Preparation of an Ultrafiltration (UF) Membrane with Narrow and Uniform Pore Size Distribution via Etching of SiO2 Nano-Particles in a Membrane Matrix

The UF membrane with a narrow and uniform pore size distribution and a low tendency to foul has significant applications in wastewater treatment. A major hindrance in the preparation of the UF membrane with these features is the lack of a scalable and economical membrane fabrication method. Herein, we devise a new strategy to prepare a high-quality polyvinylidene fluoride/polymethyl acrylate/cellulose acetate (PVDF/PMMA/CA) blend UF membrane via a combination of the etching mechanism with the traditional Loeb–Sourirajan (L-S) phase inversion method. Different concentrations of silicon dioxide (SiO2) nanoparticles (NP) in the membrane matrix were etched by using a 0.2 M hydrofluoric acid (HF) solution in a coagulation bath. This strategy provided the membrane with unique features along with a narrow and uniform pore size distribution (0.030 ± 0.005 μm). The etched membrane exhibits an increase of 2.3 times in pure water flux (PWF) and of 6.5 times in permeate flux(PF), with a slight decrease in rejection ratio (93.2% vs. 97%) when compared to than that of the un-etched membrane. Moreover, this membrane displayed outstanding antifouling ability, i.e., a flux recovery ratio (FRR) of 97% for 1000 mg/L bovine serum albumin (BSA) solution, a low irreversible fouling ratio of 0.5%, and highly enhanced hydrophilicity due to the formation of pores/voids throughout the membrane structure. The aforementioned features of the etched membrane indicate that the proposed method of etching SiO2 NP in membrane matrix has a great potential to improve the structure and separation efficiency of a PVDF/PMMA/CA blend membrane.

antifouling property and high permeability, which are essential for membranes to satisfy the requirement of industrial separation processes.

Materials
The membrane materials PVDF/PMMA(Mw = 4000-5000)/CA-398 were obtained from Shanghai 3F New Material Ltd., China, η = 1.4-1.9/Beijing Organic Chemical Engineering Factory/Britain Nelsons Acetate. Solvents: N-methyl-2-pyrrolidone (NMP)/HF were procured from Tianjin Fuchen Chemical Reagent Factory, China. Additive PVP was purchased from Beijing Chemical Engineering Factory. Bovine serum albumin (BSA) (Mw = 67,000 Da), used as a reagent for determining the observed retention of the membrane, was acquired from Beijing Microorganism Culture Medium Manufacturing Corporation China and its isoelectric point of pH is 4.8. Deionized-water was used in all experiments.

Preparation of PVDF/PMMA/CA Blend Membranes
The detailed fabrication procedure is shown in Figure 1. Initially, different concentrations of SiO2 NPs (0.3, 0.5, 0.7 and 1 wt%) were dispersed into a specific amount of NMP via vigorous sonication for 1 h. After attaining the uniform dispersion of NP, the stoichiometric amounts of PVDF, PMMA and CA were dissolved in the solutions according to a certain ratio (11.2:2.4:2.4) under intensive stirring. After complete dissolution of the polymers, a measured amount of PVP (3 wt%) was added and the resulting solution was intensively stirred for 2 h in order to achieve a homogeneous mixture of all ingredients. Then the casting solutions were placed in the vacuum oven at 65 °C for 4 days.
Later, the casting solutions were cast on clean glass plates with a casting blade under 25 °C and 30% humidity, and the plates were gently immersed into a coagulation bath containing DI-water/0.2 To the best of the authors' knowledge, the etching of SiO 2 NPs in s membrane matrix combined with the L-S phase inversion method to prepare PVDF/PMMA/CA blend membranes has not been reported yet in the literature. This technique may be a feasible approach for the preparation of highly porous PVDF/PMMA/CA membranes to achieve a uniform pore size distribution, superior antifouling property and high permeability, which are essential for membranes to satisfy the requirement of industrial separation processes.

Materials
The membrane materials PVDF/PMMA(Mw = 4000-5000)/CA-398 were obtained from Shanghai 3F New Material Ltd., China, η = 1.4-1.9/Beijing Organic Chemical Engineering Factory/Britain Nelsons Acetate. Solvents: N-methyl-2-pyrrolidone (NMP)/HF were procured from Tianjin Fuchen Chemical Reagent Factory, China. Additive PVP was purchased from Beijing Chemical Engineering Factory. Bovine serum albumin (BSA) (Mw = 67,000 Da), used as a reagent for determining the observed retention of the membrane, was acquired from Beijing Microorganism Culture Medium Manufacturing Corporation China and its isoelectric point of pH is 4.8. Deionized-water was used in all experiments.

