High performance polyethersulfone microfiltration membranes having high flux and stable hydrophilic property

Page 1

Hh

Ha

b

a

ARRAA

KMPPgPL

1

iwi[mivsmbol

cite

0d

Journal of Membrane Science 342 (2009) 153–164

Contents lists available at ScienceDirect

Journal of Membrane Science

journa l homepage: www.e lsev ier .com/ locate /memsci

igh performance polyethersulfone microfiltration membranesaving high flux and stable hydrophilic property

eru Susanto a,b, Nico Stahra a, Mathias Ulbricht a,∗

Lehrstuhl für Technische Chemie II, Universität Duisburg-Essen, 45117 Essen, GermanyDepartment of Chemical Engineering, Universitas Diponegoro, Samarang, Indonesia

r t i c l e i n f o

rticle history:eceived 21 March 2009eceived in revised form 7 May 2009ccepted 21 June 2009vailable online 27 June 2009

eywords:icrofiltration membrane

luronic®

a b s t r a c t

High performance polyethersulfone microfiltration membranes with a high flux and stable hydrophiliccharacter were successfully prepared by vapor induced phase separation coupled with non-solventinduced phase separation method. Pluronic®, a poly(ethylene oxide)-b-poly(propylene oxide)-b-poly(ethylene oxide) triblock copolymer and triethylene glycol (TEG) were used as hydrophilicitymodification agent and non-solvent, respectively. Casting solution composition (non-solvent and additivecontent) and process condition (exposure time before coagulation and relative humidity) were optimizedto obtain these membranes. The resulting membrane characteristics, which include hydraulic perme-ability, pore size distribution, surface hydrophilicity, chemical composition and membrane morphology,

oly(ethylene glycol)-b-poly(propylenelycol)-b-poly(ethylene glycol)olyethersulfoneow fouling membrane

were investigated. The membrane performance was examined by investigation of adsorptive fouling andmicrofiltration using solutions of bovine serum albumin as the model system. The results suggest thatthe non-solvent content (TEG) and the exposure time were the most critical parameters to obtain high-flux membranes, whereas the concentration of Pluronic was important to obtain hydrophilic property.The increase in resistance towards both adsorptive and microfiltration fouling was clearly observed. Thesurface chemistry and wettability of the resulting membranes did not change after incubating in water

days.

(40 ◦C) for a period of 10

. Introduction

Microfiltration (MF) has increasingly been accepted as promis-ng technology in a wide range of applications including water and

astewater treatments, dairy, biotechnological and pharmaceuticalndustries, food and beverage processing, and medical applications1–4]. Along with its many successes, significant loss of perfor-

ance with respect to flux and often selectivity attributed to foulings recognized as the biggest problem. This limitation not only pre-ents a more widespread commercial applicability of MF but alsohortens the membrane life due to chemical cleaning. Althoughany methods such as adjusting hydrodynamic conditions have

een proposed to overcome the problem of fouling [5,6], the heartf MF processes is the membrane itself. Therefore, preparation of

ow fouling MF membranes is very important.Beside mechanical, chemical and thermal stabilities, important

haracteristics for high performance MF membranes are high fluxn combination with desired selectivity and low fouling. Because ofheir mechanical strength, thermal and chemical stability as well asxcellent film forming properties, sulfone polymers, e.g., polyether-

∗ Corresponding author. Tel.: +49 201 1833151; fax: +49 201 1833147.E-mail address: [email protected] (M. Ulbricht).

376-7388/$ – see front matter © 2009 Elsevier B.V. All rights reserved.oi:10.1016/j.memsci.2009.06.035

© 2009 Elsevier B.V. All rights reserved.

sulfone (PES), have been used very often for the fabrication of highperformance commercial MF membranes. High flux MF membranescan be made from PES but the main limitation of this material is thatit is prone to foul due to its relatively hydrophobic character. Toincrease the performance, hydrophilization by blending PES withhydrophilic additive during preparation via non-solvent inducedphase separation (NIPS) is of very high relevance from practicalpoint of view. Unfortunately, this blending method causes signifi-cant changes in composition and properties of the casting solution,leading to different membrane structure. As a consequence, themembrane properties can be quite different from the unmodifiedreference material. Thus, introducing a hydrophilic modificationagent into casting solutions of PES and preserving the desired porestructure and high flux of the resulting membranes are of greatinterest.

In general, characteristics of polymeric membranes preparedby NIPS are influenced by the casting solution properties, the sol-vent and non-solvent system, the additive used, the coagulationbath characteristics and the exposure time and conditions for the

“proto-membrane” before precipitation. Synergistic effects can beobtained via engineering those variables but contradictory effectsare also possible. Membranes made via conventional NIPS have usu-ally abrupt asymmetric structure resulting in low flux. Many effortshave been made to manufacture high flux MF membranes. Common

Page 2

1 mbran

sia[

mamcttmnpipciwtmImmhfacflopiaundcflntSic2nsapsiabavrbfibbf

biAbp

54 H. Susanto et al. / Journal of Me

trategies include addition of non-solvent and pore forming agentn the casting solution, addition of solvent in the coagulation bathnd controlled exposure of the proto-membrane before coagulation7].

Preparation conditions for integral highly asymmetric PES MFembranes having high flux have been proposed by Wrasidlo [8]

nd Zepf [9]. The membranes were prepared by dissolving theembrane polymer in solvent/non-solvent systems followed by

asting and coagulating in non-solvent. In this regard, the type andhe concentration of non-solvent in the solvent/non-solvent sys-em are very important, in particular for the porosity as well as the

echanical strength of the resulting membrane. The presence ofon-solvent results in thermodynamically unstable condition of theolymer solution leading to rapid precipitation during coagulation

n the non-solvent. Eventually, high overall porosity and a macro-orous skinless structure at the interface (asymmetric structure)an be obtained. It should be noted that the asymmetric structuren this context is different from that of asymmetric UF membranes,

hich show an abrupt discontinuity in pore structure between theop layer and the supporting layer. Zepf [9] had improved Wrasidlo’s

ethod by decreasing the temperature for casting and coagulation.n addition, they also reduced the exposure time to the environ-

ent between casting and coagulation. Thereby, the pore size in theembrane surface region could be increased. Greenwood et al. [10]

ad successfully prepared high flux skinless PES MF membranesrom a casting solution containing PES, a solvent and a glycol suchs triethylene glycol (TEG) as non-solvent. The precipitation bathontained water and TEG. Very regular pore structures and highuxes of the MF membranes could be obtained. However, the usef a huge amount of glycol in the coagulation medium makes therocess expensive and hard to be practically applied. Immersion

n non-solvent bath after humid air exposure could also producefinal morphology that is more open and has larger pores at thepper surface of the film [11]. Very recently, diethylene glycol ason-solvent was also used in combination with the solvent N,N-imethylacetamide for preparation of PES MF membranes via NIPSombined with vapor induced phase separation (VIPS) [12]. High-ux membranes could be obtained only at a high mass ratio ofon-solvent/solvent resulting in membranes with cellular pores onhe top surface and sponge-like structure over the cross-section.hin et al. [13] used 2-methoxyethanol (2-ME) as a non-solventn the polymer solution during PES MF membrane formation byombination of VIPS and NIPS. They claimed that the hydrophilic-ME could draw water vapor into the cast film to induce sponta-eous emulsification and initiate the formation of a cellular surfacetructure. The non-solvent content and the exposure time in humidir were the main variables to adjust the pore structure. The mor-hology of the resulting membranes changed from finger-like toponge-like structures as the concentration of non-solvent wasncreased. Nevertheless, the harmful properties of 2-ME may havenegative impact especially if it still remains in the polymer mem-rane. The use of water as non-solvent in the polymer solution couldlso be found in the literature [14]. However, because water is aery strong non-solvent, undesirable skin formation leading to fluxeduction is possible. In sum, high flux PES MF membranes coulde prepared by phase separation method with some solvent modi-cations as compared with standard NIPS protocols. Nevertheless,ecause no hydrophilic agent is incorporated, the resulting mem-rane is most probably still hydrophobic and eventually prone tooul.

