Hydrophobic modification of a PVDF hollow fiber membrane by plasma activation and silane grafting for membrane distillation

As a thermally driven membrane process, membrane distillation (MD) has some advantages, such as lower operating temperature than distillation, lower operation pressure than reverse osmosis, and high rejections of the nonvolatile compounds (Ray et al. 2020; Madalosso et al. 2021). MD can be applied to treat various wastewater, such as desalination brine, textile wastewater, and oily wastewater (Yan et al. 2021). During the MD process, the vapor crosses the microporous hydrophobic membrane while the nonvolatile compounds are kept on the feed side. Thus, the membrane acts as a barrier and plays a vital role in MD efficiency.

Increasing demand for MD technology accelerates the development of hydrophobic membrane fabrication methods, such as 3D imprinting, electrospinning, and attaching nanomaterials (Eykens et al. 2017; Ray et al. 2020). Water contact angles (WCAs) of the membranes prepared by these methods are generally more than 120° and even up to 160°, and such excellent hydrophobicity enhances the anti-wetting and anti-fouling properties of the membrane (Zhang et al. 2017, 2021). However, most of these methods are still in the laboratory. The industrial scaling-up of such membranes seems a long way off. In this study, we will focus on developing a simple and scalable method to prepare hydrophobic membranes.

Among the hydrophobic materials used in the MD process, polyvinylidene fluoride (PVDF) is widely used and outstanding for its high chemical stability, thermal stability, mechanical strength, and feasibility of forming membranes via phase inversion (Kang & Cao 2014; Lee et al. 2019). In order to fabricate microporous membranes, some pore-forming additives (e.g., polyvinylpyrrolidone (PVP) and LiCl) are added (Hou et al. 2009; Hamzah & Leo 2017). However, these additives remain in the membrane after phase inversion, and enhance the membrane hydrophilicity (Ray et al. 2020). The WCAs of such PVDF membranes stay around 90°, which leads to membrane wetting problems during the long-term running (Teoh & Chung 2009).

Besides membrane wetting, membrane fouling is another common problem in the MD process. For example, some natural organic matter (e.g., humic acid and polysaccharide) can easily attach to the membrane surface and pores, causing deterioration of water flux and salt rejection (Hou et al. 2015), which is an effective way to enhance membrane hydrophobicity to overcome the fouling and wetting problems (Eykens et al. 2017; Mousavi et al. 2021). Some commonly used hydrophobic modification methods include electrospinning, coating, chemical vapor deposition, and plasma (Jeong et al. 2016; Hamzah et al. 2018; Madalosso et al. 2021). Of all these methods, plasma treatment is convenient to change the surface properties without influencing the bulk materials, is easy to control, can be repeated, and is environmentally friendly (Boulares-Pender et al. 2013; Kravets et al. 2022). Plasma treatment always improves surface pore size, porosity, and roughness, and creates hydroxyl groups and other oxygen-containing groups (Gryta 2021a; Madalosso et al. 2021). Plasma treatment is always used to enhance surface hydrophilicity (Gryta 2021b; Sharma et al. 2021). Thus, further grafting of hydrophobic monomers is necessary to enhance the hydrophobicity (Deng et al. 2020).

Fluoroalkyl silanes are known for their low surface free energy with one end of hydrophobic fluoroalkyl, making them suitable hydrophobic monomers (Tran et al. 2021; Zhang et al. 2021). It is convenient to combine the plasma treatment and silane grafting, which can improve both the porosity and hydrophobicity of the membranes, and eventually improve the MD performance (Yang et al. 2011; Xu et al. 2015; Liu et al. 2016). Many factors, such as plasma activation, silane concentration, and the type of gas may influence the membrane properties (Xu et al. 2015; Liu et al. 2016; Lee et al. 2019).

In this study, hydrophobic modification of the PVDF hollow fiber membrane was carried out using a two-step method of plasma activation and silane grafting. The PVDF hollow fibers were firstly irradiated with plasma and then soaked into the grafting solution. The influence of plasma power, activation time, type of gas, and the silane type and concentration on the membrane hydrophobicity were investigated. Two organosilane monomers were compared, including methyl trichloroalkyl silane (MTCS) and 1H, 1H, 2H, 2H-perfluorooctane trichlorosilane silanes (PTCS). The modified membranes were further characterized for their physicochemical properties and to determine the performance of direct contact membrane distillation (DCMD). The anti-wetting and anti-fouling properties of humic acid were studied during a 20 h running time of DCMD. The possible factors influencing the performance of the modified membrane were then discussed. This hydrophobic modification is simple and scalable, which resulted in higher salt rejection and anti-fouling properties during the MD treatment. However, the permeability should be improved in further study.

PVDF with a molecular weight of 500 kDa was obtained from Sinochem Blue Sky Fluorine Materials Co., Ltd, China. PVP (k30) was purchased from Sinopharm Chemical Reagent Co., LTD. MTCS and PTCS were purchased from Shanghai McLean Co., Ltd. N,N-dimethylacetamide (DMAc, > 99%) and n-hexane were purchased from Rowen Reagent Co., Ltd, China. Humic acid sodium was purchased from Aldrich Reagents. Ultrapure water with conductivity < 2 μS cm⁻¹ and deionized water with conductivity < 10 μS cm⁻¹ were produced by the ultrapure water purification system (Millipore).