Preparation of PVDF/PMMA/CA Blend Membranes
The detailed fabrication procedure is shown in Figure 1. Initially, different concentrations of SiO 2 NPs (0.3, 0.5, 0.7 and 1 wt%) were dispersed into a specific amount of NMP via vigorous sonication for 1 h. After attaining the uniform dispersion of NP, the stoichiometric amounts of PVDF, PMMA and CA were dissolved in the solutions according to a certain ratio (11.2:2.4:2.4) under intensive stirring. After complete dissolution of the polymers, a measured amount of PVP (3 wt%) was added and the resulting solution was intensively stirred for 2 h in order to achieve a homogeneous mixture of all ingredients. Then the casting solutions were placed in the vacuum oven at 65 • C for 4 days.
Later, the casting solutions were cast on clean glass plates with a casting blade under 25 • C and 30% humidity, and the plates were gently immersed into a coagulation bath containing DI-water/0.2 M HF solution at 25 • C for 30 mins. Finally, the membrane were stored in DI-water at room temperature for 2 days prior to filtration test in the stirred cross flow cell. The resultant membranes with DI-water as coagulant were M@SiO 2(0) , M@SiO 2(0.3) , M@SiO 2(0.5) , M@SiO 2(0.7) M@SiO 2(1) while the membranes with HF solution as coagulant were named M@SiO 2(0) HF, M@SiO 2(0.3) HF, M@SiO 2(0.5) HF, M@SiO 2(0.7) HF and M@SiO 2(1) HF, respectively.

Design of Coagulation Bath for Etched Membranes
The coagulation bath was comprised of 0.2 M of HF solution for the etched membranes.

Basic Chemical Reaction
On immersing cast films into the coagulation bath, the following chemical reaction occurred between SiO 2 NPs in the membrane matrix and 0.2 M HF solution in the coagulation bath.

Characterization of Etched UF Membranes
SEM images of all the synthesized membranes were obtained by using a scanning electron microscopy (SEM, S-4800, Hitachi Limited Inc., Tokyo, Japan). Snapping of the membranes was completed under liquid nitrogen to give a clean and concordant cut. Double-sided adhesive tape was used to mount the membrane samples on sides of brass plate and then samples were sputtered with Au before observation.
The sessile-drop method (Zwick/Roell BL-GRS500N) was utilized to measure the water contact angles (WCA) at 25 • C and 50% relative humidity. Before measurements, all the membrane samples were dried for 24 h at room temperature. In order to minimize the systematic error and to attain the accuracy, five measurements on different locations of each sample were taken and the mean of resulting values was computed.
The liquid-liquid displacement technique was applied to analyze the pore size distribution by utilizing n-butyl alcohol-water as a solvent pair [37]. The pore size distribution functions were calculated by using Equations (1) and (2): where, r, σ, θ and J i represent the pore radius, the surface tension of n-butyl alcohol-water, the polymer-n-butyl alcohol contact angle, and the flux measured at the i th increment where the applied pressure is P i , respectively [38,39]. The dry-wet method was used to measure the porosity of the membrane samples by using Equation (3): where P o is the porosity of the membrane (%), W 2 is the weight of the wet membrane (g), W 1 is the weight of the dry membrane (g), V o is the volume of the membrane (cm 3 ) and d water is the water density at room temperature (g·cm −3 ).

The Performance of Etched UF Membranes
Permeation tests were performed using a cross-flow stirred cell having an effective area of 24 cm 2 . One g/L BSA protein was used as a feed solution at 0.1 M Pa and 25 • C. Rejection was checked on UV-Vis spectrophotometer (Shimadzu, Japan) at a wavelength of 280 nm. The equations employed to compute the resultant flux and rejection are as follows [2,10]: where J P , ∆V, ∆t (h), P and A (m 2 ) denote the permeate flux (L/(m 2 h)), the permeate volume (L), the filtration time interval (h), the driving pressure (Mpa) and the effective area (m 2 ), respectively.
where the feed solution and permeate concentrations are denoted by C f (g/L) and C p (g/L), respectively.

