters of magneticnanoparticles. Future studies will be focused on (1) theoptimization of surface chemistry and diameter selection ofmagnetic nanoparticles, (2) the investigation of their perfor-mance in forward osmosis for seawater desalination, and (3)the investigation of their sustainability in forward osmosisprocesses.

Acknowledgment

The authors would like to thank the Environmental & WaterIndustry Development Council (MEWR 651-06-158 and R-279-000-271-272), the National University of Singapore (NUS) (R-279-000-249-646), the King Abdullah University of Science andTechnology (KAUST), and Saudi Arabia (R-279-000-265-597).The authors also acknowledge Hydration Technologies Inc. forproviding the membranes. Special thanks are due to Prof. T.Alan Hatton of the Massachusetts (Cambridge, MA), Dr. QianYang, Dr. May May Teoh, Ms. Honglei Wang, Mr. PanuSukitpaneenit, and Mr. Zhengzhong Zhou for their valuablesuggestions.

Nomenclature

A ) effective membrane surface area, m2

Jv ) product water flux, L m-2 h-1

∆t ) operation time interval, h∆V ) water permeation volume, LC ) molar concentration of surface-capping material on magnetic

nanoparticles, mol/LMw ) molecular weight of surface-capping material, g/molw ) weight percentage of surface-capping material on magnetic

nanoparticlesF ) density of a magnetic nanoparticle solution, g/L

Literature Cited

(1) Robert, L.; Elimelech, M. Global Challenges in Energy and WaterSupply: The Promise of Engineered Osmosis. EnViron. Sci. Technol. 2008,42, 8625–8629.

(2) Shannon, M. A.; Bohn, P. W.; Elimelech, M.; Georgiadis, J. G.;Marias, B. J.; Mayes, A. M. Science and technology for water purificationin the coming decades. Nature 2008, 452, 301–310.

(3) Morin, O. J. Design and operating comparison of MSF and MEDsystems. Desalination 1993, 93 (1-3), 69–109.

(4) Wade, N. M. Distillation plant development in dual-purpose powerand desalination plants. Desalination 1999, 123, 115–125.

(5) Sourirajan, S. ReVerse Osmosis; Academic Press, Inc.: New York,1970.

(6) Loeb, S. Osmotic power plants. Science 1975, 189, 654–655.(7) Cath, T. Y.; Childress, A. E.; Elimelech, M. Forward osmosis:

Principles, applications, and recent developments. J. Membr. Sci. 2006, 281,70–87.

(8) Wright, J. C.; Johnson, R. M.; Yun, S. I. DUROS osmoticpharmarceutical systems for parenteral & site-directed therapy. DrugDeliVery Technol. 2003, 3, 64–73.

(9) Yang, Q.; Wang, K. Y.; Chung, T. S. A novel dual-layer forwardosmosis membrane for protein enrichment and concentration. Sep. Purif.Technol. 2009, 69, 269–274.

(10) Wang, K. Y.; Chung, T. S.; Qin, J. J. Polybenzimidazole (PBI)nanofiltration hollow fiber membranes applied in forward osmosis process.J. Membr. Sci. 2007, 300, 6–12.

(11) Yang, Q.; Wang, K. Y.; Chung, T. S. Dual-layer hollow fibers withenhanced flux as novel forward osmosis membranes for water reuse.EnViron. Sci. Technol. 2009, 43 (8), 2800–2805.

(12) Beaudry, E. G.; Lampi, L. A. Membrane technology for directosmosis concentration of fruit juice. Food Technol. 1990, 44, 121–129.

(13) McCutcheon, J. R.; McGinnis, R. L.; Elimelech, M. A novelammonia-carbon dioxide forward (direct) osmosis desalination process.Desalination 2005, 174, 1–11.

(14) Lu, A. H.; Salabas, E. L.; Schuth, F. Magnetic nanoparticles:Synthesis, protection, functionalization, and application. Angew. Chem., Int.Ed. 2007, 46, 1222–1244.

(15) Wang, W.; Xu, Y.; Wang, D. C.; Li, Z. Recyclable nanobiocatalystfor enantioselective sulfoxidation: Facile fabrication and high performanceof chloroperoxidase-coated magnetic nanoparticles with iron oxide core andpolymer shell. J. Am. Chem. Soc. 2009, 131 (36), 12892–12893.

(16) Kim, G. C.; Li, Y. Y.; Chu, Y. F.; Cheng, S. X.; Zhuo, R. X.;Zhang, X. Z. Nanosized temperature-responsive Fe3O4-UA-g-P(UA-co-NIPAAm) magnetomicelles for controlled drug release. Eur. Polym. J. 2008,44, 2761–2767.

(17) NanoMagnetics, Water purification, 2005. http://www.softmachines.org/wordpress/?p)68.

(18) Adham, S.; Oppenheimer, J.; Liu, L.; Kumar, M. DewateringreVerse osmosis concentrate from water reuse using forward osmosis;WateReuse Foundation: Alexandria, VA, 2007.

