Preparation and Characterization of Hydrophilically Modified ... - MDPI

membranes Article

Preparation and Characterization of Hydrophilically Modified PVDF Membranes by a Novel Nonsolvent Thermally Induced Phase Separation Method Ningen Hu 1 , Tonghu Xiao 1, *, Xinhai Cai 1 , Lining Ding 2 , Yuhua Fu 1 and Xing Yang 3, * 1 2 3

*

Faculty of Materials Science and Chemical Engineering, Ningbo University, Ningbo 315211, China; [email protected] (N.H.); [email protected] (X.C.); [email protected] (Y.F.) Tri-Tech Chemical Co. Pty Ltd, 5-11 Normanby Ave, Sunshine VIC 3020, Australia; [email protected] Institute for Sustainability and Innovation, College of Engineering and Science, Victoria University, P.O. Box 14428, Melbourne 8001, Australia Correspondence: [email protected] (T.X.); [email protected] (X.Y.); Tel.: +86-136-8589-2736 (T.X.); +61-9919-7690 (X.Y.)

Academic Editor: Klaus Rätzke Received: 14 October 2016; Accepted: 14 November 2016; Published: 18 November 2016

Abstract: In this study, a nonsolvent thermally-induced phase separation (NTIPS) method was first proposed to fabricate hydrophilically-modified poly(vinylidene fluoride) (PVDF) membranes to overcome the drawbacks of conventional thermally-induced phase separation (TIPS) and nonsolvent-induced phase separation (NIPS) methods. Hydrophilically-modified PVDF membranes were successfully prepared by blending in hydrophilic polymer polyvinyl alcohol (PVA) at 140 ◦ C. A series of PVDF/PVA blend membranes was prepared at different total polymer concentrations and blend ratios. The morphological analysis via SEM indicated that the formation mechanism of these hydrophilically-modified membranes was a combined NIPS and TIPS process. As the total polymer concentration increased, the tensile strength of the membranes increased; meanwhile, the membrane pore size, porosity and water flux decreased. With the PVDF/PVA blend ratio increased from 10:0 to 8:2, the membrane pore size and water flux increased. The dynamic water contact angle of these membranes showed that the hydrophilic properties of PVDF/PVA blend membranes were prominently improved. The higher hydrophilicity of the membranes resulted in reduced membrane resistance and, hence, higher permeability. The total resistance Rt of the modified PVDF membranes decreased significantly as the hydrophilicity increased. The irreversible fouling related to pore blocking and adsorption fouling onto the membrane surface was minimal, indicating good antifouling properties. Keywords: nonsolvent thermally-induced phase separation (NTIPS); PVDF/PVA blend membrane; hydrophilic modification; membrane resistance; antifouling property

1. Introduction The worldwide problems associated with the shortage of clean water have driven the rapid development of waste water treatment technologies. Membrane-based processes received much attention. In particular, microfiltration (MF) and ultrafiltration (UF) are the most popular methods for portable water purification due to the high efficiency, low costs, ease of implementation and low environmental impact [1,2]. MF (pore size range of 0.1 µm–10 µm) and UF (pore size range of 0.01 µm–0.1 µm) membranes are effectively used in wastewater pretreatment for filtering organic micropollutants. Amongst the commonly-used materials for MF/UF applications, poly(vinylidene fluoride) (PVDF) is one of the most widely-used membrane materials in water treatment due to its excellent chemical and thermal stability, as well as high mechanical strength [3,4]. Membranes 2016, 6, 47; doi:10.3390/membranes6040047

www.mdpi.com/journal/membranes

Membranes 2016, 6, 47

2 of 16

Unfortunately, its lower surface energy and relatively high hydrophobicity lead to the increase of the transmembrane pressure of the PVDF membrane, irreversible fouling and rapid flux decline compared to hydrophilic membranes when treating waste water containing organic pollutants [4]. Subsequently, deteriorating membrane performance over time and frequent interruption of operation for membrane cleaning are the major drawbacks for MF/UF applications, leading to significant energy consumption and high cost. Attention has been paid to improving membrane performance to mitigate membrane fouling and reduce operation cost [5]. Usually, fouling is caused by the deposition of organic and inorganic compounds on the membrane surface and into its pores. These compounds commonly result in the formation of a cake and gel layer, adsorption and pore blockage [5–10]. It is known that a hydrophilic surface offers a better antifouling property for the membrane [11,12]. A pure water layer could easily form on a highly hydrophilic membrane surface for preventing the adsorption and deposition of hydrophobic pollutants and, hence, can reduce fouling [4]. Therefore, hydrophilic modification of MF/UF membranes becomes the main method to improve anti-fouling properties [4,11]. Among different methods, blending modification by hydrophilic polymers is the most reported approach due to its simplicity and effectiveness. In the literature, nonsolvent-induced phase separation (NIPS) [13–16] and thermally-induced phase separation (TIPS) [17,18] are mostly commonly adopted to prepare PVDF membranes. In general, the NIPS method requires complicated control of the solvent exchange rate by simultaneously varying several process parameters, such as the dope composition, additives, coagulation medium, quenching bath temperature and evaporation time, to obtain PVDF membranes with the desired morphology and good performance [19,20]. In comparison, the TIPS process is relatively simple for obtaining a membrane with higher overall porosity, better mechanical strength and a narrower pore size distribution [4,21]. However, since it is difficult to form an asymmetric PVDF membrane structure with a skin layer using TIPS [4], NIPS is more commonly used to fabricate the PVDF membrane. Thus far, the PVDF/hydrophilic polymer blend membrane prepared by the TIPS method has not been reported in the literature. In TIPS, the polymer is dissolved in the diluent at high temperature, and phase separation is induced by cooling the dope solution [22]. Hence, the selection of diluent is crucial for membrane formation, as it serves as both a nonsolvent for the polymer at the quenching temperature (e.g., room temperature) and a strong solvent at high temperature (i.e., membrane fabrication temperature) to induce the TIPS mechanism [22]. Moreover, the dissolution temperature of PVDF and membrane preparation temperature are normally higher than the melting point of PVDF. For example, 240 ◦ C was used as the membrane fabrication temperature by Ji et al. [23]. However, at such a high fabrication temperature, the hydrophilic polymers, such as polyvinyl alcohol (PVA) and polyvinylpyrrolidone (PVP), could be partly oxidized, cross-linked, thermally decomposed or depolymerized [24], which will strongly influence the stability and repeatability of the dope solution. In other words, the high temperature of the TIPS process is not suitable for the fabrication of hydrophilically-modified PVDF membranes. In our previous work [25], a novel water-soluble diluent ε-caprolactam (CPL) was first used to dissolve PVDF and obtained homogeneous PVDF casting solutions at a much lower fabrication temperature of 130 ◦ C by a nonsolvent thermally-induced phase separation (NTIPS) method, which has combined the advantages of NIPS and TIPS. With the appropriately-selected diluent for the PVDF polymer, the low solution temperature can substantially eliminate the effects of high temperature on the stability of the hydrophilic polymer used for fabricating hydrophilically-modified PVDF membranes with the desired pore structure and performance. In this study, to obtain the desired membrane characteristics under simpler fabrication conditions, a nonsolvent thermally-induced phase separation approach (NTIPS) [25] is first used to prepare hydrophilically-modified PVDF/PVA blend membranes. ε-caprolactam is used as the diluent of PVDF at a fabrication temperature of 140 ◦ C. The PVDF/PVA blending ratio and total polymer concentration of the casting solution are optimized. The pore structure and filtration performance of the PVDF/PVA blend membranes are studied. Bovine serum albumin (BSA) is used as a modelled foulant to study