Polyvinylpyrrolidone (PVP) and poly(ethylene glycol) (PEG) have

een intensively used as hydrophilic macromolecular additive dur-

ng preparation of PES membranes by phase separation methods.ddition of PVP could not only suppress the macrovoid formationut also enhance the membrane hydrophilicity [15–17]. Similarhenomenon was also observed after addition of PEG into cast-

e Science 342 (2009) 153–164

ing solution [14,18,19]. Kraus et al. [20,21] proposed a method forpreparation of hydrophilic PES membranes by addition of PEG intothe casting solution. Recently, Wang et al. invented methods forpreparation of highly asymmetric, hydrophilic polysulfone-basedMF membranes using PVP and PEG as hydrophilic modifier [22]. Ingeneral, previous studies showed that incorporation of hydrophilicmacromolecular agent into polymer solution is a simple techniqueto increase the hydrophilicity of PES MF membranes. Unfortunately,two limitations are observed: (i) incorporation can result in eitherisotropic structure or abrupt asymmetric structure, the latter lead-ing to low flux, (ii) the immobilization stability of modifying agentin the polymer membrane matrix is relatively low leading to a lossof the hydrophilic character. As reason for the latter effect, eitherextraction or chemical degradation of hydrophilic macromolecularadditive during cleaning process has been reported [23,24]. In orderto improve the stability, cross-linking structure of PVP through-out the bulk of the polysulfone membrane has been proposed [25].However, many steps are needed and this makes the process com-plex and difficult. Very recently, we investigated the stability ofthree macromolecular additives with similar molar mass, i.e., PEG,PVP and an amphiphilic triblock copolymer poly(ethylene glycol)-b-poly(propylene glycol)-b-poly(ethylene glycol) (Pluronic®), in thepolymer matrix of PES UF membranes prepared under identicalNIPS conditions [26]. The results showed that Pluronic was themost stable macromolecular additive and made the PES membranesmore hydrophilic. The successful application of Pluronic as surfacemodifier for PES UF membranes had also been reported by the groupof Jiang and co-workers [27,28].

The objective of this study was to prepare high flux PES MFmembranes with incorporating of Pluronic into the membranematrix as stable hydrophilic agent. In particular, we investigatedthe effect of non-solvent content, exposure time in humid air, andPluronic concentration on the resulting membrane characteris-tics and performances. Furthermore, we optimized the preparationconditions to obtain hydrophilic PES MF membranes with skin-less and sponge-like pore structure. The membranes should havea hydraulic permeability which is competitive with commer-cially available high performances MF membranes, i.e., beyond20,000 L m−2 h−1 bar−1.

2. Experimental

2.1. Materials

Commercial PES (Ultrason E 6020 P) donated by BASF (Lud-wigshafen, Germany) was used and dried at 120 ◦C for atleast 4 h before use. N-methyl-2-pyrrolidone (NMP) was pur-chased from Merck (Hohenbrunn, Germany). Pluronic® F127 (Plu)(MW ∼ 12,600 g/mol) and PE6400 (MW ∼ 2900 g/mol) were fromBASF (Mount Olive, NJ, USA) and BASF (Ludwigshafen, Germany),respectively (Pluronic® PE6400 is identical with Pluronic® L64 soldin USA). Triethylenglycol (TEG) was purchased from Arcos (Geel,Belgium). Bovine serum albumin (BSA) was purchased from ICNBiomedicals, Inc. (California, USA). Potassium dihydrogen phos-phate (KH2PO4) and disodium hydrogen phosphate dihydrate(Na2HPO4·2H2O) were purchased from Fluka Chemie AG (Buchs,Germany). Nitrogen gas purchased from Messer Griesheim GmbH(Krefeld, Germany) was ultrahigh purity. Water purified with aMilli-Q system from Millipore (Burlington, MA, USA) was used forall experiments.

2.2. Membrane preparation

First, solvent (NMP) and non-solvent (TEG) were mixed withstirring. PES (10 wt.%) and block copolymer additive (at certain con-

Page 3

H. Susanto et al. / Journal of Membran

coctpompbspcsatwWdm(twitceitfl

2p

crF1ifiap5ttgau

R

During static adsorption tests, a pre-weighed membrane with

Fig. 1. Chemical structures of the triblock copolymer additives used.

entration) were then dissolved in NMP–TEG mixture. Two typesf block copolymer with different molar masses and different PEGhain lengths were used (Fig. 1). Polymer solution without an addi-ive was also prepared for control experiments. The homogenousolymer solution was left without stirring until no bubbles werebserved. The viscosity of representative polymer solutions waseasured by using the rheometer Low Shear LS40 model from

roRheo (Althengstett, Germany). The membranes were preparedy using a computer-controlled casting/coagulation machine, con-tructed in our group and built in the university work shop. Theolymer solution was cast with a thickness of 200 �m using a steelasting knife on a glass substrate (casting speed 80 mm/s) and thenubjected to humid air (relative humidity, RH = 50–60 or 65–75%,djusted in a box over the casting line by regulating the flow ofhe air saturated with water vapor by passing it through four gasashing bottles containing water) for a specific time (∼0–5 min).e could not perform an experiment with RH higher than 75%

ue to limitations of the equipment used. Thereafter, the nascentembrane was solidified in a coagulation bath containing water

20 ◦C ± 1) for 1 h. It should be noted that ∼0 min exposure meanshat no additional exposure time was applied, i.e., the substrateith the cast film was not passed through the humidifier box before

t had been completely immersed in the coagulation bath. Becausehe morphology of the membrane is dependent on the membraneasting thickness [29], this parameter was kept constant for allxperiments. The resulting membranes were washed by immersingn a large excess of water for ∼24 h before drying. It should be men-ioned that all membranes including the samples used for waterux measurements were dried.

