Polyethersulfone (VERADEL 3000P with MW = 63,000 g mol^-1) from Solvay Advanced Polymer was dried for 4 h in an oven at 100 °C before utilizing it for fabrication of the dope solution. N-methyl-2-pyrrolidone (NMP) with purity > 99% from Labsynth Produtos para Laboratórios Ltda. was used as received as the solvent for the polymer. Tap water was used as the nonsolvent agent, as bore fluid and in coagulation bath.

The dope solution was prepared with 16 and 20% (weight/weight) concentration of PES, with the remaining concentration (84 and 80%, respectively) being of NMP. The polymer was slowly added to the solvent, taking 1 h for complete addition, at room temperature. Then, the solution remained for about 18 h at 200 r.p.m. stirring, assuring complete solubilization of the polymer and homogenization, to be finally degassed in an ultrasonic bath for 1 h, eliminating any bubbles of air trapped into solution.

The solution was spinning with a solution flow rate of 2 mL min^-1 and a bore fluid flow rate of 4 mL min^-1, using dry-jet wet spinning method with an air gap of 2 cm and take-up speed of 4 mL min^-1, at room temperature. A schematic representation of the spinneret and fabricating hollow fiber membrane processes is shown in Fig. 1.

Figure 1.
Schematic representation of fabricating hollow fiber membrane system.

Hollow fiber membranes were stored in demineralized water to keep their integrity. To realize some characterization procedures, these were dried following the steps described elsewhere²⁵.

The produced membranes were investigated about its morphology, permeability, mechanical resistance and MWCO (molecular weight cutoff). The data obtained were treated to remove outliers and statistical analysis of the results was done using Minitab^® 18.1, from Minitab, Inc., assuming a significance level of 0.05 and equality of variances.

2.4.2 Pure water permeability

Pure water permeability (PWP) measurements were estimated in an experimental setup, where a module made from low-density polyethylene (LDPE) was utilized with two hollow fiber membranes with 30 cm effective lengths, folded in half and fed topside in an inside out mode.

A syringe pump with a constant flow rate of 0.5 mL min^-1 supplied demineralized water to the module and the water production measurement was started after thirty minutes to achieve steady-state conditions. Records were made every five minutes for one hour, similar to the procedure previously described²⁶. The permeation flux (J_w) through the membrane was calculated following the Eq. 1:

(1)

where: J_w = Water permeability (L m^-2 h^-1 bar^-1); V = Volume of permeate (L); A = Inner surface area (m^-2); t = Time (h); ΔP = Transmembrane pressure (bar).

2.4.4 Molecular weight cutoff (MWCO)

The MWCO was investigated using 200 ppm polyethylene glycol (PEG) solution with molecular weights of 10 kDa, 32 kDa, 90 kDa and 150 kDa, into the same system and module described in 2.4.2.

The concentration of PEG in permeate and in retentate was analyzed by a UV-VIS spectrophotometer (WUV-M51, Weblaborsp) at a wavelength of 254 nm and the MWCO was determined from the rejection of PEG solution, following Eq. 2:

(2)

where: R = Rejection (%); C_p = Permeate concentration (Da); C_r = Retentate concentration (Da).

To evaluate the impact of polymer concentration on the morphology of hollow fiber membranes, SEM was used to observe its enlarged cross section and outside face of the samples. Figures 2 and 3 show both membranes resulted in an outside dense skin and a double layered finger-like structure pore, but it is possible to see that the 16% PES concentration membrane has a bigger aperture at the end of the pores, nearby their center.

Double layer finger-like and dense skin were expected microstructure characteristics for these membranes. They occur because water is a strong nonsolvent for the polymer, that provides a fast coagulation, and this formation is a consequence of it, like reported previously^11,27,28.

The morphological analysis indicates that the permeation flux tends to be better in the 16% PES concentration membrane. This tendency is verified by pure water permeability investigation results, but it can happen because its pores had an opening larger than the other membrane, facilitating the flow of the fluid through it.

Figure 2.
SEM images of 16% PES sample: a) full cross section of the hollow fiber membrane, b) and c) cross section of the hollow fiber membrane with different magnifications, and d) outer skin.

Figure 3.
SEM images of 20% PES sample: a) full cross section of the hollow fiber membrane, b) and c) cross section of the hollow fiber membrane with different magnifications, and d) outer skin.

The results of demineralized water permeation flux for the hollow fiber membranes produced are shown in Fig. 4. It shows an average permeability about 220 L m-² h^-1 bar^-1 for the 16% PES concentration membrane and 174 L m-² h^-1 bar^-1 for the 20% PES concentration membrane. These results agree with the SEM analysis showed above and also with reports in the literature^29,30.

Figure 4.
Pure water permeability of 16% PES and 20% PES membranes.

Table 1 shows statistical analysis of the results of permeability presented. According to ANOVA evaluation, there is a statistical difference between them.

Table 1.
ANOVA statistical analysis of membranes PWP.

The results of the mechanical properties investigation for membranes are summarized in Tab. 2.

Table 2.
Mechanical resistance properties of membranes.

These results show there is a small variance between the average of tensile strength, elongation at break and Young's modulus. With the analysis of results, it is possible to note there is no relevant difference between the 16 and 20% PES membranes (Tabs. 3 to 5).

Table 3.
ANOVA statistical analysis of membranes tensile strength.

Table 4.
ANOVA statistical analysis of membranes elongation at break.

Table 5.
ANOVA statistical analysis of membranes Young’s Modulus.

The data obtained for MWCO were recorded up to PEG 150 kDa MW and the membranes were not able to remove 90% of this substance, which is the concept of MWCO. Thus, to estimate this important characteristic of membranes and knowing that the typical rejection curve is sigmoidal^31,32, an extrapolation was made with the results (Fig. 5 and 6).

Based on the point that represents 90% of removal in these curves, values of 271 kDa and 279 kDa are the MWCO estimated of 16% PES membrane and 20% PES membrane, respectively. The statistical analysis does not show a relevant difference between both results, as demonstrated in Tab. 6.

Figure 5.
16% PES membrane MWCO.

Figure 6.
20% PES membrane MWCO.

Table 6.
ANOVA statistical analysis of membranes MWCO.

karen.gonzaga@email.com