3 of 16

the fouling propensity of the membranes. The membrane fouling resistance analysis is conducted to confirm the advantages of hydrophilically-modified PVDF membranes via the NTIPS method. 2. Materials and Methods 2.1. Materials The PVDF (Model: 1015) polymer was supplied by Solvay Co. Bovine (Shanghai, China) serum albumin (BSA, MW = 67,000, Biochemical reagents) was purchased from Aladdin Industrial (Shanghai, China). Polyvinyl alcohol (PVA, Model: 1788) and ε-caprolactam (CPL, 99.5%), both supplied by Aladdin Industrial, were used as the hydrophilic modification polymer and diluent, respectively. 2.2. Preparation of Hydrophilically-Modified PVDF Membranes The PVDF, PVA and CPL were mixed in a container proportionately to prepare a casting solution. The composition of the solutions for various PVDF/PVA blend membranes is shown in Table 1. The solution was heated up in an oil bath under the protection of nitrogen at 140 ◦ C and stirred at a constant speed of 120 rpm to form a homogeneous dope solution. The solutions were degassed at the preparation temperatures and then were rapidly casted on the glass plate by an automated high-temperature casting machine described elsewhere [25], which was preheated to 140 ◦ C. The nascent membrane was quickly and smoothly immersed into a water coagulant bath (25 ◦ C). After the nascent membrane was completely solidified, the membrane was transferred into a flowing water bath to remove residual diluent and subsequently stored in DI water before use. Table 1. Composition of casting solutions for various PVDF/PV blend membranes. CPL, ε-caprolactam. Membrane ID

Total Polymer (PVDF/PVA) Concentration (wt %)

PVDF/PVA Blend Ratio

CPL Concentration (wt %)

S1 S2 S3 S4 S5 S6 L M N

16 18 20 22 24 26 20 20 20

8:2 8:2 8:2 8:2 8:2 8:2 7:3 9:1 10:0

84 82 80 78 76 74 80 80 80

2.3. Membrane Characterization The membrane surface and cross-sectional morphologies were observed using a scanning electron microscope (Model: TM3000, Hitachi, Tokyo, Japan). The membrane samples were fractured in liquid nitrogen. All samples were coated with a thin layer of gold in standard high vacuum conditions before scanning. The mechanical properties of the membranes were measured via tensile strength using a tensiometer (Model: 5542, Instron Corp., Boston, MA, USA). Five pieces of membrane samples under each fabrication condition were tested to ensure reproducibility. Dynamic water contact angles of the membranes were measured with an angle meter (Model: JC2000D2, Shanghai Zhongchen Company, Shanghai, China) to evaluate the membrane hydrophilicity. DI water was dropped on the sample surface at five different sites. Repetition of water contact angle measurements was done with three membrane samples under the same fabrication conditions.


4 of 16

The membrane porosity was tested according to the method described in the literature [26]. The membranes were weighed when wet and were later dried in an oven. The porosity (Pr ) was calculated with the following equation: Pr = (Ww − Wd ) /dw Am Lm × 100%

(1)

where Ww is the weight of the wet membrane (g), Wd is the weight of the dry membrane (g), dw is the water density (g/cm3 ) and Am and Lm are the membrane area (cm2 ) and thickness (cm), respectively. The pore size distribution of the membranes was determined by the liquid-liquid displacement method based on an isobutanol-DI water system. The detailed experimental procedure can be found elsewhere [27–30]. 2.4. Antifouling Performance and Membrane Fouling Resistance Analysis For membrane performance evaluation, a flat-sheet membrane testing cell (MSC 300, Shanghai Mosu Science Company, Shanghai, China) was used to measure water flux under a pressure of 0.1 MPa. The effective membrane area was 35 cm2 . The water flux, J, was calculated by the following equation: J = V/ ( At)

(2)

where J is the water flux (L·m−2 ·h−1 ), V is the volume of permeated water (L), A is the effective membrane area (m2 ) and t is the filtration time (h). The membranes were pre-pressurized by filtering DI water for 0.5 h until the flux reached a plateau, and then, three steps of filtration were performed. Firstly, a 30-min period of recording the initial water flux (Ji ) was measured with DI water at 0.1 MPa. Secondly, the BSA filtration test was carried out for 1 h, and the membrane fouling step water flux (Jf ) was measured by filtering the BSA (1 g/L) solution at 0.1 MPa. Thirdly, at the end of the BSA fouling run, membrane “physical” cleaning was carried out, and then the post-cleaning water flux recovered (Jr ) was measured with DI water at 0.1 MPa. For membrane cleaning, the membrane surface was flushed with DI water under stirring (200 rpm) condition for 5 h. Normalized fluxes in Steps 2 and 3, i.e., Jf /Ji and Jr /Ji , were used to evaluate the antifouling performances of the currently-developed membranes [31–34]. According to Darcy–Poiseuille’s law [35], in the filtration process, the membrane total resistance Rt (m−1 ) can be divided into the intrinsic membrane resistance Rm (m−1 ) and the fouling resistance Rf (m−1 ). Rm is the initial hydraulic resistance, calculated from Darcy’s law Equation (3) [35] using the initial water flux (Ji ) measured: Ji = ∆P/µRm (3) where ∆P (Pa) is the transmembrane pressure and µ (Pa·s) is the viscosity of the feed solution. For further interpretations, fouling resistance Rf can be divided into a reversible resistance Rrevf (m−1 ) and an irreversible resistance Rirrf (m−1 ) [36]. Rirrf can be calculated applying Equation (4) to the recovered water flux (Jr ), whereas Rrevf is calculated according to Equation (5): Jr = ∆P/µ( Rm + Rirr f )

(4)

J f = ∆P/µRt = ∆P/µ( Rm + R f ) = ∆P/µ Rm + Rrev f + Rirr f

(5)

where Rt is the total resistance (m−1 ). Rrevf is due to concentration polarization and the formation of a cake layer on the membrane surface, removable by physical cleaning; Rirrf is due to pore blocking, and adsorption and can only be suppressed by chemical cleaning [37], which however was not investigated in this study.


5 of 16

3. Results and Discussion 3.1. Investigation on Total Polymer Concentration Membranes 2016, 6, 47

5 of 16

Newly-prepared membranes S1 –S6 (Table 1) with different total polymer concentrations at a Newly-prepared membranes 1–S6 (Table 1) with different total polymer concentrations at a PVDF/PVA blend ratio of 8:2 were Sused to investigate the influence on membrane structure and PVDF/PVA blend ratio of 8:2 were used to investigate the influence on membrane structure and filtration performance. filtration performance.

3.1.1. Effect of Total Polymer Concentration on Membrane Morphology 3.1.1. Effect of Total Polymer Concentration on Membrane Morphology

Figure 1 shows the the cross-section morphologies (b) and (c) are 1 –S 6 ; Columns Figure 1 shows cross-section morphologiesof of membranes membranes SS 1–S 6; Columns (a), (a), (b) and (c) are the full cross-section, the cross-section adjacent to the top surface and the cross-section adjacent the full cross-section, the cross-section adjacent to the top surface and the cross-section adjacent toto the bottom respectively. thesurface, bottom surface, respectively.

Figure 1. SEM images of PVDF/PVA membranesSS11–S66: :(a) (b)(b) cross-section adjacent Figure 1. SEM images of PVDF/PVA membranes (a)full fullcross-section; cross-section; cross-section adjacent to the top surface; (c) cross-section adjacent to the bottom surface. to the top surface; (c) cross-section adjacent to the bottom surface.


