Evaluation of the physical properties and filtration efficiency of PVDF/PAN nanofiber membranes by using dry milk protein

In engineering applications, especially ultrafiltration (UF) applications, it is very important to use polyacrylonitrile (PAN) and poly (vinylidene fluoride) (PVDF) nanofiber membranes. In this study, membrane nanofibers made of pure PAN, PVDF: PAN blends, and pure PVDF (M1, M2, M3, M4, M5, and M6), were produced by the electrospinning technique with different contents of PVDF in each blend. The prepared membranes were characterized by field emission scanning electron microscopy (FE-SEM), atomic force microscopy (AFM), contact angle measurements, x-ray diffraction (XRD), Fourier transform infrared (FT-IR) spectrophotometry, and differential scanning calorimetry-thermogravimetric analysis (DSC-TGA). In terms of the physical properties, the viscosity of the membranes increased with an increase in the content of PVDF in the blends compared with the viscosity of the pure polymer solutions. This led to increases in nanofiber diameter, pore size, and porosity by 261.664%, 875.107%, and 114.41%, respectively, when the content of PVDF increased from 20% (M2) to 80% (M5); this was also accompanied by an increase in the surface wettability of the membrane depending on its contact angle. In addition, the thermal properties and crystallinity of PAN improved after increasing the PVDF content from 20% (M2) to 60% (M4). Moreover, the filtration efficiency of the membranes was measured to determine the per cent reduction in pure water flux, reduction in mean depth (RMD) before and after using dry milk protein, the flux recovery ratio and porosity, giving values of 15.68%, 82.51%, 84.32%, and 67.79%, respectively, for the M4 membrane.


Introduction
The electrospinning technique is considered one of the most important techniques for producing nanofibers from polymeric solutions depending on the electric field because it is low in cost and produces a uniform and continuous nanofibers with high porosity [1]; the process of directing the nanofibers horizontally using a rotating collector or electrically with a pair of conductive substrates separated, which led to the enhancement of the mechanical and crystalline properties of the fibers or the calcining the nanofibers after the addition of singlecrystal (SC) ceramic materials [2].
Many physical techniques and chemical methods to treat water and remove its pollutants, such as (sedimentation, filtration, and centrifugation), while chemical methods include chemical precipitation and electrochemical treatment [3][4][5].
Ultrafiltration (UF) technology is one of the most important physical technologies for drinking water purification. Although there are many ways to improve filtration efficiency, UF technology has attracted great interest among researchers due to its economic importance. Because it uses a new, highly effective polymeric membrane and has a large surface area compared to the surface area of conventional membranes [6]. This nonwoven material has a porous structure. Porosity has a distinctive role and is considered one of the main

Preparation of the polymeric solutions
A DMF/acetone mixed solvent was prepared with a weight ratio 50/50 to dissolve 12 wt% PVDF powder. In addition, N, N-dimethylformamide (DMF) was used to dissolve 12 wt% PAN powder. To eliminate bubbles inside the solution and avoid defects during the electrospinning process, continuous magnetic stirring was used for all solutions with sonication at 30°C-45°C for one hour. Then, four types of PVDF: PAN blends were prepared with weight ratios of 20:80, 40:60, 60:40, and 80:20. The polymeric solutions were mixed according to the above ratios at 30°C-35°C for one hour before being subjected to the electrospinning process.

Fabrication of nanofibrous membranes
The compositions of the pure membranes and the different PVDF: PAN blend membranes are shown in table 1. The polymeric solutions were placed in a syringe with a capacity of 1 ml ending with a needle with an inner diameter of 0.63 mm and in an MS-2200-Daiwha syringe infusion pump. The syringe pump was operated with a feeding rate of 1 ml hour −1 , an electric voltage of 20 kV, and a distance between the needle and the rotating collector of 15 cm, the humidity of electrospinning less than 40%.

