Nanoporous Polyether Sulfone Membrane, Preparation and Characterization: Effect of Porosity and Mean Pore Size on Performance

: Flat sheet membranes were prepared by phase inversion technique using polyether sulfone (PES) dissolved in dimethylacetamide (DMAc) with and without adding polyvinylpyrrolidone (PVP) or polyethyleneglycol (PEG). The characteristics of the prepared membranes were evaluated using Scanning Electron Microscope (SEM) images, Atomic Force Microscopy (AFM), and Optical Contact Angle (OCA) measurements, and porosity tests. The porosity test and SEM images show that increasing additives to a certain value increases the porosity of the membrane. Also, as the coagulation bath temperature is increased, the porosity of the membrane is increased. The roughness of the membrane is increased by increasing the additive concentration. The analysis of AFM images confirms the nanoporous structure of the prepared membranes, and that the membranes with appropriate pore size distribution can be prepared by the applied method. Permeability tests using single-layer membranes show that the direct relationship between porosity and the flux of pure water or salt solution is dominated by the effect of applied additive while the salt rejection shows an inverse relationship with the mean pore size regardless of the applied additive. The salt permeation flux is a function of total porosity while the salt rejection is a function of surface porosity. Pervaporation tests show that both permeation flux and enrichment factor depend on the total porosity of the support membrane.


INTRODUCTION
Growing attentions to membrane processes in recent years for different applications such as removal of volatile organic compounds from wastewater [1][2][3], dehydration of solvents [4,5], and organic-organic separations [6,7] demand exact understanding about the performance of the membranes. The membrane process operates based on different driving forces i.e. the processes such as nanofiltration, ultrafiltration, microfiltration, and reverse osmosis utilize total pressure driving force while the processes such as gas separation and pervaporation operate based on the partial pressure driving force [8,9]. Therefore, the membrane properties such as its morphology have distinctive effects on the process operation. These properties are affected in a single-layer membrane by preparation procedure of the membrane [10], choosing the number and types of non-solvents [11], solvents and additives [12], crosslinking agents [13], and polymer inherency [14]. The effects of membrane properties on operational performance have been investigated qualitatively for the filtration process [15,16]. The ultrafiltration Poly(vinyl chloride) membranes *Address correspondence to this author at the Chemistry and Chemical Engineering Research Center of Iran, Tehran, P.O. Box: 14335-186, Iran; Tel: +9821 44 787 751; Fax: +9821 44 787 781; E-mail: mortaheb@ccerci.ac.ir prepared by combination of thermally induced phase separation and non-solvent-induced phase separation techniques indicated that the changes in the contents of diluents changes the pore size and porosity of the membranes and consequently their rejections and fluxes [17]. However, there are few researches, in which the effects of pore size distribution on the performance of composite membranes are studied.
A membrane for pervaporation process can be made of a thin dense layer supported by a porous sublayer. Theoretically, the active top layer in a composite membrane should provide appropriate selectivity and sufficient permeation flux while the support layer must exhibit outstanding mechanical strength without interfering in the mass transfer. However, several studies have demonstrated that the support layer significantly affects the pervaporation performance [18]. Tan et al. made different polysulfone membranes as support layers by varying polymer concentration in the casting solution and the evaporation time [19]. The results indicated that the support layer with significant resistance can dominate the pervaporative performance of the composite membrane and that the inhomogeneity of the support layer may produce extra mass transfer resistance when the resistance of support layer cannot be neglected [19]. Bode and Hoempler [20] by applying a mass transfer model concluded that about 80% of mass transfer resistance is comprised in the support layer. In a later research, it was found that the permeation rate in the pervaporation can be enhanced significantly when the intrusion of cast top layer in the pores of support layer is prohibited [21,22]. Trifunovic et al. studied the effect of polyetherimide and polyacrylonitrile membranes as the supports of composite membranes on the mass transfer rate in hydrophobic pervaporation for removal of alcohols and esters [23]. It was observed that the dependency of pervaporative separation is not only related to the geometric characteristics of the support but also to its nature [23]. It has been concluded that permeate-side pressure losses in the support pores can be important [24]. The performance of a pervaporation membrane having a bigger pore size and porosity was found to be superior to that of a membrane with smaller ones. In addition, it was found that the controlling step of mass transfer is changed from concentration polarization to membrane resistance by increasing the feed flowrate [25]. The water selective pervaporative performance of an ultrathin polysiloxaneimide (PSI) top layer on polyetherimide (PEI) support in a composite hollow fiber was attributed to the influence of hydrophilic PEI layer in a research by Zhu et al. [26].
None of the above researches have studied the effect of support layer porosity and pore size distribution on the performance of the composite membranes.
Polyethersulfone (PES) is considered as a suitable material for support membrane due to its favorable mechanical strength, resistance to acids and alkalis, and high thermal and chemical resistances. In the present research, membranes are prepared by blends of PES as the membrane matrix, and polyvinyl pyrrolidone (PVP) or polyethylene glycol (PEG) as the hydrophilic additives. These membranes are characterized by structural analyses, and their porosity and pore size distributions are determined. The membranes are then applied in permeability and pervaporation tests, and the effects of those parameters on the operation performance are investigated [27,28].