.3. Hydraulic permeability measurement and filtrationrocedures

All experiments were carried out by using a dead-end stirredell (Amicon cell model 8010, Millipore Corp.) connected to a feedeservoir (∼450 mL) and pressurized by nitrogen from a gas tank.irst, each membrane was compacted by filtration of pure water at–3 bar (depending on its flux) for about 0.5 h. Hydraulic permeabil-

ty was measured at different transmembrane pressures and at leastve measurements from different membrane samples were aver-ged. Microfiltration experiments at a constant transmembraneressure (0.2 bar) were conducted using a BSA solution (0.1 g/L, pHin phosphate buffer) as the feed. The balance was connected to

he PC to record online the weight of permeate and the flux washen calculated. The permeate flux profile over time was investi-ated. BSA concentrations were determined by measuring its UV

bsorbance at 280 nm. The apparent BSA rejection was calculatedsing the following equation:

(%) = 1 − Cdownstream

Cupstream× 100 (1)

e Science 342 (2009) 153–164 155

2.4. Membrane morphology

The top surface and cross-section morphology of the mem-branes were observed by using a Quanta 400 FEG (FEI)environmental scanning electron microscope at standard high-vacuum conditions. A K 550 sputter coater (Emitech, UK) was usedto coat the outer surface of the sample with gold/palladium. Forcross-section analysis, the membranes were broken in liquid nitro-gen and sputtered for 1.5 min, while for analysis of outer membranesurface, sputtering was done for 0.5 min.

2.5. Pore size distribution

The pore size distribution of the membrane and the averagepore size were determined by gas flow/liquid displacement methodusing the Capillary Flow Porometer CFP-34RTG8A-X-6-L4 (PMI Inc.,Ithaca, NY, USA). The membrane samples with diameter of 25 mmwere characterized by using the “dry up-wet up” method, i.e.,the gas flow was measured as a function of the transmembranepressure, first through a dry membrane and then after wettingthe membrane with 1,1,2,3,3,3-hexafluoropropene (“Galwick”, PMI;surface tension 16 dyn cm−1). The pore size distribution was esti-mated using the PMI software.

2.6. Contact angle (CA)

An optical contact angle measurement system (OCA 15 Plus, Dat-aphysics GmbH, Filderstadt, Germany) was used to measure CA ofmembranes. A static captive bubble method, which is preferredfor porous membrane surfaces, was chosen [30]. Membranes wereinverted (active layer to the bottom) in pure water at a temperatureof 21 + 1 ◦C. An air bubble (5–10 �L) was injected from a syringewith a stainless steel needle onto the sample surface under water.Multiple contact angle values were measured and average valueswere obtained from at least five bubbles at different locations onthe membrane surface.

2.7. Membrane chemical composition

The membrane surface chemistry was analyzed by attenuatedtotal reflection (ATR) Fourier transform infrared (FTIR) spec-troscopy, by using the instrument Varian 3100 Excalibur series(equipped with a MCT detector, Ge crystal, 60◦). A total of 64scans were performed at a resolution of 4 cm−1 and the tem-perature of 21 + 1 ◦C. The Varian’s Resolution Pro 4.0 was usedto record the membrane spectra versus the corresponding back-ground spectra. The content of Pluronic in the membrane polymermatrix was investigated by using NMR spectroscopy. The 1H NMRspectrum of a piece of dried membrane, dissolved in deuterateddimethyl sulfoxide (d-DMSO), was recorded at 300 MHz by usingthe instrument DRX 300 (Bruker). Because the peak area (deter-mined by integration method) is proportional to the number ofprotons, with the known number of protons in the repeating units(8 H for PES and 4 H for the PEG blocks in Pluronic), molar massof repeating units and composition of the block copolymer, themass content of Pluronic in the PES membrane can be deter-mined.

2.8. Protein adsorption

outer surface area of 3.14 cm2 was immersed in 5 mL BSA solu-tion (1 g/L, pH 5 in phosphate buffer) in a closed container for3 h. The membrane was then removed, rinsed with water, driedand weighed again. The concentration of protein solution after

Page 4

1 mbrane Science 342 (2009) 153–164

aom

2

m(dc(c

3

3i

stw0mt

wac(3aiaio2nsPe

Fphpp

Fig. 3. Water permeability as function of exposure time for the membranes preparedfrom a polymer solution of 10/30/60 wt.% (PES/NMP/TEG). The relative humidity

56 H. Susanto et al. / Journal of Me

dsorption test was also measured. Quantification of the amountf protein bound to the membrane was done by gravimetric andass balance calculation methods.

.9. Stability test

The stability of macromolecular additive in the membrane poly-er matrix was examined by incubating the membrane sample

∼3.80 cm2) in water (20 mL, 40 ◦C) in a closed container for up to 10ays. Water is usually used for membrane washing before chemicalleaning will be performed. Contact angle and surface chemistryby ATR-IR) were evaluated to investigate changes in membraneharacteristics.

. Results and discussion

.1. Effects of non-solvent content and exposure time withoutncorporating additive

First, the effects of non-solvent (TEG) content in the castingolution and of exposure time in the absence of Pluronic addi-ive were investigated. TEG contents of 0, 30, 45 and 60 wt.%ere applied, while exposure time was varied within the range of

–5 min. Figs. 2 and 3 show the water permeability of the resultingembranes as function of TEG content and exposure time, respec-

ively.As clearly seen in Fig. 2, the effect of TEG content on

ater permeability is dependent on exposure time to humidir. For preparation with immediate coagulation, no signifi-ant effect of TEG content on water permeability was observedall membranes had their water permeability within the range30–350 L m−2 h−1 bar−1). By contrast, when exposure to humidir was performed, a significant effect of TEG was observed. Thencrease in TEG content increased the hydraulic permeability, and

very strong increase was observed when the TEG content wasncreased from 45 to 60%. This phenomenon agrees well with thebservation of Shin et al. during preparation of PES membrane using

-ME as non-solvent [13]. Our observation showed that we couldot add TEG beyond ∼65 wt.% into the casting solution becauseuch solvent–non-solvent mixtures did not completely dissolve theES. Interestingly, similar behavior was observed for the effect ofxposure time (cf. Fig. 3). A significant effect of humid air exposure

ig. 2. Water permeability as function of TEG content for the membranes pre-ared from a polymer solution without an additive (Pluronic) and exposed toumid air with RH = 50–60% for ∼0 and 1 min (∼0 min exposure means that theroto-membrane after casting was immediately immersed in coagulation bath). Theolymer solution with 0% of TEG was 10% of PES and 90% of NMP.

was 50–60% (∼0 min exposure means that the nascent membrane after casting wasimmediately immersed in coagulation bath). The polymer solution with 0% of TEGwas composed of 10% of PES and 90% of NMP.

time on hydraulic permeability was found only if TEG was presentin the casting solution. Strong increase in permeability was foundwithin the range ∼0 to 1 min exposure. Beyond 1 min exposure, theincrease in permeability could still be observed but it showed atrend to level off. Because too much shrinkage of the membraneswas observed, we do not include the results for exposure timesbeyond 5 min. In general, in order to obtain PES membranes havinghigh hydraulic permeability from the PES/NMP/TEG system, bothnon-solvent content (TEG) and long enough humid air exposuretime should be applied. Exposure to humid air is aimed to bringthe proto-membrane in a thermodynamically meta-stable condi-tion. This condition can result in membranes with large pore size(as a result from rapid solidification during immersing in coagula-tion bath) as well as skinless structure (as a result from slow phaseseparation already before immersion in the coagulation bath). Inthis experiment, the meta-stable condition could be achieved bywater absorption in the nascent membrane (it should be noted that,due to low vapor pressure, evaporation of solvent does not playa role in this case). The amount of water absorption depends onthe absorbent and the contact time. Because TEG is hygroscopic innature, it should have ability to absorb water from humid air. Theresults suggest that the solution of PES in NMP also has a slight abil-ity to absorb water as indicated by the slight increase in hydraulicpermeability with increasing exposure time for membranes pre-pared without addition of TEG (cf. Fig. 3). Hence, contact time isanother important factor. It is obviously seen in Fig. 2 that althoughthe hygroscopic TEG is added to polymer solution even up to 60%,an increase in hydraulic permeability is not observed if the nascentmembrane is not exposed to humid air.