Membrane characterization
Each nanofiber membrane's topography and morphology were studied and investigated using field emission scanning electron microscopy (FE-SEM) (MIRA3 TESCAN, FRANCE). The surface topography of the pure and blended membrane nanofibers was measured by atomic force microscopy (Naio AFM 2022, Nanosurf AG, Switzerland), and the roughness surface properties were measured by Mountains SPIP ® Academic 9.3.10249 software. The crystal properties of the nanofiber membranes were evaluated by x-ray diffraction (XRD) (D8 Bruker, Germany), and data were collected at 0.02-degree intervals with a count of 0.5 s per step in the 2°range of 5°-60°. A Fourier transform infrared (FT-IR) spectrophotometer (Spectrum Two, PerkinElmer, USA) was used to determine the interactions between the polymer blends. Differential scanning calorimetrythermogravimetric analysis (DSC-TGA, SDT Q600 V20.9 Build 20) showed that the average error of thermal effects was ±3%, and the melting enthalpy was calculated using the Netzsch Proteus program. At 25°C and 5 h before the test, samples of the polymeric solutions were placed in the test chamber to reach a constant temperature. Then the viscosity of each sample was measured using a Brookfield digital viscometer DV-III Ultra at different velocities in the range of 10-100 rpm. The surface tension of all pure polymer and blend solutions was measured at room temperature and using a surface interfacial tensiometer (TEN202). Contact angles were measured by adding a drop of pure water to the membrane surface using a G10 KRUSS (Germany) and determining the wettability of the membrane surface.
2.5. Measuring the filtration efficiency of the membranes 2.5.1. Porosity and average pore size The porosity of the membrane was determined by using the dry-wet method by immersing the membrane samples in pure water for 24 h. Then, these samples were wiped with tissue paper, and the mass (g) of the membrane was determined at equilibrium with pure water (W wet). The samples were placed in a vacuum oven at a temperature of 70°C for 4 h until a constant mass (g) was reached, recorded as the dry mass (W dry). The porosity (%) of the membrane was calculated according to the following relation [25] w w here 'ε is the porosity (%)', V is the membrane volume (cm 3 ), and ρw is the density of water (1 g cm −3 ). Moreover, the average pore radius of the membrane was calculated using the Guerout-Elford-Ferry equation [25]: Where r is the radius of the membrane (m), η is the viscosity of water at 25°C, I is the membrane thickness (m), Q is the permeate water volume per time (m 3 /s), A is the area of the membrane (m 2 ), and is the operating pressure (atm).
2.5.2. Pure water flux, permeability, and protein rejection Permeability and protein rejection are the most important parameters to determine membrane performance. Reverse osmosis (RO) was used for the PVDF: PAN membranes where partial permeability occurred during separation. Constant operating parameters of room temperature, a specified time (20 min), an operating membrane pressure of 30 psi, and a membrane area of 0.00025 m 2 were used. The pure water flux before using dry milk protein (Jw1) was calculated by using equation (3) [26]: Here, (Jw1) is the pure water flux before using dry milk protein (L/m 2 h), V is the volume permeate (L), A is the membrane area (m 2 ), and t is the time (h). Non-fat dry milk is pumped into the filter system at a concentration of (500 ppm) and under the same conditions in the first case, the flux of dry milk protein (Jp) is calculated by using equation (4) [26]: Here, (Jp) is the protein flux (L/m 2 h), and V is the volume of milk solution permeate (L). The permeate sample was collected every 30 min. The concentration of dry milk in feed and permeate was analysed by UV-Vis spectrophotometer (Shimadzu UV-2700 UV-vis spectrophotomter) at a wavelength of 280 nm. Protein rejection was calculated according to equation (5) [25]: In this equation, SR% is the percent solute rejection, and Cf and Cp are the concentrations of dry milk in the feed and permeate, respectively. The membrane nanofiber separation efficiency can be determined through the permeability of pure water and the use of dry milk protein, according to equation (6) [25]: Where Pw is the permeability of pure water (Lm −2 h −1 atm −1 ), V is the volume of permeate (L), Δt is the time of permeation (h), and ΔP is the operating pressure drop (atm). The rejected dry milk protein remaining in the membrane resulted from the protein solution passing through the membrane, after which the membrane was cleaned using 2% citric acid under 30 psi of pressure for 30 min. Pure water was passed through the membrane again under the same conditions to calculate the pure water flux after dry milk protein (Jw2) was used. The flow recovery ratio (FRR) and the total fouling ratio (Rt) can be determined, with which the fouling characteristics of the membrane can be evaluated [26].
Since FRR is the flux recovery ratio, Jw1 is the initial pure water flux, and Jw2 is the pure water flux after fouling [26].
Rt is the total fouling ratio, and Jp is the fouled flux. The pure water flux, permeability, and protein rejection tests were repeated three times for each membrane sample.

