Introduction

Membranes are playing more and more important roles in myriad aspects for sustainable development1,2,3,4,5. As emerging artificial biomimetic membranes, smart gating membranes with porous substrates and functional gates, whose permeation properties can be dramatically controlled or adjusted by the gates in response to mild chemical and/or physical stimuli in the external environments, are attracting ever-increasing interests from various fields5,6,7. Such smart gating membranes could find myriad applications in numerous fields including water treatment8,9,10, controlled release6,11, chemical/biological separations12,13, chemical sensors and valves14,15, tissue engineering16 and so on. Easy fabrication, high flux, significant response and strong mechanical strength are critical for the versatility of such smart gating membranes, because these attributes ensure low cost and easy mass-production as well as satisfactory performances of membranes for practical applications. However, current smart gating membranes are still suffering from complicated and difficult-to-scale-up fabrication process, low flux, poor response or weak mechanical property, which severely limits their applications.

Considerable efforts have been directed at addressing these problems by develo** diverse strategies to introduce responsive domains into membrane materials for fabrication of smart gating membranes. All the methods for preparing smart gating membranes can be classified into two main types according to time order of introduction of responsive domains, i.e., the responsive domains can be introduced into membrane materials after or before membrane formation. It is very popular to fabricate smart gating membranes by introducing responsive domains into membrane materials after membrane formation, i.e., responsive domains are introduced into or onto preformed porous membrane substrates by certain modification methods including chemical grafting5,6,9,12,13,14,17 and physical coating or pore-filling10,Supplementary Table S1 and Fig. S1 for details). The normalized thermo-responsive coefficient, which is the ratio of membrane resistance at low temperature to that at high temperature, can be used to compare the responsive performances of different membranes at different temperatures directly. For the previous thermo-responsive membranes prepared by introducing thermo-responsive domains into membrane materials before membrane formation via LIPS, either the maximum normalized fluxes or the maximum normalized thermo-responsive coefficients are limited (Supplementary Table S1 and Fig. S1). For the membranes prepared with grafted thermo-responsive copolymers (“Series 1” in Supplementary Table S1 and Fig. S1), although the maximum normalized fluxes are very large, the maximum normalized thermo-responsive coefficients are not high (typically less than 3.0)20. For the membranes prepared by blending membrane-forming materials with thermo-responsive polymers as additives (“Series 2” in Supplementary Table S1 and Fig. S1), both the maximum normalized fluxes (typically lower than 870 L m−2 h−1 bar−1) and the maximum normalized thermo-responsive coefficients (typically less than 1.8) are very limited21,22. For the membranes prepared by blending membrane-forming materials with thermo-responsive nanogels as additives (“Series 3” in Supplementary Table S1 and Fig. S1), although the maximum normalized thermo-responsive coefficients could be as high as 5.9, the maximum normalized fluxes are very low (typically less than 700 L m−2 h−1 bar−1)23. Excitingly, for our membranes prepared via VIPS with nanogel content of 17.00%, the maximum normalized flux and the maximum normalized thermo-responsive coefficient are as high as 4300 L m−2 h−1 bar−1 and 6.0 respectively (Supplementary Table S1 and Fig. S1). The results verify that, by constructing the above-mentioned unique architecture inside the membranes via VIPS, our smart gating membranes are able to achieve ever better comprehensive performances on the flux and responsive characteristics.

Furthermore, the thermo-responsive gating characteristics of the composite membranes for diffusional permeation of solute molecules with different molecular weights are investigated (Fig. 5, Supplementary Fig. S2 and S3). The results show that the value of the diffusion coefficient of the same solute decreases rapidly with lowering the temperature, which is responding to the changing trend of the flux (Fig. 5a). Then, with increasing the molecular weight of the solute, the diffusion coefficient (D) turns down, owing to the increasing of the Stokes-Einstein radius of the solute for diffusion (Fig. 5a).

Figure 5
figure 5

Trans-membrane diffusional permeation performances.

(a) The thermo-responsive diffusional coefficients of solute molecules with different molecular weights. (b) The thermo-responsive diffusion factor (RD(39/20)) of solute molecules with different molecular weights, in which RD(39/20) is defined as the ratio of the diffusion coefficient of the solute at 39 °C to that at 20 °C.

