Investigation of abnormal thermoresponsive PVDF membranes on casting solution, membrane morphology and filtration performance

Abstract

An interesting PVDF membrane with unusual thermoresponsive behavior was prepared by the incorporation of P(OEGMA-co-VTMOS). The P(OEGMA-co-VTMOS) copolymer was first in situ synthesized in a PVDF/triethyl phosphate (TEP) casting solution. A P(OEGMA-co-VTMOS) network was assembled in the PVDF membrane through hydrolysis and condensation during phase inversion. PVDF/P(OEGMA-co-VTMOS) in organic solvent demonstrated a typical LCST around 35 °C, reflecting the reversible transition of the coil-to-globule conformation. Microphase separation was responsible for the appearance of turbidity of the casting solution. FTIR, XPS, TGA and SEM confirmed the surface enrichment of the copolymer, especially in the membrane bottom. The hydrophilicity and protein anti adsorption were improved despite the temperature variation. In particular, AFM in aqueous media was conducted to determine the reversible morphology and water channel variation under heating and cooling. Different from common thermoresponsive membranes, the so-modified PVDF membranes exhibit a reversible abnormal change of pure water flux and BSA rejection with temperature. The intriguing thermoresponsive behavior and mechanism of the modified PVDF membrane was thoroughly investigated from the aspects of PVDF/TEP casting solution, membrane morphology in aqueous media and water/BSA filtration performance.

---

1. Introduction

Feedback-controlled membranes have been investigated considerably in recent decades, including those controlled by temperature,^1,2 pH,^3,4 light,^5,6 sugar concentration,⁷ electro-stimuli,^8,9 magnetic fields etc.^10,11 Thermoresponsive membranes are one of the most striking materials due to their easy regulation by a stimulus. So far, thermoresponsive behavior has been widely studied in a variety of settings, including controlled drug delivery,^12,13 bioseparation,^14,15 filtration,^16,17 and smart surfaces.^18,19 The thermal phase transition behavior of poly(N-isopropylacrylamide) (PNIPAM) was first revealed by Scarpa et al. in 1967.²⁰ The most commonly accepted theory was thought to be the competition between intermolecular and intermolecular hydrogen bonding in aqueous solution around the lower critical solution temperature (LCST) of about 32 °C. Below the LCST, PNIPAm chains tend to form intermolecular bonds with water, and result in a swelling and stretching state and enhance the hydrophilicity. Above the LCST, the intermolecular bonding with water was disrupted, and PNIPAm tends to form intermolecular hydrogen bonds between C=O and N–H groups, resulting in a compact and collapsed conformation to enhance the hydrophobicity accordingly.

The hydrophilicity/hydrophobicity of the porous membrane could be regulated by the thermal response behavior, which was applied in oil/water separation,²¹ controlled release²² and smart ultra-/micro-filtration.^16,23 For example, Lei Jiang’s group first realized the reversible switching between superhydrophilicity and superhydrophobicity on a roughness-enhanced PNIPAm-modified surface.²⁴ Afterwards, the block copolymer PMMA-b-PNIPAm was directly coated onto a common industrial steel mesh to design a temperature-controlled mixed water/oil on–off switch, which collected water and oil from the mixture below and above the LCST separately.²⁵ It was thought that the intermolecular hydrogen bonding between PNIPAm chains and water molecules lead to hydrophilicity/oleophobicity below the LCST, while the intramolecular hydrogen bonding between C=O and N–H groups in PNIPAm chains lead to hydrophobicity/oleophilicity above the LCST. John H. Xin prepared an electrospun regenerated cellulose nanofibrous membrane. PNIPAm chains/brushes were grafted on the membrane surface via surface-initiated atom transfer radical polymerization.²⁶ The as-prepared membrane shows a switchable wettability that absorbed about 4 times its own weight for water from a water/hexane mixture at 22 °C and about 3–5 times its own weight for oils from an oil/water mixture at 40 °C.

A thermoresponsive surface is appealing for biological applications such as microgels, microcarriers and membranes for drug delivery,^2,27–29 cell culture³⁰ and controlled release of proteins.²² PNIPAm-4-acryloylbenzophenone (pNIPAM-ABP) was rooted in the mesoporous CaCO₃ core by chemical crosslinking to form both temperature and pH sensitive microgels, which are attractive for drug delivery applications.²⁸ A thermoresponsive affinity membrane was prepared to achieve hydrophobic adsorption and hydrophilic desorption of BSA by changing the environmental temperature across the LCST.³¹ It was found that SiO₂ nanoparticles generating a nano-structured surface also played an important role in adsorption/desorption. The PNIPAM-grafted membrane exhibited a enhanced BSA adsorption value 2.95 mg m⁻² at 40 °C compared to the control membrane 0.5 mg m⁻², however, about 90% BSA is desorbed from the membrane surface when decreasing the temperature to 20 °C. Furthermore, the porous Nylon-6 membrane modified by PNIPAM-GMA-β-cyclodextrin demonstrated thermo-responsive and molecular-recognition gating characteristics for VB₁₂ through host–guest interaction.³²