Membrane Preparation
PES flakes were dried in an oven to remove their adsorbed water in advance. Homogeneous casting solutions were made by dissolving PES, and PVP or PEG in DMAc. The compositions of casting solutions were adapted by changing the percentage of PES and the additive. Three concentrations of PES (15,18, and 22 wt%), three concentrations of PVP (3, 5, and 10 wt.%), and two concentrations of PEG (5 and 10 wt%) were tested for preparation of the membranes. Asymmetric flat membranes were prepared using phase inversion technique. The solutions of mixed polymers were well homogenized by a magnet stirrer and were cast smoothly on a level glass plate by a casting knife at room temperature. The film thickness was set to 200 µm for all the membranes. The layered polymers on the glass plates were immediately immersed in the coagulation bath for all the trials. The membranes were then stored in distilled water for 48h to allow the water-soluble components in the membranes to be leached out. As the final stage, the membranes were washed again with distilled water and were dried by filter papers to be stored for at least 48 h at room temperature.
For preparation of the active layer, PDMS was dissolved in n-heptane to form a 50 wt% solution. The crosslinking agent, TEOS, and the catalyst, DBTL, with respectively 0.1/0.02 weight ratios of the polymer were then dissolved in the solution. The solution was stirred vigorously for 2 h at room temperature. Prior to the coating step, the PES support was pre-wetted by glycerol in a pan to reduce mass transfer resistance due to penetration of PDMS solution into the porous substrate. Excess glycerol on the PES support surface was then wiped off with a filter paper. The support layer was put into a glass mold, and the PDMS solution was spread out uniformly on its surface. The active layers were allowed to be partially crosslinked at room temperature in 24 h. The membrane was put into an oven set to 70 °C afterward for 4 h to complete the crosslinking.

Characterization of PES Single Layer Membranes
SEM was used to observe the structure of the support layer (VEGA 3 SBH\\Tescan Brno, Check). PES membranes were snapped in liquid nitrogen to give a clean break. The samples were then placed on a sample stand and coated with gold by sputtering.
Atomic Force Microscopy (Dual scope C-26 DME, Denmark) was applied to measure the pore sizes in the membrane. The distribution of pores in the membrane was then determined by using Sigmaplot mathematical software.
The contact angle and hence wettability behavior of the prepared membranes were measured by an Optical Contact Angle measuring instrument (OCA-20, Dataphysics, Germany).
A viscometer (AMVn1569 Anton Paar Austria) was used to measure the solution viscosity.
The prepared PES membranes were immersed in solution of 30 wt% glycerol for 5 days at room temperature. The weight of membranes was then recorded and porosity was calculated using the following equation [29]: Here P is the membrane porosity, Q 0 and Q 1 are the weights of wet and dried membranes (g), respectively, A is the membrane surface area (cm 2 ), h is the membrane thickness (cm), and ρ is the density of liquid (g/cm 3 ). In order to minimize the experimental error, the membrane porosity of each sample was measured three times.

Water Permeability Test
Water permeability tests through the prepared support membranes were performed by a cross-flow setup under steady state conditions. The setup consists of a reservoir, a pump, valves, pressure regulators, and a membrane cell as illustrated in Figure  1. The prepared membranes were cut into the desired size for fixing on the setup. The membrane cell was initially pressurized with distilled water to a transmembrane pressure of 200 kPa. Water permeability is calculated as follows [30]: where L p is the pure water permeability, V is the volume of water permeated through the membrane (L), A is the membrane area (m 2 ), Δt is the operation time (h), and ΔP is the transmembrane pressure (bar).

