Introduction
With the recent advances in surface functionalization methods, the fabrication of porous domains with special wettability has become one of the leading research areas in the field of membrane science.1–7 Currently, one of the primary topics in this area is the fabrication of Janus membranes −membranes with asymmetric wettability. Janus membranes bring novel functionalities, not achievable using conventional membrane processes, to membrane-based processes.8–11 Given this asymmetric wettability, Janus membranes offer an additional means to regulate the transport of species in a designated direction.12 This unique feature makes these membranes an attractive candidate for many applications such as ion gating,4 oil/water separation,3 interfacial mass transport,13 energy harnessing,14anti-gravity capillary water pumps,15 bubble aeration,16 and water desalination.17
To date, extensive research exists on the fabrication of Janus membranes for different applications.18–22 Fabrication methodologies are classified into two main categories, interfacial and non-interfacial strategies.23 Interfacial methods typically add a new layer or interface to the substrates. As a result, additional resistance is introduced to the transport of species across the membrane. In comparison, non-interfacial strategies are not limited to the surface. The non-interfacial strategies refer to the techniques in which asymmetric wettability is formed within the porous substrate. Chemical vapor deposition and atomic layer deposition are among the promising non-interfacial strategies for altering the wettability of porous media.23,24 However, it is still challenging to limit the wettability modification to the surface or a specific region within the porous domains.23,25 Therefore, despite the promising advantages of Janus membranes, the complexities involved with the fabrication schemes of these membranes have hindered their development on a large scale.
Here, we developed a simple method, based on the selective deposition of fluoropolymers onto porous media. We utilized initiated chemical vapor deposition (iCVD) to synthesize polymeric domains from the vapor phase and coated the outermost layers of various hydrophilic porous substrates. By infiltrating nonvolatile liquids containing polymerization inhibitors, we limited the deposition of the fluoropolymer to the surface and prohibited the bridging of polymer films over the pores of the substrate. We examined the performance of the fabricated Janus membranes in both static and dynamic tests.
Experimental Methods
Materials and chemicals
Hydrophilic polyvinylidene fluoride (Durapore®, 0.22 μm, 47 mm) and cellulose acetate (0.45 μm, 47 mm) membranes were purchased from Millipore Sigma. Perfluorobutane sulfonyl fluoride (PBSF, 96%) and hydroquinone (HQ) (>99%) were purchased from Sigma-Aldrich. Isopropyl alcohol (IPA, ACS grade), glycerol (certified ACS), and sodium chloride (NaCl, ACS grade) were purchased from Fisher Scientific. Ethylene Glycol (99.5%) was purchased from ACROS Organics. Hexafluoropropylene oxide (HFPO) was purchased from Oakwood chemical. Deionized (DI) water was obtained from a Simplicity® ultrapure water purification system (Millipore, Billerica, MA). Ultrapure nitrogen was purchased from Matheson Gas Company.
Initiated chemical vapor deposition (iCVD) of PTFE
The PTFE coating was deposited onto hydrophilic substrates using a custom-made iCVD reactor with a volume of 3600 cm3. For each deposition, a hydrophilic membrane, prefilled with glycerol/inhibitor mixture (5000 ppm), was placed on the reactor stage, which was cooled to 15 oC via a recirculating chiller (Thermo Fisher Scientific). The monomer, HFPO, was metered into the reactor using a mass flowmeter (MKS Instruments) with a flow rate of 12 sccm. The PBSF jar was maintained at 27 oC to provide a constant flow rate of 1 sccm, adjusted using a precision needle valve (Swagelok). A pressure transducer (MKS Instruments) and an automated butterfly valve, connected to a vacuum process controller (MKS Instruments), were used to maintain the reactor pressure at the set point (300-1200 mTorr). For activating the initiator through thermal pyrolysis, a filament array of phosphor bronze (Goodfellow), suspended 2 cm above the substrate, was heated resistively to 350oC.