Full size image

As mentioned above, for the similar purpose, a coefficient called thermo-responsive diffusion factor (RD(39/20)) is defined as the ratio of the diffusion coefficient of the solute at 39 °C to that at 20 °C. When the molecular weight of the solute increases from 1355 to 40000, the value of RD(39/20) undergoes a process of rising from 3.3 to 22.5 first and then falling to 11.25 later (Fig. 5b). For VB12, because the molecular size is small, it is easy for the VB12 molecules to permeate through the membrane pores whether the temperature is 39 °C or 20 °C (Supplementary Fig. S3a) and the trans-membrane permeability of VB12 is affected by the size change of the diffusion channels to a certain extent. However, for the 4000 and 10000 (MW) FITC-dextrans, at 20 °C, as the molecular size is larger than the “closed” pore size, the molecules are excluded by the membranes; while at 39 °C, the size of the these molecules becomes smaller than the “open” pore size and then the solute molecules can permeate easily through membranes (Supplementary Fig. S3b). As a result, the value of RD(39/20) increases remarkably. In the case of 40000 FITC-dextran with the largest molecule size in this study, even at 39 °C, the permeation of the solute molecule is still affected by the size exclusion of the membrane pores (Supplementary Fig. S3c), because the molecular size is so larger that the 40000 FITC-dextran molecules cannot permeate through the membrane easily. For the solute molecule with molecular weight of 10000 (g/mol), the ratio of the diffusion coefficient of the solute at 39 °C to that at 20 °C is as high as 22.5, which verifies the fabricated membranes are “smart” and highly potential in separations and controlled release.

Mechanical properties

Our smart gating membranes with enough in situ self-assembled PNIPAM nanogels as thermo-responsive gates exhibit excellent mechanical properties (Fig. 6). On the condition of exposure time of 20 min and vapor at 25 °C and 70% (RH), Our smart gating membranes prepared via VIPS have much better mechanical properties than the membranes prepared via LIPS (Fig. 6a,b). To compare the mechanical properties of our membranes prepared via VIPS with those prepared via LIPS, PES membranes with equal contents of nanogels are prepared via LIPS as references. Although the thicknesses of casted solution films are all 200 μm, the thicknesses of dried membranes prepared via VIPS are 64 ± 4 μm while those prepared via LIPS are 98 ± 5 μm. Because the membranes prepared via VIPS have symmetric porous structures24 while those prepared via LIPS have asymmetric porous structures23, the membranes prepared via VIPS are denser throughout the whole membrane thickness that those prepared via LIPS. As a result, the membranes prepared via VIPS are mechanically stronger than those prepared via LIPS. For the membranes prepared via LIPS, no matter how the nanogel content varies, the largest tensile strain at break is less than 8.0% and the largest tensile stress at break (σb) is smaller than 3.8 MPa; however, for our membranes prepared via VIPS, the tensile strains at break are all about 23.0% and the tensile strengths at break are all higher than 9.4 MPa (Fig. 6a,b). More importantly and surprisingly, with increasing the nanogel content from 4.25% to 17.00%, the tensile strengths at break of our membranes prepared via VIPS increase from 9.4 MPa to 13.0 MPa (Fig. 6b). The mechanical properties of membranes prepared with different VIPS parameters are also tested. The membranes prepared with the exposure time of 20 min have higher tensile strengths at break and the tensile strains at break than those prepared with the exposure time of 2 min (Fig. 6c,d). Among the membranes prepared by VIPS, the membranes prepared with 2 min, 15 °C and 70% (RH) have a typical structure like those prepared with LIPS and have a mechanical property like those prepared by LIPS (Fig. 6a–d). The membranes prepared by higher RH with the limited exposure time of 2 min have a better mechanical property (Fig. 6c,d), which implies that higher RH speeds up the process of pore coarsening. It should be noted that with enough exposure time and fixed nanogel content, the mechanical properties of membranes vary little (Fig. 6c,d).

Figure 6
figure 6

Mechanical properties of membranes.

(a) Typical tensile stress versus tensile strain curves of membranes, in which “V-0” and “L-0” stand for membranes prepared by VIPS and LIPS respectively with nanogel content being 0% and “V-1” and “L-1” for nanogel content being 4.25%, “V-2” and “L-2” for nanogel content being 8.50%, “V-3” and “L-3” for nanogel content being 12.75% and “V-4” and “L-4” for nanogel content being 17.00%. (b) Effect of nanogel content on the tensile strength at break (σb) of membranes. (c) Typical tensile stress versus tensile strain curves of membranes prepared with different conditions. (d) Effects of preparation conditions on the tensile strength at break (σb) of membranes. (e) Effect of nanogel content on the relative density (ρ*/ρs) of membranes, in which ρ* and ρs represent bulk density and true density respectively. (f) Comparison of calculated tensile strength at break (σb) of membranes from equation (1) with experimental data. Error bars indicate standard deviation, n = 5.