The spinning conditions are shown in Table 1. The homogeneous dope solution of 5 mL was loaded into a syringe and pressed into the spinneret at a flow rate of 6 mL·min⁻¹. The fibers were soaked in the coagulation bath for 10 min to complete the phase inversion. Then, the fibers were stored in deionized water for 1 day to remove the residual solvent. Finally, the wet fibers were dried at room temperature for the test or further hydrophobic modification.

The plasma treatment was carried out using a plasma system of PlasmaFlecto 10 (Plasma Technology, Germany). The dry PVDF hollow fibers were placed in the plasma chamber. N₂, O₂, or air gas was introduced into the chamber at 0.5–1 bar after the chamber was vacuumed to 0.2 mbar. After the plasma treatment, the fibers were immediately immersed in the silane solution in a sealed plastic container for 6 h. Silane was dissolved in n-hexane at 0.1–1.0 wt% for grafting. The grafted fibers were then rinsed three times with n-hexane to remove any unreacted silane. Finally, the modified fibers were dried at room temperature.

The effects of plasma operation parameters and the silane concentration on the membrane hydrophobicity were investigated in four sets of experiments, as shown in Table 2. After the grafting, the hydrophobicity of the fibers was tested using a WCA. The optimized grafting parameters were chosen based on the hydrophobicity of the modified membranes. Three membrane samples were made using these optimized grafting parameters, including the plasma-treated membrane – M1, MTCS-grafted membrane – M2, and PTCS-grafted membrane – M3. The pristine PVDF membrane was called M0.

After 20 g·L⁻¹ of NaCl solution was run for an 8-h DCMD test, the feed was replaced to carry on the fouling test. The fouling feed was the solution of 35 g·L⁻¹ of NaCl and 30 mg·L⁻¹ of humic acid. This test ran for 20 h, and the water flux and salt rejection were recorded. At the end of the test, the feed of the humic acid solution was removed, and the whole system was flushed with deionized water for half an hour. Then 20 g·L⁻¹ of NaCl solution was used as the feed and ran for a 3-h DCMD test. The water flux and salt rejection were recorded.

As the plasma activation time varied from 0.5 to 4 min, the WCAs of the modified membranes first increased and then decreased (Figure 3(b)). The highest WCAs were obtained at 2 min. Plasma treatment for 0.5–2 min can produce rougher and more porous membrane surfaces, while the 4-min plasma treatment resulted in a less porous surface (Supplementary material, Figure S2). Thus, 2-min plasma treatment was chosen.

With the plasma power varying from 150 to 300 W, the WCAs of the modified membranes slightly increased from 110° to 116° (Figure 3(c)). The effects of plasma power were not significant, and the power of 300 W was used in our following experiments.

With the silane concentration increasing from 0.1 to 1.0 wt%, the WCAs of the MTCS- and PTCS-grafted membranes increased from 107.4° to 114.4° and from 101.6° to 118.8°, respectively (Figure 3(d)). It is possible to deduce that more silane grafted onto the membrane at higher concentration, which made the surface more hydrophobic (Hamzah & Leo 2017; Thi My Hanh et al. 2022).

To make the PVDF hollow fiber surface more hydrophobic, the following conditions were chosen: plasma treatment of 300 W for 2 min under air and then grafted with 1 wt% MTCS or PTCS solution.

The pristine membrane – M0, plasma-treated membrane – M1, MTCS-grafted membrane – M2, and PTCS-grafted membrane – M3 were characterized using FTIR, XPS, SEM, and dynamic contact angle. The physicochemical properties of the four membranes were compared as follows.

The element compositions on the membrane surfaces were analyzed using XPS spectra (Figure 4(b) and Table 3). Compared with the pristine membrane – M0, the other three modified membranes have an obvious increase in O1s peak height and oxygen content, indicating that the oxygen-containing groups were introduced by the plasma treatment. No silicon is detected in M0 and M1, while the contents of silicon range 2.6–4.3% in M2 and M3. These changes in element composition prove the successful grafting of silane in M2 and M3.

The AFM images of the four membranes are shown in Supplementary material, Figure S3, and roughness data are shown in Table 4. The roughness of M0 decreased after plasma treatment, which then increased after grafting. The roughness values (R_a and R_q) of M2 and M3 were similar.

The bulk pore size and porosity of the membranes are listed in Table 4. Compared with the pristine membrane M0, M1 had a larger pore size of 0.395 μm. M1 and M0 showed similar porosity of 79.22–80.59%. These results are different from the changes of surface pore observed in the SEM images that an obvious increase in pore size and porosity was found in M1 (Figure 5). The plasma treatment mainly affects the top surface of the membrane and has little effect on the bulk material (Boulares-Pender et al. 2013; Lee et al. 2019).