SEM images presented in Fig. 4 support the preceding explana-tions. For the same concentration of non-solvent TEG, increasingexposure time to humid air results in membranes having greaterpore size. Similar phenomenon was observed by increasing TEGcontent for the same exposure time. In agreement with hydraulicpermeability results, both parameters clearly influenced the cross-section structure of the resulting membrane. Indeed, evolutionfrom finger-like to sponge-like structures could be observed by con-

trolling both the TEG content and the humid air exposure time.Moreover, membranes prepared from a polymer solution with TEGcontent of 60% and 1–5 min exposure (image for 3 min exposure isnot shown) showed a skinless porous top layer and a macrovoid-free, very homogenous cross-section pore structure.

Page 5

H. Susanto et al. / Journal of Membrane Science 342 (2009) 153–164 157

F ifferenm

3s

PPta

ig. 4. SEM images of the PES membranes prepared from polymer solutions with dorphology (left panel) and cross-section morphology (right panel).

.2. Incorporating amphiphilic PEG block copolymer into castingolution

The effects of additive concentration were first studied by usingluronic® F127 with different concentrations (1, 3 and 5 wt.%). TheES and NMP concentrations were fixed at 10 and 65 wt.%, respec-ively. Fig. 5 presents the results plotted as permeability versusdditive content at 1 min exposure time (it is worth mention that no

t non-solvent (TEG) contents and exposure times to humid air: membrane surface

significant difference in permeability was observed for 3 min expo-sure; therefore the data are not included). The data for membranesprepared without addition of TEG (i.e., from casting solutions with

variable NMP content) are also included.

In the absence of TEG, the addition of Pluronic increased thehydraulic permeability. The presence of Pluronic in casting solu-tion led to an increase of the viscosity and could apparently alsoincrease the water absorption ability during exposure to humid

Page 6

158 H. Susanto et al. / Journal of Membrane Science 342 (2009) 153–164

Fmt

acdbltwtmtts(wacwtims

cswt

Fig. 6. Hydraulic permeability as function of Pluronic® PE6400 concentrationfor the membranes prepared from a polymer solution with concentration of

TS

N

#########

to

ig. 5. Hydraulic permeability as function of Pluronic® F127 concentration for theembranes prepared with 1 min exposure to humid air (RH = 50–60%). The “x” in

he legend represents the Pluronic concentration in the casting solutions.

ir, leading to higher hydraulic permeability. However, if a highoncentration (5%) was used, the hydraulic permeability tended toecrease. The penetration of water into the proto-membrane maye slowed down at too high concentration of Pluronic due to a too

arge increase in viscosity. For example, the viscosity of casting solu-ion without Pluronic was 0.104 Pa s, and it increased to 0.228 Pa sith addition of 5 wt.% Pluronic. No significant effect of the addi-

ion of Pluronic on hydraulic permeability was observed when aoderate concentration of TEG (20–25 wt.%) had been added to

he casting solution. This observation can be explained because inhe presence of both Pluronic and TEG, the viscosity of polymerolution was significantly higher than if only Pluronic was presentPES/NMP/Plu). Even though a different Pluronic was used, Table 1,hich shows the increase in viscosity in the presence of Pluronic

nd TEG, supports this explanation. On the other hand, a moderateontent of TEG (20–25%) was not sufficient to absorb the amount ofater to induce the transition to the high flux macroporous struc-

ures (cf. Figs. 2 and 4). Even though exposure time to humid air wasncreased up to 5 min, a significant further increase of hydraulic per-

eability could not be achieved (the data for 5 min exposure wereimilar to the ones for 1 min exposure, i.e., ∼4000 L m−2 h−1 bar−1).

This confirms the conclusion drawn above that sufficient TEG

ontent and exposure time are two crucial parameters, and bothhould be fulfilled to obtain high-flux membrane. In addition, theater permeability of PES membrane could be increased with addi-

ion of Pluronic but all results were below the target set relative

able 1tatic water contact angle (measured by captive bubble method), ATR-IR and 1H NMR ana

o. PES(wt.%)

NMP(wt.%)

PluronicPE6400(wt.%)

TEG(wt.%)

Viscosity(Pa s)

Exposure time(min)a

Membranecodeb

Wa(L m

1 10 90 0 0 0.104 1 A0T0t1Hm ∼2 10 30 0 60 1.240 0 A0T60t0Hm ∼3 10 30 0 60 1 A0T60t1Hm ∼34 10 30 1 59 1.303 1 A1T59t1Hm ∼35 10 30 3 57 1.587 1 A3T57t1Hm ∼36 10 30 5 55 1.731 1 A5T55t1Hm ∼47 10 30 5 55 3 A5T55t3Hm ∼8 10 30 5 55 5 A5T55t5Hm ∼99 10 30 5 55 1 A5T55t1Hh ∼a All membranes were exposed to humid air with RH = 50–60%; only membrane #9 wab Because the concentration of PES was always the same (10%) and the concentration of

he code only covers the concentration of additive Pluronic (A) and TEG (T), exposure timr exposure time to humid air, whereas m and h as subscript to H indicate the relative hu

c Calculated from 1H NMR via calculating the area ratio of proton peaks for PES (at 8–7

PES/NMP/Plu/TEG = 10/30/x/60 − x wt.% and different exposure times into humid air(50–60% RH). The “x” in the legend represents the Pluronic concentration in the dopesolutions.

to commercial MF membranes (cf. Section 1). Of course, increas-ing the TEG content would be a straightforward option to increasehydraulic permeability. However, due to dissolution problems forthis block copolymer (Pluronic® F127) in combination with PES, thecontent of TEG could not be increased any more. One reason for thisincompatibility is presumably the high PEG content. Therefore, weused an analogous triblock copolymer with a lower molar mass anda lower PEG content which is also a liquid, i.e., Pluronic® PE6400(cf. Fig. 1).

