Journal of Membrane Science 505 (2016) 53–60

Contents lists available at ScienceDirect

Journal of Membrane Science

http://d0376-73

n CorrE-m

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

Upgrading polysulfone ultrafiltration membranes by blending withamphiphilic block copolymers: Beyond surface segregation

Yuqing Chen, Mingjie Wei, Yong Wang n

State Key Laboratory of Materials-Oriented Chemical Engineering, College of Chemical Engineering, Nanjing Tech University, Nanjing 210009, Jiangsu,PR China

a r t i c l e i n f o

Article history:Received 20 November 2015Received in revised form11 January 2016Accepted 16 January 2016Available online 19 January 2016

Keywords:Block copolymersPhase inversionPolyethylene glycolPolysulfoneSurface segregation

x.doi.org/10.1016/j.memsci.2016.01.03088/& 2016 Elsevier B.V. All rights reserved.

esponding author.ail address: yongwang@njtech.edu.cn (Y. Wan

a b s t r a c t

Surface segregation of amphiphilic copolymers in the phase inversion process has long been used toimprove membrane hydrophilicity. Typically, the copolymer is sparsely dosed into the casting solutionsas additives. Herein we substantially increase the dosages of amphiphilic copolymers, and obtain blendultrafiltration membranes with synergetically upgraded performances because of extra permeability-enhancing effect of the copolymer in addition to the surface segregation effect. We blend amphiphilicblock copolymer, polysulfone-block-polyethylene glycol (PSf-b-PEG), with polysulfone base polymer atvarious percentages up to 70%. There is no compatible issue between the PSf and PSf-b-PEG as they aremiscible at any blend ratio. Infrared spectroscopy and X-ray photoelectron spectroscopy confirm thesurface segregation of PEG blocks. Moreover, PSf-b-PEG evidently influences the phase separation pro-cess by slowing down the precipitation rate of the polymer solutions, thus producing membranes withthicker skin layers. Interestingly, increasing copolymer percentages result in more water-permeable PEGmicrodomains in the blend membranes and consequently enhanced water permeance. The blendmembranes exhibit simultaneously upgraded permeance, hydrophilicity, fouling resistance, and alsoperformance stability. The highest permeance reaches nearly 400 L/(m2 h bar) at a copolymer percentageof 40%, which is much higher than those of PSf membranes prepared in other works. By comparing withPEG homopolymer, we identify that the superior performances are originated also from the additionalwater permeability through PEG microdomains in addition to the effect of surface segregation of PSf-b-PEG copolymers.

& 2016 Elsevier B.V. All rights reserved.

1. Introduction

Polysulfone (PSf), as an engineering plastic with a glass tran-sition temperature of nearly 200 °C, is one of the most extensivelyused polymers for membrane separations because of its superiorthermal stability, mechanical robustness, chemical resistance, ex-cellent processibility as well as convenience in tuning membranemicrostructures [1]. Typically, PSf is manufactured into ultra-filtration (UF) membranes through the nonsolvent-induced phaseinversion (NIPS) and thus-produced PSf membranes are usuallyused in aqueous circumstances including water treatment [2],protein purification and fractionation [3], hemodialysis [4] andreverse/forward osmosis (as the substrates for thin-film compo-sites) [5]. However, as can be easily seen from the multiple aro-matic rings and scarcity of polar groups in its molecule PSf isstrongly hydrophobic in nature. Membranes produced purely from

g).

neat PSf suffer from very limited permeability, poor fouling re-sistance, and shortened lifetimes [6]. Consequently, polar additivesare frequently introduced into the casting solutions in the pro-duction of PSf membranes, or post modifications are used to im-prove their hydrophilicity [7].

A number of physical or chemical approaches has been used toenhance the hydrophilicity of PSf membranes [8], mainly includingadding inorganic nanofillers into membrane matrix [9,10], plasmaetching to generate polar groups on the surface [11] and directlygrafting some hydrophilic groups onto the membranes surface[1,12–14]. Besides, using hydrophilic additives is most technicallyand commercially feasible since most PSf membranes are preparedby the NIPS process in which additives can be readily co-dissolvedinto the casting solutions [15]. The extra advantage of hydrophilicadditives is improving the flux of water. It is believed that the poresize in the selective layer, porosity, and pore interconnection will beenlarged with the incorporation of hydrophilic additives into themembranes as the subtle interactions among polymer, solvent, andnonsolvent are changed with the presence of the additives [16].

Y. Chen et al. / Journal of Membrane Science 505 (2016) 53–6054

Water-soluble/dispersible polymeric additives, such as poly-ethylene glycol (PEG) or polyethylene oxide (PEO) (PEG and PEOshare the same molecular backbone and thus close physico-chemical properties but they have different end groups as they aresynthesized from different monomers) [17], polyvinylpyrrolidone(PVP) [18], polybenzimidazole (PBI) [19], and polyaniline (PANI)[20–22] are most frequently included into the recipes of PSfmembranes. In this regard, PEG is of particular interest because ofits strong hydration effect, biocompatibility, as well as easy avail-ability with variable molecular weights [15,23–25]. Typically,higher percentages of polymeric additives lead to membraneswith a stronger hydrophilicity and permeability. However, over-dosed additives may not be compatible with the casting solutionsand ruin the phase inversion process as well as the membraneperformances. An even worse situation for these hydrophilic ad-ditives, which are usually water-soluble, is that a large portion ofthem will leach out during the phase inversion process and thesubsequent operations in aqueous circumstances, leading to de-caying performances in terms of both hydrophilicity and perme-ability [16,26].