(19) Hearn, J. D.; Lovett, A. J.; Smith, G. D. Ozonolysis of oleic acidparticles: Evidence for a surface reaction and secondary reactions involvingCriegee intermediates. Phys. Chem. Chem. Phys. 2005, 7, 501–511.

(20) Li, Z.; Chen, L.; Bao, H. B.; Gao, M. Y. One-pot reaction tosynthesize water-soluble magnetite nanocrystals. Chem. Mater. 2004, 16(8), 1391–1393.

(21) Cai, W.; Wan, J. Q. Facile synthsis of superparamagnetic magnetitenanoparticles in liquid polyols. J. Colloid Interface Sci. 2007, 305, 366–370.

(22) Ge, J. P.; Hu, Y. X.; Biasini, M.; Dong, C. L.; Guo, J. H.;Beyermann, W. P.; Yin, Y. D. One-step synthesis of highly water-solublemagnetite colloidal nanocrystals. Chem.sEur. J. 2007, 13, 7153–7161.

(23) Zhang, T. R.; Ge, J. P.; Hu, Y. X.; Yin, Y. D. A general approachfor transferring hydrophobic nanocrystals into water. Nano Lett. 2007, 7,3203–3207.

(24) Zhang, Q.; Cheng, S. X.; Zhang, X. Z.; Zhuo, R. X. Water-solublePolymer Protected Lipofectamine 2000/DNA Complexes for Solid-PhaseTransfection. Macromol. Biosci. 2009, 9, 1262–1271.

(25) Sun, S. H.; Zeng, H.; Robinson, D. B.; Li, G. X. MonodisperseMFe2O4 (M)Fe, Co, Mn) nanoparticles. J. Am. Chem. Soc. 2004, 126, 273–279.

(26) Qin, C.; Li, C.; Hu, Y. Z.; Shen, J. F.; Ye, M. X. Facile synthesisof magnetic iron oxide nanoparticles using 1-methyl-2-pyrrolidone as afunctional solvent. Colloids Surf. A 2009, 336, 130–134.

(27) Palma, R. D.; Peeters, S.; Bael, M. V.; Rul, H. V.; Bonroy, K.;Laureyn, W.; Mullens, J.; Borghs, G.; Maes, G. Silane ligand exchange tomake hydrophobic superparamagnetic nanoparticles water-dispersible. Chem.Mater. 2007, 19, 1821–1831.

(28) Mi, B. X.; Elimelech, M. Chemical and physical aspects of organicfouling of forward osmosis membranes. J. Membr. Sci. 2008, 320, 292–302.

(29) Zydney, A. L. Stagnant film model for concentration polarizationin membrane systems. J. Membr. Sci. 1997, 23, 275–282.

(30) Wang, Y.; Chung, T. S.; Wang, H.; Goh, S. H. Butanol isomerseparation using polyamide-imide/CD mixed matrix membranes via per-vaporation. Chem. Eng. Sci. 2009, 64, 5198–5209.

ReceiVed for reView February 27, 2010ReVised manuscript receiVed April 13, 2010

Accepted April 28, 2010

IE100438X


10788 Chem. Commun., 2011, 47, 10788–10790 This journal is c The Royal Society of Chemistry 2011

Cite this: Chem. Commun., 2011, 47, 10788–10790

Facile synthesis of thermosensitive magnetic nanoparticles as ‘‘smart’’

draw solutes in forward osmosisw

Ming Ming Ling, Tai-Shung Chung* and Xianmao Lu*

Received 1st July 2011, Accepted 19th August 2011

DOI: 10.1039/c1cc13944d

Thermosensitive superparamagnetic nanoparticles were synthe-

sized by a one-step thermal decomposition method and successfully

recycled as a ‘smart’ draw solute in forward osmosis processes

for water reuse without losing performance efficiency.