Membrane characterization
Scanning electron microscopy (SEM) images were taken using the FEI Helios NanoLab 660 microscope. To reduce the moisture content, we dried the samples in a vacuum oven set at 60 oC for three hours. After drying, the samples were cut to size, mounted on the SEM stub, and coated with 60 nm of gold, using a Ted Pella sputtering machine (108-Auto). Fourier-transform infrared spectroscopy (FTIR) measurements were performed using the attenuated total reflection (ATR) module of Bruker Alpha-p. A small (1 cm × 1 cm) sample was cut to size and placed on the diamond crystal of the ATR module. The FTIR measurements were performed using 24 high-resolution scans on each sample with a resolution of 4 cm-1. The liquid contact angles on the membrane surfaces were measured using an optical tensiometer (Rame-hart, Model 590) and the sessile drop method. A 5 μL liquid droplet was placed on the dried membrane sample. We performed the measurements on three random points on each sample and presented the data with one standard deviation. The liquid entry pressure (LEP) and gas permeation measurements were performed using a custom-made porometer setup described elsewhere.26 The overhead space of the membrane in the filter holder was filled with DI water. Subsequently, the pressure behind DI water was increased gradually. The LEP is reported as the pressure at which gas flow is detected by the flowmeter, placed inline at the outlet of the filter holder. We performed three measurements and presented the data with one standard deviation. For the gas permeation test, we flowed ultrapure nitrogen into the membrane at different pressures ranging from 2 to 40 kPa. The nitrogen permeation through the membrane was recorded using a digital flowmeter (Omega Engineering, FMA1820A), connected to a computer. We characterized three membranes and presented the data with one standard deviation. The DCMD performances for the Janus membranes were evaluated using a laboratory-scale DCMD unit described elsewhere.27 We placed the membrane into a custom-built cell with a channel dimension of 26 mm in length, 26 mm in width, and 3 mm in depth. The effective membrane area exposed to feed and distillate streams was about 6 cm2, with the hydrophilic side facing the feed stream. The temperatures of feed and permeate streams were maintained at 70 °C and 20 °C, respectively. The water vapor flux across the membrane (J) was determined by the gravimetric method.27
Results and Discussion
Figure 1 illustrates the representative images of the membrane surfaces and the schematic of the procedure used to impart asymmetric wettability into hydrophilic substrates. Our fabrication scheme had three consecutive steps, shown in Figure 1 (a). In Step (i), the support was filled with glycerol/inhibitor (HQ) mixture. The detailed procedure for filling the support with liquid mixture can be found in Section S1, Supporting Information. Subsequently, in Step (ii), the filled membrane was placed on the cooled stage (15 oC) of the iCVD reactor, and the deposition of polytetrafluoroethylene (PTFE) is performed at a pressure of 900 mTorr for different times (e.g., five minutes). After the PTFE deposition, in Step (iii), the glycerol/inhibitor mixture was removed from the sample by placing the substrate in an isopropyl alcohol (IPA) bath. To visualize the glycerol/inhibitor removal, we placed the coated support on the top of an IPA/Water (50/50 v.%) bath, shown in Movie S1, Supporting Information. It can be clearly seen that the glycerol/inhibitor mixture is diffusing from the membrane into the bath. Figure 1 (b) shows a top-down scanning electron microscopy (SEM) image of the PVDF support. In that image, the top surface of the PVDF support has an open and interconnected structure. Figures 1 (c) and (d) also show the top-down and cross-section SEM images of the dried Janus membrane, respectively, fabricated through the described process. As shown, the deposited PTFE has a particulate-like structure on the top of the sponge-like structure of hydrophilic PVDF support; in the image we separated these areas using a dashed line. Moreover, the SEM images show that the top surface of membranes remained open after PTFE deposition. As a control experiment, we performed the same operation on porous PVDF supports, under the same operational conditions for iCVD, without wetting the membranes. The results indicate that the substrate, not prefilled with our liquid mixture, was blocked, and nitrogen did not flow through these substrates.