The grafted membranes M2 and M3 showed smaller pore sizes and lower porosity, especially higher reduction for M2. Combined with the surface changes in SEM images, it can be deduced that the silane grafting not only covered the surface pores, but also blocked the bulk pores. The decrease in pore size and porosity in M2 and M3 may cause flux reduction in the MD process (Hamzah & Leo 2017; Madalosso et al. 2021).

LEPw represents the minimum pressure of water penetration into the membrane pores. The membrane with larger LEPw is supposed to run a longer time before wetting (Madalosso et al. 2021). The LEPw of the pristine membrane M0 was 1.2 bar. The modified membranes M2 and M3 had larger LEPw of >2.0 bar than M0. LEPw is strongly correlated with the WCA and pore size. M2 and M3 had larger WCAs and smaller pore sizes than M0, which led to the increase of LEPw. Conversely, the plasma-treated membrane M1 had smaller WCA and bigger surface pore size. Thus, the LEPw of M1 was the lowest, only 0.8 bar. The LEPw data indicate that M2 and M3 may show better anti-wetting properties.

The mechanical properties are determined by tensile strength, which is an important factor showing the stiffness of the hollow fibers. The tensile strength of all membranes ranged from 3.24 to 3.48 MPa (Table 4). No significant differences in the tensile strength were observed among all the membranes. It indicates that the plasma treatment and silane grafting would not change the stiffness of the PVDF hollow fibers.

The salt rejection of M0 kept steady at around 99.40%, which increased to 99.95 and 99.60% for M2 and M3, respectively (Figure 7(b)). High and stable salt rejection was found in M2, which may be attributed to the higher hydrophobicity and smaller pores. Both water flux and salt rejection data show that M2 had better DCMD performance than M3. These data are in the range of previous reports that the water flux of PVDF hollow fibers ranged 5–15 kg·m⁻²·h⁻¹ with salt rejections above 99% during the MD process (Xu et al. 2015; Madalosso et al. 2021).

Different from M0, the modified membranes M2 and M3 showed steadier water flux and salt rejection during the whole 20 h. The flux reductions were only 13 and 3% for M2 and M3, respectively. The salt rejection of M2 and M3 was kept steady above 99.5% during the whole process. These data indicate that the fouling on M2 and M3 is less serious than M0.

After the 20 h of fouling test, the membranes were flushed with deionized water and tested using 20 g·L⁻¹ of NaCl solution. The FRR achieved 100% for M2 and M3, while less than 5% for M0. The salt rejections were around 99.8% for M2 and M3, but no salt rejection was found with M0 (Table 5). The foulants on M2 and M3 can be easily flushed away, which was not the case for M0. The reversible fouling on M2 and M3 should be just on the membrane surface, while the irreversible fouling on M0 should be attributed to the fouling on the surface and in pores. The foulants on M0 caused the membrane wetting and resulted in bad DCMD performance.

The WCAs of all membranes decreased after the fouling test. It is usually considered that membranes with a contact angle greater than 90° are hydrophobic, while membranes with a contact angle less than 90° are hydrophilic (Duong et al. 2018). The fouled M0 became hydrophilic with the WCA dropping to 75°. The fouled M2 and M3 were still hydrophobic with WCAs nearly or larger than 90°. The humic acid fouling on the membrane surface caused reduction in the WCAs. More humic acid accumulated on the membrane M0, causing a darker colored surface of M0 than M2 and M3 (Supplementary material, Figure S4).

Therefore, the modified membrane showed better anti-fouling properties than the pristine membrane. However, the flux reduction after grafting should be paid attention to, and further optimization on the grafting needs to be carried out.

In this study, the hydrophobic modification of the PVDF hollow fiber membrane was carried out by plasma activation and grafting of MTCS and PTCS. The XPS, FTIR, and SEM data proved the successful grafting of silane onto the membrane surface. To achieve a more hydrophobic surface, the optimized conditions included plasma treatment of 300 W for 2 min under air and then grafted with 1 wt% MTCS or PTCS solution. The hydrophobicity of the modified membranes increased with WCA increasing from 88° to 112°–116°. Meanwhile, the pore size and porosity decreased after grafting.

These property changes complied with the decrease of water flux and increase of salt rejection during the DCMD test. The water flux decreased by 35–65% for the modified membranes during DCMD test of treating 20 g·L⁻¹ of NaCl, while the rejection increased from 99.4% of the pristine membrane to 99.6–99.9% of the grafted membranes. However, during the fouling test, the modified membrane showed more excellent anti-fouling properties. The pristine membrane had serious fouling with significant flux and rejection reduction, whereas the modified membranes showed steady water flux and rejections, and had 100% flux recovery after flushing.

Plasma activation and grafting of MTCS is an effective way to hydrophobically modify PVDF hollow fibers, which show excellent anti-fouling properties. But the flux reduction should be further improved.

This work was supported by the National Natural Science Foundation of China [NSFC U 1867208] and LingChuang Research Project of China National Nuclear Corporation.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.