Osmotically driven processes stand at the forefront to meet

our ever-increasing global demand for clean water. Forward

Osmosis (FO) is broadly acknowledged as the key technology

to integrate future sea water desalination and waste water

reclamation processes to solve the global water scarcity problem.1,2

In recent years, FO technology has been intensively studied for

its use in desalination, water reuse, as well as protein enrichment

processes3 and power generation.4 Forward osmosis utilizes

the osmotic pressure difference of two solutions separated by a

semi-permeable membrane to induce spontaneous movement

of water molecules from the less concentrated solution (feed

solution) to the other solution (draw solution) while most

solutes are rejected by the FO membrane.1,2 Contrary to

reverse osmosis and other water production processes, FO

discloses an innovative scenario to obtain water from concentrated

solutions without any external aid including hydraulic

pressure or heat. Since energy consumption is increasing

exponentially as energy sources become scarcer and natural

disasters affect the supply of water to millions of people each

year, FO provides a versatile technology which minimizes

energy consumption while maintaining high rejection of

solutes and low membrane fouling propensity.5

Eligible FO membranes of high water permeate and low

reverse salt flux play an important role in FO processes.6–8

Likewise, the selection of a suitable draw solute can greatly

influence the efficiency of FO. In general, entitled draw solutes

in FO for water production possess the qualities of being able

to generate high osmotic pressures and easy recovery of the

water obtained. A variety of chemical compounds have been

attempted as draw solutes in FO applications but the progress

is well behind the development of FO membranes.2,9–13

Ammonium bicarbonate induces reasonable FO fluxes and

the water product must be obtained via thermal decomposition

of the ammonium salt at about 60 1C.9 Highly water-soluble

superparamagnetic nanoparticles were recently discovered as

a new type of draw solutes and exhibit the advantages

of extremely low reverse flux compared to traditional

chemicals.11,12 Nanoparticles with hydrophilic surface func-

tionality and high surface area-to-volume ratio may generate

high osmotic pressures for desalination and water reclamation

purposes. Moreover, the nanoparticles can be readily regenerated

using more efficient and conventional methods, such as magnetic

fields. However, magnetic nanoparticles were found to aggregate

to much larger sizes under a high-strength magnetic field,

causing a significant decrease in efficiency. Stimuli-responsive

polymer hydrogels have been explored as a draw agent,13 but

their FO performance is poor at room temperature and the

dewatering process needs a hydraulic pressure of 30 bar at

elevated temperatures.

Therefore, the objectives of this communication are, for the

first time, (1) to overcome the aforementioned hurdles of magnetic

nanoparticles and stimuli-responsive polymer hydrogels, and

(2) to conceptually demonstrate magnetic nanoparticles with

thermo-responsive properties for FO processes and easy

regeneration without performance deterioration. At tempera-

tures above the Lowest Critical Solution Temperature (LCST),

thermosensitive magnetic nanoparticles tend to agglomerate

together resulting from the coil-to-globule transition of the

thermosensitive polymer on their surfaces.14–22 As such, a

magnetic field of much lower strength can facilitate the

separation of thermosensitive magnetic nanoparticles when

the temperature is above the LCST and significantly decrease

the possibility of irreversible agglomeration caused by high-

strength magnetic fields. As a result, thermosensitive magnetic

nanoparticles can effectively function as ‘smart’ draw solutes

in FO without particle size changes upon magnetic separation.

So far, thermosensitive magnetic nanoparticles are mainly

synthesized by the method of either ligand exchange of particles

initially capped with oleic acid16 or by polymerization on the

surface of magnetic nanoparticles prepared from co-precipitation

reaction.21 The ligand exchange method suffers from the

disadvantages of a large amount of solvent consumption and a

low yield while the polymerization method is more complicated

and involves many reaction parameters. More importantly, the

Department of Chemical and Biomolecular Engineering, NationalUniversity of Singapore, Singapore 117576.E-mail: [email protected], [email protected];Fax: +65-67791936w Electronic supplementary information (ESI) available: Experimental(synthesis and characterization of thermosensitive magnetic nanoparticles; FO tests using PNIPAM/TRI-MNP as draw solute), schematicdiagram of FO process, TGA and size distribution of PNIPAM/TRI-MNP. See DOI: 10.1039/c1cc13944d

ChemComm Dynamic Article Links

www.rsc.org/chemcomm COMMUNICATION

http://dx.doi.org/10.1039/c1cc13944d

http://dx.doi.org/10.1039/c1cc13944d

This journal is c The Royal Society of Chemistry 2011 Chem. Commun., 2011, 47, 10788–10790 10789

resultant particle size and uniformity of particles through these

two methods are not satisfactory as draw solutes in FO processes.

In order to fulfil the required characteristics of draw solutes in

FO, thermosensitive magnetic nanoparticles must possess both

highly hydrophilic groups to induce osmotic pressures and

extremely thermosensitive polymers on the surface to facilitate

easy regeneration via low-strength magnetic field separation. Here

we report a facile synthesis of thermosensitive magnetic nano-

particles by a one-step thermal decomposition method with

improved solubility in water and their application as ‘smart’ draw

solutes in FO. In this work, poly(N-isopropylacrylamide) is

selected as the stabilizer and triethylene glycol as the organic

solvent. The detailed synthesis procedures can be found in ESI.wPoly(N-isopropylacrylamide) as the basic thermosensitive polymer

possesses the coordination capacity with transition metal ions

owing to its amide groups23 and hence it can be added directly

as the nanoparticle surface capping agent in the thermal

decomposition reaction without modification. Triethylene

glycol is chosen as the solvent because of its high boiling

point24 and modest interaction with poly(N-isopropylacrylamide)

at elevated temperatures. More importantly, it can functionalize

the nanoparticle surface with improved hydrophilicity.24 Both

poly(N-isopropylacrylamide) and triethylene glycol would

compete to bind on the magnetic nanoparticle surface during

the reaction to inhibit the nanoparticle growth because they

both possess the available ligands to form strong chemical

coordination with iron cations on nanoparticle surfaces. In

addition, the copious CQO groups along the chain of poly

(N-isopropylacrylamide) result in multiple anchor points on the

iron oxide core surface, which contribute to the robust coating of

poly(N-isopropylacrylamide) on the nanoparticles. Therefore, in

order to perform as ‘smart’ draw solutes in FO, thermosensitive

magnetic nanoparticles functionalized with poly(N-isopropyla-

crylamide) and triethylene glycol (PNIPAM/TRI-MNP) exhibit

enhanced hydrophilic properties and optimal responses to dual

stimuli (heat and magnetic field) when compared to conventional

thermosensitive nanoparticles.

