Temperature- and pH-responsive properties of poly(vinylidene fluoride) membranes functionalized by blending microgels

Abstract

A novel blend membrane with temperature- and pH-responsive properties was prepared by the physical blending of poly(vinylidene fluoride) (PVDF) bulk material, poly(N-isopropylacrylamide) (PNIPAAm) microgels and poly(acrylic acid) (PAA) microgels with a simple and practical procedure, which is suitable for industrial scale production. The composition and structure of the blend membrane were investigated by Fourier transform infrared spectroscopy (FT-IR), X-ray photoelectron spectroscopy (XPS) and field-emission scanning electron microscope (FESEM). The results indicated that the incorporation of PAA and PNIPAAm microgels improved the pore size, porosity and hydrophilicity of the blended membrane, leading to a higher water flux and remarkable antifouling property. In particular, the blend membrane exhibited temperature- and pH-sensitive characteristics. The dual responsive feature makes it easy to control the blend membrane’s rejection properties as well as the water flux and helps the membrane retain preferable mass transfer and separation property when responding to one of the investigated stimuli.

---

1. Introduction

Poly(vinylidene fluoride) (PVDF) membranes have been continuously investigated due to their outstanding mechanical strength, thermal stability and good chemical resistance.^1–4 However, the performance of the PVDF membrane is restricted by its initially formed structure and that the distinctly hydrophobic membrane can easily be contaminated by proteins and natural organic matters.^5,6 Introducing stimuli-responsive polymers is one of the hydrophilic modification methods. More importantly, such polymers can endow the membrane with reversible changes in the pore dimensions/size, surface wettability, and permeability in response to an environmental signal such as pH, temperature, ionic strength, chemical cues, and light or magnetic fields.^7–11 Therefore, the mass transfer and separation properties of the membranes can be easily regulated by adjusting the external stimuli. Because of all stimuli, temperature and pH are convenient to control, temperature- or pH-sensitive membranes have been well investigated.^9,10,12

Recently, intelligent PVDF membranes introducing temperature-sensitive poly(N-isopropylacrylamide) (PNIPAAm) or pH-sensitive poly(acrylic acid) (PAA) have been reported. Hu and Dickson prepared pH-sensitive membranes by in situ cross-linking PAA inside PVDF microporous substrate membranes.¹³ Another way is to fabricate the pH-responsive PVDF membrane by the radical grafting of PAA onto the porous PVDF membrane using supercritical carbon dioxide (scCO₂) as a solvent, as reported by Ferro.¹⁴ Iwata et al. studied a temperature-sensitive membrane by plasma grafting N-isopropylacrylamide onto the porous PVDF membrane surface.¹⁵ However, the limited modification in the membrane pores, in the above mentioned methods, possibly leads to the less improved membrane performance because these methods usually succeed in enriching the introduced polymers on the top and/or bottom surface of the membrane. Additionally, the high cost of membrane production owing to the special processing required will limit the commercial application of these membranes.¹⁶ In contrast, blending modification is a simple and practical method, suiting an industrial scale production and can achieve the designed functions along with the membrane preparation.

Furthermore, an improved property of these PVDF membranes with single responsivity always comes with another declined property when responding to the external stimuli. For example, an increased water flux of the membrane is always accompanied with the decrease in the rejection and fouling resistance when the hydrophilic segments become hydrophobic in response to the stimuli. It is necessary to keep the improved property of the membrane and avoid the seriously decreased performance in the membrane process. Fabricating a smart PVDF membrane with dual or multiple response properties by introducing two or more stimuli-responsive polymers is expected to meet this requirement.

Microgels show a reversible volume variation by swelling and shrinking in response to temperature or pH stimuli.¹⁷ Blending PVDF with both temperature- and pH-sensitive microgels can achieve the adjustable pore size stemming from the configuration change of the microgels, resulting in the regulation of the solution permeability and solute separation. Moreover, the dual response properties may provide more tools for regulating the membrane structure and performance and help the membrane retain preferable mass transfer and separation properties when in response to one of the stimuli conditions.

However, to our best knowledge, no study has been reported on preparing such a pH- and temperature-sensitive membrane by blending microgels into PVDF. In the present study, a dual stimuli-responsive blend PVDF membrane was prepared by blending PVDF powder, PNIPAAm and PAA microgels based on a phase inversion process. The effects of the microgels on the structure, hydrophilicity, filtration and anti-fouling properties of the membranes were investigated in detail.

2. Experimental

2.1. Materials

N-Isopropylacrylamide (NIPAAm) was purchased from Tokyo Chemical Industry Co. Ltd. (analytical reagents) and purified by recrystallization from a solution of toluene/hexamethylene (50 : 50 vol%). Acrylic acid (AA), N,N-dimethylformamide (DMF), potassium peroxodisulfate (KPS), and N,N-methylene-bis(acrylamide) (MBAA) were of analytical grade and supplied from Tianjin Guangfu Fine Chemical Research Institute, in which AA was purified by vacuum distillation. 2,2′-Azobisisobutyronitrile (AIBN) was purchased from Tianjin Yingda Rare Chemical Reagents Factory (analytical reagents) and used as received. Acetonitrile was provided by Tianjin Kemiou Chemical Reagent Co. Ltd. (analytical reagents) and purified by distillation. Ethylene glycol dimethacrylate (EGDMA) was purchased from J&K Scientific Ltd. and used as received. Bovine serum albumin (BSA) with a molecular weight of 68 000 was obtained from Solarbio Science & Technology Co. Ltd. PVDF powders (M_w = 3.52 × 10⁵, M_w/M_n = 2.3, Solvay Company, Belgium, Solef 1010) were used without any further purification. De-ionized water was used in all experiments. All the other chemicals were of analytical reagents and used as received.

2.2. Synthesis of PNIPAAm and PAA microgels

The cross-linking polymer PNIPAAm microgels were prepared by a surfactant-free emulsion polymerization method: NIPAAm (0.92 g), MBAA (0.06 g), and KPS (0.04 g) were dissolved in 50 mL of de-ionized water in a three-neck flask. Nitrogen gas was bubbled for 15 min to remove dissolved oxygen in the reaction mixture. The reaction mixture was maintained at 70 °C for 5 h at an agitation speed of 400 rpm. The obtained mixture was cooled to ambient temperature and centrifuged, and the precipitate was resuspended in de-ionized water with ultrasonic bathing and centrifuged once again; this operation step was repeated three times. The resulting PNIPAAm microgels were dried in vacuo at 50 °C for 24 h.