6 of 16

Membranes 2016, 6, 47 6 of 16 It is observed in Figure 1b that the cross-section structure adjacent to the top surface gradually changes from long finger-like pores to cellular pores with increasing total polymer concentration from It is observed in Figure 1b that the cross-section structure adjacent to the top surface gradually S1 –S6 . The finger-like pores eventually disappeared with the further increase in the total polymer changes from long finger-like pores to cellular pores with increasing total polymer concentration concentration over 22 wt % (S4 ). This is because at a low total polymer content, the casting solution from S1–S6. The finger-like pores eventually disappeared with the further increase in the total has low viscosity, and molecules move freely [38], which is beneficial to the double-diffusion polymer concentration over 22 wt % (Smore 4). This is because at a low total polymer content, the casting between water and solvent resulting in the formation of large finger-like pores. As the total solution has low viscosity, and molecules move more freely [38], which is beneficial to polymer the concentration increased to 22water wt %and (S4solvent ) and above, both theformation viscosity of oflarge the casting solution double-diffusion between resulting in the finger-like pores. and As the the total polymer concentration increased to 22 wt % (S 4 ) and above, both the viscosity of the casting mass transfer resistance of the solvent exchange increased. This has impeded the interactions between solution and the mass(water) transferwhen resistance of the solvent exchange increased. This the bath. solvent and non-solvent the nascent membrane was immersed into has the impeded coagulation interactions between non-solvent (water) when the nascent membrane was The crystallization rate ofsolvent PVDF and would be reduced, hence resulting in the formation of immersed a cellular pore into the coagulation crystallization rate of PVDF reduced, hence in the role structure adjacent to thebath. top The surface of the membrane. Thewould NIPS be mechanism playsresulting an important formation of a cellular pore structure adjacent to the top surface of the membrane. The NIPS in forming the cellular morphology of the top surface, which was also observed in the literature [39]. mechanism plays an important role in forming the cellular morphology of the top surface, which was Different from the cross-section structure adjacent to the top surface, Figure 1c shows that also observed in the literature [39]. the cross-sectional structure in the sublayer adjacentto to surface of thethat membrane Different from the cross-section structure adjacent thethe top bottom surface, Figure 1c shows the gradually changes from a bicontinuous network pore to a cellular pore structure as the total polymer cross-sectional structure in the sublayer adjacent to the bottom surface of the membrane gradually concentration increases from S1 –Snetwork of the bottom surface is mainly formed via changes from a bicontinuous porestructure to a cellular pore structure as the total polymer 6 . The pore concentration increases from S1–Sin 6. The pore structure the bottom surface is mainly formed the the TIPS mechanism, as observed a previous studyof[25]. Detailed explanations on the via membrane TIPS mechanism, observed in a previous study formation mechanismaswill be provided in Section 3.2.[25]. Detailed explanations on the membrane formation mechanism will be provided in Section concentration 3.2. Figure 2 illustrates the effect of total polymer on membrane porosity. It is observed Figure 2 illustrates the effect of total polymer concentration on membrane porosity. It is observed that as the total polymer concentration increased from 16 wt %–26 wt % at the same blend ratio, the that as the total polymer concentration increased from 16 wt %–26 wt % at the same blend ratio, the porosity of membrane decreased from about 86% down to 50%. The results are consistent with the porosity of membrane decreased from about 86% down to 50%. The results are consistent with the variation in membrane structure indicated in Figure 1. variation in membrane structure indicated in Figure 1.

Figure 2. Effect of total polymer concentration on membrane porosity.

Figure 2. Effect of total polymer concentration on membrane porosity.

3.1.2. Effect of Total Polymer Concentration on Membrane Pore Size and Pore Size Distribution

3.1.2. Effect of Total Polymer Concentration on Membrane Pore Size and Pore Size Distribution

Pore size distribution curves and mean pore size of membranes S1–S6 are shown in Figure 3 and

Tablesize 2, respectively. With the increase of total concentration,S1the of the in membrane Pore distribution curves and mean porepolymer size of membranes –S6curves are shown Figure 3 and pore size distribution shift toward the left, and the mean pore size decreases. Overall, the width of pore Table 2, respectively. With the increase of total polymer concentration, the curves of the membrane the pore size distribution becomes narrower with increasing polymer concentration, indicating size distribution shift toward the left, and the mean pore size decreases. Overall, the width of thea pore more uniform pore structure. This corresponds to the decrease of the mean pore size from 81.7 nm size distribution becomes narrower with increasing polymer concentration, indicating a more uniform down to 27.6 nm as the polymer concentration increased from 16 wt %–26 wt %. Membranes with UF pore structure. This corresponds to the decrease of the mean pore size from 81.7 nm down to 27.6 nm (0.01 μm–0.1 μm) and MF (0.1 μm–10 μm) pore size ranges could be easily fabricated by adjusting as thethe polymer concentration increased from 16 wt %–26 wt %. Membranes with UF (0.01 µm–0.1 µm) dope composition to meet the separation requirements. and MF (0.1 µm–10 µm) pore size ranges could be easily fabricated by adjusting the dope composition to meet the separation requirements.


7 of 16


7 of 16


7 of 16

Figure 3. Pore size distribution curves of PVDF/PVA blend membranes S1–S6.

Figure 3. Pore size distribution curves of PVDF/PVA blend membranes S1 –S6 . Table 2. Mean pore size of PVDF/PVA blend membranes S1–S6. Table 2. Mean pore size ofcurves PVDF/PVA blendblend membranes S1 –SS61.–S6. Figure 3. Pore size distribution of PVDF/PVA membranes Membrane ID S1 S2 S3 S4 S5 S6 (Polymer Concentration) 16 wt %size of 18S wt % 20blend wt %membranes 22 wt %S1–S624 wtS% 26 wt % PVDF/PVA . Membrane ID Table 2. Mean S1pore S S S6 2 3 4 5 Mean pore size (nm) 81.7 73.6% 64.6 48.0 36.2wt % 27.6 (Polymer Concentration) 16 wt % 18 wt 20 wt % 22 wt % 24 26 wt % Membrane ID S1 S2 S3 S4 S5 S6 (Polymer Concentration) 16 wt % 18 wt % 2064.6 wt % 22 wt % 24 wt36.2 % 26 wt 27.6 % Mean pore 81.7 73.6 48.0 3.1.3. Effect of size Total(nm) Polymer Concentration on Water Flux Mean pore size (nm) 81.7 73.6 64.6 48.0 36.2 27.6

The effect of total polymer concentration on membrane flux is shown in Figure 4, in which the 3.1.3. of the Total Polymer Concentration Fluxof the total polymer concentration. This is waterEffect flux of membrane decreases withon theWater increase mainly dueof tototal the decreasing pore size, as indicated in Table 2. At is a total polymer concentration of The effect polymer concentration on membrane flux shown in Figure in which The effect of total polymer concentration on membrane flux is shown in Figure 4, in4,which the the −1. 26flux wt %, the water flux of the PVDF/PVA blend membraneofS6the is about 370 L·m−2·hconcentration. waterwater of the membrane decreases with the increase total polymer is flux of the membrane decreases with the increase of the total polymer concentration. This This is mainly due due to the decreasing pore inTable Table2.2.AtAta total a total polymer concentration of mainly to the decreasing poresize, size,as asindicated indicated in polymer concentration of −. 2 ·h−1 . −2·· −1 %, water the water of the PVDF/PVAblend blend membrane about 370 L· m h 26 wt26%,wtthe fluxflux of the PVDF/PVA membraneS6Sis is about 370 L m 6

3.1.3. Effect of Total Polymer Concentration on Water Flux

Figure 4. Effect of total polymer concentration on water flux for PVDF/PVA blend membranes S1–S6.

3.1.4. Effect of Total Polymer Concentration on Membrane Mechanical Properties

Figure 4. Effect of total polymer concentration on water flux for PVDF/PVA blend membranes S1–S6.