The core material of PNIPAM/TRI-MNPs is composed of

magnetite, which is verified by the X-ray diffraction (XRD)

patterns as displayed in Fig. S2 (ESIw). It can be seen that the

position and relative intensity of the diffraction peaks match well

with the standard XRDmagnetite data (JCPDS file No. 19-0629).

Moreover, the black colour of a dry PNIPAM/TRI-MNPs

sample also indicates magnetite as the core material rather

than maghemite with similar XRD patterns. Fig. 1 shows

TEM pictures of PNIPAM/TRI-MNPs dispersed in water

before and after magnetic separations. The nanoparticles of

sphere shape are observed and they exhibit a slightly poly-

dispersed distribution with a diameter range of 9 3 nm which

is close to the crystallite size of the core metal oxide of 8.2 nm

obtained from XRD data using the Scherrer formula, indicat-

ing the single crystal of each nanoparticle.25 However, the

particle sizes determined from TEM and XRD are smaller

than the size distribution (Fig. S4, ESIw) obtained from the

nanoparticle size analyzer, which can be ascribed to some

degrees of nanoparticle aggregation and hydration of a hydro-

philic polymeric layer on the nanoparticles in the aqueous

solution. Magnetic nanoparticles synthesized in triethylene

glycol alone (TRI-MNP) were also characterized to affirm

the existence of poly(N-isopropylacrylamide) on the PNI-

PAM/TRI-MNPs. TGA measurements were conducted to

determine the polymer coverage on nanoparticle surfaces. As

shown in Fig. S3 in ESIw, the major weight loss of both the

nanoparticles spans from 150 1C to 700 1C. It is observed that

PNIPAM/TRI-MNPs exhibit a different pattern compared to

TRI-MNPs, which indicates the existence of poly(N-isopropyl-

acrylamide) on PNIPAM/TRI-MNP surfaces compared to

TRI-MNPs capped with only triethylene glycol. PNIPAM/

TRI-MNPs have a much greater polymeric coverage (38%)

than TRI-MNP (20%) on nanoparticle surfaces owing to the

decomposition of poly(N-isopropylacrylamide) of high molecule

weight (10 K). Magnetic properties of PNIPAM/TRI-MNPs

and TRI-MNPs were also examined. Fig. 2 compares the

magnetization hysteresis loops of both the nanoparticles at

room temperature. The magnetic nanoparticles possess a

superparamagnetic property as both the hysteresis loops show

zero coercivity and magnetic remanence. The saturation

magnetization of PNIPAM/TRI-MNPs (40 emu g1) is lower

than TRI-MNPs (50 emu g1), which further confirms that

poly(N-isopropylacrylamide) was successfully grafted on the

Fig. 1 TEM images of PNIPAM/TRI-MNP before (a, b) and after

magnetic separation for times (c, d).

Fig. 2 VSM loops of PNIPAM/TRI-MNP and TRI-MNP.

10790 Chem. Commun., 2011, 47, 10788–10790 This journal is c The Royal Society of Chemistry 2011

nanoparticles. These experimental results also accord with the

phenomenon from other researches that the superparamagnetic

property of magnetic nanoparticles is strongly dependent on the

amount of polymer grafted.22 The thermosensitive property of

PNIPAM/TRI-MNPs was examined by measuring the particle

size through a series of temperature changes. Sharp and rever-

sible changes of particle size were observed at 35 1C upon heating

and cooling (Fig. 3). The reversible thermosensitive properties in

terms of particle size and osmotic pressure are essential for the

regeneration of PNIPAM/TRI-MNPs with higher recovery rates

in modest energy consumptions.

To evaluate the thermosensitive magnetic nanoparticles as

draw solutes in FO, PNIPAM/TRI-MNPs were tested in FO

systems at room temperature and regenerated via low-strength

magnetic separation at temperatures above its LCST for 5 runs.

The detailed experimental setup and procedures are included in

the ESIw (Fig. S2). The size measurements of PNIPAM/TRI-

MNPs were conducted after regenerations and no increment in

particle size was observed even after 5 runs (Fig. S3, ESIw). TheTEM images in Fig. 1c and d show the recycled PNIPAM/TRI-

MNPs and no obvious particle changes were found either. The

reversible thermosensitive property of the PNIPAM/TRI-MNPs

synthesized in our approach and the low-strength magnetic

separations contribute to the intactness and robustness of PNI-

PAM/TRI-MNPs as a ‘smart’ draw solute in FO. Thereafter, as

displayed in Fig. 4, the FO performance of PNIPAM/TRI-

MNPs can be well maintained through 5 runs in both Pressure

Retarded Osmosis (PRO) mode (draw solution facing the dense

layer of a FO membrane) and Forward Osmosis (FO) mode

(feed solution facing the dense layer of a FO membrane).