Full size image

As mentioned above, our membranes prepared via VIPS have symmetric cellular-like structures. For cellular solids, the mechanical properties are mainly affected by the most important structural characteristic parameter that is called the relative density28. The relative density of a cellular solid is the density ratio of the cellular material (i.e. bulk density ρ*) to the solid of which it is made (i.e. true density ρs). The smaller the relative density (ρ*/ρs) is, the larger the porosity of the porous membrane. With increasing the nanogel content from 4.25% to 17.00%, the ρ*/ρs value of the membrane prepared via VIPS increases from 0.26 to 0.33 (Fig. 4e). The results indicate that, by adding more nanogels, although the membrane pores are enlarged and get more interconnected with each other (Fig. 2c–f), the membrane porosity is decreased slightly, which means the pore walls become denser. As a result, the tensile strength at break of the membrane prepared via VIPS increases with increasing the nanogel content from 4.25% to 17.00%. Because of the open-cellular structures of the membranes prepared via VIPS with addition of enough nanogels, the following equation can be used to calculate the tensile strength of the membrane from the relative density28:

where is the tensile strength of the membrane and is the yield strength of the pore wall material (PES). The calculated data of the tensile strengths of membranes prepared via VIPS with different nanogel contents fit in well with the experimental data (Fig. 6f). Both the experimental and calculated results exhibit an important and exciting phenomenon, which is that the mechanical properties of our smart gating membranes with in situ self-assembled nanogels as functional gates are enhanced with increasing the nanogel content. That is, all the flux, responsive and mechanical properties of our smart gating membranes can be simultaneously enhanced without any conflict.

Discussion

We have demonstrated simple and controllable fabrication of a novel type of smart gating membranes with simultaneous large flux, significant response and excellent mechanical properties, by constructing self-assembled responsive nanogels in situ on membrane pore surfaces as functional gates via a VIPS process. The generated membrane pores are three-dimensionally interconnected inside the membranes and the self-assembled nanogels on the membrane pore surfaces serve as responsive gates. With the proposed unique architecture, factors conducive to improving all the flux, responsive and mechanical properties are simultaneously introduced into the smart gating membranes. The flux, responsive and mechanical properties of the smart gating membranes can be easily customized by adjusting the nanogel content and the effects of preparation conditions on the structures and performances of the composite membranes are systematically investigated. By using a proper recipe with enough nanogel content, a smart gating membrane could have all the high flux, significant response and strong mechanical properties. Such a combination of high flux, significant responsive characteristics and strong mechanical properties, along with an easy one-step method of fabrication, makes our smart gating membranes ideal candidates for further investigations and applications. The strategy of self-assembling nanogels in situ on the pore surfaces via VIPS and the simple fabrication procedure presented here circumvent the difficulties in simultaneously improving flux, responsive and mechanical properties of the smart gating membranes. Due to the excellent concurrent flux, responsive and mechanical properties, the smart gating membranes with in situ self-assembled responsive nanogels as functional gates will provide ever better performances in myriad applications including water treatment, controlled release, chemical/biological separations, chemical sensors, chemical valves and tissue engineering and may open up new fields of application for smart gating membranes. Furthermore, the proposed novel strategy can be used to fabricate various kinds of functional porous materials with pores immobilized or modified by various kinds of responsive or even non-responsive nanoparticles for numerous applications, including smart gating membranes5,6, anti-fouling membranes5,29 and functional cellular solids28 or foams30 and so on, which might be a fertile area of research.

Methods

Fabrication of nanogels and membranes

Monodisperse homogenous poly(N-isopropylacrylamide) (PNIPAM) nanogels were synthesized by precipitation polymerization25. Typically, monomer N-isopropylacrylamide (NIPAM), crosslinker N,N-methylenebisacrylamide (MBA) and initiator ammonium persulfate (APS) were mixed in a molar ratio of 100:5:2 and dissolved in 200 ml deionized (DI) water with the molar concentration of NIPAM being 0.1 mol L−1. To observe the morphology of nanogels in water with confocal laser scanning microscope (CLSM), fluorescence dye methacryloxy thiocarbonyl rhodamine B (Polyfluor 570, Polysciences) was added in the monomer aqueous solution with a concentration of 3.0 mmol L−1. The monomer solution was bubbled with nitrogen gas for 30 min to remove the dissolved oxygen and then was kept in a water bath at 70 °C for precipitation polymerization for 4 h. After reaction, the PNIPAM nanogels were thoroughly purified by repeating centrifugation at 8000 rpm and redispersed in deionized water to remove the residual unreacted components. Finally, the nanogels were freeze-dried at −35 °C for 48 h. The morphology of the PNIPAM nanogels in dried state was observed by field-emission scanning electron microscope (FESEM, JSM-7500F, JEOL). The thermo-responsive hydrodynamic diameters of nanogels in water at temperatures ranging from 20 to 45 °C were measured by dynamic light scattering (DLS, Zetasizer Nano ZS90, Malvern) equipped with a He-Ne light source (λ = 633 nm, 4.0 mW). Before each datum collection, the highly diluted PNIPAM nanogel dispersion in DI water was allowed to equilibrate for 20 min at each predetermined temperature. The morphology of the nanogels dyed with Polyfluor 570 in water at room temperature was observed by CLSM (SP5-II, Leica), with red fluorescent channel excited at 543 nm.