Figure 4. Effect of total onconcentration water flux foron PVDF/PVA blendtensile membranes S1 –S 6. Figure 5 shows thepolymer effect ofconcentration total polymer the membrane strength for 3.1.4. Effect of Total Polymer Concentration on Membrane Mechanical Properties PVDF/PVA blend membranes S1–S6. In general, the tensile strength of the membranes increases from 3.1.4.0.5 Effect of Total Concentration on Membrane Mechanical Properties MPa–4.5 MPaPolymer with increasing total polymer concentration from 16 wt % (S1)–26 wt % (S6). The Figure 5 shows the effect of total polymer concentration on the membrane tensile strength for enhancement of mechanical strength is due to the change of the membrane structure from a long Figure 5 shows effect ofStotal polymer concentration on of the membrane strength PVDF/PVA blendthe membranes 1–S6. In general, the tensile strength the membranestensile increases from for 0.5 MPa–4.5 MPa with increasing polymer concentration from 16 wt % (S 1)–26 wt % (S 6).increases The PVDF/PVA blend membranes S1 –Stotal . In general, the tensile strength of the membranes 6 mechanical strength istotal duepolymer to the change of the membrane long fromenhancement 0.5 MPa–4.5of MPa with increasing concentration from 16structure wt % (Sfrom )–26a wt % (S ). 1

6

The enhancement of mechanical strength is due to the change of the membrane structure from a long

Membranes 2016, 6, 47 Membranes 2016, 6, 47

8 of 16 8 of 16

finger-like pore to a cellular pore structure, which led to decreased pore size and a narrower pore size finger-like pore Membranes 2016, 6, 47to a cellular pore structure, which led to decreased pore size and a narrower 8pore of 16 distribution, as the total polymer concentration increased (Figure 1). This is also consistent with the size distribution, as the total polymer concentration increased (Figure 1). This is also consistent with decreasing porosity, asashown inpore Figure 2, which was also observed inin previous [25]. finger-like poreporosity, to cellular structure, led toalso decreased pore size andwork awork narrower the decreasing as shown in Figure 2,which which was observed previous [25]. pore size distribution, as the total polymer concentration increased (Figure 1). This is also consistent with the decreasing porosity, as shown in Figure 2, which was also observed in previous work [25].

Figure 5. Effect of total polymer concentration on membrane tensile strength for PVDF/PVA blend

Figure 5. Effect of total polymer concentration on membrane tensile strength for PVDF/PVA blend membranes S1–S6. membranes S1 –S6 . Figure 5. Effect of total polymer concentration on membrane tensile strength for PVDF/PVA blend

3.2. Investigation membranes Son 1–Sthe 6. PVDF/PVA Blend Ratio

3.2. Investigation on the PVDF/PVA Blend Ratio With a pureon PVDF membraneBlend N as the benchmark, hydrophilically-modified PVDF/PVA blend 3.2. Investigation the PVDF/PVA Ratio

With a pureMPVDF N as the benchmark, hydrophilically-modified PVDF/PVA blend membranes and Smembrane 3 (Table 1) with the same total polymer concentration of 20 wt % at varying membranes M and S (Table 1) with the total polymer concentration ofPVDF/PVA 20 wt structure % at varying 3 With a blend pure PVDF as the hydrophilically-modified blend PVDF/PVA ratiosmembrane 9:1 (M) andN8:2 (Ssame 3) benchmark, were fabricated for comparing the membrane PVDF/PVA blend ratios 9:1 (M) and 8:2 (S ) were fabricated for comparing the membrane structure membranes and S3 (TableThe 1) effect with the total polymer of 20 wt %mechanical at varying and filtrationMperformance. of the PVDF/PVA blendconcentration ratio on the membrane 3same and filtration performance. The of (S the PVDF/PVA blend ratio the membrane membrane mechanical PVDF/PVA ratios 9:1 (M) and 8:2 3) were fabricated for comparing the structure properties isblend shown in Figure 6,effect in which the tensile strength of the blendon membrane decreases from and filtration The effect ofthe thetensile PVDF/PVA blend on8:2. theWhen membrane mechanical about 1.7 MPaperformance. down to 0.8 MPa increasing blend ratio from 10:0 to the blend ratio of from properties is shown in Figure 6, inwith which strength ofratio the blend membrane decreases properties shown Figure in which the tensile strength offrom thestrength blend decreases from ratio increased toMPa 7:3 6, (membrane L; Table 1), the tensile the When PVDF/PVA blend aboutPVA 1.7 further MPaisdown toin0.8 with increasing blend ratio 10:0 membrane toof8:2. the blend about 1.7 MPa down to 0.8 MPa with increasing blend ratio from 10:0 to 8:2. When the blend ratio ofblend membrane was as weak as 0.16 MPa, and the structure strength of blend membrane was poor for the of PVA further increased to 7:3 (membrane L; Table 1), the tensile strength of the PVDF/PVA PVA further test; increased 7:3 the (membrane L; lower Table than 1), the strength of the PVDF/PVA blend performance hence,toonly blend ratio 7:3tensile was considered appropriate. membrane was as weak as 0.16 MPa, and the structure strength of blend membrane was poor for the membrane was as weak as 0.16 MPa, and the structure strength of blend membrane was poor for the performance test; hence, only the blend ratio lower than 7:3 was considered appropriate. performance test; hence, only the blend ratio lower than 7:3 was considered appropriate.

Figure 6. Effect of the PVDF/PVA blend ratio on tensile strength (total polymer concentration of 20 wt %). Figure 6. Effect the PVDF/PVA ratio on tensile strength (total polymer concentration Figure 6. Effect of theofPVDF/PVA blendblend ratio on tensile strength (total polymer concentration of 20 wt %). of 20 wt %).


9 of 16 9 of 16

3.2.1. Effect of PVDF/PVA Blend Ratio Ratio on on Membrane Membrane Morphology Morphology PVDF/PVA Blend The SEM SEM images images ofof the thesurface surfaceofofmembranes membranes S3 are shown in Figure 7. N, N, MM andand S3 are shown in Figure 7. The The morphologies thesurface top surface (top layer) of PVDF/PVA blend membranes Mratio (blend morphologies of theoftop (top layer) of PVDF/PVA blend membranes M (blend 9:1)ratio and 9:1) and Sratio ratiorough 8:2) are rough with microvoids; whilePVDF the pure PVDF membrane N has a S3 (blend 8:2) are with microvoids; while the pure membrane N has a smooth top 3 (blend smooth surface without observable With increasing PVAthe content, the topbecomes surface surface top without observable microvoids.microvoids. With increasing PVA content, top surface becomes rougher moreThis porous. is due to the integration enhanced integration between theofinterface of rougher and moreand porous. is dueThis to the enhanced between the interface PVDF and PVDF the PVA content increases [11]. all Meanwhile, all of the bottom surface (sublayer) PVA asand the PVA PVA as content increases [11]. Meanwhile, of the bottom surface (sublayer) morphologies morphologies exhibit a network bicontinuous network porous and are significantly different the exhibit a bicontinuous porous structure andstructure are significantly different from the topfrom surface top surface morphologies. The combined results from Figure 7a,b suggest that the top layer and the morphologies. The combined results from Figure 7a,b suggest that the top layer and the bottom layer bottom layer of thewere membrane formed by different phasemechanisms. separation mechanisms. of the membrane formedwere by different phase separation Specifically,Specifically, during the during the formation of PVDF/PVA blend membrane, TIPS and occurred NIPS occurred simultaneously when formation of PVDF/PVA blend membrane, TIPS and NIPS simultaneously when the the homogeneous solution immersed the coagulation of water, which serves as homogeneous dopedope solution was was immersed into into the coagulation bath bath of water, which serves as both both a coagulant (in NIPS) a coolant TIPS) . The solvent contactwith withthe theouter outersurface surface of of the a coagulant (in NIPS) andand a coolant (in (in TIPS) . The solvent inincontact casted layer (facing the coagulant side; Section 2) would exchange with water and then trigger the occurrence process, which determines the structure of the top layertop (as layer shown(as in Figure occurrence ofofthe theNIPS NIPS process, which determines the structure of the shown7a). in Meanwhile, for the innerfor surface of thesurface casted of layer glass plate;the Section the Section heat exchange Figure 7a). Meanwhile, the inner the(facing casted the layer (facing glass 2), plate; 2), the ◦ C) and ◦ C), coolant occurred between high temperature casting solution (140 coolant (140 water°C) (25and inducingwater TIPS heat exchange occurred between high temperature casting solution and subsequently formation of the bicontinuous pore structure of the bottom pore sublayer (as shown in (25 °C), inducing the TIPS and subsequently the formation of the bicontinuous structure of the Figure The TIPS mechanism contributed to the solid-liquid separation crystallization bottom7b). sublayer (as shown in Figure 7b). Theheavily TIPS mechanism contributed heavilyand to the solid-liquid in the polymer-rich phase and in eventually to the formation of the network to porous structure of the separation and crystallization the polymer-rich phase and eventually the formation bottom the membrane. This occurred compared to the mass-transfer dominant networksurface porousof structure of the bottom surface ofbecause the membrane. This occurred because compared NIPS in the dominant top layer, NIPS the heat transfer (dominant TIPS) much faster for forming to the process mass-transfer process in the top layer,inthe heatistransfer (dominant in TIPS)the is bottom surface [25]. This has explained the fundamental difference in the top and bottom much faster for forming thehence bottom surface [25]. This hence has explained the fundamental difference surface morphologies. in the top and bottom surface morphologies.