In conclusion, we have successfully synthesized thermosensitive

superparamagnetic nanoparticles with improved hydrophilicity in

one step. The resultant PNIPAM/TRI-MNPs exhibit uniform

particle sizes of less than 20 nm and excellent stability in water.We

have demonstrated that PNIPAM/TRI-MNPs can be recycled as

a ‘smart’ draw solute in FO processes without losing performance

efficiency as a result of their reversible thermosensitive property

facilitating the magnetic separation of low strength to assure the

integrity of nanoparticle draw solutes. The FO performance can

be enhanced with nanoparticle surface engineering and decreasing

particle sizes. It is believed that thermosensitive magnetic nano-

particles hold great potential as a novel draw solute in FO

processes for water reuse, desalination, protein dehydration and

biomedical applications.

The authors would like to thank the Singapore National

Research Foundation (NRF) for support through the Competitive

Research Program for the project entitled, ‘‘New Advanced

FOmembranes and membrane systems for wastewater treatment,

water reuse and seawater desalination’’ (R-279-000-336-281;

R-279-000-337-281). The authors also acknowledge Hydration

Technologies Inc. for providing the membranes.

Notes and references

1 R. L. McGinnis and M. Elimelech, Environ. Sci. Technol., 2008,42, 8625.

2 T. S. Chung, S. Zhang, K. Y. Wang, J. C. Su and M. M. Ling,Desalination, DOI: 10.1016/j.desal.2010.12.019.

3 M. M. Ling and T. S. Chung, J. Membr. Sci., 2011, 372, 201.4 K. Gerstandt, K. V. Peinemann, S. E. Skilhagen, T. Thorsen andT. Holt, Desalination, 2008, 224, 64.

5 B. X. Mi and M. Elimelech, Environ. Sci. Technol., 2010, 44, 2022.6 K. Y. Wang, T. S. Chung and J. J. Qin, J. Membr. Sci., 2007,300, 6.

7 S. Zhang, K. Y. Wang, T. S. Chung, H. M. Chen, Y. C. Jean andG. Amy, J. Membr. Sci., 2010, 360, 522.

8 J. C. Su, Q. Yang, J. F. Teo and T. S. Chung, J. Membr. Sci., 2011,355, 36.

9 J. R. McChtcheon, R. L. McGinnis and M. Elimelech, Desalination,2005, 174, 1.

10 P. McCormicka, J. Pellegrino, F. Mantovani and G. Sarti,J. Membr. Sci., 2008, 325, 467.

11 M. M. Ling, K. Y. Wang and T. S. Chung, Ind. Eng. Chem. Res.,2010, 49, 5869.

12 Q. C. Ge, J. C. Su, T. S. Chung and G. Amy, Ind. Eng. Chem. Res.,2011, 50, 382.

13 D. Li, X. Y. Zhang, J. F. Yao, G. P. Simon and H. T. Wang,Chem. Commun., 2011, 47, 1710.

14 X. D. Xu, C. S. Cheng, Z. C. Wang, G. R. Wang, S. X. Cheng,X. Z. Zhang and R. X. Zhuo, J. Polym. Sci., Part A: Polym.Chem., 2008, 46, 5263.

15 H. Cheng, J. L. Zhu, Y. X. Sun, S. X. Cheng, X. Z. Zhang andR. X. Zhuo, Bioconjugate Chem., 2008, 19, 1368.

16 I. Robinson, C. Alexander, L. D. Tung, D. G. Fernig and N. T.K. Thanh, J. Magn. Magn. Mater., 2009, 321, 1421.

17 S. Kalele, R. Narain and K. M. Krishnan, J. Magn. Magn. Mater.,2009, 321, 1377.

18 C. F. Lee, C. C. Lin, C. A. Chien and W. Y. Chiu, Eur. Polym. J.,2008, 44, 2768.

19 R. Narain, M. Gonzales, A. S. Hoffman, P. S. Staytonand K. M. Krishnan, Langmuir, 2007, 23, 6299.

20 I. Robinson, C. Alexander, L. T. Lu, L. D. Tung, D. G. Fernig andN. T. K. Thanh, Chem. Commun., 2007, 4602.

21 P. L. Golas, S. Louie, G. V. Lowry, K. Matyjaszewski andR. D. Tilton, Langmuir, 2010, 26, 16890.

22 B. Y. Du, A. X. Mei, P. J. Tao, B. Zhao, Z. Cao, J. J. Nie, J. T. Xuand Z. Q. Fan, J. Phys. Chem. C, 2009, 113, 10090.

23 Z. Li, L. Zhao, H. B. Bao and M. Y. Gao, Chem. Mater., 2004, 16,1391–1393.

24 C. Wei and J. Wan, J. Colloid Interface Sci., 2007, 305, 366.25 H. P. Klug and L. E. Alexander, X-Ray Diffraction Procedures,

Wiley, New York, 1959.

Fig. 3 PNIPAM/TRI-MNP size changes at different temperatures.

Fig. 4 FO performance of recycled PNIPAM/TRI-MNP.

This journal is c The Royal Society of Chemistry 2013 Chem. Commun., 2013, 49, 8471--8473 8471

Cite this: Chem. Commun.,2013,49, 8471

Hydroacid complexes: a new class of draw solutes topromote forward osmosis (FO) processes†

Qingchun Ge and Tai-Shung Chung*

A new class of draw solutes from hydroacid complexes is presented.