A solution to this leaching problem is to use amphiphilic co-polymers containing both hydrophobic segments and hydrophilicsegments. The hydrophobic parts are believed to be anchored tothe membrane skeleton by chain entanglement and the hydro-philic ones are forced to segregate on the pore walls in the phaseinversion process [27]. To this end, both graft and block copoly-mers (BCPs) have been used and enhanced hydrophilicity havebeen clearly evidenced in most cases although the improvement inwater permeance was frequently not observed. For the specificcase of PSf membranes discussed in the present work, in additionto usages of PEG or PEO homopolymers as additives where the lossof these additives are considered to be unavoidable [16], copoly-mers of PSf and PEG are expected to promise stable improvementof the membrane performances as the PSf components in the co-polymer will be fully compatible and miscible with PSf homo-polymers which are the base material for the membranes. Han-cock et al. firstly used polysulfone-block-polyethylene oxide BCPs,PSf-b-PEO, as additives to create blend semipermeable membranesfor medical purpose [28,29]. Several years later, Mayes and cow-orkers synthesized graft copolymers of PSf and PEG, PSf-g-PEG,and blended them into the PSf casting solutions to prepare PSfmembranes [1]. In both cases, the enhanced hydrophilicity and theconsequently improved resistance to protein adsorption were re-markable. However, both groups did not investigate the separationperformances of the produced membranes. In addition, Zhu andcoworkers also observed clearly enhanced hydrophilicity of PSfmembranes by combining other PSf-based amphiphilic BCPs intothe casting solutions. However, the contribution of the combiningPSf copolymers on the membrane performances remained un-elucidated as the role of PEG homopolymers used as co-additivesin their work cannot be excluded [30].

Interestingly, in previous works amphiphilic copolymers wereused “really” as additives or modifiers as they were dosed in thecasting solutions with small percentages, typically less than20 wt% in the total polymer concentrations [1,28–30]. This is areasonable and practicable strategy if these copolymers are me-chanically weak as materials, or poorly compatible with the basepolymers, or costly in price. The small dosages of copolymers maystill deliver the function to enhance the membrane hydrophilicityas surface segregation does not require large amount of additives.However, it remains open what will occur when the dosage ofcopolymers is increased be comparable to that of the base polymergiven that affordable, mechanical strong copolymer is available. Inthis situation, the copolymer is not only the additive in the com-mon sense as it is also a major component of the producedmembranes which should be considered as blend membranes. The

copolymer blended in the membrane casting solutions with do-sages comparable to that of the base polymer will significantlyinfluence the phase diagram of the solutions. Therefore, it mayoffer us a new strategy to tune the membrane microstructures andperformances in addition to the well-studied effect of surfacesegregation.

Considering that PSf-b-PEG BCPs can be conveniently synthe-sized with good yields and this copolymer largely maintains thegood mechanical robustness of PSf homopolymers [1,28–30], inthe present work, we blend PSf-b-PEG BCPs with PSf homo-polymer at various percentages up to 70% to prepare blend UFmembranes through the NIPS process. The surface segregation alsooccurs, leading to enhanced surface hydrophilicity. Furthermore,the heavily dosed BCPs substantially change the microstructuresand separation performances of the PSf membranes by formingwater-permeable PEG microdomains. As a consequence of the dualeffect of the blended BCPs, the permeance, hydrophilicity, foulingresistance, and also the performance stability of the blend mem-branes are synergistically improved. Specifically, the blend mem-branes with 40% content of PSf-b-PEG exhibit a permeance up toten times higher than PSf membranes prepared using differentadditives and also similar or even higher retentions. Surface seg-regation of amphiphilic copolymers has long been evidenced toenhance the membranes hydrophilicity and permeability. How-ever, the present work reveals that another permeability-enhan-cing effect of the amphiphilic copolymers (permeation throughwater-permeable microdomains) concurs when the dosages of thecopolymer are comparable to that of the base membrane-formingmaterials in addition to the effect of surface segregation. This newrole of amphiphilic copolymers has never been reported beforeand it offers us new possibilities for the development of advancedmembranes with the assistance of amphiphilic copolymers.

2. Experimental

2.1. Materials

PSf (P3500 LCD, average molecular weight 22000 Da) suppliedby Solvay, was used as the base polymer in the membrane castingsolutions. Reagent grade N-methyl-2-pyrrolidone (NMP) was usedas the solvent to dissolve the polymers without further purifica-tion. Reagent grade PEG homopolymer (average molecular weight400 Da) or PSf-b-PEG BCP (the weight ratio of PEG blocks in thecopolymer is 49%) were used as the blend components in castingsolutions. Deionized water (conductivity: 8.20 μs/cm, Wahaha)was used as the main nonsolvent in the coagulation bath. Bovineserum albumin (BSA) was purchased from Sigma-Aldrich.

2.2. Preparation of PSf/PSf-b-PEG and PSf/PEG blend membranes

The casting solutions of 16% in weight were prepared by dis-solving PSf and PSF-b-PEG in NMP at desired proportions (BCPpercentages ranging from none to 70% in the total amount ofpolymers) under mechanical stirring at 70 °C for 4 h. The castingsolutions were kept for at least 48 h at 70 °C in vacuum oven toremove air bubbles. After removal of the air bubbles, the liquiddope was cast onto a glass plate with a casting knife to obtain acasting film of 250 μm thick, then coagulated in deionized water atroom temperature. The obtained PSf/PSf-b-PEG membrane wasthoroughly washed with water to remove the residual NMP. Themethod of preparing PSf/PEG membranes is the same as the PSf/PSf-b-PEG membranes, by replacing PSf-b-PEG block polymer withPEG homopolymer.

Fig. 1. FTIR-ATR spectra of membranes prepared from pure PSf and from the blendof PSf and PSf-b-PEG.