In case of smart micro- or ultra-membranes involved with PNIPAM, both pore filling and surface grafting methods usually cause increased permeability and decreased rejection with increasing temperature.^33,34 An isoporous polystyrene-b-poly(4-vinylpyridine) (PS-b-P4VP) membrane with a pore size around 0.05–0.1 μm modified by PNIPAM also exhibited pH- and thermo- double-sensitivity.³⁵ The substrate PS-b-P4VP membrane was first coated with dopamine, and then PNIPAM-NH₂ was grafted on the membrane or pore surface through a Michael-addition reaction. The pores were open and resulted in higher water flux above LCST due to the contraction and collapse of PNIPAM. A thermoresponsive nanofiltration membrane was developed from a BPPO membrane functionalized by ionic liquid and PNIPAM. The membrane showed a similar tunability for the pure water flux and rejection to Crocein Orange G, Toluidine Blue and polyethylene glycol (PEG-1000).³⁶ PNIPAM was also grafted onto ultrathin nanoporous silicon nitride membranes (NSiNMs) to control the release of vitamin B12 and fluorescein isothiocyanate-dextrans as a thermo-responsive nano-valve.³⁷ A pore filling method was adapted to synthesize the PNIPAM-functionalized PVDF membrane,³⁸ which exhibited a temperature controllable flux and dextran rejection as a valve to regulate filtration properties.

From the critical review on responsive membranes, we can see that the overwhelming majority of previously reported thermoresponsive polymers are based on PNIPAM. Actually, various thermoresponsive polymers with better biocompatibility have demonstrated similar promise for the preparation of adaptive materials. For example, OEGMA based polymers combine the advantages of PEG (i.e., biocompatibility) and thermoresponsive polymers (i.e., LCST behavior in water) in a single macromolecular structure. Moreover, they have inherent advantages as compared to traditional PNIPAM such as: an excellent bio-repellency below the LCST (i.e., anti-fouling behavior); reversible phase transitions (i.e., no marked hysteresis); and bio-inert properties (i.e., no specific interactions with biological materials).³⁹ In the present work, we aim to fabricate a novel thermoresponsive PVDF membrane in situ modified by P(OEGMA-co-VTMOS). The P(OEGMA-co-VTMOS) copolymer can be in situ synthesized in PVDF/TEP solution. The LCST and microphase separation are responsible for the turbidity switch of the polymer in the organic solvent. Different from most smart micro- or ultra-membranes reported before, the thermoresponsive PVDF membrane exhibited a decreased permeability and enhanced selectivity with increasing temperature, which was elucidated by FTIR, TGA, XPS, SEM, AFM, contact angle, protein adsorption and rejection experiments, respectively. Beside, the membrane exhibited a good protein resistance despite the temperature variation.

2. Experimental

2.1 Materials

PVDF (Kynar 761A, Arkema), triethyl phosphate (TEP, Sinopharm Chemical reagent Co.), oligo(ethylene glycol) methacrylate (OEGMA) (M_n = 475 g mol⁻¹, Aladdin), 2,2′-azobis(2-methyl propionitrile) (AIBN, 99%, Aladdin), vinyltrimethoxysilane (VTMOS) (98 wt%, Aldrich), bovine serum albumin (BSA, M_w = 67 kDa, 96 wt%, Aladdin) and deionized water were all directly used without further purification.

2.2 Synthesis and characterization of copolymer P(OEGMA-co-VTMOS)

4 g OEGMA and 0.35 g VTMOS was first dissolved in 85 g TEP at 80 °C, and the solution was vigorously stirred under a nitrogen atmosphere for 1 hour to remove the oxygen. 0.02 g AIBN was added into the solution as the initiator to start the copolymerization. The amphiphilic copolymer P(OEGMA-co-VTMOS) in TEP was obtained after 18 hours.

The temperature sensitivity of copolymer P(OEGMA-co-VTMOS) in TEP was characterized with Zetasizer Nano ZS (Britain Malvern) by monitoring the average diameter of the copolymer at 633 nm under the heating and cooling mode. The solvent refractive index, dielectric constant and viscosity at 25 °C were set to 1.430, 13.2 and 1.6 cP, respectively. The sample was heated from 20.0 °C to 60.0 °C at 5.0 °C min⁻¹, and then cooled from 60.0 °C to 20.0 °C at 5.0 °C min⁻¹.