Figure 7. surfaces of of N, N, M M and and S S33:: (a) Figure 7. SEM SEM images images of of membrane membrane top top and and bottom bottom surfaces (a) top top surface surface (top (top layer); layer); (b) bottom surface (sublayer). (b) bottom surface (sublayer).

The SEM images of the cross-section of membranes N, M and S3 are shown in Figure 8; Rows (a), The SEM images of the cross-section of membranes N, M and S3 are shown in Figure 8; Rows (a), (b) and (c) are the full cross-section, the cross-section adjacent to the top surface and the cross-section (b) and (c) are the full cross-section, the cross-section adjacent to the top surface and the cross-section adjacent to the bottom surface, respectively. Specifically, in Figure 8a, the finger-like pores beneath adjacent to the bottom surface, respectively. Specifically, in Figure 8a, the finger-like pores beneath the the top surface for the pure PVDF membrane N were formed mainly based on the NIPS mechanisms. top surface for the pure PVDF membrane N were formed mainly based on the NIPS mechanisms. This is This is because during membrane formation, the water (nonsolvent) quickly diffuses into the polymer because during membrane formation, the water (nonsolvent) quickly diffuses into the polymer solution solution beneath the top surface; hence, the rapid exchange rate of water and diluent CPL leads to beneath the top surface; hence, the rapid exchange rate of water and diluent CPL leads to the formation the formation of the finger-like structure beneath the top surface. As the PVA content increases for the PVDF/PVA blend membranes M and S3, the number of the finger-like pores beneath the top


10 of 16


10 of 16

of the finger-like structure beneath the top surface. As the PVA content increases for the PVDF/PVA blend andisSdue of the finger-like poresPVA beneath surfacedopes decreases. 3 , thetonumber surfacemembranes decreases. M This the addition of hydrophilic intothe thetop polymer that This is due to the addition of hydrophilic PVA into the polymer dopes that increases the precipitation increases the precipitation rate of the casting solution [14]. This may result in the formation of cellular rate of the casting [14]. This may result in the formationinofthe cellular via typical morphologies via solution typical liquid-liquid (L-L) phase separation NIPSmorphologies process [21]. Consistent liquid-liquid (L-L) phase separation in the NIPS process [21]. Consistent with the observation with the observation in Figure 7, the cross-sectional morphologies adjacent to the bottom surface in of Figure 7, the cross-sectional morphologies adjacent to the bottom surface of all membranes exhibit all membranes exhibit a bicontinuous network porous structure. This is induced by the TIPS amechanism, bicontinuous network structure. is induced by the TIPS is mechanism, in which the in which theporous crystallization for This macromolecules (i.e., PVDF) much slower as a result crystallization for macromolecules PVDF) is muchweight slowercompound as a result CPL. of chain of chain conformation compared (i.e., to low molecular As conformation CPL rapidly compared to low molecular weight compound CPL. As CPL rapidly crystallizes at its original crystallizes at its original locations in a homogenous solution, this effectively suppresses thelocations growth in homogenous this effectively suppresses growthThus, of PVDF crystals and subsequently of aPVDF crystals solution, and subsequently dictates the pore the structure. as CPL is dissolved in water, dictates pore structure. Thus, aslean CPLphase, is dissolved in water, thebecomes space occupied by pores the polymer the spacethe occupied by the polymer or CPL-rich phase, membrane with a lean phase, or CPL-rich phase, becomes membrane pores with a well-connected bicontinuous network well-connected bicontinuous network structure [25]. Hence, both surface and cross-sectional structure [25]. Hence, both1,surface cross-sectional morphologies in Figures 7 and 8 confirm that morphologies in Figures 7 and 8and confirm that the formation mechanisms of 1, the blend membranes the of the blendNTIPS. membranes are combined NIPS and TIPS, namely NTIPS. are formation combined mechanisms NIPS and TIPS, namely

Figure 8. 8. SEM SEM images images of of membrane membrane cross-section cross-section of of N, N, M, M, SS33:: (a) (a) full full cross-section; cross-section; (b) (b) cross-section cross-section Figure adjacent to to the the top top surface; surface; (c) (c) cross-section cross-section adjacent adjacent to to the the bottom bottom surface. surface. adjacent

3.2.2. Effect Effect of of the the PVDF/PVA PVDF/PVA Blend Distribution 3.2.2. Blend Ratio Ratio on on Membrane Membrane Pore Pore Size Size and and Pore Pore Size Size Distribution The pore pore size size distribution distribution curves curves of of membranes membranes N, N, M M and and SS3 are depicted in Figure 9, and the The 3 are depicted in Figure 9, and the corresponding mean pore size is given in Table 3. As the PVA content increases, the the pore pore size size corresponding mean pore size is given in Table 3. As the PVA content increases, distribution becomes becomes wider, corresponding to the mean increasing pore size from distribution wider, corresponding to the increasing pore size mean from 34.3 nm (N)–64.6 nm 34.3 nm (N)–64.6 nm (S 3) in Table 3. This is consistent with the morphological observations from (S3 ) in Table 3. This is consistent with the morphological observations from Figures 7 and 8. Figures 7 and 8.


11 of 16


11 of 16

Figure 9. Pore size distribution curves of membranes N, M and S3 (total polymer concentration of 20 wt %).

Figure 9. Pore size distribution curves of membranes N, M and S (total polymer concentration of Figure 9. Pore size distribution curves of membranes N, M and S3 (total3polymer concentration of 20 wt %). 20 wt %). Table 3. Mean pore size of membranes N, M and S3.

Table 3.ID Mean pore pore size size of of membranes M N, M M and and SS33.. S3 Table 3. Mean N, Membrane N membranes (PVDF:PVA Blend Ratio) (10:0) (9:1) (8:2) Membrane ID S3 N M Mean pore size (nm) 34.3 41.6 64.6 Membrane ID N M S 3

(PVDF:PVA Blend Ratio) (10:0) (9:1) (8:2) (PVDF:PVA Blend Ratio) (10:0) (9:1) (8:2) Mean pore size (nm) 34.3 41.6 64.6 3.2.3. Effect of the PVDF/PVA Blend Ratio on Water Flux Mean pore size (nm) 34.3 41.6 64.6 Figure 10 PVDF/PVA illustrates theBlend effect of the PVDF/PVA blend ratio on the water flux. As the PVDF/PVA 3.2.3. Effect of the Ratio on Water Flux blend ratio decreases from 10:0 to 8:2, the water flux increases significantly. The enhanced 3.2.3. Effect of the PVDF/PVA Blend Ratio on Water Flux Figure 10 illustrates the effect of the PVDF/PVA blend ratioinon the water As the PVDF/PVA permeability is mainly due to the larger pore size (as shown Table 3) and flux. the wider pore size Figure 10 illustrates the effect of the PVDF/PVA blend ratio on the water flux. As the PVDF/PVA with thefrom increase PVDF/PVA results are also consistent with the blenddistribution ratio decreases 10:0of to 8:2, the blend waterratio. fluxThe increases significantly. The enhanced of these membranes shown in increases Figures 7 and i.e., with the number ofpermeability surface blendstructural ratio decreases from 10:0 8:2, the water flux significantly. enhanced permeability isvariation mainly due to to the larger pore size (as shown in8, Table 3)The and the wider pore size poresdue and pore size increasing, water fluxin increases. substantial on membrane is mainly to the larger pore size (as shown Tableratio. 3)The andThe the results widerimprovement pore size distribution with the distribution with the increase of the PVDF/PVA blend are also consistent with permeability is also attributed to the higher hydrophilicity of the blend membranes, which will bethese increase ofvariation PVDF/PVA blendmembranes ratio. The results areinalso consistent thewith structural variation structural of these shown Figures 7 andwith 8, i.e., the number of of surface discussed in Section 3.2.4.

membranes shown Figures 7 the andwater 8, i.e.,flux withincreases. the number surface pores and pore size increasing, pores and pore size in increasing, Theofsubstantial improvement on membrane the water fluxisincreases. The substantial improvement on membrane permeability is alsowhich attributed to permeability also attributed to the higher hydrophilicity of the blend membranes, will be the higherin hydrophilicity discussed Section 3.2.4. of the blend membranes, which will be discussed in Section 3.2.4.