Viscosity and contact angle
Viscosity is one of the main parameters influencing the properties of nanofiber membranes, such as the fiber diameter, porosity, and pore size of the membrane fibers, especially when blending polymers such as PVDF and PAN [27,28]. Increasing the content of PVDF in the PVDF: PAN solutions led to an increase in the blend solution viscosity. Since the viscosity of pure PVDF is higher than that of pure PAN, a decrease in the homogeneity of the polymer solutions occurs with increasing PVDF content in the mixture. These results agree with those of a previous study [26]. The highest value of viscosity was observed with the M4 and M5 membranes. Figure 1 shows the viscosity and water contact angles of pure PVDF, pure PAN, and different PVDF: PAN blends as a function of the PVDF content in the membrane. To endow the films with better properties and performance, it is necessary to determine whether the films are hydrophobic or hydrophilic. An increase in surface wettability indicates a decrease in the contact angle on the film surface. Many research studies have focused on the contact angles of PVDF films and pure PAN. An increase in the contact angle depends on increases in the concentrations of the polymeric solutions [29]. In addition, the surface of the PAN membrane was more hydrophilic than the surface of the PVDF membrane, and the contact angles of the blended membranes were less than those of the pure membranes. Thus, the blended membranes had a more hydrophilic surface than the surfaces of the pure polymer membranes. The membranes M3 and M4 surfaces displayed the best hydrophilicity, as shown in figure 1.

Morphological properties
To evaluate the properties of the electrospun membranes, the microscale test was performed on the produced membrane samples on single fibers because of the difficulty and high cost of this type of assessment [28]. Field emission scanning electron microscopy (FE-SEM) was used to observe the upper surface at a certain magnification to determine the morphological properties of the membranes, such as diameter (nm), pore volume (μm3), and porosity (%), which were measured by ImageJ software. The porosity of the membranes can be calculated by comparing the average pore volume with the volume from the FE-SEM image using the following equation [30]: v v 100 9 p total ( ) * According to previous studies, there is a direct relationship between the diameter of the fibers and the porosity and diameter of the pores: the porosity increases with an increase in the fiber diameter, or increasing the diameter of the fiber increases the size of the pores [31]. Figure 2 shows the average nanofiber diameters and pore volumes of the pure and blended membranes as a function of PVDF content in PVDF: PAN nanofibers M1, M2, M3, M4, M5, and M6. The viscosity results clearly show that the viscosity of polymeric solutions rises with an increase in PVDF content, and since there is a direct relationship between viscosity and fiber diameter [32], significant increases in the diameter, pore size, and porosity from 275.403 ± 60.875 to 996.038 ± 410.944 nm, 0.233 ± 0.00416 to 2.272 ± 1.068 μm 3 , and 6.71 ± 0.836 to 14.387 ± 4.7% by increment (261.664%, 875.107%, and 114.41%), respectively, were observed when the content of PVDF increased from 20% (M2) to 80% (M5) as a result of the increased solidification rate with increasing PVDF content.
On the other hand, from the FE-SEM images at the 2 μm and 20 μm scale and energy dispersive x-ray spectroscopy (EDX) analysis, it is possible to display the atomic results of nitrogen, carbon, oxygen and fluorine, where the nitrogen is from PAN, and the fluorine is from the PVDF polymer. These results also confirmed that the atomic percentage of fluorine increased with the addition of PVDF to the blend, as shown in figure 3. Membranes M3 and M4 also contain more balanced ratios between the weight and atomic percentages of   nitrogen and fluorine in PAN and PVDF, respectively. These results indicate that the above membrane surfaces have the best wettability. Figure 4 shows the FTIR spectra of the samples in the range from 500 to 4000 cm −1 to determine the structural characterization of PAN, PVDF, and the PVDF/PAN blend. The FTIR spectrum of the PAN nanofibers showed peaks at approximately 2925 cm −1 , 2244 cm −1 , 1735 and 1,666 cm −1 , which were assigned to the stretching vibrations of CH2, -C≡N, -C=O, and -C=N groups, respectively. The peaks at approximately 1460 and 1270 cm −1 represent the different vibrations of CH and C-O, and the peak at approximately 3618 cm −1 represents the C-H stretching band [33][34][35]. Additionally, the FTIR spectrum of pure PVDF nanofibers presented peaks at approximately 487 and 531 cm −1 , representing the α-phase, and the peaks at approximately 838, 1184, and 1274 cm −1 correspond to the β-phase and the α-phase and β-phase indicate that the PVDF nanofibers are semi-crystalline [36]. The characteristic peak is approximately 1,173 cm −1 is the symmetric stretching of the -CF2 group, and the peak at 876 cm -1 is specific to the C-F stretching vibration of the amorphous phase [37]. The spectral peaks at approximately 2240, 1735, and 1250 cm −1 are due to -C≡N, -C=O, and C-O in the PAN polymer, while the peaks at approximately 840 and 880 cm −1 denote the C-C-C and CF stretching vibrations of PVDF. In the blend, we noted that the nitrite group peak shifted from 2244 to 2240 cm -1 when the atomic % of fluorine increased in the PVDF: PAN blend. We also found new peaks at approximately 1182 and 1665 cm −1 , assigned to the starching vibrations of C-N and C=N groups [38,39].