Fig. 6 shows the hydraulic permeability of the membranes pre-pared from polymer solutions with addition of Pluronic® PE6400and a high content of TEG for 1 and 3 min exposure time. Inter-estingly, all membranes showed hydraulic permeabilities beyondthe target value 20,000 L m−2 h−1 bar−1. The membrane preparedfrom a polymer solution with 5% Pluronic and 3 min exposure timeshowed a hydraulic permeability of ∼95,000 L m−2 h−1 bar−1. Asobserved and discussed in Section 3.1, the content of non-solvent(here TEG and Pluronic) and the exposure time play a critical role forachieving high water permeability. These observations are stronglysupported by the images from visualization of membrane surface

and cross-section structure (Fig. 7). Indeed, evolution of membranestructure can clearly be obtained by controlling the content of non-solvent (TEG and Plu) and the exposure time. Similar to the resultsfor membranes prepared from PES solutions without addition of

lyses of the membranes prepared using different conditions.

ter permeability−2 h−1 bar−1)

Air captive bubblecontact angle (◦)

ATR-IR relative peakincrease at∼1105/1578 cm−1 (%)

Pluronic contentin polymermembrane (%)c

1,200 63.7 ± 4.1 ∼0 n.d.1,140 65.2 ± 4.0 n.d. n.d.1,240 62.5 ± 3.5 ∼0 ∼08,620 55.2 ± 3.2 4.5 3.75,250 49.6 ± 2.9 10.6 6.50,270 40.3 ± 3.3 13.4 9.9

81,740 43.6 ± 2.8 13.7 7.44,960 47.2 ± 3.1 14.1 n.d

41,330 40.1 ± 3.4 13.3 9.7

s exposed to humid air with RH = 65–75%.NMP can be calculated with known PES, additive Pluronic and TEG concentrations,

e (t) and relative humidity (H). The number (subscript) indicates the concentrationmidity used, i.e., medium (50–60%) and high (65–75%), respectively.ppm) and for Pluronic (at 1 ppm).

Page 7

H. Susanto et al. / Journal of Membrane Science 342 (2009) 153–164 159

F P/Plus

Ps

wnsetdfmh

a1a

ig. 7. SEM images of the PES membranes prepared from a polymer solution of PES/NMurface morphology (left panel) and cross-section morphology (right panel).

luronic (cf. Section 3.1), the membranes, which have completeponge-like structure, showed the highest hydraulic permeability.

In sum, PES membranes with high flux can be manufacturedith incorporation of the amphiphilic additive Pluronic and the

on-solvent TEG to the casting solution and a relatively long expo-ure time to humid air before coagulation. Nevertheless, as thexposure time was increased the surface hydrophobicity (see Sec-ion 3.3.2 for more details) as well as the degree of shrinkage (weid not investigate this effect in detail) would also increase. There-

ore, we tried to keep the exposure time as short as possible. Thisay be achieved with applying an atmosphere with higher relative

umidity (RH).The hydraulic permeability data obtained from the experiments

t two different RHs, medium (50–60%) and high (65–75%), withmin exposure time, indicated a slight increase in water perme-bility with increasing relative humidity. However, the increase was

/TEG = 10/30/5/55 wt.% (Pluronic® PE6400) for different exposure times: membrane

within the range of standard deviation (data not shown). Neverthe-less, SEM images showed that a sponge-like structure was obtainedeven though a slight tendency to form an asymmetric structure wasstill observed (Fig. 8). Thus, by increasing the water content of airwe could obviously shorten the required exposure time in order toobtain high flux PES membranes.

In the next sections, pore size distribution, membrane sur-face hydrophilicity and membrane composition of selected PES MFmembranes were further investigated; their preparation conditionsalong with the pure water permeability can be found in Table 1.

3.3. Membrane characterizations

In the following sections, a code, which indicates composi-tion of casting solution and preparation condition (cf. Table 1),is given to the membranes so that the reader can easily know

Page 8

160 H. Susanto et al. / Journal of Membrane Science 342 (2009) 153–164

F solutioi ared wa ht pan

tuhoe51k

3

tm

Fwp

ig. 8. SEM cross-section images of the PES membranes prepared from a polymernto humid air with high RH (65–75%) (left panel; A5T55t1Hh). The membranes prepnd longer (3 min; right panel) exposure to humid air with normal RH (50–60%, rig

he composition of casting solution and preparation conditionsed. For example, code A5T55t3Hm means that the membranead been prepared from a casting solution having compositionf PES/NMP/Additive Pluronic®/TEG = 10/30/5/55 wt.% with 3 minxposure time to humid air of a medium relative humidity of0–60% (note that the concentration of PES was always the same,0%, and that the concentration of NMP can be calculated with thenown PES, additive Pluronic and TEG concentrations).

.3.1. Pore size distributionFig. 9 shows the pore size distribution, which is based on dewet-

ing the pores as function of transmembrane pressure, for selectedembranes. Indeed, the results agree well with previous water

ig. 9. Pore size distribution of selected membranes: (a) PES membranes preparedith different Pluronic concentration (Pluronic® PE6400) and (b) PES membranes

repared with different relative humidity of humid air.

n of PES/NMP/Plu/TEG = 10/30/5/55 wt.% (Pluronic® PE6400) with 1 min exposureith the same polymer composition for the same (1 min; middle panel; A5T55t1Hm)

el; A5T55t3Hm) are included for comparison.

permeability measurements. As already indicated by hydraulic per-meability, membrane A5T55t3Hm had the greatest average poresize, ∼0.90 �m, among the membranes tested. The same prepara-tion conditions with lower Pluronic content resulted in membraneA1T59t1Hm with smaller average pore size, 0.47 �m (Fig. 9(a)). Sim-ilar pore size distribution was observed for membranes preparedwith the exposure time 1 min even though different relative humid-ity has been applied; both membranes A5T55t1Hm and A5T55t1Hhhad an average pore diameter of ∼0.55 �m (Fig. 9(b)).

3.3.2. Membrane surface hydrophilicityTable 1 shows also the contact angle (CA) values measured with

captive bubble method. It is important to mention that measur-ing CA by sessile drop method with water resulted in completespreading after a few seconds for all membranes prepared fromPES solutions having high Pluronic content and high water perme-ability. Therefore, the captive bubble was then used.

The PES membranes prepared from a polymer solution withoutaddition of Pluronic had lower CA (∼64◦) than typically measuredfor non-porous PES film (∼76◦) [31]. A porous structure of the outermembrane surface, where during the CA measurement the poresare filled with water, is the reason for this difference. While all themembranes prepared without addition of Pluronic showed similarCA, the effect of Pluronic on surface hydrophilicity can clearly beobserved. The increase in Pluronic concentration lead to increasedsurface hydrophilicity as noticed by decreasing CA from ∼55◦ (forthe membranes prepared with 1% Pluronic) to ∼40◦ (for the mem-brane prepared with 5% Pluronic). However, the CA value tendedto increase as the exposure time was increased (cf. membranes#6, #7 and #8). Different conformations of the hydrophilic andhydrophobic blocks of the copolymer at the membrane surfacemight be formed. Recently, this has been studied with advancedspectroscopy for PES and Pluronic 127 in model films [32]. On theone hand, the desired effect, a more hydrophilic surface would bedue to PEG blocks exposed to the surface of a PES matrix withembedded poly(propylene glycol) (PPG). One could speculate thatupon contact with the aqueous coagulation bath, this orientationwould be preferred. However, apparently, the hydrophobic PPGblocks tended to orient to the interface of the proto-membrane toair with increasing the exposure time, so that upon solidificationa more PPG rich surface would fixed. Even though humid air hasbeen applied, this is not enough to suppress this rearrangement.Similar observations had been obtained previously by Suk et al. dur-ing membrane preparation for membrane distillation [33]. In orderto minimize this effect, membrane A T t H (#9) was prepared

5 55 1 hwith shorter exposure time (1 min). The result showed that a lowercontact angle (∼40◦) could be obtained. In general, the effect of sur-face hydrophilization of PES membranes by addition of Pluronic canobviously be observed. More importantly, the CA can be controlledby adjusting the Pluronic concentration and exposure time.