The cross-linked PAA microgels were synthesized by a distillation–precipitation polymerization method in acetonitrile. A mixture of 2 g AA, 0.28 g EGDMA, 0.046 g AIBN and 160 mL acetonitrile was prepared in a three-neck flask and purged with nitrogen gas for 20 min. The reaction mixture was maintained at 80 °C for 2 h with an agitation speed of 500 rpm. After polymerization, the resulting mixture was precipitated by centrifugation, and the obtained sedimentation was purified by three cycles of centrifugation, decantation, and resuspended in acetonitrile. The PAA microgels were dried in vacuo at 50 °C for 36 h.

2.3. Preparation of PVDF/(PNIPAAm-PAA) microgel blend membranes

The PVDF/(PNIPAAm-PAA) blend membranes were prepared by a phase inversion method, and the PNIPAAm and PAA microgels were dispersed in 29.24 mL DMF with ultrasonic bathing for 30 min. The PVDF powder was added to the mixture, and the polymer solution was maintained at 60 °C for 2 h with an agitation speed of 400 rpm to form a casting solution. The casting solution was degassed for 1 h under vacuum and was then cast onto a glass plate using a steel knife. Further, the glass plate was immersed in de-ionized water at room temperature. The membranes were peeled off from the glass plate and then rinsed with de-ionized water to remove the residual solvent. The resultant membranes were stored in de-ionized water for further experiments. The concentration of the solid content (PVDF powder, PNIPAAm and PAA microgels) in the casting solutions was 18 wt%, and the concentration of the microgels content (PNIPAAm and PAA) in the solid content was 20 wt% (Table 1). The obtained membranes were labeled as BM-n, in which n (n = 100-0, 75-25, 50-50, 25-75, 0-100) was the weight ratio of PAA to PNIPAAm microgels.

Table 1 Composition of the casting solution for preparing the blend membranes

Membrane label	PAA (g)	PNIPAAm (g)	PVDF (g)	DMF (mL)
BM0-0	0.00	0.00	5.00	29.24
BM100-0	1.00	0.00	4.00	29.24
BM75-25	0.75	0.25	4.00	29.24
BM50-50	0.50	0.50	4.00	29.24
BM25-75	0.25	0.75	4.00	29.24
BM0-100	0.00	1.00	4.00	29.24

---

2.4. Microgels and membranes characterization

Prior to use, the microgels and membrane samples were dried in a vacuum oven at 60 °C for 24 h. Attenuated total reflectance Fourier transform infrared spectroscopy (ATR-FTIR) of the PNIPAAm microgels, PAA microgels and the membranes were recorded using a Bruker TENSOR37 instrument.

The zeta potential of PAA microgels was measured with a Zetasizer Nano ZS instrument (Malvern, UK) by measuring the electrophoretic mobility of the spheres using de-ionized water as the electrolyte at room temperature. In addition, the average sizes of the PAA and PNIPAAm microgels were measured by a Zetasizer Nano ZS instrument at a constant concentration of 1 wt% under different temperature (20–45 °C) or pH (1–7) conditions.

X-ray photoelectron spectroscopy (XPS) was employed to study the surface elemental composition of the PVDF/(PNIPAAm-PAA) microgel blend membranes. XPS was conducted on a K-alpha spectrometer (Thermo Fisher, UK) using a monochromatic Al Kα X-ray source (1486.6 eV photons) under vacuum.

The surface and cross-section microstructures of the membranes were observed using field emission scanning electron microscope (FESEM, Hitachi S-4800) at 10 kV.

2.5. Porosity and water contact angle

The membrane porosity ε was determined by measuring the weight of the membranes before (W_wet) and after (W_dry) freeze drying. Prior to freeze drying, the membranes were immersed in de-ionized water for 24 h at 25 °C, and the surface of the membranes was dried with filter paper. The membrane porosity ε was calculated by the following equation:¹⁸where ε is the membrane porosity; W_wet is the weight of the wet membrane (g); W_dry is the weight of the dry membrane (g); ρ_water is the density of water (g mL⁻¹); A_membrane is the area of the membrane (cm²); and h_membrane is the thickness of the membrane (cm). The thickness of the test membranes was 1.3 × 10⁻² cm in this study.

The surface hydrophilic properties of the membranes were determined by the dynamic contact angles and measured by a contact angle goniometer (JYSP-180, Jinshengxin, Testing Machine Co., China). The contact angle was recorded every 5 s at room temperature.

2.6. Pure water flux through the PVDF/(PNIPAAm-PAA) microgel blend membranes

The pure water flux of the membranes was measured by a stainless steel plate-and-frame cross-flow membrane filtration with one membrane cell. The effective area of the membrane was 15.8 cm² in all experiments. Prior to the experiment, the membranes were pre-pressurized under 0.1 MPa for 90 min until the flux reached a steady state. The water flux was calculated by the following equation:where F is the water flux (L m⁻² h⁻¹); V_permeation is the permeate volume (L); A_membrane is the effective membrane area (m²); and t is the filtration time (h). The temperature of the water was maintained using a heat exchanger connected with a thermostat bath. The pH of the feed solution was adjusted using 0.1 M NaOH and 0.1 M HCl.

Based on the water flux, the average pore size d of the membrane could be calculated by the Guerout–Elford–Ferry equation:¹⁹

where η is the water viscosity; l is the membrane thickness (m); and ΔP is the operation pressure (0.1 MPa).

2.7. Rejection properties of the PVDF/(PNIPAAm-PAA) microgel blend membranes

In the rejection performance tests, a BSA solution (1 g L⁻¹) was permeated through the membranes for 0.5 h at a given temperature or pH, and the BSA protein rejection ratio (R) was calculated by eqn (4):where C_f and C_p are BSA concentrations of feed and permeate solutions, respectively. The concentration of BSA solutions was recorded using a UV-vis spectrophotometer (TU-1810PC, Beijing Purkinje General Instrument Co., Ltd., China).