Antifouling Performance Measurements
In order to study the antifouling performance of membranes, BSA as a typical foulant was utilized as feed solution. Firstly, to measure the pure water flux (J W ) of the membrane, DI-water was used at 0.1 MPa. Secondly, 1.0 g/L BSA solution (pH = 7) was filtrated through cross-flow cell for 80 min at 0.1 MPa and the permeate flux (J p ) was calculated. Thirdly, the cleaning of the fouled membrane was completed with DI-water for 20 min at a stirring speed of 600 rpm, and then the PWF (J R ) of the membrane washed with DI-water was calculated again. The above-mentioned filtration cycle was repeated thrice for each membrane [40,41].
Equation (6) was used to compute the flux recovery ratio (FRR).
Three resistance parameters were characterized to make a thorough study on the membrane fouling behavior. The total fouling ratio (R t ), irreversible fouling ratio (R ir ) and reversible fouling ratio (R r ) were computed by utilizing the following equations: where R t is the total decrease in flux degree because of the total membrane fouling. While R r and R ir are the sum of R t , R ir is the fouling caused due to the deposition of foulant (protein) on the membrane surface and membrane pores, and R r is the fouling caused by concentration polarization. The permeate flux and rejection for BSA were performed at a subsequent time to assess the long-term stability of the fabricated membranes.

The Pore Size Distribution
The membranes which possess a large number of pores with narrow mean pore radii and a uniform pore size distribution can exhibit superior filtration efficiency [4,5]. The mean pore radii and pore size distributions of the prepared membranes with different coagulation baths (0.2 M HF solution/DI-water) and different concentrations of SiO 2 NPs (0, 0.3, 0.7 and 1 wt%) are given in Figure 2. A sharp decrease in the mean pore radii (0.06 µm/0.05 µm/0.03 µm) of the etched membranes (M@SiO 2 (0.3) HF/M@SiO 2 (0.7) HF/M@SiO 2 (1) HF) with an increase in SiO 2 concentration can be seen in Figure 2a 1 , a 2 and a 3 . This significant reduction in pore size is credited to the generation of pores/voids during etching mechanism. This indicates that an increase in SiO 2 NP concentration in the membrane matrix increased the number of pores, which resulted in high porosity and reduced mean pore size.
during etching mechanism. This indicates that an increase in SiO2NP concentration in the membrane matrix increased the number of pores, which resulted in high porosity and reduced mean pore size.
On the other hand, the un-etched membrane M@SiO2 (1) prepared with DI-water (coagulation bath) possesses bigger mean pore radii (1.16 μm) and its pore radius distribution range (1.16 ± 0.060 μm) is also broader (Figure 2b) than the etched membrane (0.030 ± 0.005 μm) with the same SiO2 concentration. Moreover, the pore size distribution of the former membrane is also more uniform and it had the smallest mean pore radii which resulted in higher flux. Contrastingly, the bigger mean pore radii (1.16 μm) in the case of the latter membrane resulted in less permeation due to clogging of pores by protein. When NPs were etched in the membrane matrix, pore connectivity and asymmetry were significantly improved. As a result, a well-connected microporous membrane with high porosity was obtained and its average pore size was reduced to around 0.03 μm. The mean pore size and porosity of the fabricated membranes are given in Table 1.

Surface Hydrophilicity
The surface hydrophilicity is always considered a crucial parameter for the performance of membranes, as it plays a vital role in enhancing the permeation and antifouling property. The membrane with superior hydrophilicity has the ability to adsorb more water molecules and resist the attachment of pollutants (e.g., macromolecules, ions, etc.) to its surface [35]. Therefore, the wettability of the prepared membranes is tested by measuring their water contact angles (WCA) (Figure 3). A decrease in water contact angle with an increase in SiO2NPs (0.3 to 1 wt %) is observed in the case of the etched membranes. Among all the prepared etched membranes, M@SiO2 (1) HF exhibited the lowest WCA value of 52.9° due its excellent hydrophilicity and higher porosity (89%). As etching of SiO2NPs results in the formation of pores/voids, a larger number of voids were formed due to higher concentration of SiO2NPs in the membrane matrix which, in turn, increased the porosity with extra pore channels for water molecules to penetrate the membrane [42]. On the other hand, un-etched membranes M@SiO2(0.7)HF/M@SiO2(1)HF showed the highest WCA values (71.2° and 70.7°, respectively) due to the clogging of the valleys with SiO2 NPs, which were removed by HF in the case of the etched membranes. Thus, etching of SiO2NPs in the membrane matrix can significantly improve the membrane surface properties, which is vital for their application. On the other hand, the un-etched membrane M@SiO 2 (1) prepared with DI-water (coagulation bath) possesses bigger mean pore radii (1.16 µm) and its pore radius distribution range (1.16 ± 0.060 µm) is also broader (Figure 2b) than the etched membrane (0.030 ± 0.005 µm) with the same SiO 2 concentration. Moreover, the pore size distribution of the former membrane is also more uniform and it had the smallest mean pore radii which resulted in higher flux. Contrastingly, the bigger mean pore radii (1.16 µm) in the case of the latter membrane resulted in less permeation due to clogging of pores by protein. When NPs were etched in the membrane matrix, pore connectivity and asymmetry were significantly improved. As a result, a well-connected microporous membrane with high porosity was obtained and its average pore size was reduced to around 0.03 µm. The mean pore size and porosity of the fabricated membranes are given in Table 1.