Page 9

mbran

3

stopswittTbpttmetP

msonae3lpr1stI

H. Susanto et al. / Journal of Me

.3.3. Membrane chemical compositionCharacterization of membrane surface chemistry by ATR-FTIR

pectroscopy indicated that no additional peak was observed forhe PES membranes prepared with added Pluronic® compared tonly PES membrane. The reason for this result would be overlap-ing bands of the strongest bands for Pluronic (ether) with the verytrong ether bands of PES (data not shown). Similar observationas also made in our previous study [26]. However, a significant

ncrease in absorbance at wave number of ∼1105 cm−1 relativeo aromatic C–H band at 1578 cm−1, which is assigned to addi-ional intensity of C–O bond stretch (from Pluronic), was observed.his confirms the presence of the additive in the polymer mem-rane. 1H NMR spectra support this interpretation, because a neweak was observed (cf. below). It is obviously seen in Table 1hat the increase in IR peak rises with increasing Pluronic concen-ration. This implies that the Pluronic concentration at the outer

embrane surface increased. By contrast, no significant effect ofxposure time on the increase in IR peak was observed (notehat we can here not distinguish the ether bands of PEG andPG).

To know the content of the additive in membrane polymeratrix, 1H NMR spectroscopy was used to analyze membranes dis-

olved in d-DMSO. Fig. 10 shows an examplaric 1H NMR spectrumf a PES membrane prepared with the additive Pluronic. Reso-ances for the aromatic protons of PES are assigned to both peakst 8–7 ppm (note that PES is a purely aromatic polymer). The pres-nce of additive can be seen by appearance of peaks at ∼1.1, ∼3.4 and.5 ppm which are attributed to the protons from methyl, methy-

ene and methyne groups, respectively. The Pluronic content in theolymer membrane matrix was determined by calculating the area

atio of proton peaks for PES (at 8–7 ppm) and for Pluronic (atppm). It was found that as the Pluronic content in the polymer

olution was increased, this area ratio also increased indicating thathe content of Pluronic in the polymer membrane also increased.t is reasonable that a fraction of the Pluronic from the blend with

Fig. 10. 1H NMR spectrum of a d-DMSO solution of a PES me

e Science 342 (2009) 153–164 161

the membrane polymer would be dissolved in the water duringcoagulation, while this is not possible for PES. This is the reason forthe reduced Pluronic/PES ratio in the solid membrane comparedto the casting solution (the maximum weight fraction would be33.3% for entries #6 to #9; cf. Table 1). Comparing IR-ATR and 1HNMR results suggests that both methods show similar trends butdifferent values. The reason for this difference would be the differ-ent sample analysis by the both two methods. IR-ATR analyzed theouter membrane surface region in a depth of about 2 �m, whereasthe 1H NMR analyzed the bulk composition of the membrane. Dueto the problem with overlapping IR peaks mentioned above andbecause no calibration for the ATR measurement was available, adiscussion about the degree of surface enrichment of the additiveis not possible.

3.4. Membrane performance examinations based on adsorptiveprotein fouling and microfiltration

The membrane performance was investigated with respect tostatic adsorptive fouling and microfiltration. Static adsorptive foul-ing was studied by soaking the membranes in protein (BSA) solution(1 g/L) for 3 h. The amount of BSA bound to the membrane wasdetermined by gravimetric and mass balance methods (cf. Section2.8). The results are presented in Table 2. As expected, both methodsshowed that the addition of Pluronic decreased the amount of pro-tein adsorbed by the membrane. Moreover, the amount of proteinbound to the membrane decreased with increasing Pluronic con-tent. Slightly different absolute values but identical trends wereobserved for both methods, only data for membrane A0T60t0Hm

(#2) showed larger differences. Heterogeneity of membrane sam-

ples is a possible reason. Further evaluation of the data was doneby estimating the surface coverage of the membrane by protein(amount of protein bound/specific surface area of membrane).For example, for membrane A5T55t3Hm (#7) with a specific sur-face area of 37.6 m2/g (determined by BET surface area analysis),

mbrane prepared with addition of Pluronic® PE6400.

Page 10

162 H. Susanto et al. / Journal of Membrane Science 342 (2009) 153–164

Table 2Amount of protein bound to various Pluronic-modified PES membranes after incu-bation in 1 g/l BSA solution (pH 5).

No. Membrane Gravimetric method (mg/g)a Mass balance (mg/g)a

#3 A0T60t1Hm 10.30 7.62#4 A1T59t1Hm 5.48 5.63#5 A3T57t1Hm 4.61 4.40#6 A5T55t1Hm 2.05 2.46#7 A5T55t3Hm 2.35 2.76#

t

afoswbeatmbiocP[imfhismroififospmma

fi(tsb

totPmobpoons

Fig. 11. Normalized flux (relative to pure water flux) during microfiltration of BSAsolutions (0.1 g/L in phosphate buffer 0.05 M, pH 5) at a transmembrane pressure of0.2 bar.

Table 3Apparent protein rejection during BSA microfiltration (0.1 g/L, pH 5) for 2 h for vari-ous Pluronic-modified PES membranes.

No. Membrane Protein rejection, initialstage of filtration (%)

Protein rejection, finalstage of filtration (%)

#3 A0T60t1Hm 46 57#4 A1T59t1Hm 43 51#5 A3T57t1Hm 40 54###

9 A5T55t1Hh 1.98 2.11

a The amount of protein bound is expressed in terms of weight of protein relativeo weight of membrane.

verage protein surface densities of 6.2 and 7.2 ng/cm2, calculatedrom gravimetric and mass balance methods, respectively, werebtained. For the unmodified membrane A0T0t1Hm (#1) with apecific surface area of 15.2 m2/g, average protein surface densitiesere 67.8 and 50.1 ng/cm2, calculated from gravimetric and mass

alance methods, respectively. The latter value is smaller than cov-rage of planar surfaces by a BSA monolayer (typical values for BSAre in the range 220–365 ng/cm2 [34]). This may be due to par-ial inaccessibility of membrane pores for the protein. However, a

arked decrease in adsorbed protein density on the pore surfacey about one order of magnitude has been achieved by incorporat-

ng the PEG-based additive. The mechanisms of protein resistancef solid surfaces with grafted PEG (here immobilized via the blockopolymer) have been widely investigated (cf., e.g., [35–37]). ForES/Pluronic blend membranes this has also been discussed before26,27,32]. Table 2 also shows that none of the membranes wasnert to protein deposition. The porous structure of membrane

ay have influence on solute deposition. We had shown by per-orming simultaneously diffusion/adsorption experiment using aydrophilic membrane and a hydrophilic solute that solute binding

n the membrane was still observed [31]. More importantly, it is rea-onable to speculate that significant but incomplete shielding of theembrane surface by Pluronic PEG groups, leading to significant

eduction of protein adsorption but not to complete resistance, wasbtained in this modification. Similar observation was made during

nvestigation of protein adsorption on polystyrene surfaces modi-ed by poly(ethylene glycols) including Pluronics, self-assembled

rom aqueous solution [36]. For the same Pluronic content, it wasbserved that membrane A5T55t3Hm prepared with 3 min expo-ure showed greater protein adsorption than membrane A5T55t1Hm

repared with 1 min exposure. This may be related to the fact thatembrane A5T55t3Hm had a higher contact angle which indicates aore hydrophobic character (cf. Table 1), leading to higher protein

dsorption.To investigate microfiltration performance, dead-end stirred

ltration was performed with constant transmembrane pressure0.2 bar). The results are presented in terms of permeate flux rela-ive to initial water flux (Fig. 11). It was observed that all membraneshowed similar behavior, i.e., permeate flux dropped rapidly in theeginning of filtration.