With hydroacid complexes as draw solutes in FO, superior performance

is achieved in terms of high water fluxes and negligible reverse

solute fluxes. The characteristics of expanded configurations, abundant

hydrophilic groups and ionic species are essential for hydroacid

complexes as competent draw solutes.

Freshwater scarcity is a global issue with the rapid growth ofpopulation and economics. Nowadays the supply of potablewater is increasingly relying on science and technology.1–7

Reverse osmosis (RO) is the most widely used membranetechnology in seawater desalination to produce clean water.1

However, due to the high energy consumption in the ROprocess, forward osmosis (FO) has received much attentionbecause of its unique characteristics of high water recovery, lowoperation pressure and minimal brine discharge.2–7 It maypotentially alleviate the stress of freshwater scarcity. Despitethese advantages, however, the unavailability of cost-effective,easy recovery and high performance draw solutions is still a bigobstacle to the advancement of FO.6,8

Given the importance of FO, the exploration of draw solutionshas received great attention. Various draw solutes have beenproposed over the past few decades.5–8 Commercial compoundsincluding mixed gases,9,10 sugars,11,12 inorganic and organicsalts13–15 have been extensively used in FO. Draw solutions fromthese compounds create reasonably high water fluxes but withsignificant reverse solute fluxes,15 which not only contaminatethe product water but also increase the replenishment cost ofdraw solutes. To overcome these drawbacks, synthetic drawsolutes have been explored.16–21 However, most of them havelow osmotic pressures unless at high concentrations whereconcentration polarization may become a serious issue. To solve

this problem, a new class of draw solutes, hydroacid complexes,is explored in this study.

Coordination complex is a material consisting of metal(s)and ligand(s). The configuration of the metal center expandswhen it bonds to ligands. This feature may lead to a low reverseflux in FO and easy solute regeneration in posttreatment whencomplexes are used as draw solutes. Meanwhile, they can bedesigned freely according to the required functions by varyingeither the metals or ligands. Such characteristics make complexesappropriate candidates to be draw solutes. In this work, a seriesof cupric and ferric complexes with hydroxyl acids of citric acid(CA), malic acid (MA) and tartaric acid (TA) as ligands weresynthesized according to a modified method.22 Fig. 1 shows arepresentative structure of a ferric CA complex (Fe–CA), whilestructures of other complexes with their precursor ligands areillustrated in Table S1 (ESI†).

The coordination of the acids to the Cu2+ or Fe3+ metal corewas confirmed by FTIR spectroscopy (the spectra of CA complexesare given in Fig. S1, ESI†). Characteristic absorptions at B3400,1608–1720 and 1396–1427 cm1 correspond to O–H, CQO andC–O groups, respectively,16 indicating the presence of COOH. Theband at B570 cm1 corresponds to the characteristic absorptionof the metal–oxygen bond.16 All analytic results verify the successfulcoordination between hydroxyl ligands and metal centers.

The composition of the complexes was disclosed by TGAmeasurements (Fig. S2, ESI†). The thermal decomposition of CAand MA complexes takes place in two major stages: (i) dehydrationof the hydrated complexes to form unhydrated intermediates(45–140 1C); (ii) decomposition of the unhydrated intermediates

Fig. 1 Structure of Fe–CA.

Department of Chemical & Biomolecular Engineering, National University of

Singapore, Singapore 117576, Singapore. E-mail: [email protected];

Fax: +65 6779-1936

† Electronic supplementary information (ESI) available: Experimental (synthesis,characterization and relative viscosity of hydroacid complexes; FO set up andtests), structures and molecules, FTIR spectra, TGA spectra and weight losses. SeeDOI: 10.1039/c3cc43951h

Received 26th May 2013,Accepted 22nd July 2013

DOI: 10.1039/c3cc43951h

www.rsc.org/chemcomm

ChemComm

COMMUNICATION

8472 Chem. Commun., 2013, 49, 8471--8473 This journal is c The Royal Society of Chemistry 2013

to form metal oxides (250–350 1C).23,24 The weight loss in thefirst stage corresponds to the departure of H2O molecules eithercoordinating to the metal cores or present in the complexes viaH bonds. The weight loss in the second stage is primarily due tothe decomposition of the unhydrated complexes to form metaloxides by losing H2O, CO or CO2 molecules. Unlike CA and MAcomplexes, TA complexes do not contain water. Therefore, only onestep of decomposition was observed in the range of 230–320 1C. Forall complexes, the observed weight losses closely approximate thecalculated ones (Table S2, ESI†).

Fig. S3 (ESI†) shows the relative viscosity (Zr) of all complexesat different concentrations. Zr increases with an increase in theconcentration. Comparing the respective cupric and ferricseries, Zr only changes slightly when varying the ligands fromCA to MA to TA. At the same concentration, the ferric complexwith a more complicated molecular structure has a larger Zr

than its cupric analogue. However, compared to polyacrylic acidsodium (PAA-Na) draw solutes,17 all of these complexes exhibitinsignificant Zr. This indicates that the adverse effect of drawsolute viscosity on FO performance may be negligible whenusing these complexes as draw solutes. This hypothesis isverified by subsequent FO experimental results.