2.8. Antifouling properties of the PVDF/(PNIPAAm-PAA) microgel blend membranes

In order to evaluate the antifouling properties of the PVDF/(PNIPAAm-PAA) microgel blend membranes, the water flux (J_w1) was obtained with water at 0.10 MPa; then, a BSA solution (1 g L⁻¹) flowed through the PVDF/(PNIPAAm-PAA) microgel blend membranes for 0.5 h, and the flux (J_BSA) was measured. After the membranes were washed for 1 h in de-ionized water under stirring, the water flux (J_w2) was recorded again. The relative flux recovery ratio (F_R) was calculated by eqn (5):where J_w1 is the initial water flux; J_w2 is the water flux measured after the membranes were washed; F_R is the relative flux recovery ratio.

The J_BSA/J_w1 and F_R values were used to evaluate the antifouling properties of membranes. The antifouling capacity of the membrane can be further evaluated by the recoverable flux decline (R_r), the non-recoverable flux decline (R_ir, which is caused by pore plugging and adsorption or deposition of a foulant on the membrane surface or wall pore) and the total fouling ratio (R_t), which were calculated by the following equations:²⁰

3. Results and discussion

3.1. Characterization of PNIPAAm and PAA microgels

Fig. 1S† shows the FT-IR spectra of the microgels. The detailed band assignments are listed in Table 2. The prominent absorption bands of the PNIPAAm microgels at 1646 and 1546 cm⁻¹, can be attributed to amide I (C=O stretching) and amide II (CN and CH stretching).²¹ There are three strong bands at 1360–1460 cm⁻¹ associated with CH₂ and CH₃. The PAA microgels have a strong band at 1709 cm⁻¹, which is attributed to C=O stretching.²² Two prominent bands at 1415 and 1240 cm⁻¹ are associated with COH in-plane deformation and C–O stretching, respectively. In addition, there is a wide band at 798 cm⁻¹, which is attributed to an out-of-plane OH⋯O deformation and indicates the existence of strong interchain hydrogen bonds.²²

Table 2 FT-IR band assignments for the PNIPAAm and PAA microgels

PNIPAAm	Wavenumber (cm^−1)	Group	PAA	Wavenumber (cm^−1)	Group
A1	3437	N–H	B1	2960	CH, CH2 stretch
A2	3310	N–H	B2	1709	COOH stretch
A3	3065	Amide B	B3	1452	COH in-plane deformation
A4	2968	CH3	B4	1415	COH in-plane deformation
A5	2929	CH2	B5	1240	CO stretch
A6	1646	Amide I	B6	1150	Ketone
A7	1546	Amide II	B7	798	OH–O out-of-plane deformation
A8	1458	CH2
A9	1387	CH3
A10	1368	CH2
A11	1280	Amide III
A12	1173	CH3 skeletal
A13	1132	CH3 rocking

---

The morphologies of the PNIPAAm and PAA microgels are revealed by FESEM (Fig. 1). As can be seen, the obtained microgels are spherical. The size of the PNIPAAm microgels is about 410 ± 70 nm and that of PAA microgels is about 250 ± 80 nm.

Fig. 1 FESEM (a) and the size distribution (b) of the PNIPAAm and PAA microgels, respectively.

Fig. 2A shows the average size of the PNIPAAm microgels measured by a Zetasizer Nano ZS instrument under different temperature conditions. The average size of the PNIPAAm microgels shows an obvious sensitivity to temperature, suggested by a decrease in the average size from 890 to 420 nm upon increasing the temperature from 20 to 45 °C because of the PNIPAAm collapsing. In the curve of Fig. 2A, the temperature at around 32 °C, which has the largest inclined rate can be regarded as a reference point of the apparent LCST of the PNIPAAm microgels.

Fig. 2 The average size of the PNIPAAm microgels under different temperature conditions at pH = 7 (A); the average size and Zeta potential of the PAA microgels under different pH conditions at 25 °C (B).

Fig. 2B shows the average size and zeta potential of the PAA microgels as a function of pH. As shown, the average size of the PAA microgels increases from 214 to 280 nm, while the zeta potential decreases from −1.71 to −20.1 mV with an increase of pH from 1 to 7. These results indicate that there are more and more carboxyl acid groups of the PAA microgels dissociated into carboxylate ions.^23,24 The solvation or hydration of PAA makes the microgels size increase. The apparent pK_a of PAA microgels, about 4, can also be similarly obtained by a maximum-slope method from the curves in Fig. 2B.

In addition, the average size values of the PNIPAAm (890 nm) and PAA (280 nm) microgels measured with a Zetasizer Nano ZS instrument at 25 °C, pH = 7 are larger than those obtained by FESEM (410 and 250 nm). This may be caused by the swelling or aggregated state of the microgels in the water dispersing medium used in the Zetasizer Nano ZS measurements.

3.2. Chemical composition of the blend membranes

The PVDF/(PNIPAAm-PAA) microgels blend membranes were prepared by immersing the casting solution of PVDF/(PNIPAAm-PAA) microgels/DMF in water. According to the phase-inversion mechanism,²⁵ the membrane pores will form during solidification because of segregation between the hydrophilic and hydrophobic polymers. Additionally, membrane surface enrichment of the hydrophilic polymers can be achieved by a migration of the hydrophilic microgels from the casting solution onto the interface between the casting solution and water. The chemical composition of the PVDF/(PNIPAAm-PAA) microgel blend membranes was determined by ATR-FTIR spectra and XPS.

As shown in ATR-FTIR spectra (Fig. 3), the obvious absorption bands, found in the pure PVDF membranes and blend membranes, at 1110–1280 cm⁻¹ and 1346–1461 cm⁻¹ are characteristic of CF₂ and CH₂ groups, respectively. When compared with the spectrum of the pure PVDF membrane, three new peaks occur at about 1709 cm⁻¹, 1646 cm⁻¹ and 1546 cm⁻¹ in the spectra of the PVDF/(PNIPAAm-PAA) microgel blend membranes. The peak at 1709 cm⁻¹ is attributed to the C=O stretching of carboxyl acid groups in PAA. Meanwhile, the peak at 1646 cm⁻¹ is associated with the second amide C=O stretching and at 1546 cm⁻¹ is assigned to the CN and CH stretching of the O=C–NH groups in the PNIPAAm chains. Therefore, we have proved the successful incorporation of the PAA and PNIPAAm microgels into the blend membranes. It should be noted that the characteristic peaks of the PAA microgels are more remarkable than those of the PNIPAAm microgels even at low weight ratios of PAA:PNIPAAm microgels. This may be caused by the stronger trend of surface enrichment of PAA microgels, which can also be supported by the XPS results shown below.