Y. Chen et al. / Journal of Membrane Science 505 (2016) 53–60 55

2.3. Characterization

The Fourier transformation infrared attenuated total reflection(FTIR-ATR) spectra of membrane samples were obtained fromdried membrane samples (Nicolet 8700 FTIR spectrometer) in therange of 4000–400 cm�1. A contact angle goniometer (DropmeterA-100, Maist) was applied to obtain the water contact angle. Foreach sample, at least 3 sites were tested and the average watercontact angle was acquired. Scanning electron microscopy (SEM)investigations on samples sputter-coated with Au/Pd alloy wereperformed with a field emission SEM (Hitachi S4800) operated at5 kV. Cross-sectional specimens were prepared by fracturing inliquid nitrogen. X-ray photoelectron spectroscopy (XPS) was per-formed with an ESCALAB 250 XPS system (Thermo Scientific)using a monochromatic Al Kα X-ray source. To compensate thesurface charge effects, all binding energies in the spectra werereferenced to the C 1 s neutral peak at 285.0 eV.

2.4. Filtration and separation tests

Water flux as well as retention of BSA was determined using astirred filtration cell (Amicon Model 8010, Millipore). For waterflux tests, membrane compaction was carried out at 0.1 MPa fortypically about 15 min to obtain a constant water flux, then waterflux was measured at the same pressure. For retention tests, BSAwere dissolved in deionized water at concentrations of 0.5 g/L.Retention rates for BSA were determined by measuring the con-centrations of BSA in solution using a UV�vis spectrometer (Na-noDROP 2000C, Thermo Scientific). For this purpose, we evaluatedthe relative intensities of the characteristic BSA peak at 280 nm.The water flux and the pure water permeability (PWP) were cal-culated using the following equations:

=× ( )JV

A t 11

=× × ∆ ( )

VA t P

PWP 2

where J1 represents the water flux (L/(m2 h)), V (L) is the volume ofthe pure water penetrating through the membrane, A is the ef-fective membrane area (m2), t is the operation time (h). The soluterejections of UF membranes were calculated by the followingequations:

=−

×( )

RC C

C100%

3p f

p

where Cp and Cf are the BSA concentrations (g/L) in the permeateand the feed solutions, respectively.

2.5. Antifouling tests

The antifouling property of the membrane was determinedusing the reported procedure [31]. Briefly, after filtration of thefeed BSA solution, the membrane was washed with deionizedwater for about 0.5 h, then the water flux of cleaned membranes J2(L/(m2 h)) was measured again. Finally, the membrane antifoulingproperty was determined by calculating flux recovery ratio (FRR)using the following equation:

= ×( )

J

JFRR 100%

42

1

3. Results and discussions

3.1. Preferential segregation of PSf-b-PEG on the membrane surfaces

We co-dissolved PSf homopolymers and PSf-b-PEG BCPs inNMP with various BCP content ranging from none to 70% to pre-pare the casting solutions for the blend membranes. We foundthat the two polymer components were miscible at randomblending ratios and no macrophase separation can be observed inany of the blend solutions investigated in this work. The stableblend solutions should be attributed to the high compatibility ofthe PSf homopolymer and the PSf-based BCPs and the high com-patibility between A polymer and AB block copolymer has longbeen confirmed by differential scanning calorimetry (DSC) analysis[32]. Specifically for the blend systems of PSf and PSf-b-PEG, thePSf block in BCPs can be solubilized in the PSf homopoymers whilethe other block is thus stabilized in the solution because of itscovalent junction to the PSf block.

To confirm the presence of PSf-b-PEG in blend membranes, wecompared the chemical compositions of blend membranes andneat PSf membranes by the FTIR in the ATR mode. Fig. 1 exhibitsthe IR spectra of the membranes prepared from neat PSf and fromblended PSf/PSf-b-PEG with the BCP content of 50%. All peaksemerging in the neat PSf membrane are also present in the blendmembranes. Importantly, there apparently appear two new peaksin the spectrum of the blend membrane. The peaks centeringaround 946 cm�1 and 2864 cm�1 are originated from the ethergroups (C–O–C) and C–H stretching, respectively, which are char-acteristic peaks of PEG [12,33]. Therefore, the incorporation of PEGcomponents in the blend membranes is clearly evidenced.

As the detection depth of IR is around 0.5–5 μm [34], it mainlyprovides information of the membrane bulk. Therefore, to obtainthe information near the very top surface of the membrane, weused X-ray photoelectron spectroscopy (XPS) to analyze blendmembranes with different BCP percentages as XPS is believed todetect the material surface with a depth of around 5–10 nm [35].As S element exists only in the PSf but not in PEG, blending PSf-b-PEG with PSf reduces the O/S ratio and higher contents of BCP leadto lower O/S ratio. The averaged O/S ratios of the blend mem-branes can be easily obtained by considering the content of PSf inthe BCP and also the percentages of the BCP in the blend mem-branes. The calculated O/S ratios are obtained based on the as-sumption that the two polymer components are homogeneously

Table 1The bulk and surface O/S ratios of membranes with various percentages of PSf-b-PEG.

Sample Bulk Surface

Neat PSf membrane 4.00 8.47Blend membrane with 40% PSf-b-PEG 5.91 9.69Blend membrane with 70% PSf-b-PEG 11.56 12.6

Y. Chen et al. / Journal of Membrane Science 505 (2016) 53–6056

mixed throughout the entire thickness of the membranes. As givenin Table 1, the O/S ratio increases from 4.00 to 5.91 and 11.56 asthe BCP percentage is increased from none to 40% and 70%, re-spectively, because of the dilution effect of the BCP to the contentof S in the blend membranes. In contrast, the O/S ratios on themembrane surfaces can be directly obtained by comparing thecontent of O and S as measured by XPS analysis. The O/S ratio onthe surface of the neat PSf membrane is 8.47, which is much higherthan the bulk value (4.00). Such an enrichment of O element onthe surface of the neat PSf membrane should be ascribed to theoxygen-containing contaminants adsorbed on the sample surfacewhich is detectable in highly sensitive XPS analyses [36]. For theblend membranes with 40% and 70% BCP percentages, the surfaceO/S ratios increase to 9.69 and 12.60, respectively, evidently higherthan that of the neat PSf membrane. The increased surface O/Sratios imply that the PEG chains preferentially segregate onto themembrane surface.