2.3 Preparation of PVDF membranes

The thermoresponsive PVDF membranes were prepared via a typical phase inversion method.⁴⁰ After thorough dissolution of 15.0 g PVDF in 85.0 g TEP at 80 °C, a mixture composed of 4.0 g OEGMA and 0.35 g VTMOS was added into PVDF solution followed by the addition of 0.02 g AIBN dissolved in 2.0 g TEP to initiate the polymerization. The polymerization was carried out at 80 °C with vigorous stirring under a nitrogen atmosphere for 18 h. The solutions were kept still overnight at 25 °C and 80 °C to eliminate air bubbles. The solutions at 80 °C and 25 °C were spread uniformly onto clean glass plates by a casting knife with a thickness of 300 μm. The nascent membranes were immediately immersed into pure water baths at 25 °C and 80 °C. After complete coagulation, the membranes were transferred to a fresh water bath at 60 °C for 24 h and then freeze dried. The surface contacting the glass plate was defined as the bottom, while the surface contacting the air was defined as the top. The PVDF membranes modified by copolymer were coded as B1–B4. For comparison, the neat PVDF membranes without copolymer were coded as A1–A4. The detailed conditions of the eight membranes are shown in Table 1.

Table 1 The neat A1–A4 and modified PVDF membranes B1–B4 prepared at different condition

PVDF membrane	Code	Casting solution temperature/(°C	Coagulation bath temperature/°C
Neat	A1	25	80
	A2	80	80
	A3	25	25
	A4	80	25
Modified	B1	25	80
	B2	80	80
	B3	25	25
	B4	80	25

---

2.4 Membrane characterization

FTIR spectra of the neat and modified PVDF membranes were measured using a Thermo-Nicolet 6700 FTIR spectrometer (US) over a range of 4000–400 cm⁻¹. The membrane surface composition was determined by XPS analysis (AXIS UTLTRADLD, Japan), using Al-Kα as radiation resource. The take-off angle of the photoelectron was set at 90°. Thermal studies were carried out to calculate the copolymer content in the membrane using a TGA (Mettler Toledo, Switzerland) at a heating rate of 10 °C min⁻¹ with a sample weight of approximately 4–5 mg under nitrogen atmosphere.

The morphologies of the top, cross section and bottom surface of the membranes were characterized and analyzed by a scanning electron microscope (SEM, Hitachi S-4800, Japan). Samples of the cross section were fractured by liquid nitrogen. All membrane samples were sputtered with gold for 90 s before observation. The underwater AFM images of bottom and top surface of the membranes were observed at 30 °C and 60 °C through a scanning probe microscope (SPM, Agilent 5500), respectively. The detailed data of the pore size and pore size distribution was characterized via the BET method by automatic surface area and micropore analyser (ASAP2020-HD88, USA) in nitrogen at 25 °C. The contact angle variation with drop age was recorded by a water contact angle system (OCA20, Dataphysics, Germany).

Protein adsorption experiments were carried out in physiological saline by selecting a bovine serum protein (BSA) as the model protein. The 3 cm × 3 cm A2 and B2 were added to 70 mL of a BSA solution at 1 g L⁻¹. The sample was kept at 25 °C, then heated to 50 °C and cooled to 25°. The membranes were maintained at the determined temperature for 6 h for adsorption equilibrium, and the BSA concentration of the solution was measured by a UV-VIS-NIR spectrometer (Lambda 950, Perkin Elmer, US) at 280 nm. The adsorption of BSA on the membrane was calculated based on the reduction of BSA in the solution.

The homemade cross flow water flux system with temperature control equipment was used to measure the performance of the membranes. The measuring protocol was as follows: for the first 30 min, the membrane (the effective area was 4.9 cm²) was under a pressure of 0.15 MPa to compromise the compacting effect, and then the pure water flux was measured five times every 5 min at 0.1 MPa. The pure water flux was measured at 25 °C to 85 °C according to the following equation.

where V is the volume of the permeate water (L), A is the valid area of the membrane (m²) and t is the time (h).

Afterwards, the pure water was changed to 1.0 g L⁻¹ BSA solution at 0.1 MPa for the BSA rejection test. The flux of concentrations of protein in the feed and permeate solutions collecting were also measured by UV-VIS-NIR spectrometer (Lambda 950, Perkin Elmer, US) at 280 nm. The rejection of BSA (R (%)) can be defined as follows:

where A₁ represents the absorption value of the permeate, and A₀ represents the absorption value of the feed.

3. Results and discussion

3.1 Thermoresponsive behavior of P(OEGMA-co-VTMOS) in TEP

The P(OEGMA-co-VTMOS) copolymer was first in situ synthesized in PVDF/TEP and converted into the PVDF membrane via phase inversion. The copolymer undergoes a hydrolysis reaction of –Si–O–CH₃ and subsequent condensation to form the crosslinking network during the phase inversion and hydrothermal treatment. Therefore P(OEGMA-co-VTMOS) are persistently rooted in the PVDF membrane. The P(OEGMA-co-VTMOS) synthesis and membrane formation mechanism is proposed in Fig. 1.

Fig. 1 Synthesis mechanism of the P(OEGMA-VTMOS) network.