Fig. 3 ATR-FTIR spectra of the pure PVDF membrane and the microgel blend membranes.

To further characterize the surface composition of the blend membranes, XPS was carried out and the C1s core-level spectra are shown in Fig. 2S.† Three curve-fitted peak components with BE at 290.5, 285.8, and 284.6 eV were assigned to the CF₂, CH₂, and CH end group species of the pure PVDF, respectively.^12,26 The peak ratio for the CH₂ species and CF₂ species is about 1.04, in agreement with the stoichiometry of PVDF. In comparison with the pure PVDF membrane, the blend BM100-0 membrane shows two new chemical species with BE at 288.5 and 286.2 eV, which are assigned to the OH–C=O and C–O species of the PAA microgels. While the blend BM0-100 membrane shows another two new chemical species with BE at 287.4 and 285.9 eV, which are assigned to the NH–C=O and C–N species of the PNIPAAm microgels.^27,28 The four chemical species of OH–C=O, C–O, NH–C=O and C–N can be observed in the blend membranes with both PAA and PNIPAAm microgels.

The relative content of the PNIPAAm and PAA microgels on the membrane surface can be evaluated from the XPS-derived atomic ratios using eqn (9) and (10), which are based on the fact that the fluorine atoms are from PVDF, the oxygen atoms stem from both the PAA and PNIPAAm microgels and the ratio of nitrogen to oxygen atoms is 1 : 1 in the PNIPAAm microgels. A theoretical content of microgels can be calculated from the composition of the casting solution. The XPS measured and the theoretical values revealing the composition of the membrane surface are shown in Table 3.

Table 3 [N]/[F] and ([O] − [N])/[F] molar ratios on the surface of the PVDF/(PNIPAAm-PAA) microgels blend membranes

Sample	Molar ratio
	[N]/[F]^a	[N]/[F]^b		([O] − [N])/[F]^a	([O] − [N])/[F]^b
a Calculated based on the microgels/PVDF ratio in casting solution.b Calculated based on the XPS data.
BM100-0	0		0	0.2222	0.3637
BM75-25	0.0177		0.0179	0.1667	0.3043
BM50-50	0.0354		0.0607	0.1111	0.2214
BM25-75	0.0531		0.0732	0.0556	0.0590
BM0-100	0.0708		0.0856	0	0

---

It can be seen that the [N]/[F] or ([O] − [N])/[F] molar ratios of the membrane surface increase, respectively, with the increase of PNIPAAm or PAA microgels in the casting solution. More importantly, the [N]/[F] and ([O] − [N])/[F] molar ratios on the membrane surface are much larger than that found in the casting solution, showing that more microgels are distributed on the membrane surface because of a trend in the surface enrichment of the microgels. Furthermore, the increase in PAA microgel content on the surface, when compared with the casting solution, is greater than that of the PNIPAAm microgels. For example, in the BM50-50 membrane, the [N]/[F] ratio on the membrane surface is 71% larger than that found in the casting solution, while the ([O] − [N])/[F] ratio is 99% larger. These results further demonstrate a strong trend of membrane surface enrichment of the PAA microgels.

3.3. Morphology and structure of the blend membranes

The surface and cross-section morphologies of the pure PVDF membrane and blend membranes are revealed by FESEM. The surface of the pure PVDF membrane (Fig. 4a) is dense without pores; however, the blend membranes show a porous top surface with pore sizes of about 100–420 nm. In addition, parts of the top surface pores are formed by the transfer of PNIPAAm and PAA microgels from the top surface into the aqueous medium. Comparably, the PAA microgels are more easily enriched on surface and thus transfer to water, leading to larger surface pores on the membrane with more PAA microgels. Therefore, the size of the top surface pore increases with an increasing PAA content in the membranes. It should be noted that the pores and the depressions possibly stem from the transfer of PNIPAAm and PAA microgels from the top surface into the aqueous medium during the phase separation process, and they are not connecting the interior membrane pores. Thus, such large pores on the surface cannot be used to reflect the mass transfer or separation properties of the blend membrane. The sizes of the effective pores existing on the skin layer of the membrane, which are not examined by FESEM are calculated based on the water flux data (Table 4).

Fig. 4 Top surface (a) and cross-section (b) FESEM pictures of the pure PVDF membrane and the PVDF/(PNIPAAm-PAA) microgel blend membranes. The top right corner is a partial enlarged view of the cross-section.

Table 4 Average pore sizes of the membranes based on the water flux experiments

Sample	Average pore size (nm)
	dw,T=20°C^a	dw,T=45°C^a	dB,pH=1^b	dB,pH=7^b
a Calculated from the eqn (3) based on water flux experiment at pH 7 and 0.1 MPa.b Calculated from the eqn (3) based on water flux experiment at 25 °C and 0.1 MPa.
PVDF	6.5	6.5	6.45	6.3
BM100-0	14.2	14.3	23.2	14.4
BM75-25	15.7	19.9	22.7	15.9
BM50-50	17.1	25.1	22.3	18.0
BM25-75	18.1	35.8	21.5	19.1
BM0-100	19.2	38.1	20.7	20.6

---

The cross-section FESEM micrographs (Fig. 4b) of the pure PVDF membranes show a typical unsymmetrical structure, including a dense separation layer on the top surface of the membrane, a finger-like middle layer and a sponge-like bottom layer. In contrast, the microgel blend membranes have a longer finger-like structure with macrovoids running throughout the entire sponge-like layer. A large amount of microgels immobilized inside the blend membrane can also be observed at higher magnification.