The preferential segregation of PEG chains on the membranesurface suggests an improved surface hydrophilicity as PEG ishighly hydrophilic. We also tested the water contact angles (CAs)of the membranes to reveal the change of surface hydrophilicitywith the incorporation of PSf-b-PEG. As shown in Fig. 2, there is aslight decreasing trend in the change of CAs of the membranesurfaces with the percentages of PSf-b-PEG and the CA decreasesfrom more than 70° to ca. 55° when the BCP percentages increasesfrom 0% to 70%, indicating that the hydrophilicity of the blendmembranes increase with dosing of PSf-b-PEG. We noticed thatCAs did not follow a perfect linear decreasing trend with the in-crease of copolymer dosages. The fluctuation in WCAs occurs be-cause CAs are the macroscopic representation of the surfacechemistry and roughness of the membrane which may not changein the same direction with rising copolymer dosages. The de-creasing degree of CAs is less pronounced compared to otherstudies [19], however, it does not necessarily imply a weaker effectin improving the membrane permeances, which will be discussedlater.

Fig. 2. Water contact angles of blend membranes with various BCP percentages.

3.2. Effect of BCP percentages on membrane microstructures andperformances

We then investigate the microstructures of the blend mem-branes. As shown in Fig. 3, the membranes produced from differ-ent percentages of PSf-b-PEG all exhibit an asymmetric structurewith a denser skin at top and a porous sublayer. The sublayershave finger-like cavities as well as macrovoid structure. The for-mation of such a structure can be attributed to the fact that NMPhave high mutual affinity to water which is used as the nonsolventduring the phase inversion), which results in instantaneous de-mixing of the casting solutions [37]. Moreover, it is also evidentthat that the size of the finger-like pore enlarged with the increaseof the BCP percentage. In addition, the thickness of the top skinlayer also increases with the BCP percentages, which becomesmore pronounced when the BCP percentage is larger than 40%(Fig. 3f–h). The changes in the membrane microstructure arecaused by the evolution of the subtle interactions among thepolymer/solvent/nonsolvent system initiated by the incorporationof hydrophilic PEG components in the casting solution, thuschanging the precipitation rate of the polymers.

The cross sections of the blend membranes are predominantlyoccupied by finger-like pores, suggesting that instantaneous de-mixing is still the main precipitation mechanism for casting solu-tions with various BCP percentages. However, the incorporation ofPSf-b-PEG, especially at higher percentages, reduces the differencein chemical potentials between the casting solutions (PSf, PSf-b-PEG, and NMP) and the precipitation bath (water) as PEG blockshave a high affinity toward the nonsolvent, water. Consequently,similar to adding minor amounts of solvent into the precipitationbath [38], the precipitation rates of the blend solutions slow downcompared to casting solutions containing no PSf-b-PEG. In otherwords, there is a tendency of transition from instantaneous de-mixing to delayed demixing with increasing BCP percentages in thecasting solutions. For example, in the extreme case with the castingsolution containing only PSf-b-PEG BCP and no PSf homopolymerthe produced membranes are totally dense in which all the finger-like pores and macrovoids are completely suppressed as a result ofthe delayed demixing dominating the phase separation process. Forthe casting solutions containing no more than 70% BCP, demixingoccurs in a somewhat retarded way, resulting in a relatively denseand thick skin layer [37], and higher BCP percentages retard thedemixing to larger extents, thus producing thicker skin layers.Thicker skin layers slows down the solvent exchange [39]. Pro-longed solvent exchange through the skin formed when the castingsolution is immersed in water allows longer duration for the poly-mer-lean phases to grow and coalesce, thus forming larger finger-like pores [17,40]. That explains why the membranes with higherBCP percentages have large pore sizes.

As the thickness of the skin layer plays an important role indetermining the permeability of the membranes, we have a closerexamination on the thickness of the skin layers and correlate itwith the membrane permeance. We measured the pure water fluxand BSA rejection of membranes with various BCP percentages. Asclearly shown in Fig. 4, the flux dramatically increases with BCPpercentages up to 40% and then decreases sharply with furtherrising BCP percentages to 70%. The flux goes from 37 L/(m2 h bar)for the neat PSf membrane without dosing of the BCP to ap-proximately 400 L/(m2 h bar) for the blend membrane with 40%BCP percentage, and drops back to about 54 L/(m2 h bar) at theBCP percentage of 70%. In contrast to the significant changes ofpermeance, the rejection to BSA of these membranes maintains ata stable high level of around 96% regardless of the BCPpercentages.

Such changes of the permeances are the synergetic result of thehydrophilicity and microstructure of the membranes induced by

Fig. 3. The cross-sectional SEM images of the blend membranes with various BCP percentages: (a) 0%, (b) 10%, (c) 20%, (d) 30%, (e) 40%, (f) 50%, (g) 60%, (h) 70%.

Fig. 4. Pure water permeability and BSA retention of the blend membranes withvarious BCP percentages.

Fig. 5. The thickness of the skin layer of blend membranes with various BCPpercentages.