X-ray diffraction (XRD) analysis
The crystal size of pure nanofibers and their blends depends on the half-width (b) of the peaks (2θ = 20°to 23°) and is calculated using the Debye-Scherrer equation [40].
Where L(hlk) is the average size of crystallites, K represents the Scherrer constant (0.98), λ denotes the wavelength (Å), β denotes the full width at half maximum (FWHM), and max q is the angle for the maximum peak (rad). While the inter-chain separation length (Å) was determined based on the analysis of the most intense crystalline peak according to the following equation [40]: Where R is the inter-chain separation length, the distance between planes (the d-spacing) was determined from the Bragg equation [40]: Where n is the deflection integer.  Table 2 shows the crystalline properties results, such as average crystalline size L(h k l), the distance between planes (the d-spacing), the inter-chain separation length( R), and Lattice strain (e) for the manufactured membranes. All sample membranes showed broad peaks at 2θ of 20.4533, 23.7750, 22.816, 20.5024, 20.269, and 20.265 for M1, M2, M3, M4, M5, and M6, respectively. The decrease in the full width at half maximum (FWHM) when the PVDF content in the PVDF: PAN nanofibers increased from 1.2 (M1) to 0.4344 (M4). The decrease in FWHM caused an increase in the average crystalline size and decreased the inter-chain separation length(R) and Lattice strain (e) of the membrane until M5 compared with the crystalline properties of pure PAN. The best crystalline properties for the blended membranes were observed with M4 and M5. Increasing the content of PVDF in PVDF: PAN caused an increase in crystalline properties because PVDF is semi-crystalline and contains both the α-phase and β-phase, which enhanced the polarity of the electrons around the fluorine atom [36].On the other hand, the distance between planes of all membranes had similar and close results, indicating no significant change in the crystalline phase of PVDF and its blends in all nanofiber membranes. M4 is the best nanofiber membrane with a large crystalline size, low inter-chain separation length, and low lattice strain.

Thermal properties
To characterize the membranes that were manufactured and to determine which has the best performance in terms of thermal properties, differential scanning calorimetry-thermogravimetric analysis (DSC-TGA) was used over the heating range of 30.8 to 980.90°C, as shown in figure 6. The exothermic DSC thermograms showed that the thermal behavior of the nanofibers changed in the membrane samples. It was observed that increasing the content of PVDF in the PVDF/PAN nanofibers leads to an improvement in resistance of the thermal decomposition, as the temperature of decomposition increased from 323.28 to 486°C for pure PAN [41] to pure PVDF, while the required heat capacity for fusion increased from 145.6 to 706.6 J/g for M1 to M5 since the thermal stability of PVDF is higher than that of PAN. In other words, the TGA thermographs showed that the   [38,42]. The results show that the best weight loss (%) during the thermogravimetric analysis was M4.
3.6. Effect of dry milk protein on the filtration efficiency of the membranes 3.6.1. Effect of dry milk protein on surface topography Surface roughness and mean pore depth are important parameters that AFM studied. Figure 7 shows the effect of dry milk protein on the surface topography via 2D and 3D AFM images of the pure and blended membrane nanofibers before and after dry milk protein was used. The AFM images prove a clear change in the topography of the membrane surface after using dry milk protein, as the gaps between the nanofibers were closed or reduced in size in most of the membranes with dissolved protein materials, especially membranes M3 and M4. Table 3 shows the surface roughness, mean pore depth and percent reduction in mean pore depth of the pure and blended membranes before and after using dry milk protein. The root mean square (RMS) and the mean roughness (Sa) on the surfaces of the blended membranes were higher than those on the pure membranes because surface roughness depends on the increase in the diameter of the nanofiber [43]. Additionally, the RMS values for PAN and PVDF were 278.0 and 384.1 nm, respectively, while the RMS of M5 was 592.9 nm, and membranes with less surface roughness have better antifouling properties [44]. In addition, the highest percent reduction in mean depth (RMD) is desirable, as it indicates the ability of the membrane surface to reject protein materials and provide high filtration efficiency. The RMD% of M3 and M4 were 75.17 and 82.51%, respectively, because these membranes' surfaces have the best wettability according to a good balance between the fluorine in PVDF and nitrogen in PAN elements, which discuss in previous sections.