Surface Hydrophilicity
The surface hydrophilicity is always considered a crucial parameter for the performance of membranes, as it plays a vital role in enhancing the permeation and antifouling property. The membrane with superior hydrophilicity has the ability to adsorb more water molecules and resist the attachment of pollutants (e.g., macromolecules, ions, etc.) to its surface [35]. Therefore, the wettability of the prepared membranes is tested by measuring their water contact angles (WCA) (Figure 3). A decrease in water contact angle with an increase in SiO 2 NPs (0.3 to 1 wt %) is observed in the case of the etched membranes. Among all the prepared etched membranes, M@SiO 2 (1) HF exhibited the lowest WCA value of 52.9 • due its excellent hydrophilicity and higher porosity (89%). As etching of SiO 2 NPs results in the formation of pores/voids, a larger number of voids were formed due to higher concentration of SiO 2 NPs in the membrane matrix which, in turn, increased the porosity with extra pore channels for water molecules to penetrate the membrane [42]. On the other hand, un-etched membranes M@SiO 2(0.7) HF/M@SiO 2(1) HF showed the highest WCA values (71.2 • and 70.7 • , respectively) due to the clogging of the valleys with SiO 2 NPs, which were removed by HF in the case of the etched membranes. Thus, etching of SiO 2 NPs in the membrane matrix can significantly improve the membrane surface properties, which is vital for their application.

Morphological Investigation
Surface and cross-sectional photographs of all the fabricated membranes are given in Figure 4. The results for the M@SiO2(0) and M@SiO2(0) HF membranes illustrate that both of the membranes possess a smooth surface without any visual defect. This indicates that the coagulant has no effect on the appearance of membrane surface in terms of pore size and porosity in the absence of SiO2 NPs in the membrane matrix. Similarly, cross-sectional observations reveal that both of the membranes also possess almost the same morphological structure. They have a typical asymmetric structure with a dense top surface and un-straight finger-like projections in the sub-layer. Similarity in the crosssectional as well as surface observations of both membranes indicates that the hydrofluoric acid as a coagulant has no effect on the morphology of the membrane when the concentration of SiO2NPs is 0 wt%, and the structure remains well preserved. In contrast, after the incorporation of SiO2NPs (1 wt%) in the casting solution, the membrane was found to have a porous structure with the straightthrough finger-like projections and macro-voids. The reason is mainly because of the accelerated solvent (NMP) and non-solvent (DI-water) exchange rate during the phase inversion process due to embedded hydrophilic SiO2 NPs in casting solution, which resulted in macro-voids and finger like pores [43]. On the other hand, the etching of the SiO2 NPs (1 wt%) in the membrane matrix using a 0.2 M HF solution as a coagulant significantly affected the morphology of the M@SiO2(1) HF membrane. The M@SiO2(1) HF membrane exhibited un-straight finger-like projections with a highly porous top surface due to the formation of voids/pores. These voids/pores are the result of chemical reaction between SiO2 NPs in the membrane matrix and the 0.2 M solution of HF in the coagulation bath which promoted the formations of narrower and large pore channels by accelerating the solvent and non-solvent exchange rate [44] and increased the thermodynamic instability of the cast film. As a consequence, the instantaneous de-mixing occurred in the coagulation bath, which resulted in unstraight-through finger-like projections, which provided the additional pathways for water transportation and increased the filtration efficiency [45]. Therefore, the etching mechanism does not destroy the structure of the membrane and its morphology is well preserved.