Y. Chen et al. / Journal of Membrane Science 505 (2016) 53–60 57

the incorporation of PSf-b-PEG into the casting solutions. We firstdiscuss the effect of membrane microstructure. No pores can bedetected under SEM on the top surface of all the blend membranesinvestigated here, which explains their high rejections to BSA.Moreover, similar rejections also imply that their pore sizes andporosities do not substantially change with BCP percentages, andtherefore, the thickness of the skin layers having similar porousproperties determines the permeance of the membranes. Asshown in Fig. 5, the thickness of the skin layers, which can beeasily determined from the SEM images as there are clearboundaries between the dense layers and the macroporous sub-layers of the membranes, maintains stable at �0.30 μm at the BCPpercentages up to 40%. The thickness starts to significantly

increase at BCP percentages higher than 40% and reaches to�2.55 μm at the BCP percentages of 70%, which is nearly ninetimes thicker than that of membranes with lower BCP percentages.Thicker skin layers pose increasing flow resistance and conse-quently lead to smaller permeances. Therefore, if we only considerthe change of thickness of the skin layers, the incorporation of PSf-b-PEG has no or little influence on water permeance at low BCPpercentages (o40%) but this influence becomes significantly ne-gative at BCP percentages larger than 40%. However, microphaseseparation of macroscopically miscible blends of PSf/PSf-b-PEG

Table 2Comparison of performances of the blend membrane prepared with 40% BCPpercentages with that of other membranes previously reported.

Additives PWP (L/(m2 h bar) BSA rejection(%)

FRR (%) Ref

NPPCSa 115 94.0 74.0 [44]PANIb 115 98.0 78.5 [21]PBIc 178 68.7 93.5 [19]TiO2 nanoparticles 31 93.0 68.0 [45]PSF-b-PEG 396 96.0 85.0 This

study

a N-propylphosphonic chitosan.b polyaniline nanoparticles.c poly[2,2′-(m-phenylene)-5,5′-dibenzimidazole].

Y. Chen et al. / Journal of Membrane Science 505 (2016) 53–6058

will take place and form PEG microdomains dispersed in PSf ma-trix because of the strong repulsion between PSf and PEG as re-vealed by the large difference in the polar contribution in theirsolubility parameters [41]. As it has been demonstrated that PEGmicrodomains are permeable to water molecules because of thehigh water affinity of PEG chains [42], the PEG microdomains areexpected to serve as additional transport paths for water mole-cules and enhance water permeance. However, this effect oftransport paths is only valid for PEG microdomains connectingpores in the skin layer as PEG microdomains completely envelopedin the membrane skeleton will not allow accessibility of watermolecules. According to the phase diagram of block copolymers[43], the sizes of the PEG microdomains will be progressively in-creased and develop from spheres to cylinders with the increase ofthe volume fractions of PEG blocks in the blend. At a PSf-b-PEGpercentage of 40% in the blend, the volume fraction of PEG can beroughly estimated to be �20%, which produce a cylindrical mor-phology of the PEG microdomains. Clearly, PEG microdomainswith larger sizes or high aspect ratios will have more chances toconnect membrane pores, therefore, membranes with higher PSf-b-PEG percentages will produce more water-permeable PEG mi-crodomains, and consequently exhibit increasing water perme-ability. In addition, the increased surface hydrophilicity with risingBCP percentages should also contribute to the enhanced the waterpermeability but its contribution should be very limited asmembrane hydrophilicity is only slightly increased with BCP per-centages as discussed in Fig. 2. The two competing effects of thethickness of skin layer and the water permeability of PEG micro-domains eventually lead to firstly increased and then droppedwater permeance with increasing BCP percentages and the max-imal permeance is obtained at the BCP percentage of 40%.

The fact of that membrane surface hydrophilicity increaseswith BCP percentages implies that the blend membranes will ex-hibit stronger fouling resistance. We tested the flux recovery ratio(FRR) of the membranes after protein filtration. As shown in Fig. 6,the FRR rises evidently from 71% to 79% and 85% when the BCPpercentage increases from none to 10% and 20%, respectively. Thenthe FRR keeps stable around 85% with the BCP percentages furtherincrease to 50%. Such a change of FRRs with BCP percentages canbe readily understood as initially the increasing BCP percentagesprogressively enhance the surface hydrophilicity of the mem-branes which consequently adsorb gradually less proteins. There-fore, the FRRs keep increasing with the BCP percentages not largerthan 20%. However, further increasing BCP percentages may notsubstantially further reduce the protein adsorption, therefore, the

Fig. 6. Flux recovery ratio of blend membranes with different BCP percentages.

fouling resistance and consequently the FRRs level off.As there are a large number of works reporting the modifica-

tion of PSf UF membranes using a diversity of additives, wecompare the performances of our blend membranes with 40% BCPpercentages with those of previous works. As listed in Table 2, therejection and FRR of our membrane is among the best while thepermeance is the highest, which is approximately 2–10 timeshigher than that of others. In addition, the blend membranes alsoshow excellent pressure resistance. As shown in Fig. 7, the purewater flux almost linearly increases with the trans-membranepressure in the range of 0.02–0.2 MPa. The result suggests that theincorporation of PSf-b-PEG into the membrane does not weakenthe structure of membranes at all although finger-like pores in thesublayer of the membrane is enlarged.

3.3. Comparison between PSf-b-PEG copolymer and PEG homo-polymer in blend membranes

To understand the role of PSf-b-PEG in the casting solutions tochange the microstructure and performances of the PSf mem-branes, we also prepared PSf membranes with PEG homopolymeras the blend components and compared the two membranes, thatis, the PSf/PSf-b-PEG membrane and the PSf/PEG membrane, withthe same content (3.2 wt% in the entire casting solutions) of PEGchains in the casting solutions. Both membranes exhibit similarsurface and cross-sectional morphology as examined by SEMprobing. We also investigated the permeation properties of bothmembranes. As can be seen from Fig. 8, the initial sharp decline influx for both nascent membranes can be observed due to

Fig. 7. The pressure dependent pure water flux of the blend membrane with 40%BCP percentages.