The thermoresponsive behaviors of OEGMA based polymers in aqueous media are already heavily investigated. Unexpectedly, the P(OEGMA-co-VTMOS) copolymer/PVDF in organic solvent TEP also exhibited a thermoresponsive behavior. From the optic pictures of three different solutions in Fig. 2(a), we can clearly observe that the transparent P(OEGMA-co-VTMOS)/PVDF solution becomes turbid with increasing the temperature from 25 °C to 85 °C. Besides, the turbidity change of the P(OEGMA-co-VTMOS)/PVDF solution was reversible in long term storage. However, the turbidity change of the pure P(OEGMA-co-VTMOS) solution was not observed due to the absence of micro-phase separation with PVDF. In comparison, the pure PVDF solution remains clarified despite heating or cooling. The turbidity phenomenon indicated that the solution was temperature sensitive. From the viewpoint of thermodynamics, below LCST, the polymer solution is enthalpically favorable and the mixed molecule entropy is less dominant, thereby causing a miscible and homogeneous polymer solution. Above LCST, the enhanced molecular assembly of copolymer and TEP caused negative entropy and a positive free energy of mixing, and therefore manipulated the microphase separation.⁴¹ The P(OEGMA-co-VTMOS) copolymer undergoes a reversible conformation change responding to the temperature. The conformation variation of the copolymer induced microphase separation and resulted in the occurrence of the cloud point. Similar to the amphiphilic random copolymer in aqueous media,⁴² the P(OEGMA-co-VTMOS) can also be considered as a amphiphilic copolymer containing the water soluble polymer POEGMA and the hydrophobic segment VTMOS. The two different segments demonstrated a different affinity to the organic solvent TEP. The coil-to-globule transition can be confirmed by the DLS results as shown in Fig. 2(b). As the P(OEGMA-co-VTMOS) solution is heated from 20 °C to 70 °C and then cooled to 20 °C, the diameter of the copolymer aggregates and contracts from about 17.0 to 8.5 nm and then expands back to 17.5 nm. The sharp decrease and increase of the polymer size indicates the LCST is around 35 °C. The thermoresponsive assembly of P(OEGMA-co-VTMOS)/PVDF in organic solvent TEP was proposed in Fig. 3. Below LCST, P(OEGMA-co-VTMOS) the copolymer intends to form a solvation interaction with the surrounding TEP molecules and results in a coiled state, displaying a compatible entanglement with PVDF in TEP. Above the LCST, the solvation interactions between the copolymer and TEP are weakened and therefore it is more favourable for TEP molecules to be expelled from the coiling copolymer. The copolymer chains are desolvated and agglomerate to form a globular state. Furthermore, the globule-like copolymer above the LCST triggered micro-phase separation with PVDF and caused the final turbidity.

Fig. 2 (a) The optical pictures of the polymer solutions at different temperatures; and (b) DLS of P(OEGMA-co-VTMOS) in TEP.

Fig. 3 The thermoresponsive assembly of P(OEGMA-co-VTMOS)/PVDF in TEP.

3.2 Chemical composition of PVDF membranes

Fig. 4(a) shows the FTIR spectra of the membranes. Generally all membranes manifest a typical absorption of PVDF at 1402 cm⁻¹, 1170 cm⁻¹ and 1180 cm⁻¹, assigned to CH₂ deformation vibration, CF₂ stretching vibration and the skeleton of C–C vibration, respectively, implying that the bulk chemistry of PVDF membrane was not influenced. Nevertheless, in contrast to the neat PVDF membranes (A1–A4), the modified PVDF membranes (B1–B4) showed new absorption at 841 cm⁻¹ and 1109 cm⁻¹ attributed to the functional group –Si–O– of the VTMOS segment. Meanwhile, all modified membrane spectra show another two new bands at 1248 cm⁻¹ (C–O stretching) and 1727 cm⁻¹ (symmetrical C=O stretching), respectively. The C=C stretching absorption peak at 1636 cm⁻¹ of the monomer was absent from the modified membranes.

Fig. 4 (a) FTIR spectra of the P(OEGMA-co-VTMOS), neat (A1–A4) and modified PVDF membranes (B1–B4); (b) XPS wide-scan spectra for top surface of neat (A2) PVDF membrane, top and bottom surface of modified (B1–B4) PVDF membranes; (c) TGA curves for P(OEGMA-co-VTMOS), neat (A2) PVDF membrane and modified (B1–B4) PVDF membranes; (d) contact angle changes with drop age of PVDF membranes; (e) BSA adsorption on A2 and B2 at different temperature.

The surface composition of the PVDF membrane was determined by XPS. Fig. 4(b) presents the XPS wide-scan spectra of the neat (A2) PVDF membrane, and the top and bottom surface of modified PVDF membranes (B1–B4). Besides the two signals attributed to the C and F elements of PVDF, two new signals attributed to the O and Si elements indicated the existence of P(OEGMA-co-VTMOS) in the modified PVDF membranes (B1–B4).