Indeed, the presence of Pluronic additive increased the rela-ive flux indicating that higher resistance towards fouling has beenbtained. It was also clearly observed that as the Pluronic con-ent was increased, the permeate flux increased. For the sameluronic content, membrane A5T55t1Hm (#6) showed higher per-eate flux than membrane A5T55t3Hm (#7). The reason for this

bservation is that the higher hydraulic permeability of the mem-rane A5T55t3Hm (#7) (cf. Table 1) causes a larger amount of protein

ermeated through the membrane pores leading to higher amountf fouling by deposition during the filtration experiment. Thisbservation also suggests that the resulting fouling is influencedot only by the membrane surface chemistry but also by the poretructure. Further, membrane A5T55t1Hh (#9) showed higher rel-

6 A5T55t1Hm 41 507 A5T55t3Hm 31 439 A5T55t1Hh 41 51

ative flux than membrane A5T55t1Hm (#6) indicating less fouling.Because both membranes had similar hydraulic permeability, themembrane hydrophilicity may here be the proper reason. In fact,membrane A5T55t1Hm (#9) was more hydrophilic than membraneA5T55t1Hm (#6) (cf. Table 1). The membrane prepared from a solu-tion without an additive (A0T60t1Hm) had a permeate flux of only∼20% relative to the initial water flux, whereas the membranes pre-pared with 5% Pluronic had a stable relative flux within the range(50–60%).

Rejection data presented in Table 3 show that all membranes(except membrane A5T55t3Hm) showed similar protein rejection.These results are in agreement with pore size distribution resultswhere membrane A5T55t3Hm showed the largest average pore size(cf. Fig. 9). Another observation was that the protein rejectionincreased with filtration time, which is probably due to the effect ofprotein deposition on/in the membrane (and this had been evokedas a significant contribution to flux decline; cf. above). In sum, theresults suggest that the addition of Pluronic not only increases thehydraulic permeability (cf. Section 3.2) but also increases the overallmembrane separation performance in microfiltration.

3.5. Stability test study

For long term applications, the performance of microfiltrationmembrane will be influenced by the stability of the additive in themembrane polymer matrix. In this part, the stability of membrane-immobilized Pluronic in water was investigated. Surface chemistrycharacterizations by IR-ATR and contact angle were applied for eval-uation, and only membrane A T t H (#7) was used for this study.

5 55 3 m

Fig. 12 shows the values of contact angle and IR absorbance ratio(measure of Pluronic content; cf. Section 3.3.3). It was observed thatno change in CA occurred over 10 days of incubation. The fluctua-tion is within the range of standard deviation. This indicates that the

Page 11

H. Susanto et al. / Journal of Membran

Fba4

swAIa

4

pPgmmsHpasaiFcmsatoosEiocfidccts(b

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

ig. 12. Stability test of Pluronic modified membrane A5T55t3Hm (#7), investigatedy measuring the ATR-IR absorbance ratio as measure of additive content and their captive bubble contact angle, both as a function of incubating time in water at0 ◦C.

tability of hydrophilic property obtained from addition of Pluronicas quite high. Leaching out of Pluronic seemed not to occur. IR-TR results support this observation, i.e., no significant change in

R absorbance for the introduced functional additive was observedfter incubation in water.

. Conclusions

High flux hydrophilic microfiltration membranes have beenrepared from PES by a combination of VIPS and NIPS methods.oly(ethylene glycol)-b-poly(propylene glycol)-b-poly(ethylenelycol) triblock copolymers (Pluronic®) have been chosen asodification agent because of their stable incorporation within theembrane polymer matrix. The characteristics, performance and

tability of Pluronic-modified PES membranes were investigated.igh performance membranes can be prepared from castingolymer solutions with high non-solvent (TEG) content and usingrelatively long exposure in humid air. Water permeability, pore

ize and membrane morphology can effectively be controlled bydjusting both variables. Due to better compatibility with the cast-ng solutions, Pluronic® PE6400 was better suited than Pluronic®

127 for the preparation of high-flux membranes with isotropicross-section morphology. The hydrophilicity of the resultingembranes increased as the concentration of Pluronic in casting

olution was increased. However, at the same content it decreaseds the exposure time in humid air was further increased. These lat-er effects can be related to the amphiphilic, surfactant propertiesf the triblock copolymer. Performance evaluation via investigationf adsorptive and microfiltration fouling using BSA showed that aignificant increase in resistance toward fouling can be obtained.ven though none of the resulting membranes was completely

nert to protein adsorption, much lower protein adsorption wasbserved for the membranes prepared with addition of Pluronicompared to the membrane prepared without Pluronic. Micro-ltration experiments demonstrated that the antifouling effectsue to the presence of Pluronic in the membrane polymer couldlearly be observed as noticed by their higher permeate fluxes

ompared to the membrane without Pluronic. In general, based onhe results presented, the membrane prepared from the castingolution having a composition of PES/NMP/TEG/Plu=10/30/55/5wt.%) with 3 min exposure in humid air (50–60% RH) showed theest performance.

[

e Science 342 (2009) 153–164 163

Acknowledgements

The authors gratefully acknowledge Smail Boukercha (Anor-ganische Chemie, Universität Duisburg Essen) for his contributionto the SEM measurements. The authors also thank BASF, Lud-wigshafen, Germany, for supplying the PES and Pluronic 6400.

References

[1] R.W. Baker, Membrane Technology and Applications, 2nd ed., John Wiley & Sons,Ltd., Chichester, 2004.

[2] H. Nelson, R. Singh, R. Toledo, N. Singh, The use of a submerged microfiltra-tion system for regeneration and reuse of wastewater in a fresh-cut vegetableoperation, Sep. Sci. Technol. 42 (2007) 2473.

[3] L.V. Saboya, J.L. Maobois, Current developments of microfiltration technologyin the dairy industry, Lait 80 (2000) 541.

[4] R. van Reis, A. Zydney, Bioprocess membrane technology, J. Membr. Sci. 297(2007) 16.

[5] N. Hilal, O.O. Ogunbiyi, N.J. Miles, R. Nigmatullin, Methods employed for controlof fouling in MF and UF membranes: a comprehensive review, Sep. Sci. Technol.40 (2005) 1957.

[6] G. Belfort, R.H. Davis, A.L. Zydney, The behaviour of suspensions and macro-molecular solutions in crossflow microfiltration, J. Membr. Sci. 96 (1994) 1.