Fig. 8. The time-dependent pure water permeability of the PSf/PSf-b-PEG and PSf/PEG membranes.

Fig. 9. The PEG percentage in membranes of PSf/PSf-b-PEG and PSf/PEG after im-mersion in DI water for 24 h.

Y. Chen et al. / Journal of Membrane Science 505 (2016) 53–60 59

compaction, which is very common for polymeric membranes[16,33] After around 15 min the flux becomes steady for the PSf/PSf-b-PEG membranes and it remains almost constant thereafter.However, in contrast, for the PSf/PEG membranes the flux keepsdeclining for a much longer duration as long as 60 min, indicatingits worse compaction resistance. Moreover, the steady flux of thePSf/PEG membranes is more than 50% lower than that of PSf/PSf-b-PEG membranes, as also can be determined from Fig. 8. We notethat the PSf/PEG membrane either does not show a higher BSArejection (95%) than the PSf/PSf-b-PEG membrane (96%) althoughit has a much lower permeance.

As mentioned above, the PEG homopolymer might move to thecoagulation bath during the phase inversion process and even willbe drained away when operated in water treatment. The purposeto incorporation PSf-b-PEG rather than PEG homopolymer in thecasting solutions is to prepare membranes with long-lasting andstable structure and surface chemistry as the BCP will not bewashed out. To investigate the possible washing out of PEGhomopolymer, we prepared a set of membranes blended with PEGhomopolymer and the contents of PEG in the casting solutionswere set to be equal to the contents of PEG blocks in the castingsolutions for the preparation of membranes blended with PSf-b-PEG. The nascent membranes were separately immersed in deio-nized water for one day and then we checked the relative contentof PEG chains in the membrane by IR analysis. The relative contentof PEG chains, either in the form of PEG homopolymer or PEGblocks, can be obtained by comparing the intense of the IR peak at946 cm–1 and the one at 1580 cm–1 (H946/H1580) as they arecharacteristic peaks for PEG and PSF, respectively. It is expectedthat higher content of PEG chains in casting solutions will lead tocorrespondingly higher content of PEG in the formed membrane ifno leaching out occurs. As shown in Fig. 9, there is a significantincrease of H946/H1580 with the PEG content increasing from 0.8%to 3.2% and to 5.6% for PSf/PSf-b-PEG membranes which corre-spond to BCP percentages of 10%, 40%, and 70%, respectively.However, for PSf/PEG membranes H946/H1580 maintains almoststeady when the PEG contents in the casting solutions increase byfour or seven folds, suggesting that the PEG homopolymers keep anearly constant content regardless their initial content in thecasting solutions. Moreover, the H946/H1580 values of the PSf/PSf-b-PEG membranes are always much higher than that of the PSf/PEGmembranes with the same content of PEG chains in the castingsolutions. Both observations indicate that a significant portion ofthe PEG homopolymers added into the casting solutions has beenwashed during the membrane formation process and subsequentusage in water because of the highly water solubility of PEG

homopolymers. However, in vivid contrast, as PSf-b-PEG macro-molecules cannot be dissolved in water they will survive the phaseinversion process and be permanently locked in the membraneskeleton and no leaking out in applications will occur for it, thusrendering long-lasting improvement to the membrane perfor-mances. Therefore, we conclude that the superior performances ofthe blend membranes are originated from the amphiphilic natureof PSf-b-PEG where the PEG blocks are permanently residing onthe pore walls through the anchoring effect of PSf blocks en-tangled with the PSf base forming the membrane skeleton.

4. Conclusions

Amphiphilic PSf-b-PEG BCP is blended with PSf homopolymerbase material at dosages significantly higher than traditional usesto prepare PSf UF membranes with upgraded performances by thenonsolvent-induced phase separation technique. PSf homo-polymer is miscible with PSf-b-PEG BCP at arbitrary blend ratio,and the PEG blocks are preferentially segregated to membranesurfaces as revealed by FTIR and XPS analysis. In addition to theeffect of surface segregation, heavily dosed amphiphilic PSf-b-PEGBCPs influence membrane microstructures and performances byforming water-permeable PEG microdomains. The BCP percentageof 40% gives the highest permeance up to 400 L/(m2 h bar), whichis two to ten times higher than PSf UF membranes prepared inother works. Moreover, the rejection and resistance to proteinfouling of that membrane are also excellent compared to that ofothers. Parallel tests on membranes blended with PEG homo-polymer and PSf-b-PEG BCP reveal that PEG homopolymer isreadily leached out in water during the membrane formation andapplication whereas PSf-b-PEG BCP is permanently incorporated inthe membrane skeleton. Therefore, blend membranes with PSf-b-PEG exhibit long-lasting and superior performances in multipleaspects including permeance, selectivity, fouling resistance, etc.

Acknowledgements

Financial support from the National Basic Research Program ofChina (2015CB655301), the Natural Science Foundation of JiangsuProvince (BK20150063), and the Project of Priority AcademicProgram Development of Jiangsu Higher Education Institutions(PAPD) is gratefully acknowledged.

Y. Chen et al. / Journal of Membrane Science 505 (2016) 53–6060

References

[1] J.Y. Park, M.H. Acar, A. Akthakul, W. Kuhlman, A.M. Mayes, Polysulfone-graft-poly(ethylene glycol) graft copolymers for surface modification of polysulfonemembranes, Biomaterials 27 (2006) 856–865.