The element mass concentration is given in Table 2. Compared to the top surface of A2, the F concentration is slightly decreased and the C concentration of the top surface of B1–B4 is almost unchanged. The oxygen and silicon mass concentrations on the top surface of all modified membranes are lower than those of the bottom surface due to the more porous structure, which is also consistent with SEM results. The B2 membrane showed higher Si and O content on both the top and bottom surfaces. For example, 2.75% O and 0.97% Si on the top, as well as 5.10% O and 1.19% Si on the bottom were measured, respectively. Besides, the TGA curves in Fig. 4(c) also confirmed the highest P(OEGMA-co-VTMOS) content of 9.0 wt% in the B2 membrane. B1 and B3 showed a similar content of copolymer, about 6.5 wt%. Moreover, approximately 3.0% content of the copolymer is present in B4, which also contains a lower content of 1.08% O and 0.41% Si on the top and 2.87% O and 0.6% Si on the bottom. For a B2 membrane with a higher casting solution temperature and a higher coagulation bath temperature, the contracted copolymer was easier to hydrolyze and crosslink in the membrane bulk and surface. The B4 membrane behaved in the opposite way. Combing with FTIR, XPS and TGA results, it was inferred that P(OEGMA-co-VTMOS) was successfully incorporated into PVDF membranes and demonstrated surface enriching assembly.

Table 2 Element mass concentration as calculated by XPS analysis

Sample	XPS mass concentration (%)
	C	F	O	Si
A2 top	42.79	57.21	0.00	0.00
B1 top	42.40	55.55	1.38	0.67
B2 top	42.25	54.03	2.75	0.97
B3 top	42.19	55.21	1.82	0.78
B4 top	42.56	55.95	1.08	0.41
B1 bottom	40.57	53.91	4.64	0.88
B2 bottom	40.77	52.94	5.10	1.19
B3 bottom	40.73	53.88	4.40	0.99
B4 bottom	41.91	54.62	2.87	0.60

---

The surface assembly of copolymer P(OEGMA-co-VTMOS) will improve the hydrophilicity accordingly, as shown in Fig. 4(d). All modified membranes (B1–B4) exhibited a declined water contact angle compared to the neat PVDF membrane. The contact angle of the A2 membrane is maintained at 113°, in contrast, the contact angle of B2 membrane top surface decreases from 76° to 87° within 60 s. The improved hydrophilicity was mainly attributed to the surface enrichment of the copolymer verified by XPS and TGA results. The fouling resistance to protein was also revealed, as shown in Fig. 4(e). The neat PVDF membrane A2 displayed a serious BSA adsorption around 372 ± 10 μg cm⁻². In comparison, BSA adsorption on the B2 membrane at 25 °C was only 24 ± 3 μg cm⁻², indicating a 94% reduction of protein adsorption. The adsorbed protein amount of the B2 membrane remained almost unaffected by increasing and decreasing the temperature. The temperature had no significant effect on BSA adsorption. The main force of the rejection of the protein comes from the highly repulsive hydration interaction of the hydrophilic surface. The surface enrichment of P(OEGMA-co-VTMOS) played an crucial role in repelling the protein adsorption, despite the thermoresponsive extended or collapsed conformation.⁴³

3.3 Morphologies of PVDF membranes

The morphologies of the neat membranes (A1–A4) and modified membranes (B1–B4) were observed by SEM. In Fig. 5(a), the top surfaces of modified PVDF membranes are more porous than neat PVDF membrane due to the outflow of unreacted monomer OEGMA or VTMOS partially acting as a pore-forming agent. B2 possessed more micropores than other membranes and the mean pore size is around 0.5 μm from the amplified picture. For the modified PVDF membranes, the amphiphilic copolymer enhanced the instability of the casting solution in thermodynamics, and furthermore, hydrophilic POEGMA chains accelerated the phase separation and subsequently caused a more porous surface than the neat membranes. In addition, P(OEGMA-co-VTMOS) of membrane B2 migrated faster to the surface to minimize the interface free energy at higher casting solution temperature and higher water bath temperature.

Fig. 5 (a) SEM images of the top surfaces of the neat (A1–A4) and modified (B1–B4) PVDF membranes; (b) SEM images of the cross section of the neat (A1–A4) and modified (B1–B4) PVDF membranes. (c) SEM images of the bottom surfaces of membranes.

Fig. 5(b) shows SEM images of the cross section of PVDF membranes. All membranes demonstrated an asymmetric porous structure with a thin skin layer and an interconnected porous sublayer, which was caused by the special effect of non-solvent induced phase inversion using TEP as the solvent.⁴⁴ Compared with the neat PVDF membrane A1–A4 with a thicker skin layer of 50 μm, the skin layer of B1–B4 was reduced. In addition, few P(OEGMA-co-VTMOS) aggregates around 1 μm were distributed in the upper skin layer of the B1–B3 membrane. For the B4 membrane, a reduced amount of P(OEGMA-co-VTMOS) aggregates can be observed. The relatively higher casting solution temperature (80 °C) leads to a smaller P(OEGMA-co-VTMOS) assembly, besides, the lower temperature (25 °C) is unfavorable for hydrolysis and crosslinking of the silane coupling agent VTMOS. It becomes more difficult to form the P(OEGMA-co-VTMOS) network and most of the smaller P(OEGMA-co-VTMOS) aggregates flow away from the membrane during phase inversion and subsequent hydrothermal treatment.