All complexes except Cu–TA are in anionic forms. Theirsodium salts have good water solubility under neutral conditionsand produce considerable osmotic pressures in water. Fig. S4(ESI†) compares their osmotic pressures and shows that osmoticpressure increases with increasing solute concentration. Theosmotic pressure of CA and its complexes follows the order ofFe–CA > Cu–CA > CA. In the studied complexes, Fe–CA has thehighest osmotic pressure, which is determined by its structure.According to the van’t Hoff equation,8 the presence of more ionicspecies in an aqueous solution gives a higher osmotic pressure.Since the number of ionic species increases from a ligandprecursor to its complexes and the ferric complexes have moreionic species than the cupric analogues, ferric complexes havehigher osmotic pressures.

The configurations of these bulk complexes are either tetrahedral(Cu2+ complexes) or octahedral (Fe3+ complexes). The characteristicsof their expanded structures along with low viscosity and highosmotic pressure make them good candidates to be draw solutes.Fig. 2 shows their FO performance using HTI membranes underboth pressure retarded osmosis (PRO) (draw solution against theselective layer of membranes) and FO (draw solution against thesupport layer of membranes) modes. Water flux increases withincreasing solute concentration, consistent with the changetendency of osmotic pressure (Fig. S4(a), ESI†). The water fluxesunder the PRO mode consistently outperform those under theFO mode, as observed elsewhere.5–8,10,25 The increment ofreverse solute flux with the increase in solute concentration isvery small. The highest reverse flux is below 0.2 g MH when thesolute concentration is up to 2.0 M, which is one order ofmagnitude lower than that of NaCl draw solutions with compar-able water fluxes under the same conditions.21 The ratio ofreverse solute flux (Js) to water flux (Jw), Js/Jw, is lower than0.015 g L1. This indicates that the complex loss in recoveringwater via FO is insignificant and the replenishment cost tomaintain a constant concentration is negligible. Clearly, these

complexes are superior to the inorganic draw solutes such asMgCl2, NaCl, and NH4HCO3 when comparing their losses in FO(0.015 vs. 0.57–2.48 g L1)15 with comparable water fluxes. Inaddition, the Fe–CA complex outperforms most draw solutesfrom synthetic materials developed recently such as imidazolederivatives,18 nanoparticles,16,21 organic salts,14 and others,19,20

in terms of higher water fluxes and negligible reverse fluxes.The performance of cupric and ferric complexes were also

evaluated using a self-made cellulose acetate (CA) hollow fibermembrane which has a smaller pore size and a thinner, denseselective layer than the HTI membrane.25 For comparison, NaClwas included as a benchmark draw solute. Fig. 3 shows thatwater fluxes for all compounds increase with an increase intheir concentration. Cupric and ferric complexes comprisingCA have the highest water fluxes in their respective series.Meanwhile, the water fluxes of ferric complexes increase fasterthan those of cupric complexes upon varying the ligand fromTA to MA to CA. In addition, ferric complexes produce muchhigher water fluxes than their cupric analogues due to thepresence of more ionic species in the ferric molecules. Similarto the observation in HTI membranes, reverse solute fluxes for allcomplexes are negligible (Fig. 3(c) and (d)). These advantagesdemonstrate that an FO process using this type of complexesmay provide solutions not only for water reuse but also for theseparation of some special compounds such as protein enrichmentwhere trace impurity due to reverse solute diffusion may denaturethe target products.26

Compared to NaCl, cupric complexes produce comparableor slightly higher water fluxes. Ferric complexes, however,outperform NaCl greatly. All complexes exhibit superiority overNaCl in terms of reverse flux (Fig. 3(c) and (d)). These complexesalso surpass the previous PAA-Na system17 due to their muchbetter FO performance under similar conditions. Moreover, unlike

Fig. 2 The performance comparison of Cu–CA and Fe–CA through HTImembranes: (a) water flux (Jw); (b) reverse flux (Js); (c) Js/Jw. DI water as the feedsolution.

Communication ChemComm

This journal is c The Royal Society of Chemistry 2013 Chem. Commun., 2013, 49, 8471--8473 8473

in the PAA-Na system, the increase of water flux in the currentcomplex system is almost proportional to the complex concen-tration owing to the higher degree of dissociation. Interestingly,the FO performance via CA hollow fiber membranes (Fig. 3) ismuch better than that via HTI flat sheet membranes (Fig. 2)even though both have high water fluxes and low reverse fluxes.This is due to the tighter specifications of the CA membrane inview of the small pore size, thin dense layer and poroussublayer morphology.25 Therefore, the performance of complexdraw solutes would be further improved when a more ideal FOmembrane is used.

In view of the best performance, 2.0 M Fe–CA was chosen tostudy seawater desalination via the CA membrane using3.5 wt% NaCl as the feed. A water flux of 13.1 LMH wasachieved under the PRO mode which is better than that of9.98 LMH obtained using 2.0 M MgCl2 draw solution under thesame conditions.25 Apparently, the characteristics of expanded

structure, abundant hydrophilic groups and ionic species inaqueous solutions are crucial for draw solutes to exhibit goodFO performance.