[2] R. Rezaee, S. Nasseri, A. Mahvi, R. Nabizadeh, S. Mousavi, A. Rashidi, A. Jafari,S. Nazmara, Fabrication and characterization of a polysulfone-graphene oxidenanocomposite membrane for arsenate rejection from water, J. Environ.Health Sci. Eng. 13 (2015) 1–11.

[3] A. Chabeaud, L. Vandanjon, P. Bourseau, P. Jaouen, F. Guérard, Fractionation byultrafiltration of a saithe protein hydrolysate (Pollachius virens): effect ofmaterial and molecular weight cut-off on the membrane performances, J. FoodEng. 91 (2009) 408–414.

[4] B. Madsen, D.W. Britt, C.-H. Ho, M. Henrie, C. Ford, E. Stroup, B. Maltby,D. Olmstead, M. Andersen, Hemodialysis membrane surface chemistry as abarrier to lipopolysaccharide transfer, J. Appl. Polym. Sci. 132 (2015).

[5] S. Inukai, R. Cruz-Silva, J. Ortiz-Medina, A. Morelos-Gomez, K. Takeuchi,T. Hayashi, A. Tanioka, T. Araki, S. Tejima, T. Noguchi, M. Terrones, M. Endo,High-performance multi-functional reverse osmosis membranes obtained bycarbon nanotube �polyamide nanocomposite, Sci. Rep. 5 (2015) 13562.

[6] K. Zodrow, L. Brunet, S. Mahendra, D. Li, A. Zhang, Q. Li, P.J.J. Alvarez, Poly-sulfone ultrafiltration membranes impregnated with silver nanoparticlesshow improved biofouling resistance and virus removal, Water Res. 43 (2009)715–723.

[7] D. Rana, T. Matsuura, Surface modifications for antifouling membranes, Chem.Rev. 110 (2010) 2448–2471.

[8] V. Kochkodan, N. Hilal, A comprehensive review on surface modified polymermembranes for biofouling mitigation, Desalination 356 (2015) 187–207.

[9] Y. Yang, H. Zhang, P. Wang, Q. Zheng, J. Li, The influence of nano-sized TiO2

fillers on the morphologies and properties of PSF UF membrane, J. Membr. Sci.288 (2007) 231–238.

[10] S.R. Panda, S. De, Preparation, characterization and performance of ZnCl2 in-corporated polysulfone (PSF)/polyethylene glycol (PEG) blend low pressurenanofiltration membranes, Desalination 347 (2014) 52–65.

[11] M.B.O. Riekerink, G.H.M. Engbers, M. Wessling, J. Feijen, Tailoring the Prop-erties of Asymmetric Cellulose Acetate Membranes by Gas Plasma Etching, J.Colloid Interface Sci. 245 (2002) 338–348.

[12] Y.-Q. Song, J. Sheng, M. Wei, X.-B. Yuan, Surface modification of polysulfonemembranes by low-temperature plasma–graft poly(ethylene glycol) ontopolysulfone membranes, J. Appl. Polym. Sci. 78 (2000) 979–985.

[13] J. Deng, L. Wang, L. Liu, W. Yang, Developments and new applications of UV-induced surface graft polymerizations, Prog. Polym. Sci. 34 (2009) 156–193.

[14] H. Yu, Y. Cao, G. Kang, J. Liu, M. Li, Q. Yuan, Enhancing antifouling property ofpolysulfone ultrafiltration membrane by grafting zwitterionic copolymer viaUV-initiated polymerization, J. Membr. Sci. 342 (2009) 6–13.

[15] J.-H. Kim, K.-H. Lee, Effect of PEG additive on membrane formation by phaseinversion, J. Membr. Sci. 138 (1998) 153–163.

[16] S. Zhao, Z. Wang, X. Wei, B. Zhao, J. Wang, S. Yang, S. Wang, PerformanceImprovement of Polysulfone Ultrafiltration Membrane Using Well-DispersedPolyaniline–Poly(vinylpyrrolidone) Nanocomposite as the Additive, Ind. Eng.Chem. Res. 51 (2012) 4661–4672.

[17] Y. Ma, F. Shi, J. Ma, M. Wu, J. Zhang, C. Gao, Effect of PEG additive on themorphology and performance of polysulfone ultrafiltration membranes, De-salination 272 (2011) 51–58.

[18] A. Higuchi, K. Shirano, M. Harashima, B.O. Yoon, M. Hara, M. Hattori,K. Imamura, Chemically modified polysulfone hollow fibers with vinylpyrro-lidone having improved blood compatibility, Biomaterials 23 (2002)2659–2666.

[19] E. Eren, A. Sarihan, B. Eren, H. Gumus, F.O. Kocak, Preparation, characterizationand performance enhancement of polysulfone ultrafiltration membrane usingPBI as hydrophilic modifier, J. Membr. Sci. 475 (2015) 1–8.

[20] Z. Fan, Z. Wang, M. Duan, J. Wang, S. Wang, Preparation and characterizationof polyaniline/polysulfone nanocomposite ultrafiltration membrane, J. Membr.Sci. 310 (2008) 402–408.

[21] Z. Fan, Z. Wang, N. Sun, J. Wang, S. Wang, Performance improvement ofpolysulfone ultrafiltration membrane by blending with polyaniline nanofibers,J. Membr. Sci. 320 (2008) 363–371.

[22] G.R. Guillen, T.P. Farrell, R.B. Kaner, E.M.V. Hoek, Pore-structure, hydrophilicity,and particle filtration characteristics of polyaniline-polysulfone ultrafiltrationmembranes, J. Mater. Chem. 20 (2010) 4621–4628.

[23] B. Chakrabarty, A.K. Ghoshal, M.K. Purkait, Effect of molecular weight of PEG

on membrane morphology and transport properties, J. Membr. Sci. 309 (2008)209–221.