As depicted in Fig. 5(c), squashed spherulites and petaloid microstructures are present on the bottom surface of the PVDF membranes. In comparison with B3–B4, B1 and B2 exhibit more spherical particles around 1 μm attaching to the bottom surface despite the casting solution temperature. The higher coagulation bath temperature (80 °C) obviously accelerated the hydrolysis and the crosslinking of the P(OEGMA-co-VTMOS) copolymer as well, therefore more P(OEGMA-co-VTMOS) particles were rooted in the porous bottom, which is also verified by XPS results. The pristine spherulites and petaloid microstructures of the bottom membrane surface are also helpful to capture more particles inside.

The thermoresponsive morphologies of modified membranes were not observed from SEM pictures due to the dehydrated sample. Therefore, the membranes were further observed in the aqueous media by AFM. The hydrated PVDF membranes were heated from 20 °C to 60 °C and then cooled to 20 °C. The corresponding morphological variation was shown in Fig. 6. When raising or lowering the temperature, both the roughness (Sq or Sa) and the grooves of A2 membranes kept almost unchanged, whether the bottom or top surface. In contrast, the Sq of B2 the membrane top increased to 111.0 from 61.0 and then decreased to 66.9 during a heating and cooling cycle. Similarly, the Sq of the modified PVDF membrane bottom increased to 127.0 from 108.0 and then decreased to 99.4. The grooves of the top surface decreased to 0.4 μm from 1 μm and then recovered to 1 μm with heating and cooling of the membrane.

Fig. 6 (a) AFM images of the top surface of the neat membrane A2 and modified membrane B2 in water at different temperatures; (b) AFM images of the bottom surface of the neat membrane A2 and modified membrane B2 in water at different temperature.

In case of the top membrane surface, at 20 °C, the B2 membrane displayed a smooth and hydrated top, indicating more P(OEGMA-co-VTMOS) bushes were extending and interacting with water on the membrane surface. At 60 °C, the P(OEGMA-co-VTMOS) brushes collapsed and contracted into the bulk of the membrane top to expose the dehydrated PVDF crystalline structure with a higher roughness. Similarly, in case of the bottom membrane surface, the iceberg-like PVDF crystalline structure was surrounded by the hydrated P(OEGMA-co-VTMOS) with lower roughness below LCST. In addition, more PVDF spherulites were exposed to the bottom to cause higher roughness above the LCST. The groove on the B2 membrane also undergoes reversible contraction and expansion during the heating and cooling cycle, indicating the reversible thermoresponsive pore size variation. The reversible morphological variations originate intrinsically from the chain conformation of the POEGMA chains.³⁹ The pedant –CH₂CH₂O– chains in the P(OEGMA-co-VTMOS) network interact with water molecules via hydrogen bonding when the temperature varies, providing the thermoresponsive membrane and pore surface.⁴³ Xiang Gao et al. determined the chain conformation of P(MEO2MA-co-OEGMA) thermoresponsive brushes grafted on silicon by neutron reflectometry. Usually, at a lower content of thermoresponsive polymer brushes, the pores of the membrane will be open due to the collapse of the brushes on the pore surface. However, at higher content of P(OEGMA-co-VTMOS) copolymer with a crosslinking network of 9.0 wt% in this case, the membrane surface and pore size will be influenced in a opposite way by the high density of P(OEGMA-co-VTMOS).⁴⁵ Below LCST, the hydrophilic chains swell and stretch to form a loose hydrated channel and extend the effective pore size, which allow the water molecules go through the membrane more freely. Above LCST, the P(OEGMA-co-VTMOS) are collapsed to form a compressed dehydrated barrier and narrow the pore size, which is therefore unfavorable for the permeation of water.