Figure 10. Effect of the PVDF/PVA blend ratio on water flux (total polymer concentration of 20 wt %).

Figure 10.Effect Effectof the of the PVDF/PVA ratio onflux water (totalconcentration polymer concentration Figure 10. PVDF/PVA blendblend ratio on water (totalflux polymer of 20 wt %). of 20 wt %).

Membranes 2016, 2016, 6, 6, 47 47 Membranes

12 of of 16 16 12

3.2.4. Hydrophilicity of PVDF/PVA Blend Membranes 3.2.4. Hydrophilicity of PVDF/PVA Blend Membranes To study the membrane hydrophilicity, the dynamic water contact angles of the PVDF/PVA To study the membrane hydrophilicity, the dynamic water contact angles of the PVDF/PVA blend blend membranes M and S3 were measured and compared against the pure PVDF membrane N. The membranes M and S3 were measured and compared against the pure PVDF membrane N. The results results are shown in Figure 11, in which the water contact angle decreases more significantly with are shown in Figure 11, in which the water contact angle decreases more significantly with increasing increasing PVDF/PVA blend ratio from pure PVDF (N) to 8:2 (S3). For the control membrane N, the PVDF/PVA blend ratio from pure PVDF (N) to 8:2 (S3 ). For the control membrane N, the contact angle contact angle decreases from the initial value ◦of about 86.5° and reaches a plateau of 73°; while that decreases from the initial value of about 86.5 and reaches a plateau of 73◦ ; while that of the blend of the blend membrane S3 decreases most drastically from 57° down to 0° in 25 s, indicating complete membrane S3 decreases most drastically from 57◦ down to 0◦ in 25 s, indicating complete penetration penetration of water into the membrane matrix. Hence, it is obvious that the hydrophilic properties of water into the membrane matrix. Hence, it is obvious that the hydrophilic properties of PVDF/PVA of PVDF/PVA blend membranes were prominently improved with the addition of PVA. blend membranes were prominently improved with the addition of PVA.

Figure 11. 11. Dynamic Dynamic water water contact contact angles angles of of membranes membranes N, (total polymer polymer concentration concentration of of Figure N, M M and and SS33 (total 20 wt wt %). %). 20

3.3. Antifouling Antifouling Performance Performance and and Membrane Membrane Resistance Resistance Analysis Analysis 3.3. Hydrophilicity is an important factor affecting the fouling behaviour and filtration performance applications. The The newly-prepared newly-preparedPVDF/PVA PVDF/PVAblend blendmembranes membranes and selected 3 were for MF/UF MF/UF applications. MM and S3Swere selected to to examine the antifouling performance, benchmarking against the pure PVDF membrane N. examine the antifouling performance, benchmarking against the pure PVDF membrane N. As shown in Figure after physical physical cleaning cleaning Figure 12, 12, the thenormalized normalizedfluxes fluxesJfJ/J f/Jii due to fouling fouling and and JJrr/J /Jii after In the the first first filtration filtration step as were used to evaluate the antifouling performance of the membranes. In described in Section 2.4, the water flux (J ) declined slightly during the first 30 min for all tested described ii) declined slightly during the first 30 membranes due due to pre-pressurizing. pre-pressurizing. However, However, in in the the second second filtration filtration step, step, the the relative relative water water flux membranes /Ji)i )decreased decreasedsignificantly significantly when BSA solution The ratios the water of the (Jff/J when thethe BSA solution waswas fed. fed. The ratios of theof water flux offlux the fouled fouled membrane to their initial fluxes, i.e., J /J , of the membranes N, M and S reached plateau membrane to their initial fluxes, i.e., Jf/Ji, of the N, M and S3 reached 3plateau values of f membranes i values(N), of 24.7% (N),and 39%46.4% (M) and (S3 ), respectively. Afterwith cleaning within DIthe water infiltration, the third 24.7% 39% (M) (S3),46.4% respectively. After cleaning DI water third filtration, the water flux of all membranes was partially restored. However, the effect of physical the water flux of all membranes was partially restored. However, the effect of physical cleaning on cleaning on flux recovery is more efficient for the blend membranes M and S . Specifically, for thePVDF pure flux recovery is more efficient for the blend membranes M and S3. Specifically, for the pure 3 PVDF membrane the water recovery /Ji stabilizes at 64.4% aftercleaning; cleaning;while whilefor for the membrane N, theN,water flux flux recovery ratioratio Jr/JiJrstabilizes at 64.4% after hydrophilically-modified PVDF membranes M and S , J /J are 77.6% and 91.1%, respectively. Hence, hydrophilically-modified PVDF membranes M and S33 rr/Ji iare 77.6% and 91.1%, respectively. Hence, hydrophilically-modified membranes, the pure is more susceptible compared totothethe hydrophilically-modified membranes, thePVDF puremembrane PVDF membrane is more to fouling caused by organic the foulants stickfoulants to the membranes’ surface moresurface tightly, susceptible to fouling causedmolecules, by organicand molecules, and the stick to the membranes’ which poses challenge for cleaning and long-term operation. The NTIPS PVDF membranes showed more tightly, which poses challenge for cleaning and long-term operation. The NTIPS PVDF great potential for mitigating organicfor fouling by appropriately controlling the fabrication conditions. membranes showed great potential mitigating organic fouling by appropriately controlling the

fabrication conditions.


13 of 16 13 of 16

To further understand the fouling characteristics of the pure PVDF membrane N and To further understand the fouling characteristics of the pure PVDF membrane N and hydrophilically-modified PVDF membranes M and S3 , the local resistances Rm , Rrevf , R13irrfof and Rt Membranes 2016, 6, 47 16 hydrophilically-modified PVDF membranes M and S3, the local resistances Rm, Rrevf, Rirrf and Rt are are calculated based on Equations (3)–(5). The calculation results are shown in Table 4 and Figure 13. calculated based on Equationsthe (3)–(5). The calculation results shown Table 4membrane and Figure N 13. and The To further understand fouling characteristics ofare the pureinPVDF 12 m−1 , The total resistance Rt t of of membranes membranes N,MMand andS3S3follows follows adecreasing decreasing order of 7.02 ×1210 −1, total resistance R N, a order of 7.02 × 10 m hydrophilically-modified PVDF membranes M and S 3 , the local resistances R m , R revf , R irrf and R t are 1 and 0.49 × 1012 m−1 , respectively. Thus, a significant reduction of R was obtained 1.59 × 1012 m12−m −1 and 0.49 × 1012 m−1, respectively. Thus, a significant reduction of Rt was obtained t 1.59 × 10 for calculated based on Equations (3)–(5). The calculation results are shown in Table 4 and Figure 13. The for membrane SS33 due to the higher PVA proportion. proportion.Consistent Consistent with flux recovery results, this membrane due higher PVA with the the flux this −1, total resistance Rt to of the membranes N, M and S3 follows a decreasing orderrecovery of 7.02 results, × 1012 m confirms the benefits of the hydrophilic modification for obtaining better filtration membranes with confirms the benefits of the hydrophilic modification for obtaining better filtration membranes with 1.59 × 1012 m−1 and 0.49 × 1012 m−1, respectively. Thus, a significant reduction of Rt was obtained for low fouling and, hence, improvedlong-term long-term performance. low fouling propensity and, hence, improved performance. membrane Spropensity 3 due to the higher PVA proportion. Consistent with the flux recovery results, this confirms the benefits of the hydrophilic modification for obtaining better filtration membranes with low fouling propensity and, hence, improved long-term performance.