This phenomenon of structural change is due to the membrane formation mechanism during the phase inversion process. Diffusion and displacement between DMF and water initially occurs when the casting solution is immersed in de-ionized water. As a result, the casting solution is turned into a polymer-lean domain and a polymer-rich domain by the exchange of DMF and water. Subsequently, the polymer-lean domain and the polymer-rich domain transfer into the finger-like pores and the membrane matrix, respectively. However, the incorporation of the microgels is beneficial for the exchange of DMF and water due to the hydrophilicity of the microgels, leading to the large macrovoid dimension in the membrane.²⁹ It is worth noting that the macrovoid structure becomes more notable when reducing the weight ratio of the PAA to PNIPAAm microgels. This may be caused by the loose framework of PNIPAAm microgels than that of PAA microgels, obtained by a lower dosage of cross-linker (PNIPAAm: 6.52 wt%; PAA: 14 wt%; the cross-linker relative to the monomers), which induces a more prominent effect of the PNIPAAm microgels in enhancing the DMF–water exchange. Thus, the macrovoid formation is developed, and the sponge-like formation is suppressed. Therefore, more pores and larger pore sizes are also formed at lower weight ratios of PAA to PNIPAAm microgels.

The porosity of the membranes was further calculated using eqn (1) based on a wet–dry weight method. The values are 61.3%, 64.7%, 72.3%, 80.4%, 87.3% and 92.5% for the PVDF, BM100-0, BM75-25, BM50-50, BM25-75 and BM0-100 membranes, respectively. Obviously, the porosity of the blend membranes is substantially improved by the incorporation of the microgels and increases with an increasing PNIPAAm content in the membranes, which is consistent with the FESEM images (Fig. 4).

3.4. Surface hydrophilicity of the blend membrane

The surface hydrophilic properties of pure PVDF and the PVDF/(PNIPAAm-PAA) blend membranes were assessed by dynamic contact angle measurements. In such measurements, the surface hydrophilicity of the membrane can be reflected by the diffusion speed of water droplets dribbled onto the membrane surface as well as the contact angle value at a fixed time. The higher diffusion speed or lower contact angle value indicates the better hydrophilicity of the membrane surface.^3,6,7

As shown in Fig. 5, during the first 5 s after the water droplet is dribbled onto the membrane surface, the contact angle value of pure PVDF membrane is 94.7°, which is higher than those of the blend membranes, and the diffusion speed of the water droplet on the blend membranes is faster than that on the pure PVDF membrane. This indicates that the incorporation of the PNIPAAm and PAA microgels can also improve the surface hydrophilicity of the membranes. Furthermore, the contact angle of the blend membranes decreases from 90.9° to 74.9°, with an increase in the weight ratio of PAA to PNIPAAm microgels in the blend membranes during the first 5 s. Besides, more PAA microgels incorporation leads to a faster diffusion speed of the water droplet. These results shows that the effect of the PAA microgels in improving the surface hydrophilicity of membranes is more remarkable than that of the PNIPAAm microgels, which is attributed to the stronger trend of surface enrichment of PAA microgels on the membrane surfaces as described above.

Fig. 5 Contact angles of the pure PVDF and PVDF/(PNIPAAm-PAA) blend membranes.

3.5. Temperature and pH-dependent water flux of the PVDF/(PNIPAAm-PAA) microgel blend membrane

Fig. 6 shows the effect of pH (in the pH range of 1–7) and temperature (in the temperature range of 20–45 °C) on water flux through the pure PVDF membrane and the PVDF/(PNIPAAm-PAA) blend membranes.

Fig. 6 Pure water flux of the pure PVDF membrane and the PVDF/(PNIPAAm-PAA) microgel blend membranes under different permeate temperature or pH conditions. (A) pH = 7, temperatures varied; (B) T = 25 °C, pH varied; (C) temperature and pH varied.

As shown, the water flux of the pure PVDF membrane is much lower than that of the microgel blend membranes because of the large pore size and high porosity provided by the microgel blend membranes are generally beneficial to the water passing through the membranes.^3,6 In the blend membranes, the BM100-0 membrane with a 100 : 0 weight ratio of PAA to PNIPAAm microgels shows the lowest water flux at pH 7 and in the temperature range of 20–45 °C. Furthermore, the water flux of the blend membranes increases with a reduction in the weight ratio of PAA to PNIPAAm microgels due to their increasing pore size and porosity. This phenomenon is consistent with the surface and cross-section FESEM micrographs (Fig. 4).

The temperature sensitivity of the water flux through the blend membranes also alters with the weight ratio of PAA to PNIPAAm microgels in the membranes. The pure PVDF membrane and BM100-0 membrane containing no PNIPAAm microgels show hardly any variation in the temperature range of 20–45 °C, meanwhile, the other blend membranes show an obvious temperature dependence under the same conditions. As shown in Fig. 6A, from 20 to 45 °C, the water flux through the blend membranes increases with increasing water temperature; in particular, it prominently increases within the range of 30–35 °C. In addition, the water flux increment becomes more remarkable with an increasing content of PNIPAAm microgels. The temperature-dependent flux is related to the change in the membrane pore size, which is affected by the conformational change of the PNIPAAm microgels on the membrane pore surface. When the temperature is lower than the LCST of the PNIPAAm microgels, the PNIPAAm microgels are hydrophilic and exist in a hydrated conformation.^{10,12,30–32} Therefore, the swollen PNIPAAm microgels, on the membrane pore surface, block the membrane pores partly and reduce the water flux through the blend membranes. When the temperature is higher than the LCST, the PNIPAAm segments collapse and the PNIPAAm microgels shrink, resulting in an increase in the membrane pore sizes and the water flux through the blend membranes.

As shown in Fig. 6B, similarly, the water flux through the pure PVDF membrane and BM0-100 membrane, containing no PAA microgels, hardly changes with an increase of pH. However, a prominent decrease in the water flux through the other blend membranes occurs within the pH range of 3–5, and the decrease becomes more remarkable with an increasing content of PAA microgels in the membrane. It is the conformation of the PAA microgels in response to pH on the membrane pore surface that causes the pH-dependent change in the pore size and the resultant water flux.^5,9,33 At a flux pH below the pK_a of PAA, the PAA microgels collapse on the membrane pore surface due to the protonation of their carboxyl acid groups, leading to an increase in the pore size and water flux. On the contrary, at a flux pH above pK_a, the carboxyl acid groups are dissociated into carboxylate ions and the PAA microgels are in a swollen state, giving a smaller membrane pore size and lower water permeation rate.^34,35