[7] H. Susanto, M. Ulbricht, Polymeric membranes for molecular separations, in:E. Drioli, L. Giorno (Eds.), Membrane Operations. Innovative Separations andTransformations, Wiley-VCH, Weinheim, 2009, p. 19.

[8] W.J. Wrasidlo, Dispersing casting of integral skinned highly asymmetric poly-mer membranes, US Patent 4,774,039 (1988).

[9] R.F. Zepf, Ultraporous and microporous membranes and method of makingmembranes, US Patent 5,171,445 (1992).

10] J.M. Greenwood, J.S. Johnson, M.J. Witham, Preparation of polyethersulfonemembranes, US Patent 6,056,903 (2000).

11] V.P. Khare, A.R. Greenberg, W.B. Krantz, Vapor-induced phase separation—effectof the humid air exposure step on membrane morphology. Part I. Insight frommathematical modeling, J. Membr. Sci. 258 (2005) 140.

12] J.F. Li, Z.L. Xu, H. Yang, Microporous polyethersulfone membranes preparedunder the combined precipitation conditions with non-solvent additives,Polym. Adv. Technol. 19 (2008) 251.

13] S.J. Shin, J.P. Kim, H.J. Kim, J.H. Jeon, B.R. Min, Preparation and characterization ofpolyethersulfone microfiltration membranes by a 2-methoxyethanol additive,Desalination 186 (2005) 1.

14] Y. Liu, G.H. Koops, H. Strathmann, Characterization of morphology controlledpolyethersulfone hollow fiber membrane by the addition of polyethyleneglycol to the dope and bore liquid solution, J. Membr. Sci. 223 (2003)187.

15] R.M. Boom, H.W. Reinders, H.H.W. Rolevink, Th. van den Boomgaard, C.A.Smolders, Equilibrium thermodynamics of a quaternary membrane-formingsystem with two polymers. I. Experiments, Macromolecules 27 (1994)2041.

16] I.M. Wienk, R.M. Boom, M.A.M. Beerlage, A.M.W. Bulte, C.A. Smolders, Recentadvances in the formation of phase inversion membranes made from amor-phous or semi-crystalline polymers, J. Membr. Sci. 113 (1996) 361.

17] R.M. Boom, I.M. Wienk, Th. Van den Boomgaard, C.A. Smolders, Microstructuresin phase inversion membranes. Part 2. The role of a polymeric additive, J. Membr.Sci. 73 (1992) 277.

18] B. Chakrabarty, A.K. Ghoshal, M.K. Purkait, Effect of molecular weight of PEGon membrane morphology and transport properties, J. Membr. Sci. 309 (2008)209.

19] J.H. Kim, K.H. Lee, Effect of PEG additive on membrane formation by phaseseparation, J. Membr. Sci. 138 (1998) 153.

20] M. Kraus, M. Heisler, I. Katsnelson, D. Velazques, Filtration membranes andmethod of making the same, US Patent 5,108,607 (1992).

21] M. Kraus, M. Heisler, I. Katsnelson, D. Velazques, Filtration membranes andmethod of making the same, US Patent 4,964,990 (1990).

22] I.F. Wang, R.A. Morris, R.F. Zepf, Highly asymmetric, hydrophilic, microfiltrationmembranes having large pore diameters, US Patent 7,125,493 (2006).

23] I.M. Wienk, E.E.B. Meuleman, Z. Borneman, A. van den Boomgaard, C.A. Smol-ders, Chemical treatment of membranes of a polymer blend: mechanism ofthe reaction of hypochlorite with poly(vinylpyrrolidone), J. Polym. Sci. Part A:Polym. Chem. 33 (1995) 49.

24] J.J. Qin, M.H. Oo, Y. Li, Development of high flux polyethersulfone hollow fiberultrafiltration membranes from a low a critical solution temperature dope viahypochlorite treatment, J. Membr. Sci. 247 (2005) 137.

25] H.D.W. Roesink, C.A. Smolders, M.H.V. Mulder, D.M. Koenhen, Process forthe preparation of hydrophilic membranes and such membranes, US Patent4,798,847 (1989).

26] H. Susanto, M. Ulbricht, Characteristics, performance and stability ofpolyethersulfone ultrafiltration membranes prepared by phase separation

method using different macromolecular additives, J. Membr. Sci. 327 (2009)125.

27] Y.Q. Wang, T. Wang, Y.L. Su, F. Peng, H. Wu, Z.Y. Jiang, Remarkable reduc-tion of irreversible fouling and improvement of the permeation properties ofpoly(ethersulfone) ultrafiltration membranes by blending with Pluronic F127,Langmuir 21 (2005) 11856.

Page 12

1 mbran

[

[

[

[

[

[

[

[

ration performance, Langmuir 23 (2007) 7818.

64 H. Susanto et al. / Journal of Me

28] Y.Q. Wang, Y.L. Su, X.L. Ma, Q. Sun, Z.Y. Jiang, Pluronic polymers andpolyethersulfone blend membranes with improved fouling resistant ability andultrafiltration performance, J. Membr. Sci. 283 (2006) 440.

29] D. Li, T.S. Chung, J. Ren, R. Wang, Thickness dependence of macrovoid evolutionin wet phase-inversion asymmetric membranes, Ind. Eng. Chem. Res. 43 (2004)1553.

30] W. Zhang, M. Wahlgren, B. Sivik, Membrane characterization by the con-tact angle technique. II. Characterization of UF-membranes and comparisonbetween the captive bubble and sessile drop as methods to obtain water contactangles, Desalination 72 (1989) 263.

31] H. Susanto, S. Franzka, M. Ulbricht, Dextran fouling of polyethersulfone ultrafil-tration membranes—causes, extent and consequences, J. Membr. Sci. 296 (2007)147.

32] Q. Shi, S. Ye, C. Kristalyn, Y. Su, Z. Jiang, Z. Chen, Probing molecular-level sur-face structures of polyethersulfone/Pluronic F127 blends using sum-frequencygeneration vibrational spectroscopy, Langmuir 24 (2008) 7939.

[

[

e Science 342 (2009) 153–164

33] D.E. Suk, T. Matsuura, H.B. Park, Y.M. Lee, Synthesis of a new type of sur-face modifying macromolecules (nSMM) and characterization and testing ofnSMM blended membranes for membrane distillation, J. Membr. Sci. 277 (2006)177.

34] K. Rezwan, L.P. Meier, M. Rezwan, J. Vrs, M. Textor, L.J. Gauckler, Bovine serumalbumin adsorption onto colloidal AlO particles: a new model based on zetapotential and UV–vis measurements, Langmuir 20 (2004) 10055.

35] H. Susanto, M. Ulbricht, Photo-grafted thin polymer hydrogel layers on PESultrafiltration membranes: characterization, stability and influence on sepa-

36] J.H. Lee, H.B. Lee, J.D. Andrade, Blood compatibility of polyethylene oxide sur-faces, Prog. Polym. Sci. 20 (1995) 1043.

37] D. Lazos, S. Franzka, M. Ulbricht, Size-selective protein adsorption topolystyrene surfaces by self-assembled grafted poly(ethyleneglycols) with var-ied chain lengths, Langmuir 21 (2005) 8774.