Figure Antifouling performance performance of membranes N, M and S3 (total of 20 wt %). Figure 12. 12. Antifouling of membranes N, M and polymer S3 (totalconcentration polymer concentration of 20 wt %). Table 4. Calculation of the membrane resistance for membranes N, M and S3. Figure 12. Antifouling performance of membranes N, M and S3 (total polymer concentration of 20 wt %). Table 4. Calculation of the membrane resistance for membranes N, M and S3 . S3 Local resistance N M Table 4. Calculation of membrane resistance for membranes N, M and S3. Rmthe (1012 m−1) 1.76 0.60 0.22 Local Resistance N M 0.02 S3 Rirrf (1012 m−1) 1.03 0.18 S3 Local resistance N M 12 −1 12 (10 Rrevf 4.22 0.81 0.24 0.22 0.60 0.22 Rm (10 m12−1mm ) −1) ) 1.76 Rm (10 1.76 0.60 R12 t (1012−m 7.02 1.59 0.49 0.02 1 )−1) Rirrf (10 m12 m 1.03 0.18 0.02 −1) Rirrf (10 1.03 0.18 Rirrf/R m 58.5% 30.0% 9.1% 12 − 1 Rrevf (10 m12 m)−1) 4.22 0.81 0.24 0.24 Rrevf (10 4.22 0.81 −1 ) −1 7.02 1.59 0.49 0.49 Rt (10R12t (10 m12 m ) 7.02 1.59 Rirrf /R 58.5% 30.0% 9.1%9.1% Rirrf 58.5% 30.0% m/Rm

Figure 13. Fouling resistance analysis for membranes N, M and S3 (total polymer concentration of 20 wt %).

Figure 13. Fouling resistance analysis polymerconcentration concentration of Figure 13. Fouling resistance analysisfor formembranes membranes N, N, M M and and SS33 (total (total polymer 20 wt %). 20 wtof%).


14 of 16

Furthermore, as the PVDF/PVA blend ratio increases from 10:0 to 8:2, the value of Rirrf /Rm decreases from 58.5% down to 9.1%, as shown in Figure 13. Thus, the proportion of irreversible fouling resistance Rirrf of the hydrophilically-modified PVDF membrane S3 is more significantly decreased compared to that of the control membrane N, resulting in less irreversible fouling, such as pore blocking and adsorption onto the membrane surface. Overall, the hydrophilic modification of the PVDF membranes exhibited good antifouling properties. 4. Conclusions From this study, the following conclusions can be drawn: (1)

(2)

(3)

(4)

(5)

A novel nonsolvent thermally-induced phase separation (NTIPS) method was successfully employed to prepare hydrophilically-modified PVDF membranes. The PVDF/PVA blend membranes exhibited improved hydrophilicity, higher water permeability and enhanced fouling propensity. As the total polymer concentrations increased, the pore size, porosity and water flux of the PVDF/PVA blend membranes reduced, and the mechanical strength was improved. The membrane pore size could be deliberately tuned to meet separation requirements. Both surface and cross-sectional morphologies suggested that the formation of the hydrophilically-modified PVDF/PVA blend membranes was due to NTIPS mechanisms. Different from the top surface structure, which was mainly formed via the NIPS mechanism, the bottom surface of all membranes exhibited a bicontinuous network induced by TIPS. The dynamic water contact angle of the modified membrane dropped more rapidly indicating improved hydrophilicity with the addition of PVA. However, the ratio of PVA to PVDF should be carefully chosen with the considerations of membrane mechanical strength and filtration performance. Membrane resistance analysis revealed that the hydrophilically-modified PVDF membranes had lower total resistance of mass transfer (hence, higher permeability) and showed great potential for mitigating irreversible fouling. High performance MF/UF membranes with the desired pore size can be achieved by optimizing fabrication parameters in NTIPS. In future research, surface functionalization could be incorporated into this work to obtain advanced composite membranes with further improved hydrophilicity and antifouling properties for wastewater treatment.

Acknowledgments: This study was financially supported by the Social Development Research Projects from Science and Technology Department of Zhejiang Province (No. 2016C33023), the General Project of Department of Education of Zhejiang Province (No. Y201533832), the Ningbo International Science Technology Cooperation Project (No. 2014D10017) and the New-Shoot Talents Program of Zhejiang Province (No. 2016R405024). This work was also sponsored by the K.C. Wong Magna Fund from Ningbo University. Xing Yang gratefully thanks the support given by City West Water, Victoria, Australia, and the Industry Fellowship by Victoria University, Australia. Author Contributions: Tonghu Xiao and Ningen Hu conceived of and defined the problem. Ningen Hu, Tonghu Xiao and Xinhai Cai developed the methodology and case studies. Ningen Hu wrote the paper. Yuhua Fu and Lining Ding contributed to the discussion. Xing Yang reviewed and contributed to the discussion and paper structure. Conflicts of Interest: The authors declare no conflict of interest.

References 1. 2.

Le-Clech, P.; Chen, V.; Fane, T.A.G. Fouling in membrane bioreactors used in wastewater treatment. J. Membr. Sci. 2006, 284, 17–53. [CrossRef] Bruggen, B.V.D.; Vandecasteele, C.; Gestel, T.V.; Doyen, W.; Leysen, R. A review of pressure-driven membrane processes in wastewater treatment and drinking water production. Environ. Prog. 2003, 22, 46–56. [CrossRef]


3.

4. 5.

6. 7.

8. 9. 10.

11. 12. 13.

14. 15.

16.

17. 18. 19. 20.

21. 22. 23.

24.