Fig. 6C shows the water flux through the BM100-0, BM50-50 and BM0-100 blend membranes under four pH and temperature conditions. The BM50-50 membrane exhibits tilted lines in both temperature and pH dimensions. This plot provides a view that blending PNIPAAm and PAA microgels gives the membrane temperature and pH-sensitive water flux. It is worth noting that the BM50-50 membrane shows relatively higher water flux at pH = 1, T = 25 °C when compared with the BM0-100 membrane and at a pH = 7, and T = 45 °C when compared with the BM100-0 membrane. Because at pH = 1, and T = 25 °C, the PNIPAAm microgels in the BM0-100 membrane are swollen due to the temperature being below the LCST leading to a lower water permeation rate, while the PAA microgels in BM50-50 collapse in response to the pH above the pK_a of PAA, which is beneficial to higher water flux. In addition, at pH = 7, and T = 45 °C, the shrunk PNIPAAm microgels make the BM50-50 have a relatively high water flux, although the PAA microgels swell and block the pore partly. Therefore, the dual responsive properties of the blend membranes helps them avoid the seriously decreased performance when in response to one of the stimuli conditions and makes them easy to adjust in terms of structure and performance.

In order to estimate the adjustability of the membrane pore size by changing the microgels conformation, the membrane thickness is assumed not to change, and no pores are completely blocked. The pore size of the membranes was quantitatively calculated using eqn (3) based on the water filtration velocity method,¹⁹ and the results are shown in Table 4. At a fixed temperature (20 or 25 °C) and pH (7), an obvious increase in the membrane pore sizes appears with an increasing weight ratio of PNIPAAm to PAA microgels in the membranes, in agreement with the FESEM observations (Fig. 4). However, the pore size based on the water flux experiments is much smaller than that obtained by FESEM analysis. This supports the view that the large pores on the membrane surface observed in the FESEM images are ineffective in filtration.

Based on the water flux data, the diameter ratio of membrane pores in the “open” (45 °C or pH 1) and “closed” (20 °C or pH 7) states caused by temperature or pH was further estimated. The temperature induced diameter ratios are 1.00, 1.01, 1.27, 1.46, 1.97 and 1.98; similarly, the pH-responsive diameter ratios are 1.00, 1.62, 1.42, 1.24, 1.13 and 1.00, for PVDF, BM100-0, BM75-25, BM50-50, BM25-75 and BM0-100, respectively. This further demonstrates that the temperature- or pH-sensitive change in the membrane pore sizes becomes more remarkable with an increase in the PNIPAAm or PAA microgel content. Obviously, the ability of temperature in tuning the membrane pore size is slightly stronger than that of pH.

3.6. Temperature- and pH-dependent rejection property of BSA solution

The rejection property, a significant parameter of membranes for treating water and wastewater, was investigated using BSA solution (1g L⁻¹) as the model foulant at different temperatures (Fig. 7A) and pH values (Fig. 7B). The pure PVDF membrane shows the highest rejection ratios among all the prepared membranes at room temperature and pH 7; meanwhile, the rejection ratios of the blend membranes decrease with an increase in the weight ratio of PNIPAAm to PAA microgels in the membranes. According to the FESEM results above (Fig. 4), the decrease of the rejection ratio should be ascribed to the increasing pore size of the membranes surface.

Fig. 7 Rejection properties of the membranes for a BSA solution (1 g L⁻¹): (A) rejection at different temperatures (pH = 7); (B) rejection at different pH values (T = 25 °C); (C) rejection at different temperatures and pH values.

As shown in Fig. 7A, the blend membranes containing PNIPAAm microgels show a distinct decrease of BSA rejection within the range of 20–45 °C, with an abrupt change at around 32 °C (consistent with the apparent LCST of PNIPAAm microgels), indicating temperature sensitivity. The sensitivity becomes more notable with an increase in the PNIPAAm microgels content in the membranes. It is the temperature induced swelling–shrinking behavior of the PNIPAAm microgels that leads to a small–large switch of the pore size on the membrane surface, which principally affects the separation capability^{5,30,32,36,37} and results in the rejection ratio change from high to low as the temperature increases.

Fig. 7B shows the effect of pH (in the pH range of 1–7) on rejection properties of the membranes. Similarly, the pH-dependent increase in the rejection ratio becomes more marked with an increase in the PAA microgels content with an abrupt change at around pH = 4 that is in accordance with the apparent pK_a of the PAA microgels. Because on one hand, there is also a change in the pore size of the membrane stemming from the pH induced swelling–shrinking behavior of the PAA microgels. On the other hand, the rejection property is also influenced by the electrostatic interactions between the PAA microgels and BSA whose isoelectric point (pI) is about 4.9.³³ When the solution pH is higher than the isoelectric point (pI), the BSA molecules carrying negative charges and the negatively charged carboxyl groups of the PAA microgels repel each other, leading to a difficulty for the BSA in passing through the membrane.

The three-dimension plot in Fig. 7C gives an intuitive view that the BM50-50 membrane blending PNIPAAm and PAA microgels exhibits the temperature- and pH-sensitive rejection properties, indicated by the tilted lines in the temperature and pH dimensions. Similarly, the BM50-50 membrane shows better rejection properties at pH = 1, T = 25 °C, where the BM100-0 membrane has a lower rejection ratio because of the shrinking behaviors of PAA at pH below the pK_a of PAA, and at pH = 7, T = 45 °C, where the BM0-100 membrane has a lower rejection ratio in response to the temperature below the LCST of PNIPAAm. Thus, when compared with the BM0-100 and BM100-0 blend membranes, which only show temperature or pH-sensitive properties, the blend membranes incorporating both PNIPAAm and PAA microgels with multiple stimuli-responsive properties can be used to control the rejection and other properties such as the water flux.

3.7. Temperature and pH reversibility and pressure stability

In order to evaluate the temperature and pH reversibility as well as the pressure stability of the membranes, the water flux through the BM50-50 membrane containing both PNIPAAm and PAA microgels was measured continuously under temperature and pH cycles at various pressures. As shown, within each temperature cycle of 45–25 °C (Fig. 8A) and pH cycle of 1–7 (Fig. 8B), the water flux shows regular stimuli-responsive changes, that is, a decrease with the decreasing liquid temperature from 45 to 25 °C at pH 7 and a decrease with the increasing liquid pH from 1 to 7 at 25 °C. It was proved that the temperature- and pH-dependent changes are also absolutely reversible. Furthermore, the linear relationship between water flux and pressure at a fixed pH or temperature indicates that the physical structure of the blend membranes is stable. These results are of significance, since the stable incorporation of microgels in the membrane providing the reversible stimuli-sensitivity make blending microgels an effective way to improve the structure and performance of membranes.