[24] S. Rekha Panda, S. De, Effects of polymer molecular weight, concentration, androle of polyethylene glycol as additive on polyacrylonitrile homopolymermembranes, Polym. Eng. Sci. 54 (2014) 2375–2391.

[25] A. Idris, N. Mat Zain, M.Y. Noordin, Synthesis, characterization and perfor-mance of asymmetric polyethersulfone (PES) ultrafiltration membranes withpolyethylene glycol of different molecular weights as additives, Desalination207 (2007) 324–339.

[26] S. Zhao, Z. Wang, J. Wang, S. Yang, S. Wang, PSf/PANI nanocomposite mem-brane prepared by in situ blending of PSf and PANI/NMP, J. Membr. Sci. 376(2011) 83–95.

[27] Y.-Q. Wang, T. Wang, Y.-L. Su, F.-B. Peng, H. Wu, Z.-Y. Jiang, Remarkable Re-duction of Irreversible Fouling and Improvement of the Permeation Propertiesof Poly(ether sulfone) Ultrafiltration Membranes by Blending with PluronicF127, Langmuir 21 (2005) 11856–11862.

[28] L.F. Hancock, S.M. Fagan, M.S. Ziolo, Hydrophilic, semipermeable membranesfabricated with poly(ethylene oxide)–polysulfone block copolymer, Bioma-terials 21 (2000) 725–733.

[29] L.F. Hancock, Phase inversion membranes with an organized surface structurefrom mixtures of polysulfone and polysulfone—poly(ethylene oxide) blockcopolymers, J. Appl. Polym. Sci. 66 (1997) 1353–1358.

[30] Y.-F. Zhao, L.-P. Zhu, Z. Yi, B.-K. Zhu, Y.-Y. Xu, Improving the hydrophilicity andfouling-resistance of polysulfone ultrafiltration membranes via surface zwit-terionicalization mediated by polysulfone-based triblock copolymer additive,J. Membr. Sci. 440 (2013) 40–47.

[31] W. Chen, Y. Su, J. Peng, Y. Dong, X. Zhao, Z. Jiang, Engineering a Robust, Ver-satile Amphiphilic Membrane Surface Through Forced Surface Segregation forUltralow Flux-Decline, Adv. Funct. Mater. 21 (2011) 191–198.

[32] W. Li, M. Yuan, M. Yang, Dual-phase polymer electrolyte with enhanced phasecompatibility based on Poly(MMA-g-PVC)/PMMA, Eur. Polym. J. 42 (2006)1396–1402.

[33] M.K. Sinha, M.K. Purkait, Preparation and characterization of novel pegylatedhydrophilic pH responsive polysulfone ultrafiltration membrane, J. Membr.Sci. 464 (2014) 20–32.

[34] S.G. Kazarian, K.L. Chan, ATR-FTIR spectroscopic imaging: recent advances andapplications to biological systems, Analyst 138 (2013) 1940–1951.

[35] B.K. Kuila, M.S. Rama, M. Stamm, Supramolecular assembly of Poly(styrene)-b-poly(4-vinylpyridine) and ferroceneacetic acid: an easy way to large-scalecontrollable periodic arrays of iron oxide nanomaterials, Adv. Mater. 23 (2011)1797–1800.

[36] L.-S. Johansson, J.M. Campbell, K. Koljonen, P. Stenius, Evaluation of surfacelignin on cellulose fibers with XPS, Appl. Surf. Sci. 144–145 (1999) 92–95.

[37] B. Chakrabarty, A.K. Ghoshal, M.K. Purkait, SEM analysis and gas permeabilitytest to characterize polysulfone membrane prepared with polyethylene glycolas additive, J. Colloid Interface Sci. 320 (2008) 245–253.

[38] Y.-H. Zhao, B.-K. Zhu, X.-T. Ma, Y.-Y. Xu, Porous membranes modified by hy-perbranched polymers: I. Preparation and characterization of PVDF membraneusing hyperbranched polyglycerol as additive, J. Membr. Sci. 290 (2007)222–229.

[39] Q.-Z. Zheng, P. Wang, Y.-N. Yang, Rheological and thermodynamic variation inpolysulfone solution by PEG introduction and its effect on kinetics of mem-brane formation via phase-inversion process, J. Membr. Sci. 279 (2006)230–237.

[40] A.J. Reuvers, C.A. Smolders, Formation of membranes by means of immersionprecipitation, J. Membr. Sci. 34 (1987) 67–86..

[41] J. Brandrup, E.H. Immergut, E.A. Grulke, A. Abe, D.R. Bloch, Polymer Handbook,4th ed., John Wiley & Sons, New York, 1999.

[42] T. Yamamoto, T. Kimura, M. Komura, Y. Suzuki, T. Iyoda, S. Asaoka,H. Nakanishi, Block copolymer permeable membrane with visualized high-density straight channels of poly(ethylene oxide), Adv. Funct. Mater. 21 (2011)918–926.

[43] F.S. Bates, A.G.H. Fredrickson, Block copolymer thermodynamics: theory andexperiment, Rev. Phys. Chem 41 (1990) 525–557.

[44] R. Kumar, A.M. Isloor, A.F. Ismail, T. Matsuura, Synthesis and characterizationof novel water soluble derivative of Chitosan as an additive for polysulfoneultrafiltration membrane, J. Membr. Sci. 440 (2013) 140–147.

[45] G. Zhang, S. Lu, L. Zhang, Q. Meng, C. Shen, J. Zhang, Novel polysulfone hybridultrafiltration membrane prepared with TiO2-g-HEMA and its antifoulingcharacteristics, J. Membr. Sci. 436 (2013) 163–173.