Salt Rejection Test
The filtration performances of the membranes were evaluated using the setup. The solution of NaCl in deionized water with a concentration of 2 g/lit was used for all the experiments. All experiments were performed under pressure of 200 kPa.
The permeation flux is calculated using Eq. (2), and the salt rejection is calculated as follows: where C f and C p are the salt concentrations in the feed and permeate streams, respectively, measured by a conductivity meter.

Pervaporation
Pervaporation experiments were carried out using a laboratory setup. The details of the setup were previously reported [31]. The flat composite membrane was fixed on the cylindrical pervaporation cell. The effective area of the membrane was 10 cm 2 and the permeate pressure was 1 mbar. The permeate was collected in a cold trap chilled by liquid nitrogen.
Pervaporative performance of the composite membrane can be assessed in terms of permeate flux (J) and enrichment factor (β) using the following equations: where M (g) is the total mass of permeate during the experimental time interval of t (h) at steady state, A (m 2 ) is the effective membrane area, and C' A and C A (ppm) are the toluene concentrations in the permeate and retentate, respectively. The toluene concentration in the permeate is calculated by: in which, m A (mg) is the mass of permeated toluene calculated from retentate volume and its composition by a mass balance, and V P (L) is the total permeate volume.
Aqueous solutions of toluene with a concentration of 150 ppm were used as the feed in the experiments. Each experiment was repeated three times. membranes were measured. As shown in Figure 2a, a wide finger-like structure is observed in the membrane with 15 wt% PES. The structure is changed to thin finger-like pores with some sponge areas between them by increasing in polymer concentration to 18 wt%. More increasing in polymer concentration will increase the sponge areas while the number of finger pores is decreased.

Effect of Polymer Concentration on Membrane Morphology
The porosities of the membranes for different PES concentrations are shown in Figure 3. Again, it is observed that the porosity of membrane is decreased significantly by increasing polymer concentration due to decrease of free areas between the polymer chains. Meanwhile, all fabricated membranes had enough mechanical strength so that the membrane with the minimum PES concentration (15 wt%) can be used for further experiments. It is noticeable that the current research is focused mainly on the membranes with PES concentration within the range of 15-22 wt% since the mechanical strength of the membrane is decreased significantly when the polymer concentration falls below 12 wt% [32], in which the formation of macrovoids renders the strength of the membrane. On the other hand, in concentrations higher than 22 wt% PES, the prepared membranes have irregular structures [33][34][35].

Effect of PEG and PVP Concentrations on Morphology of Membrane
Hydrophilic additives as materials with non-solvent properties would increase thermodynamic instability of the immersed film. These properties are caused by the strong and weak miscibilities of the additive with the solvent and the polymer, respectively. When a hydrophilic additive is added to the polymer solution, the exchange rate of solvent-water in the coagulation bath increases at the phase inversion step. Subsequently, more voids are created in the structure of the membrane, which exhibit a large channel-like or finger-like form [36][37][38]. Therefore, a thick macrovoid membrane is formed as a result of instantaneous demixing due to thermodynamic instability. On the other hand, increasing additives will increase viscosity of the solution. Since increasing viscosity is an obstacle for instantaneous de-mixing of the solvent and nonsolvent, additives in the polymer may have a bilateral effect on the morphology and porosity of the membrane i.e. membranes with either high or low porosities can be formed by adding additives to the polymer [39,40].
In the present research, PEG (400) and PVP (360,000) were added to a solution with 15wt% PES for investigating the influence of the additive concentration on porosity. The measured viscosities of primary solutions are shown in Table 1. The morphologies of the membranes prepared from PES/PEG (or PVP)/DMAc systems are shown in Figure  4. It is noticeable in this figure that the macrovoids are covered by a thin inherent dense top layer. The fingerlike structure with some sponge areas for the membrane without additive (Figure 2a) changes to a channel-like macrovoid structure by adding 5 wt% PEG (Figure 4a). When PEG concentration is increased to 10 wt%, the pore size is decreased (Figure 4b) and a thin channel-like structure with closed ends are formed possibly due to dominant kinetic effect and inhibition against instantaneous de-mixing. The porosity of the membrane is plotted against PEG concentration in Figure 5a.
As seen in Figures 2a and 4c, by increasing PVP concentration from 0 to 3 wt%, the finger-like structure of the membrane changes and regular channel-like macrovoids with closed ends are formed. As the PVP concentration is increased to 5 and 8 wt% (Figures 4d,  e), the structure of the pores changes from a long finger/channel-like to a sponge-like structure with tearlike ends. In addition, similar trend of change in the porosity as that in the membranes with PEG can be observed here that is increasing in porosity by adding the PVP concentration up to a certain level and then decreasing by further adding the additive (Figure 5b). The maximum porosity of the membrane is achieved with PVP concentration of about 3 wt%. It is also observed by comparing Figures 5a and 5b that the maximum porosity of the membrane obtained by adding PVP is higher than that obtained by adding PEG.