Fig. 8 Temperature- and pH-reversibility, and pressure stability for the BM50-50 membrane. (A) pH = 7, with temperatures and pressure varied; (B) T = 25 °C, with pH and pressure varied.

3.8. Assessment of the antifouling performance of the blend membranes

Fig. 9 shows the antifouling performance of the membranes by selecting BSA as a foulant. Clearly, the value of J_BSA/J_w1 and the relative flux recovery ratio (F_R) of the blend membranes are larger than those of the pure PVDF membrane (Fig. 9A), indicating an improvement of the antifouling property is achieved by incorporating the PNIPAAm and PAA microgels. Besides, higher values of J_BSA/J_w1 and F_R for the blend membranes can be obtained at larger weight ratios of PNIPAAm to PAA microgels, due to the increased membrane pore size and membrane porosity,^6–8 which is observed in the FESEM images.

Fig. 9 The permeability (A) and fouling resistance ratio (B) of the pure PVDF membrane and the blend membranes (T = 25 °C, pH = 7), and the fouling resistance ratio (C) of the blend membranes.

All the J_BSA/J_w1 are smaller than 100%, suggesting the permeate flux decreases when filtrating a BSA solution. As known, the concentration polarization or membrane fouling can lead to a flux decrease.^35,38 Since the concentration polarization could be neglected as the BSA solution with a high molecular weight is cross-flow filtration driven by a diaphragm pump, the flux decrease in the present system is potentially ascribed to a combination of reversible and irreversible membrane fouling. The J_w2 values obtained after washing are smaller than the J_w1 values, indicating that the washed membrane pore channels are still blocked to a certain extent by the irreversible membrane fouling.

The fouling resistance ratio, shown in Fig. 9B, was used to evaluate the reversible and irreversible fouling of the membranes. The total fouling ratio R_t decreases from 81.8% to 54.5% when increasing the content of PNIPAAm microgels from the pure PVDF to BM0-100 membrane. The notable decrease further proves better antifouling performance of the microgel blend membranes. However, it is impossible to eliminate the irreversible fouling completely. The decrease of the irreversible fouling ratio R_ir as well as the increase of the reversible fouling ratio R_r with an increasing weight ratio of PNIPAAm to PAA microgels, indicates that more BSA adsorption and deposition can be washed away by a water rinse when a larger membrane pore and a bigger membrane porosity appear. It should be noted that the surface hydrophilicity of the membrane reduces with an increasing weight ratio of PNIPAAm to PAA microgels, leading to poorer antifouling properties. However, in the case of this experiment, the large membrane pore rich in hydrophilic microgels on the pore surface and the high porosity have an advantage over the membrane surface hydrophilicity in improving the antifouling performance of the membrane.

Fig. 9C shows the three-dimension plot that exhibits the temperature- and pH-sensitive antifouling performance of the blended membranes. The F_R values of the BM100-0 membrane are below those of the BM50-50 and BM0-100 membrane. More importantly, the BM50-50 membrane shows better antifouling properties than the BM0-100 membrane at pH = 7 and T = 45 °C, where BM0-100 membrane has a lower F_R in response to the temperature being below the LCST of PNIPAAm. Thus, the blend membranes incorporating both PNIPAAm and PAA microgels with dual stimuli-responsive properties also help to avoid the seriously decreased antifouling when in response to one of the stimuli conditions.

Finally, based on the test results and discussions above, the mechanism of the temperature- and pH-sensitive properties as well as the ingenious microstructure of the blend membranes are schematically illustrated in Fig. 10. The microgels are fixed in the blend membrane bulk and on the membrane surface. At a flux pH above the pK_a of the PAA microgels or a permeation temperature below the LCST of the PNIPAAm microgels, the microgels on the membrane pore surface will swell and exist in an extended conformation, leading to a smaller pore size, lower water flux, and higher rejection ratio of BSA solution. The permeation of the BSA molecules through the blend membrane is also restricted by the electrostatic repulsion of the negatively charged PAA and BSA. In a solution with a pH below the pK_a of the PAA microgels or a permeation temperature above the LCST of the PNIPAAm microgels, the microgels segments collapse and the microgels shrink on the membrane pore surface. Then, a greater increase in the pore sizes in or near the skin layer of membrane appears, which results in a higher water flux and lower rejection ratio of the BSA solution. The electrostatic repulsion disappears due to the uncharged PAA, which is also in favor of the permeation of BSA. Consequently, the PNIPAAm and PAA microgels in the blend membranes can act as a sensor for temperature and pH, regulating the water flux, and rejection properties of the membranes by varying the temperature and pH.

Fig. 10 Schematic Illustration of the cross-section structure of the PVDF/(PNIPAAm-PAA) microgel blend membrane under different temperature and pH conditions.

4. Conclusions

PVDF/(PNIPAAm-PAA) blend membranes with the PNIPAAm and PAA microgels incorporated in surface and bulk of membranes were prepared using a phase inversion process in water at 25 °C. The blend membrane has a larger pore size, a higher porosity and an improved hydrophilicity when compared with the pure PVDF membrane. In addition, the effect of the PNIPAAm microgels in improving the membrane pore size and porosity is more remarkable than that of PAA microgels. However, the PAA microgels are more beneficial to the hydrophilic properties of the blend membrane than the PNIPAAm microgels. The water flux and BSA rejection of the blend membranes show obvious temperature- and pH-sensitive characteristics. The blended membrane shows increased water flux and reduced protein rejection ratio with the increasing temperature or decreasing pH due to the swelling–shrinking behaviors of PNIPAAm or PAA microgels in response to the temperature or pH. The abrupt changes in water flux and protein rejection appear at around 32 °C and pH 4, which is consistent with the apparent LSCT of PNIPAAm microgels and apparent pK_a of PAA microgels. The reduced water flux and increased protein rejection ratio with increasing the pH with a sharp change at around pH 4 are observed due to the swelling–shrinking behaviors of PAA microgels in response to the pH. Compared with the single response membranes, the blend membranes with dual responsive properties are easy to adjust in the structure and avoid seriously decreased performance when responding to one of the stimuli conditions. Moreover, the antifouling performance of the membrane is effectively improved by incorporating the PNIPAAm and PAA microgels, especially when increasing the content of the PNIPAAm microgels. Therefore, various filtrations with different properties can be obtained by changing the content of PNIPAAm and PAA microgels in the membranes. These novel membranes have great potential for selectively permeating and rejecting some solutes in water treatment.