3.4 Filtration performance

The pure water flux of all membranes was measured at 25 °C. Generally, as shown in Fig. 7(a), the neat PVDF membranes (A1–A4) showed a pure water flux around 220 ± 50–269 ± 49 L m⁻² h⁻¹, and all modified PVDF membranes (B1–B4) demonstrated a improved water flux around 367 ± 73–1151 ± 146 L m⁻² h⁻¹. The B2 membrane possessed a higher permeability due to the porous surface, thinner skin layer and more hydrophilic P(OEGMA-co-VTMOS) copolymer rooted in the bulk. Through the measurement of BET, the BJH desorption average pore sizes of A2 and B2 were 13.57 nm and 21.78 nm, respectively. In Fig. 7(b), the pore size of the modified membrane B2 was higher than that of A2. This is because that the hydrophilic P(OEGMA-co-VTMOS) migrated to the surface and acted as a porogenic agent. With increasing temperature, the pure water flux of the A1 membrane increased accordingly due to the decreased water viscosity. The modified PVDF membrane demonstrated an unique responsive permeability. With increasing the temperature, all modified PVDF membranes (B1–B4) exhibited a decreased water flux. For example, the water flux of the B2 membrane decreased from 1151 L m⁻² h⁻¹ to 318 L m⁻² h⁻¹ with increasing the temperature from 25 °C to 85 °C, as shown in Fig. 7(c). We also evaluated the rejection of the thermoresponsive PVDF membranes. It was found that the BSA rejection of B2 membrane increased from 60 ± 1.12% to 78 ± 0.78% and 92 ± 1.21% with increasing the temperature from 25 °C to 35 °C and 45 °C respectively in Fig. 5(d), indicating that the pore size of B2 membrane was decreased during heating, while the BSA retention of the A2 membrane remained constant around 26% with increasing temperature. Moreover, the changes of permeability and rejection are reversible with the temperature due to the reversible morphology variation, as disclosed by AFM results in Fig. 6. The modified PVDF membrane demonstrates a dehydrated barrier and narrowed water channel in aqueous media with increasing temperature, which consequently cause a decreased water flux and increased rejection. Fig. 8 shows the reversible thermoresponsive diversification mechanism of the water channels of the modified PVDF membrane in aqueous media during heating and cooling processes. P(OEGMA-co-VTMOS) copolymer coils are hydrated and stretch to form a loose and free water channel under cooling, while the globule-like copolymer was dehydrated and contracted to reduce the effective pore size under heating. The morphology variation of P(OEGMA-co-VTMOS) in water is consistent with the situation in TEP, as depicted in Fig. 3.

Fig. 7 (a) The pure water flux of PVDF membranes at 25 °C from the top; (b) the pore size distribution of A2 and B2 at 25 °C; (c) the pure water flux of A2 and (B1–B4) membrane under heating from the top; (d) the BSA rejection of A2 and B2 membrane under heating.

Fig. 8 The thermoresponsive diversification of water channels of modified PVDF membrane in aqueous media.

4. Conclusion

P(OEGMA-co-VTMOS) copolymer was in situ synthesized in PVDF/TEP via free radical polymerization to form the casting solution, which exhibited a reversible thermoresponsive behavior with the LCST around 35 °C. The thermodynamics, chain conformation and the microphase separation were responsible for the appearance of turbidity. The unique thermoresponsive PVDF membranes incorporated by P(OEGMA-co-VTMOS) were prepared by varying the casting solution temperature and coagulation bath temperature. The chemistry of the modified PVDF membrane was confirmed by FTIR, XPS and TGA. The B2 membrane showed a 9.0 wt% content of copolymer. The modified membrane demonstrated improved hydrophilicity and protein anti-adsorption. P(OEGMA-co-VTMOS) are enriched and assembled more in the bottom surface. AFM results showed that the membrane roughness increased and the surface pores were narrowed due to the dehydration and contraction of copolymer with increasing temperature in aqueous media, which consequently caused declined water permeability and enhanced rejection. More intriguingly, the thermoresponsive behavior is reversible in terms of the PVDF/TEP casting solution and membrane morphology in aqueous media, as well as water/BSA filtration performance.

Acknowledgements

This work is supported by the National Natural Science Foundation of China (51473177, 51273211).