15 of 16

Sukitpaneenit, P.; Chung, T.S. Molecular design of the morphology and pore size of PVDF hollow fiber membranes for ethanol-water separation employing the modified pore-flow concept. J. Membr. Sci. 2011, 374, 67–82. [CrossRef] Kang, G.D.; Cao, Y.M. Application and modification of poly(vinylidene fluoride) (PVDF) membranes— A review. J. Membr. Sci. 2014, 463, 145–165. [CrossRef] Kyllönen, H.; Pirkonen, P.; Nyström, M.; Nuortila-Jokinen, J.; Grönroos, A. Experimental aspects of ultrasonically enhanced cross-flow membrane filtration of industrial wastewater. Ultrason. Sonochem. 2006, 13, 295–302. [CrossRef] [PubMed] Argüello, M.A.; Álvarez, S.; Riera, F.A.; Álvarez, R. Enzymatic cleaning of inorganic ultrafiltration membranes used for whey protein fractionation. J. Membr. Sci. 2003, 216, 121–134. [CrossRef] Chen, S.C.; Fu, X.Z.; Chung, T.S. Fouling behaviors of polybenzimidazole (PBI)-polyhedral oligomericsilsesquioxane (POSS)/polyacrylonitrile (PAN) hollow fiber membranes for engineering osmosis processes. Desalination 2014, 335, 17–26. [CrossRef] Juang, R.S.; Lin, K.H. Flux recovery in the ultrafiltration of suspended solutions with ultrasound. J. Membr. Sci. 2004, 243, 115–124. [CrossRef] Muthukumaran, S.; Kentish, S.E.; Ashokkumar, M.; Stevens, G.W. Mechanisms for the ultrasonic enhancement of dairy whey ultrafiltration. J. Membr. Sci. 2005, 258, 106–114. [CrossRef] Muthukumaran, S.; Kentish, S.; Lalchandani, S.; Ashokkumar, M.; Mawson, R.; Stevens, G.W.; Grieser, F. The optimisation of ultrasonic cleaning procedures for dairy fouled ultrafiltration membranes. Ultrason. Sonochem. 2005, 12, 29–35. [CrossRef] [PubMed] Li, N.N.; Xiao, C.F.; An, S.L.; Hu, X.Y. Preparation and properties of PVDF/PVA hollow fiber membranes. Desalination 2010, 250, 530–537. [CrossRef] Rana, D.; Matsuura, T. Surface modifications for antifouling membranes. Chem. Rev. 2010, 110, 2448–2471. [CrossRef] [PubMed] Bonyadi, S.; Chung, T.S.; Krantz, W.B. Investigation of corrugation phenomenon in the inner contour of hollow fibers during the non-solvent induced phase-separation process. J. Membr. Sci. 2007, 299, 200–210. [CrossRef] Deshmukh, S.P.; Li, K. Effect of ethanol composition in water coagulation bath on morphology of PVDF hollow fibre membranes. J. Membr. Sci. 1998, 150, 75–85. [CrossRef] Edwie, F.; Teoh, M.M.; Chung, T.S. Effects of additives on dual-layer hydrophobic–hydrophilic PVDF hollow fiber membranes for membrane distillation and continuous performance. Chem. Eng. Sci. 2012, 68, 567–578. [CrossRef] Young, T.H.; Cheng, L.P.; Lin, D.J.; Fane, L.; Chuang, W.Y. Mechanisms of PVDF membrane formation by immersion-precipitation in soft (1-octanol) and harsh (water) nonsolvents. Polymer 1999, 40, 5315–5323. [CrossRef] Gu, M.H.; Zhang, J.; Wang, X.L.; Tao, H.J.; Ge, L.T. Formation of poly (vinylidene fluoride) (PVDF) membranes via thermally induced phase separation. Desalination 2006, 192, 160–167. [CrossRef] Li, D.M.; Krantz, W.B.; Greenberg, A.R.; Sani, R.L. Membrane formation via thermally induced phase separation (TIPS): Model development and validation. J. Membr. Sci. 2006, 279, 50–60. [CrossRef] Li, X.F.; Wang, Y.G.; Lu, X.L.; Xiao, C.F. Morphology changes of polyvinylidene fluoride membrane under different phase separation mechanisms. J. Membr. Sci. 2008, 320, 477–482. [CrossRef] Lin, D.J.; Chang, H.H.; Chen, T.C.; Lee, Y.C.; Cheng, L.P. Formation of porous poly(vinylidene fluoride) membranes with symmetric or asymmetric morphology by immersion precipitation in the water/TEP/PVDF system. Eur. Polym. J. 2006, 42, 1581–1594. [CrossRef] Van de Witte, P.; Dijkstra, P.J.; Van den Berg, J.W.A.; Feijen, J. Phase separation processes in polymer solutions in relation to membrane formation. J. Membr. Sci. 1996, 117, 1–31. [CrossRef] Atkinson, P.M.; Lloyd, D.R. Anisotropic flat sheet membrane formation via TIPS: Atmospheric convection and polymer molecular weight effects. J. Membr. Sci. 2000, 175, 225–238. [CrossRef] Ji, G.L.; Zhu, B.K.; Cui, Z.Y.; Zhang, C.F.; Xu, Y.Y. PVDF porous matrix with controlled microstructure prepared by TIPS process as polymer electrolyte for lithium ion battery. Polymer 2007, 48, 6415–6425. [CrossRef] Holland, B.J.; Hay, J.N. The thermal degradation of poly(vinyl alcohol). Polymer 2001, 42, 6775–6783. [CrossRef]


25.

26. 27.

28. 29.

30. 31.

32. 33. 34.

35. 36. 37. 38.

39.

16 of 16

Xiao, T.H.; Wang, P.; Yang, X.; Cai, X.H.; Lu, J. Fabrication and characterization of novel asymmetric polyvinylidene fluoride (PVDF) membranes by the nonsolvent thermally induced phase separation (NTIPS) method for membrane distillation applications. J. Membr. Sci. 2015, 489, 160–174. [CrossRef] Zhang, L.P.; Chen, G.W.; Tang, H.W.; Cheng, Q.Z.; Wang, S.Q. Preparation and Characterization of composite membranes of Polysulfone and Microcrystalline Cellulos. J. Appl. Polym. Sci. 2009, 112, 550–556. [CrossRef] Calvo, J.I.; Bottino, A.; Capannelli, G.; Hernández, A. Comparison of liquid-liquid displacement porosimetry and scanning electron microscopy image analysis to characterize ultrafiltration track-etched membranes. J. Membr Sci. 2004, 239, 189–197. [CrossRef] Sanz, J.M.; Jardines, D.; Bottino, A.; Capannelli, G.; Hernández, A.; Calvo, J.I. Liquid–liquid porometry for an accurate membrane characterization. Desalination 2006, 200, 195–197. [CrossRef] Calvo, J.I.; Peinador, R.I.; Prádanos, P.; Palacio, L.; Bottino, A.; Capannelli, G.; Hernández, A. Liquid-liquid displacement porometry to estimate the molecular weight cut-off of ultrafiltration membranes. Desalination 2011, 268, 174–181. [CrossRef] Peinador, R.I.; Calvo, J.I.; Prádanos, P.; Palacio, L.; Hernández, A. Characterisation of polymeric UF membranes by liquid–liquid displacement porosimetry. J. Membr. Sci. 2010, 348, 238–244. [CrossRef] An, Q.F.; Li, F.; Ji, Y.L.; Chen, H.L. Influence of polyvinyl alcohol on the surface morphology, separation and anti-fouling performance of the composite polyamide nanofiltration membranes. J. Membr. Sci. 2011, 367, 158–165. [CrossRef] Gohari, R.J.; Lau, W.J.; Matsuura, T.; Ismail, A.F. Effect of surface pattern formation on membrane fouling and its control in phase inversion process. J. Membr. Sci. 2013, 446, 326–331. [CrossRef] Mi, B.X.; Elimelech, M. Organic fouling of forward osmosis membranes: Fouling reversibility and cleaning without chemical reagents. J. Membr. Sci. 2010, 348, 337–345. [CrossRef] Liang, S.; Kang, Y.; Tiraferri, A.; Giannelis, E.P.; Huang, X.; Elimelech, M. Highly hydrophilic polyvinylidene fluoride (PVDF) ultrafiltration membranes via postfabrication grafting of surface-tailored silica nanoparticles. Appl. Mater. Interfaces 2013, 5, 6694–6703. [CrossRef] [PubMed] Mulder, J. Basic Principles of Membrane Technology; Springer: Kluwer Academic Publishers: Dordrecht, The Netherlands, 1996; pp. 281–311. Luján-Facundo, M.J.; Mendoza-Roca, J.A.; Cuartas-Uribe, B.; Álvarez-Blanco, S. Ultrasonic cleaning of ultrafiltration membranes fouled with BSA solution. Sep. Purif. Technol. 2013, 120, 275–281. [CrossRef] Wei, D.Q.; Tao, Y.; Zhang, Z.H.; Zhang, X.H. Effect of pre-ozonation on mitigation of ceramic UF membrane fouling caused by algal extracellular organic matters. Chem. Eng. J. 2016, 294, 157–166. [CrossRef] Shang, M.; Matsuyama, H.; Maki, T.; Teramoto, M.; Lloyd, D.R. Preparation and characterization of poly (ethylene-co-vinyl alcohol) membranes via thermally induced liquid–liquid phase separation. Appl. Polym. Sci. 2003, 87, 853–860. [CrossRef] Wang, L.; Huang, D.X.; Wang, X.D.; Meng, X.R.; Lv, Y.T.; Wang, X.; Miao, R. Preparation of PVDF membranes via the low-temperature TIPS method with diluent mixtures: The role of coagulation conditions and cooling rate. Desalination 2015, 361, 25–37. [CrossRef] © 2016 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC-BY) license (http://creativecommons.org/licenses/by/4.0/).