Dry milk protein rejection and membrane permeability
To calculate the amount of rejected dry milk protein (DM) and to determine membrane contamination, six standard solutions of DM in RO water were generated at room temperature at concentrations of 100, 200, 300, 400, 500, and 600 ppm to construct a calibration curve, as shown in figure 8(a). Linear fitting of the curve gave an R 2 value of 0.9681, as shown in figure 8(b). To calculate the protein rejection SR (%) according to equation (5) and flow recovery ratio (FRR) according to equation (6), the penetrate dry milk concentration (CP) can be calculated using the following equation: Figure 7. The effect of protein (DM) on the surface topography of the pure and blended membrane nanofibers (A). AFM images (2D and 3D) before using dry milk protein, (B) AFM images (2D and 3D) after using DM protein. Table 3. Surface roughness, mean pore depth, and percent reduction in mean pore depth before and after using dry milk protein for the pure and blended membranes.  Figure 9 shows the pure water flux before using dry milk protein (Jw1) and the pure water flux after using dry milk protein (Jw2) for the pure and blended membranes. According to the surface wettability results, the pure membranes made of PAN and PVDF have high contact angles with water because both polymers had a relatively high concentration (12 wt. %). Since the contact angle increases with increasing concentration, the polymer blends have low contact angles compared to each polymer, and the surfaces of the membranes change from hydrophobic to hydrophilic between M3 and M4. In addition to the EDX analysis, the atomic number of fluorine and the nitrile group were close between M4 and M3. Thus, M4 and M3 have the highest pure water flux before using dry milk protein as well as after, and these results prove the separation efficiency of these membranes. Therefore, blending PVDF with PAN led to an improvement in the flux of the membrane and porosity % [45].
In addition, table 4 shows the results of the flux recovery ratio, solute rejection, total fouling ratio, and porosity assessments in addition to the pore size before and after using dry milk protein. M3 and M4 have the highest recovery ratios of 72 and 84.32% due to their high porosities of 63.31 and 67.79%, respectively. Additionally, there were small changes in their pore sizes after using dry milk protein, 28 and 15.68%, respectively. These results were accompanied by lower percentages of total fouling and dissolved material rejection than the sample with the highest percentage of total fouling (pure PVDF) because the balance between the atomic and weight ratios of PAN and PVDF in the M3 and M4 blends limited the adhesion of protein material on the surface of the membrane and reduced the efficiency of the membrane in the filtration process [46]. The PVDF membrane had a high Rt (%) compared with their blend as M4; this result agrees with the previous study [47]. While the surface of the membrane M4 has the best wettability preventing protein  substances from adhesion to it and increasing the recovery ratio. Therefore, the decreased flux led to increased rejection of dissolved material on the surface of the membrane [48]. After one hour of DM application, the flux at M4 was 2,828 (L/m 2 .h.atm), while the flux at M6 was 353 (L/m 2 .h.atm), as shown in figure 10. These results prove that incorporating PAN in the PVDF: PAN membranes not only increases the hydrophilicity and enhances the flux but also helps to enhance the return of the membranes to a higher flux [29]. Finally, a membrane with good performance should have a good rejection ratio and suitable permeability [48].

Conclusions
The main problems that limit the efficiency of nanofiber membranes are the fouling that occurs after treatment with dissolved protein materials, as well as the concentration and quality of the fouling. In this study, dry milk was utilized, as it is the main source of protein materials. In addition, the incorporation of PAN nanofibers in the PVDF: PAN membranes improved the physical properties and filtration efficiency of PVDF nanofibers to improve the wettability of the membrane surfaces and convert them from hydrophobic to hydrophilic despite the increase in the viscosity of the blend. This led to increases in the diameters of the nanofibers, pore volume, porosity, and surface roughness compared with pure polymer nanofibers, resulting in an almost equilibrium state between the atomic number of fluorine and the atomic number of the nitrile group in PVDF and PAN, respectively. Additionally, blending PVDF with PAN enhanced PAN's crystalline and thermal properties, in contrast to the clear increase in the flux recovery ratio and the pure water flux before and after using dry milk protein. This was especially true for M3 and M4, with lower percentages of total fouling and rejection of dissolved materials than pure PVDF, which had the highest percentage of total fouling.