A pressure-driven nano-filtration process of 10 bar was usedto produce water and regenerate the Fe–CA draw solute from itsdiluted solution after FO tests. Fig. 4 shows that water produc-tion and Fe–CA rejection decrease upon increasing the feedconcentration. Concentration polarization may also contributeto the reduced water production. Nevertheless, a high rejectionof more than 90% was achieved when the concentrationincreases from 0.05 to 0.10 mol L1. To further improve therecycling efficiency, other types of membranes or other recycleprocesses, e.g. membrane distillation, will be explored.

We thank Singapore National Research Foundation underits Competitive Research Program entitled, ‘‘Advanced FOMembranes and Membrane Systems for Wastewater Treatment,Water Reuse and Seawater Desalination’’ (R-279-000-336-281)for the financial support. Special thanks are due to Dr Jincai Suand Mr Dianxue Lin for their valuable help.

Notes and references1 K. P. Lee, T. C. Arnot and D. Mattia, J. Membr. Sci., 2011, 370, 1.2 Y. Liu and B. X. Mi, J. Membr. Sci., 2012, 407–408, 136.3 J. J. Qin, B. Liberman and K. A. Kekre, Open Chem. Eng. J., 2009, 3, 8.4 V. Yangali-Quintanilla, Z. Li, R. Valladares, Q. Li and G. Amy,

Desalination, 2011, 280, 160.5 T. Y. Cath, A. E. Childress and M. Elimelech, J. Membr. Sci., 2006,

281, 70.6 T. S. Chung, X. Li, R. C. Ong, Q. Ge, H. L. Wang and G. Han, Curr.

Opin. Chem. Eng., 2012, 1, 246.7 S. Zhao, L. Zou, C. Y. Tang and D. Mulcahy, J. Membr. Sci., 2012, 396, 1.8 Q. Ge, M. M. Ling and T. S. Chung, J. Membr. Sci., 2013, 442, 225.9 R. A. Neff, US Pat., 3,130,156, 1964; G. W. Batchelder, US Pat.,

3,171,799, 1965.10 J. R. McCutcheon, R. L. McGinnis and M. Elimelech, Desalination,

2005, 174, 1.11 J. C. Su, T. S. Chung, B. J. Helmer and J. S. de Wit, J. Membr. Sci.,

2012, 396, 92.12 K. Stache, US Pat., 4,879,030, 1989.13 S. Phuntsho, H. K. Shon, S. K. Hong, S. Y. Lee and S. Vigneswaran,

J. Membr. Sci., 2011, 375, 172.14 K. S. Bowden, A. Achilli and A. E. Childress, Bioresour. Technol.,

2012, 122, 207.15 A. Achilli, T. Y. Cath and A. E. Childress, J. Membr. Sci., 2010,

364, 233.16 Q. Ge, J. C. Su, T. S. Chung and G. Amy, Ind. Eng. Chem. Res., 2011,

50, 382.17 Q. Ge, J. C. Su, G. Amy and T. S. Chung, Water Res., 2012, 46, 1318.18 S. K. Yen, F. M. Haja N, M. L. Su, K. Y. Wang and T. S. Chung,

J. Membr. Sci., 2010, 364, 242.19 D. Li, X. Zhang, J. Yao, G. P. Simon and H. Wang, Chem. Commun.,

2011, 47, 1710.20 M. Noh, Y. Mok, S. Lee, H. Kim, S. H. Lee, G. W. Jin, J. H. Seo,

H. Koo, T. H. Parka and Y. Lee, Chem. Commun., 2012, 48, 3845.21 M. M. Ling, K. Y. Wang and T. S. Chung, Ind. Eng. Chem. Res., 2010,

49, 5869.22 M. Matzapetakis, C. P. Raptopoulou, A. Tsohos, V. Papaefthymiou,

N. Moon and A. Salifoglou, J. Am. Chem. Soc., 1998, 120, 13266.23 L. Patron, O. Carp, I. Mindru, G. Marinescu and E. Segal, J. Therm.

Anal. Calorim., 2003, 72, 281.24 I. Quasim, A. Firdous, B. Want, S. K. Khosa and P. N. Kotru,

J. Phys. Chem. Solids, 2010, 71, 1501.25 J. C. Su and T. S. Chung, J. Membr. Sci., 2011, 376, 214.26 M. M. Ling and T. S. Chung, J. Membr. Sci., 2011, 372, 201.

Fig. 3 Performance comparison via CA membranes: (a) water fluxes of Cu2+

complexes and NaCl; (b) water fluxes of Fe3+ complexes and NaCl; (c) reversefluxes of Cu2+ complexes and NaCl; (d) reverse fluxes of Fe3+ complexes and NaCl.DI water as the feed solution under the PRO mode.

Fig. 4 Water production and solute rejection in the recycling of Fe–CA.

ChemComm Communication

psipw7thaward.org · 11/22/2013 · author's personal copy journal of membrane science383...

Documents