Notes and references

1. F. Shen, T. Xiao, L. Han and S. Chen, Desalin. Water Treat., 2015, 54, 1507–1518.
2. Q. Wu, R. Wang, X. Chen and R. Ghosh, J. Membr. Sci., 2014, 471, 56–64.
3. M. K. Sinha and M. K. Purkait, J. Membr. Sci., 2014, 464, 20–32.
4. M. Gao, X. Jia, Y. Li, D. Liang and Y. Wei, Macromolecules, 2010, 43, 4314–4323.
5. Y. Wang, S. Chen, L. Qiu, K. Wang, H. Wang, G. P. Simon and D. Li, Adv. Funct. Mater., 2015, 25, 126–133.
6. D. He, H. Susanto and M. Ulbricht, Prog. Polym. Sci., 2009, 34, 62–98.
7. K. B. Konov, D. V. Leonov, N. P. Isaev, K. Y. Fedotov, V. K. Voronkova and S. A. Dzuba, J. Phys. Chem. B, 2015, 119, 10261–10266.
8. H. Chen, G. R. Palmese and Y. A. Elabd, Macromolecules, 2007, 40, 781–782.
9. H. Chen, G. R. Palmese and Y. A. Elabd, Chem. Mater., 2006, 18, 4875–4881.
10. A. Alexeev, W. E. Uspal and A. C. Balazs, ACS Nano, 2008, 2, 1117–1122.
11. W.-C. Yang, R. Xie, X.-Q. Pang, X.-J. Ju and L.-Y. Chu, J. Membr. Sci., 2008, 321, 324–330.
12. C. He, S. W. Kim and D. S. Lee, J. Controlled Release, 2008, 127, 189–207.
13. L. Zhang, T. Xu and Z. Lin, J. Membr. Sci., 2006, 281, 491–499.
14. D. Roy, B. J. Nehilla, J. J. Lai and P. S. Stayton, ACS Macro Lett., 2013, 2, 132–136.
15. I. Poplewska, R. Muca, A. Strachota, W. Piatkowski and D. Antos, J. Chromatogr. A, 2014, 1324, 181–189.
16. B. P. Tripathi, N. C. Dubey, F. Simon and M. Stamm, RSC Adv., 2014, 4, 34073–34083.
17. A. G. Fane, R. Wang and M. X. Hu, Angew. Chem., Int. Ed., 2015, 54, 3368–3386.
18. T. Cai, K. G. Neoh, E. T. Kang and S. L. Teo, Langmuir, 2011, 27, 2936–2945.
19. D. Wu, X. Liu, S. Yu, M. Liu and C. Gao, J. Membr. Sci., 2010, 352, 76–85.
20. J. S. Scarpa, D. D. Mueller and I. M. Klotz, J. Am. Chem. Soc., 1967, 89, 6024–6030.
21. Y. Cao, N. Liu, C. Fu, K. Li, L. Tao, L. Feng and Y. Wei, ACS Appl. Mater. Interfaces, 2014, 6, 2026–2030.
22. Y. Stetsyshyn, K. Fornal, J. Raczkowska, J. Zemla, A. Kostruba, H. Ohar, M. Ohar, V. Donchak, K. Harhay, K. Awsiuk, J. Rysz, A. Bernasik and A. Budkowski, J. Colloid Interface Sci., 2013, 411, 247–256.
23. D. Wandera, S. R. Wickramasinghe and S. M. Husson, J. Membr. Sci., 2010, 357, 6–35.
24. T. Sun, G. Wang, L. Feng, B. Liu, Y. Ma, L. Jiang and D. Zhu, Angew. Chem., Int. Ed., 2004, 43, 357–360.
25. B. Xue, L. Gao, Y. Hou, Z. Liu and L. Jiang, Adv. Mater., 2013, 25, 273–277.
26. Y. Wang, C. Lai, H. Hu, Y. Liu, B. Fei and J. H. Xin, RSC Adv., 2015, 5, 51078–51085.
27. J. Zhang, Z. Cui, R. Field, M. G. Moloney, S. Rimmer and H. Ye, Eur. Polym. J., 2015, 67, 346–364.
28. F. Natalia, G. Stoychev, N. Puretskiy, I. Leonid and V. Dmitry, Eur. Polym. J., 2015, 68, 650–656.
29. Y. Hu, V. Darcos, S. Monge and S. Li, Int. J. Pharm., 2015, 491, 152–161.
30. H. I. Seo, A. N. Cho, J. Jang, D. W. Kim, S. W. Cho and B. G. Chung, Nanomed.: Nanotechnol., Biol. Med., 2015, 11, 1861–1869.
31. T. Meng, R. Xie, Y.-C. Chen, C.-J. Cheng, P.-F. Li, X.-J. Ju and L.-Y. Chu, J. Membr. Sci., 2010, 349, 258–267.
32. M. Yang, R. Xie, J.-Y. Wang, X.-J. Ju, L. Yang and L.-Y. Chu, J. Membr. Sci., 2010, 355, 142–150.
33. L. Xiao, A. Isner, K. Waldrop, A. Saad, D. Takigawa and D. Bhattacharyya, J. Membr. Sci., 2014, 457, 39–49.
34. X. Chen, Y. He, C. Shi, W. Fu, S. Bi, Z. Wang and L. Chen, J. Membr. Sci., 2014, 469, 447–457.
35. J. I. Clodt, V. Filiz, S. Rangou, K. Buhr, C. Abetz, D. Höche, J. Hahn, A. Jung and V. Abetz, Adv. Funct. Mater., 2013, 23, 731–738.
36. Y. Tang, B. Tang and P. Wu, J. Mater. Chem. A, 2015, 3, 7919–7928.
37. A. Bojko, G. Andreatta, F. Montagne, P. Renaud and R. Pugin, J. Membr. Sci., 2014, 468, 118–125.
38. X. Feng, Y. Guo, X. Chen, Y. Zhao, J. Li, X. He and L. Chen, Desalination, 2012, 290, 89–98.
39. J.-F. Lutz, Adv. Mater., 2011, 23, 2237–2243.
40. M. Tao, L. Xue, F. Liu and L. Jiang, Adv. Mater., 2014, 26, 2943–2948.
41. C. Weber, R. Hoogenboom and U. S. Schubert, Prog. Polym. Sci., 2012, 37, 686–714.
42. A. A. A. G. O. Yahayaa, S. A. Alib, B. F. Abu-Sharkha and E. Z. Hamada, Polymer, 2001, 42, 3363–3372.
43. X. Gao, N. Kucerka, M. P. Nieh, J. Katsaras, S. Zhu, J. L. Brash and H. Sheardown, Langmuir, 2009, 25, 10271–10278.
44. F. Liu, M.-m. Tao and L.-x. Xue, Desalination, 2012, 298, 99–105.
45. J. Meng, J. Li, Y. Zhang and S. Ma, J. Membr. Sci., 2014, 455, 405–414.

---