Smart gating membranes with self-assembled responsive nanogels as functional gates were fabricated from nanogel-contained membrane-forming solution via vapor-induced phase separation (VIPS) approach. The membrane-forming solution was 1-methyl-2-pyrrolidinone (NMP) containing 17.5 wt% polyethersulfone (PES, Mw = 40,000, Changchun Jilin Special Plastics). To add the nanogels into the membrane-forming solution, a certain amount of freeze-dried PNIPAM nanogels was dispersed in NMP first and then PES was added. The nanogel contents in the membranes, which were the blending mass ratios of PNIPAM nanogels to PES, were varied as 0%, 4.25%, 8.50%, 12.75% and 17.00%. The nanogel-contained membrane-forming solution was casted onto a glass plate with a thickness of 200 μm. The casting was performed inside a humidity chamber maintained at 15 °C and 70% relative humidity, 25 °C and 70% relative humidity and 25 °C and 90% relative humidity, respectively (TH-PE-100, JEIO). The casted film was kept in the humidity chamber for 2 min or 20 min and then immersed in a water bath at 22 °C to form flat membrane. As references, membranes were also prepared with the same recipes via liquid-induced phase separation (LIPS) approach, in which the casted film was immediately immersed into a water bath at 22 °C and left in water for 20 min. The microstructures of membranes were investigated by FESEM (JSM-7500F, JEOL). To observe the cross-sections, membrane samples were put into liquid nitrogen for enough time, fractured mechanically and stuck to the sample holder. All the samples were sputter-coated with gold for 60 s before observation.

Thermo-responsive gating property testing

To investigate the thermo-responsive gating characteristics of the prepared membranes, trans-membrane water fluxes at different temperatures were studied first. The water flux experiments of membranes were carried out using a filtration apparatus under a constant trans-membrane pressure of 0.2 MPa. Each membrane had been immersed in DI water over 24 h before testing the water flux. The diameter of the effective membrane area for water permeation was 40 mm. The test temperature range was chosen from 20 °C to 39 °C. In the experiments, a thermostatic unit was used to control the temperatures of the membranes and the feed water. The tests for water flux of each membrane at each temperature were carried out more than five times to obtain an average value for the water flux.

Mechanical property testing

The mechanical properties of the membranes were tested by a commercial test machine (EZ-LX, Shimadzu). The membrane samples were cut into dumbbell shapes of standardized JIS-K6251-7 sizes (length 35 mm, width 2 mm and gauge length 12 mm) with a sample-cutting machine (Dumbbell). Both ends of the dumbbell-shaped samples were clamped and stretched at a constant velocity of 20 mm min−1. At least five samples were tested for each membrane.

Trans-membrane diffusional permeation experiments

Trans-membrane diffusional permeation experiments of composite membranes that prepared with the condition of exposure time of 20 min and vapor at 25 °C and 70% (RH) were carried out. The environmental temperatures were changing between 20 °C and 39 °C. VB12 with molecular weight of 1355 (g/mol) and FITC-dextran molecules with number averaged molecular weights of 4000, 10000 and 40000 (g/mol) were chosen as the solute molecules. The feed solution was prepared by dissolving VB12 and FITC-dextran molecules in DI water with a concentration of 0.4 mmol L−1 (VB12) and 50 mg L−1 (FITC-dextrans). The diffusional permeation experiments of membranes were carried out by using a standard side-by-side diffusion cell with a thermostatic unit for controlling the environment temperature. Each test membrane was immersed in the permeant solution overnight before beginning the diffusion experiments. The concentration of VB12 in the receptor cell at regular intervals was measured by using an UV-vis Spectrometer (UV-1700, Shimadzu) at a wave length of 361 nm. The concentration of FITC-dextran in the receptor cell at regular intervals was measured by using a fluorescent photometer (RF5301PC, Shimadzu) and the excitation and emission wavelength were 480 and 520 nm respectively. Each concentration of the solutes at regular intervals was measured three times and the arithmetical mean value was calculated. The diffusivity of the solute across the membrane D, can also be calculated using a similar equation derived from Fick’s first law of diffusion as follows31:

where Ci, Ct and Cf are the initial, intermediary (at time t) and final concentrations of the solute in the receptor cell; V1 and V2 are the volumes of the liquids in the donor cell and in the receptor cell, respectively; L represents the thickness of the dry membrane; and A is the effective diffusion area of the membrane.

Additional Information

How to cite this article: Luo, F. et al. Smart gating membranes with in situ self-assembled responsive nanogels as functional gates. Sci. Rep. 5, 14708; doi: 10.1038/srep14708 (2015).