Acknowledgements

The authors acknowledge the financial sponsorship of this work by the National Natural Science Foundation of China (no. 51003076, 21204064), Science and Technology Commission Foundation of Tianjin (no. 14JCZDJC38100, 10JCZDJC22000, 12JCYBJC11200) and Universities of Science and Technology Development Fund Planning Project of Tianjin (no. 20120308).

References

1. M. L. Yeow, Y. T. Liu and K. Li, J. Appl. Polym. Sci., 2004, 92, 1782–1789.
2. J. F. Hester, P. Banerjee and A. M. Mayes, Macromolecules, 1999, 32, 1643–1650.
3. Y.-H. Zhao, Y.-L. Qian, B.-K. Zhu and Y.-Y. Xu, J. Membr. Sci., 2008, 310, 567–576.
4. A. Schulze, M. F. Maitz, R. Zimmermann, B. Marquardt, M. Fischer, C. Werner, M. Went and I. Thomas, RSC Adv., 2013, 3, 22518–22526.
5. L. Ying, E. T. Kang and K. G. Neoh, J. Membr. Sci., 2002, 208, 361–374.
6. F. Liu, Y.-Y. Xu, B.-K. Zhu, F. Zhang and L.-P. Zhu, J. Membr. Sci., 2009, 345, 331–339.
7. F. Liu, C.-H. Du, B.-K. Zhu and Y.-Y. Xu, Polymer, 2007, 48, 2910–2918.
8. J. Marchand-Brynaert, N. Jongen and J.-L. Dewez, J. Polym. Sci., Part A: Polym. Chem., 1997, 35, 1227–1235.
9. L. Ying, W. H. Yu, E. T. Kang and K. G. Neoh, Langmuir, 2004, 20, 6032–6040.
10. L. Ying, E. T. Kang and K. G. Neoh, Langmuir, 2002, 18, 6416–6423.
11. D. Roy, J. N. Cambre and B. S. Sumerlin, Prog. Polym. Sci., 2010, 35, 278–301.
12. Q. Liu, Z. Zhu, X. Yang, X. Chen and Y. Song, Radiat. Phys. Chem., 2007, 76, 707–713.
13. K. Hu and J. M. Dickson, J. Membr. Sci., 2007, 301, 19–28.
14. L. Ferro, O. Scialdone and A. Galia, J. Supercrit. Fluids, 2012, 66, 241–250.
15. H. Iwata, M. Oodate, Y. Uyama, H. Amemiya and Y. Ikada, J. Membr. Sci., 1991, 55, 119–130.
16. K. C. Khulbe, C. Feng and T. Matsuura, J. Appl. Polym. Sci., 2010, 115, 855–895.
17. C. Shi, X. Chen, L. Chen, M. Chen and R. Jiang, J. Appl. Polym. Sci., 2013, 128, 3034–3042.
18. C. Mu, Y. Su, M. Sun, W. Chen and Z. Jiang, J. Membr. Sci., 2010, 361, 15–21.
19. E. Yuliwati, A. F. Ismail, T. Matsuura, M. A. Kassim and M. S. Abdullah, Desalination, 2011, 283, 206–213.
20. Y. Hao, A. Moriya, T. Maruyama, Y. Ohmukai and H. Matsuyama, J. Membr. Sci., 2011, 376, 247–253.
21. B. Sun, Y. Lin and P. Wu, Appl. Spectrosc., 2007, 61, 765–771.
22. S. Dubinsky, G. S. Grader, G. E. Shter and M. S. Silverstein, Polym. Degrad. Stab., 2004, 86, 171–178.
23. K. A. Appiagyei, G. L. Messing and J. Q. Dumm, Ceram. Int., 2008, 34, 1309–1313.
24. J. Wang, S. Jiang, H. Zhang, W. Lv, X. Yang and Z. Jiang, J. Membr. Sci., 2010, 364, 253–262.
25. D. Parasuraman and M. J. Serpe, ACS Appl. Mater. Interfaces, 2011, 3, 4714–4721.
26. W.-F. Lee and Y.-C. Chen, Eur. Polym. J., 2006, 42, 1634–1642.
27. J.-Z. Yu, L.-P. Zhu, B.-K. Zhu and Y.-Y. Xu, J. Membr. Sci., 2011, 366, 176–183.
28. C.-W. Chen, M.-Q. Chen, T. Serizawa and M. Akashi, Adv. Mater., 1998, 10, 1122–1126.
29. R. M. Boom, I. M. Wienk, T. van den Boomgaard and C. A. Smolders, J. Membr. Sci., 1992, 73, 277–292.
30. L. Ying, E. T. Kang, K. G. Neoh, K. Kato and H. Iwata, J. Membr. Sci., 2004, 243, 253–262.
31. R. Pelton, Adv. Colloid Interface Sci., 2000, 85, 1–33.
32. L. Ying, E. T. Kang and K. G. Neoh, J. Membr. Sci., 2003, 224, 93–106.
33. N. Akkilic, M. Mustafaev and V. Chegel, Macromol. Symp., 2008, 269, 138–144.
34. Q. Wei, J. Li, B. Qian, B. Fang and C. Zhao, J. Membr. Sci., 2009, 337, 266–273.
35. P. Wang, J. Ma, Z. Wang, F. Shi and Q. Liu, Langmuir, 2012, 28, 4776–4786.
36. J. F. Hester, S. C. Olugebefola and A. M. Mayes, J. Membr. Sci., 2002, 208, 375–388.
37. T. Xiang, Q. Zhou, K. Li, L. Li, F. Su, B. Qian and C. Zhao, Sep. Sci. Technol., 2010, 45, 2017–2027.
38. R. Bian, K. Yamamoto and Y. Watanabe, Desalination, 2000, 131, 225–236.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c4ra02724h

---