Effect of Coagulation Bath Temperature on Morphology of Membrane
Increasing the coagulation bath temperature augments the thermodynamic instability, which is believed to be due to delaying in de-mixing [39][40][41]. During the coagulation process, the molecules in the cast film are exchanged with the non-solvent molecules in the coagulation bath. At constant temperature of the cast film, increase in the temperature of the coagulation bath increases the exchange rate [37]. The following SEM images verify this statement. Increasing in the bath temperature changes the membranes structure from the thin closed-end channels (Figure 6) to a structure of wide channels with the open ends ( Figure  4c). Also, the formation of macrovoid channels is approached more to the top surface of the membrane as the bath temperature is increased.
The effects of coagulation bath temperature on porosity are also shown in Figures 5a and 5b, in which higher porosities are obtained by increasing the bath temperature.

Effect of Additive Concentration on Surface Roughness of Prepared Membranes
Three-dimensional AFM images of the prepared membranes by PES, PES+PEG, and PES+PVP with different additive concentrations are shown in Figure 7.
Comparing the images reveals that the roughness of the membrane is generally decreased by using an additive. Figures 7b and c show that as the PEG concentration is increased from 5 to 10%, the number of holes, shown in dark color, is increased possibly because of leaching out of PEG from the membrane matrix during the coagulation step. Even a more decrease in the surface roughness is observed by applying PVP as the additive. However, the number of holes is decreased as the PVP weight percent is increased from 3% (Figure 7d) to 5 and 8% (Figures 7e  and 7f, respectively). This may show that less molecules of PVP leave the polymer matrix during the coagulation step because of their large molecular size.
Arithmetic mean deviation (S a ) as an index of roughness, for the prepared membranes is shown in Figure 8. The figure shows that as the additives concentration is increased, the surface roughness of the membrane is decreased. Figure 8 also shows the contact angles of the membranes. As seen, the contact angle is decreased by decreasing the surface roughness. This is in agreement with the results of previous researches, in which it was found that hydrophobic (or hydrophilic) properties are stepped down with decreasing the surface roughness [42]. Figure 9 shows a typical SEM image of fabricated composite membranes. As seen in the figure, the dense PDMS top layer with about 20 µm thickness is separated distinctly from the porous support. This means that inherent thin layer on the porous support can preserve well the sublayer from intrusion of the PDMS top layer during prewetting step.

Effect of Additive Concentration on Pore Size Distribution of Prepared Membranes
Pore sizes distribution was obtained by random selecting of linear profiles for 50 segments in different parts of the membrane surface. A pore size was then determined by peaking up a pore from the expanded curve of each profile. The pore sizes were plotted on the normal logarithmic distribution graph (Figure 10) by SigmaPlot mathematical software where the vertical axis indicates the frequency percentage of the pore size. The mean pore sizes of the membranes equivalent to the pore sizes in 50% frequency, and the standard deviations based on the following equation are listed in Table 2.

S.D =
Pore size in 50% Frequency Pore size in 84.13% Frequency (7) The results show that increasing in additive concentration decreases the mean pore size. The probability density function curves of the prepared membrane are shown in Figure 11. The figure shows that as the additive concentration in the membrane increases, the curve demonstrates a narrower distribution with a peak at a smaller pore size. The narrowest curve is corresponded to the PES membrane with 10% PVP while the widest distribution is corresponded to the PES membrane without additive. According to the results, all the membranes can be categorized in the range of nanoporous membranes.

