2.1 Materials and membrane preparation

Commercial PVDF-HFP (Mw 4.0 × 10⁵ g mol⁻¹, Mw/Mn = 3.08) polymer pellets were purchased from Sigma-Aldrich, USA. Hydrophobic SiO₂ nanoparticles supplied with courtesy from Wacker Chemie, Germany were used as solvent additives. The superhydrophobic membrane was fabricated by a NIPS process. PVDF-HFP polymer pellets and SiO₂ nanoparticles were dried at 60 °C in a vacuum oven for 10 h to remove moisture. Specific amounts of hydrophobic SiO₂ nanoparticles (e.g., 0, 1, 1.5, 2, 2.5, 3, 3.5 wt %) were added into 75 wt % (or 9.07 ml) of an NMP solvent and sonicated for 1 h. PVDF-HFP polymer pellets (15 wt % or 1.65 g) were dissolved in a mixture at 200 rpm and 70 °C for 24 h. The resulting solution was degassed for 5 h and left standing overnight. The dope solution was cast on a glass plate at 200 μm thickness gap and immersed in nonsolvent, ethanol, for 24 h. The membrane film was peeled from the glass plate and dried at room temperature for 48 h.

2.2 Membrane characterization

The surface microstructure of the membranes was observed using the Tabletop Microscope TM3030 (Hitachi, Japan). All samples were sputtered with a thin film of gold to render conductivity. The images of the membranes were taken at 15 kV accelerating voltage and 1000× magnification. Energy-dispersive X-ray spectroscopy (EDX) analysis was performed to quantify the weight percent of Si elemental composition on the membrane surface. The membrane hydrophobicity was examined via contact angle measurement using a goniometer (Rame-hart 260, USA). A 13-μL water droplet was dropped through a microsyringe onto the surface of the membrane under ambient conditions. The average value of each membrane was obtained from six measurements at different points. The geometric mean approach was used to determine the surface energy of the membrane [4]. In this method, contact angle of water and diiodomethane on the membrane surface was determined at room temperature. The surface energy was calculated from a set of two first-order linear equations. The surface roughness of the membrane was analyzed using a Phenom Desktop Scanning Electron Microscopy system with three-dimensional (3D) roughness reconstruction application. The images were reconstructed into 3D topography to determine the average surface roughness, R_a. The topographies of the membrane were measured at a scan area of 50 μm × 50 μm. For each membrane, three samples were analyzed, and the average was reported.

3.1 Scanning electron microscopy

The membrane microstructural morphology was visualized by SEM. As illustrated in Fig. 1(a), crystallization-dominated structure composed of interconnected globules was observed in the pristine membrane without silica nanoparticles. Such morphology was induced by the polymer crystallization process in a soft precipitation bath (ethanol). During immersion precipitation process, the semicrystalline PVDF-HFP polymer induced the growth of crystallites in the membrane bulk [7]. The crystallizable segments of the polymer nucleated to similar globular size and merged together to form the interlinked spherulitic microparticles.

The surface morphology changes associated with the addition of nano-SiO₂ particles are shown in Fig. 1. It was observed that the globular crystallites of the nano-SiO₂ blended membranes appeared to be more closely packed with fewer micropores. According to Sukitpaneenit and Chung [8], the size of globules is dependent on the time allowed for crystallization, whereby longer time promoted the formation of bigger globules. The addition of hydrophobic SiO₂ increased the viscosity of the polymer dope solution. This intensifies the delayed phase inversion and allows for a greater extent of polymer crystallization to occur in the coagulation bath. Subsequently, the top microstructure of the membrane becomes denser and more compact. The presence of nanoparticles on the membrane surface was examined by EDX analysis. As depicted in Fig. 1, the weight percentage of Si was increased from 2.89% to 12.11% with increased nanoparticles concentration from 1 to 3.5 wt %. This indicated the enhanced embedment of SiO₂ nanoparticles in the polymer matrix.

Fig. 2 shows the 3D surface images and average surface roughness, R_a. For the SiO₂-free pristine membrane, the R_a was lowest at 1.80 ± 0.37 μm due to low surface protrusion. Upon addition of nanoparticles, SiO₂-blended membranes induced the formation of hierarchical structures with multilevel roughness on the membrane surface, which results in increased protrusions and valley regions. With the increase in nanoparticles' concentration up to 3.5 wt %, the R_a was improved by 49.7% to 3.58 ± 0.39 μm. The significant changes in the R_a may be attributed to the localization of PVDF-HFP spherulites and nanoparticles. A similar observation in surface roughness was reported by Ghaee et al. [6]. This indicates that the nanoparticles have a positive impact on membrane surface roughness.

The surface energy of the membranes is shown in Fig. 3. The pristine membrane exhibited the highest surface energy of 9.56 mN m⁻¹, indicating the presence of the highest dispersion force and polarity on the surface. Therefore, the pristine membrane is highly prone to interact with water molecules and cause wetting. By increasing the SiO₂ content from 1 to 3.5 wt %, the surface energy was reduced from 9.29 to 3.49 mN m⁻¹. This was consistent with the Si elemental composition results. The reduction in surface energy was attributed to the presence of a nonpolar Si–O–Si chain from hydrophobic SiO₂ nanoparticles, which induced the steric hindrance effects in the structure. Therefore, nano-SiO₂ blended membranes were hindered from interacting with polar molecules and thus led to enhanced antiwettability properties.

The wetting behavior of the resulting membrane is illustrated via water contact angle measurement in Fig. 3. The pristine membrane showed the lowest water contact angle of 135°. The development of a superhydrophobic surface was hindered by high surface energy and low surface roughness. After blending of nanoparticles, all the nano-SiO₂ blended membranes revealed an increase in contact angle to the range of 138°–151°, which corresponds to a vast improvement in membrane hydrophobicity. The highest water contact angle of 151° was achieved in the membrane with 3.5 wt % of SiO₂, demonstrating superhydrophobicity. As shown in Fig. 1(g), water droplets on this superhydrophobic surface are almost spherical in shape, due to the poor adsorption of water molecules onto the surface. According to the Wenzel model, such deviation in water contact angle was attributed to the significant increase in surface roughness and reduction in surface energy [9]. The textured surface trapped air pockets between the patchwork of hierarchical surface protrusions. Meanwhile, low surface energy decreased the adhesion of water molecules on the surface of the membrane. Therefore, water molecules have less direct contact with the membrane, rendering the surface superhydrophobic.