Morphological Investigation
Surface and cross-sectional photographs of all the fabricated membranes are given in Figure 4. The results for the M@SiO 2(0) and M@SiO 2(0) HF membranes illustrate that both of the membranes possess a smooth surface without any visual defect. This indicates that the coagulant has no effect on the appearance of membrane surface in terms of pore size and porosity in the absence of SiO 2 NPs in the membrane matrix. Similarly, cross-sectional observations reveal that both of the membranes also possess almost the same morphological structure. They have a typical asymmetric structure with a dense top surface and un-straight finger-like projections in the sub-layer. Similarity in the cross-sectional as well as surface observations of both membranes indicates that the hydrofluoric acid as a coagulant has no effect on the morphology of the membrane when the concentration of SiO 2 NPs is 0 wt%, and the structure remains well preserved. In contrast, after the incorporation of SiO 2 NPs (1 wt%) in the casting solution, the membrane was found to have a porous structure with the straight-through finger-like projections and macro-voids. The reason is mainly because of the accelerated solvent (NMP) and non-solvent (DI-water) exchange rate during the phase inversion process due to embedded hydrophilic SiO 2 NPs in casting solution, which resulted in macro-voids and finger like pores [43]. On the other hand, the etching of the SiO 2 NPs (1 wt%) in the membrane matrix using a 0.2 M HF solution as a coagulant significantly affected the morphology of the M@SiO 2(1) HF membrane. The M@SiO 2(1) HF membrane exhibited un-straight finger-like projections with a highly porous top surface due to the formation of voids/pores. These voids/pores are the result of chemical reaction between SiO 2 NPs in the membrane matrix and the 0.2 M solution of HF in the coagulation bath which promoted the formations of narrower and large pore channels by accelerating the solvent and non-solvent exchange rate [44] and increased the thermodynamic instability of the cast film. As a consequence, the instantaneous de-mixing occurred in the coagulation bath, which resulted in un-straight-through finger-like projections, which provided the additional pathways for water transportation and increased the filtration efficiency [45]. Therefore, the etching mechanism does not destroy the structure of the membrane and its morphology is well preserved.

Membrane Permeation Flux
The permeation through a membrane is very important for its application. Thus, the filtration experiments were performed to evaluate the permeation properties of the modified membranes. Figure 5 shows the pure water flux, permeation flux and BSA rejection rates of the prepared membranes with different coagulation baths (DI-water/ 0.2 M HF solution).

Membrane Permeation Flux
The permeation through a membrane is very important for its application. Thus, the filtration experiments were performed to evaluate the permeation properties of the modified membranes. Figure 5 shows the pure water flux, permeation flux and BSA rejection rates of the prepared membranes with different coagulation baths (DI-water/ 0.2 M HF solution). As indicated in Figure 5a, by increasing SiO 2 concentration from 0.3 to 1 wt% in the membrane matrix, the PWF through the un-etched membranes decreased. In contrast, an increase in PWF through the etched membranes is observed with increasing SiO 2 concentration from 0.3 to 1 wt%.
This revealed that the etching of different SiO 2 NP concentrations in the membrane matrix enhanced the porosity and provided the membrane with good hydrophilicity. Consequently, a higher pure water flux(PWF)of the etched membranes was found, and it was comparatively 2.3 times that of the un-etched membranes. Similarly, a rapid rise in permeate flux (PF) was noticed with rising values of SiO 2 NP concentration (Figure 5b). For instance, a comparative rise in PF of etched membrane was 3.6 times higher with 0.3 wt% of SiO 2 NP concentration, and it was 6.5 times higher when we further increased SiO 2 NPs from 0.3 to 1 wt%. The probable cause is that the porous structure and uniform pore size distribution of the etched membranes provide additional pathways for water transportation and also decreases the tortuosity, which is in favor of PF. However, both PF and PWF have shown totally opposite trends the in case of un-etched membranes. The decline in both of these characteristics with a rise in SiO 2 NP concentration is attributed to the agglomeration of SiO 2 NPs, which may block the pathways [35]. It will result in the resistance for water transport. The flux and rejection of all the prepared membranes are presented in Table 2. To further investigate the increase in membrane permeability after the etching of SiO 2 NPs in the membrane matrix, SEM images of the membranes were taken under consideration ( Figure 4). The etching of SiO 2 NPs in the membrane matrix resulted in more porous and asymmetric membrane structures. The asymmetric membranes are fouling-resistant, which is consistent with the increased PWF and higher rejection. A performance comparison of the current study with the literature is given in Table 3.