Effect of Membrane Porosity on Pure Water Permeability
In order to verify the effect of membrane porosity on permeation flux, a series of filtration tests were performed using the explained setup. The results in Table 3 show that for the membranes containing PEG, permeation flux follows the changes in porosity of the membranes i.e. increases with increasing porosity and vice versa. However, for the membranes containing PVP, a different trend is observed that is while the porosity reaches to a maximum value by increasing PVP and then decreases, a steadily increase in permeation flux is observed with increasing PVP concentration. The measured values of surface contact angle (Figure 8) show that by increasing PVP concentration in the polymer solution, the membrane becomes more hydrophile possibly because of more residual content of PVP on the membrane's surface as well as inside the membrane. However, the contact angles of the membranes containing PEG in their polymeric solution do not change significantly since majority of PEG is leached out during immersing membrane in the water bath. Therefore, it may be concluded that the permeation flux in the membranes with higher concentration of PVP may be increased due to an enhancing adsorption effect of the PVP lumps, which is trapped in the membrane structure. These adsorption sites may produce additional paths for permeate through the membrane and then increase the permeation flux.   Figure 11: Probability density functions of membranes.

Effect of Additive Concentration on Salt Solution Flux and Salt Rejection
The effects of porosity and mean pore size of membrane containing PEG on salt solution flux and salt rejection are shown in Figures 12a and b, respectively. As seen in these figures, the salt solution flux depends to the porosity of the membrane while the salt rejection is inversely proportional to the mean pore size of the membrane. However, the dependence of the salt solution flux to total porosity is not observed in the membranes containing PVP (Figure 13a). The reason is that the membranes containing PVP show different behavior as explained before for water permeation flux due to presence of PVP in the membrane matrix. On the other hand, the salt rejection in the membrane containing PVP shows again an inverse dependency to the mean pore size (Figure 13b). As the mean pore size is an index of surface porosity, the above findings may verify that the salt rejection is a function of surface porosity while depending on the applied additive, the salt permeation flux may be a function of total porosity.

Effect of Additive Concentration in Support Layer on Pervaporation Performance
Since a single-layer PES membrane has an intrinsic top layer, it might represent suitable pervaporative performance. Therefore, the performance of singlelayer membranes with 22 wt% PES and different weight percents of additive were examined. The experiments showed that these membranes are not applicable for pervaporation process because of their high permeation flux and low selectivity. Therefore, the PES membrane was coated by a selective dense PDMS top layer for pervaporative process. The details of fabrication and specification of the composite membranes were explained elsewhere [31]. Figures 14  and 15 show the effects of mean pore size (surface porosity) and total porosity of the support membrane containing PEG or PVP on pervaporative performance of the composite membranes. As seen in Figure 14, the total permeation flux of the membrane is increased as the porosity of the support layer is increased. The figure also shows that the enrichment factor is decreased by increasing the porosity. Meanwhile, Figure 15 shows that the total flux is increased by increasing mean pore size of the support layer up to a certain value and then decreased by further increasing in the pore size while an inverse trend is observed for variation of enrichment factor with the pore size. It may be concluded that that the permeation flux and enrichment factor are influenced by the total porosity of the support layer and is hardly can be related to its mean pore size. This trend can be interpreted as the effect of porosity may be introduced directly in the total membrane resistance against mass transfer through the composite membrane while the effect of pore size of the porous support might be annulled possibly due to presence of intrinsic thin dense layer on the top of the support.

CONCLUSION
In order to enhance the operational performance of nanoporous PES membrane, it was blended with different weight percents of PEG or PVP. The following results were then obtained:

1)
The SEM images and porosity data show that increasing additives to a certain value is beneficial for increasing the porosity of the membrane. Also, as the coagulation bath temperature is increased, the porosity of the membrane is increased.
2) The increasing of additive concentration decreases roughness of the membrane. This effect is more pronounced in the case of PVP as the additive than that of PEG.

3)
Based on the observed contact angles, the membranes made by blending PES and PVP demonstrate the highest hydrophilicity, which may be the reason of steady increase in the pure water permeation of those membranes.

4)
The AFM results show that the nano-size membranes could be prepared and that the mean pore size of the membrane decreases with increasing additive concentration where the lower mean pore size results in more uniform distribution of the surface porosity.

5)
The salt rejection is a function of surface porosity while the salt permeation flux might be a function of total porosity depending on the applied additive.

6)
The permeation flux and enrichment factor are influenced by the total porosity of the support layer while there is an insignificant effect of mean pore size on these parameters.