BSA Rejection Performance
Protein is the main problematic pollutant among other organic foulants of UF membranes [53]. As the BSA rejection rate reflects the size exclusion ability of the membrane surface, a relatively higher value of BSA rejection in the present study implies that the pore size distribution and geometry of the skin layer of the membrane has significantly changed. A comparison of BSA rejection for the etched and un-etched membranes is presented in Figure 5b.
The results show that the un-etched membranes exhibit a slightly higher rejection rate with an increase in SiO 2 NPs, and the highest rejection ratio of 97% is achieved for M@SiO 2(1) . This is associated with the less porous structure and relatively un-uniform pore size distribution, which prevents the penetration of BSA molecules through the membrane by reducing the number of paths [54].
Contrastingly, a slight decrease in BSA rejection of the etched membranes with an increase in SiO 2 NPs is observed, and the lowest rejection ratio of 93.2% is achieved by M@SiO 2(1) HF. This is attributed to the more and asymmetric pore channels formed by rapid phase inversion.

Antifouling Performance
The fouling of the membrane surface by a foulant can cause a considerable decline in the permeation flux. A reduction in hydrophobic attraction between the membrane surface and the foulant through hydrophilic surface modification can overcome the fouling problem [55,56].
The antifouling capacity of all the synthesized membranes is tested and the corresponding results are given in Figure 6. The evaluation of the membrane antifouling property flux recovery ratio (FRR) is a crucial parameter, and the membranes with a higher FRR value possess excellent antifouling performance. However, a significant improvement in the antifouling capacity of the etched membranes was observed (Figure 6b). As seen in Figure 6b, the FRR of the etched membranes sharply increased with an increase in SiO 2 NP concentration in the membrane matrix. The maximum FRR is 95% (81% for without etching), which was witnessed for the membrane M@SiO 2 (1) HF. This indicates that the higher filtration efficiency can be achieved after etching SiO 2 NP in the membrane matrix by HF solution as it generates the pores/voids in the membrane structure, which increase with an increase in SiO 2 NP concentration. Therefore, the membrane structure becomes more porous, which increases the hydrophilicity of the prepared membrane. As a consequence, more water molecules are attracted to its surface to create a compact water barrier which prevents the interactions between foulants and the membrane surface [46]. To further investigate the antifouling performance of the prepared membranes, fouling resistance was tested, which includes the irreversible resistance (R ir) and reversible fouling (R r ).The results of fouling parameters (R ir ) and (R r ) are shown in Figure 6a,b.
A remarkable decrease in the R ir (0.5%/0.8%) was observed for the etched membranes M@SiO 2(0.7) HF and M@SiO 2(1) HF, and this is attributed to the superior features of the membrane surface properties (higher hydrophilicity and higher porosity), which can weaken the interactions between the membrane surface and foulants by reducing the contact area [47]. On the other hand, a sharp decline in permeation flux for the un-etched membranes with an increase in SiO 2 NP concentration was observed due to membrane fouling (Figure 5b). The un-etched membranes exhibited higher irreversible fouling because of the less porous sub-layer structure and less hydrophilicity. Adsorption of foulants on the membrane surface and in the valleys of the membrane is caused due to interactions between the membrane surface and foulants [43], and this ultimately results in the clogging of the valleys that cannot be washed simply by rinsing. A remarkable decrease in the Rir (0.5%/0.8%) was observed for the etched membranes M@SiO2(0.7)HF and M@SiO2(1)HF, and this is attributed to the superior features of the membrane surface properties (higher hydrophilicity and higher porosity), which can weaken the interactions between the membrane surface and foulants by reducing the contact area [47]. On the other hand, a sharp decline in permeation flux for the un-etched membranes with an increase in SiO2 NP

Conclusion
In the current study, a highly porous UF membrane with a narrow (0.03 µm) and more uniform pore size distribution (0.030 ± 0.005 µm) has been successfully synthesized via the combination of the etching mechanism and the L-S phase inversion method for the first time. Different concentrations of SiO 2 NPs in the membrane matrix were etched with a 0.2 M HF solution in a coagulation bath. The etching mechanism engendered voids/pores which provided the membrane with a hydrophilic and highly porous surface. As a consequence, the prepared etched MF membrane exhibited a rejection of 93.5%, with high PWF (19494 L/m 2 ·h) and enhanced permeance (10,368 L/m 2 ·h). Moreover, the membrane achieved an excellent antifouling property (i.e., FRR of 95% for a 1000 mg/L BSA aqueous solution). It can be concluded that an etched UF membrane with asymmetric and a reduced mean pore radii (0.03 µm) can be prepared via the etching mechanism combined with the L-S phase inversion method, which significantly improves the pore size distribution and porosity. Hence, the proposed method of etching SiO 2 nano-particles in membrane matrix may be considered as a feasible approach to improve the structure and separation efficiency of the PVDF/PMMA/CA blend membrane.