3.3.1. Pure Water Permeation
During water purification using membranes, pure water flux (PWF) of the membrane is determined to indicate the initial flux of the membrane for fouling monitoring [
22]. In this study, PWF of the prepared membranes was evaluated using the Sterlitech cross-flow filtration module at various pressure and the PWF calculated using Equation (1). The effective membrane area used in the evaluation was 42 cm
2, and the measurement was at room temperature after membrane compaction for 4 h at 8 bar. Deionized water was used for the pure water flux measurements at various transmembrane pressures.
Figure 7 depicts the PWF of the membranes.
As can be seen in
Figure 7, M4 has the lowest PWF when compared to those of the CNT/Psf membranes. Comparing the CNT/Psf membranes, the PWF of the membranes are in the order: M2 > M1 > M3. In addition, the PWF increased with increasing TMP (see
Figure 7). The results obtained in this study are consistent with literature [
23]. Furthermore, the pure water permeability (PWP) of the membranes were obtained from Equation (3) [
24].
The PWF at 6.9 bar was measured to be 287.7 L·m
−2·h
−1, 1098.9 L·m
−2·h
−1, 63.3 L·m
−2·h
−1, and 25.2 L·m
−2·h
−1 for M1, M2, M3, and M4, respectively. M2 displayed the highest PWF for the given TMP range, followed by M1, and M3 displayed the lowest PWF for the MMMs. Increasing the applied pressure enhanced the driving force for permeation, resulting in an increase in the permeation flux [
23]. Furthermore, the increase in the pure water flux for polymeric membranes has been attributed to the presence of hydrophilic sites in the inner walls of the filler particles as well as lower cross-linking density [
23]. The results from FTIR spectra confirmed the presence of hydroxyl groups. Therefore, it can be speculated that the composite membranes have a lower packing density compared to the pure polysulfone membranes. The incorporation of the CNTs in the polysulfone matrix influences the pure water flux to varying degrees for the MMMs synthesized by adopting the three CNT dispersion methods (as confirmed by the results for the PWF). Membrane permeability is one of the requirements for a good membrane; in this study, the pure water permeability/unit length (PWP) was defined as the ratio of the PWF and the TMP. The PWP results are shown in
Table 2.
The PWP at 6.9 bar was 41.7 L·m
−2·h
−1·bar
−1, 159.3 L·m
−2·h
−1·bar
−1, 3.7 L·m
−2·h
−1·bar
−1, and 9.2 L·m
−2·h
−1·bar
−1 for M1, M2, M3, and M4, respectively (see
Table 2). M2 displayed the highest PWP when compared to those of M1 and M3 for the TMP range (1.38–6.9 bar) investigated in this study. M3 displayed the lowest PWP relative to the MMMs and lowest PWF, indicating a direct relationship between the PWF and the PWP. The incorporation of CNTs in the polysulfone matrix influences the permeability of the membranes as well as the pure water flux, suggesting that the CNTs play an active role on the performance of the MMMs. However, a 5% CNT loading was maintained for all the MMMs prepared using the three CNT dispersion methods during synthesis, confirming that the CNT dispersion methods also impact the performance of the MMMs. The inconsistency in the results obtained for the PWF and PWP confirms the importance of the CNT dispersion during MMM synthesis.
Homogenous dispersion of multiwalled CNTs in the polymer matrix has been reported to influence the interfacial interactions between the filler material and the polymer, ultimately affecting the performance of MMMs [
10]. The presence of voids between the CNTs and the polysulfone in the MMMs could be attributed to the interfacial defects formed during synthesis, therefore influencing the permeation flux and the permeability of the membranes. Interfacial defects as a result of interface voids or rigidified polymer layers around the filler particles has been suggested to influence the permeability of MMMs by Aroon et al. [
1].
3.3.2. Oily Water Separation
The emulsified oil and water mixture, with an oil concentration of 1000 mg diesel/L water, was prepared by mixing water and diesel using a high-speed stirrer for 4 h. The initial concentration of oil is typical of oilfield-produced water from oil and gas reservoirs [
25]. The separation of the oil–water mixture was carried out using the Sterlitech cross-flow filtration module and the oil–water flux (OWF) was measured, with TMP varied from 1.38 bar to 6.90 bar. The effective membrane area was 42 cm
2, the experiments were conducted at room temperature, and the fluxes were measured after attaining a steady state after 4 h. The OWF as a function of the TMP is depicted in
Figure 8.
The OWF values at 6.9 bar were 182.4 L·m
−2·h
−1, 779.2 L·m
−2·h
−1, 15.6 L·m
−2·h
−1, and 47.6 L·m
−2·h
−1 for M1, M2, M3, and M4, respectively. M2 displayed the highest OWF for the given TMP range; the PWF for M2 was the highest (see
Table 3). However, the OWF is lower than the PWF obtained for M2 for the given TMP range. At 6.90 bar, the OWF of the CNT/Psf membranes were in descending order: M2 > M1 > M3.
The polysulfone membrane, M4, displayed the lowest OWF when compared to those of M1 and M2, but slightly higher than that of M3. Maphutha et al. [
9] reported that the OWF for their membranes are higher than that of the pure polysulfone membranes. This is in agreement with the results obtained for M1 and M2 in this study. However, the OWF obtained for M3 is not consistent with the finding from the authors, confirming further that the CNT dispersion method adopted during synthesis affects the performance of the MMMs. The PWF of M4 is higher when compared to its OWF for the TMP range, 1.38–6.90 bar. This could be attributed to hydrophobic nature of the membrane. The contact angle was measured to be 88.1° ± 2.1°, and the presence of methyl groups, which could be responsible for the hydrophobic properties of the membranes, was confirmed using the FTIR spectra as well.
The OWP at 6.9 bar was 26.4 L·m
−2·h
−1·bar
−1, 113.0 L·m
−2·h
−1·bar
−1, 2.3 L·m
−2·h
−1·bar
−1, and 6.9 L·m
−2·h
−1·bar
−1 for M1, M2, M3, and M4, respectively, as depicted in
Table 3. The PWP at 6.9 bar was 41.7 L·m
−2·h
−1·bar
−1, 159.3 L·m
−2·h
−1·bar
−1, 3.7 L·m
−2·h
−1·bar
−1, and 9.2 L·m
−2·h
−1·bar
−1 for M1, M2, M3, and M4, respectively (see
Table 2), confirming a decrease in the permeability of the MMMs during oil–water mixture separation when compared to the PWP. M3 displayed the lowest OWP of all the MMMs prepared in this study, and the OWP of M4 is greater than that of M3 (see
Table 3). Disparities are observed in the performance of the MMMs prepared using the three CNT dispersion methods, confirmed by the MMM performance.
OWP of the membrane from Maphutha et al. [
9] calculated using the reported OWF and TMP displayed a lower OWP when compared to M2 in this study. In addition, porous filler blockage has been reported to significantly reduce the performance of MMMs [
1]. The lower OWP of the membranes from Maphutha et al. [
9] could be attributed to the PVA layer on the CNT/Psf membrane layer.
The performance of the synthesized membranes was evaluated further using the oil rejection ratio (OR) calculated using Equation (4). Samples of oil–water permeate were taken during separation tests and analyzed using a UV spectrophotometer. The oil rejection results for the membranes are depicted in
Figure 9.
M3 displayed the highest OR of 99.88% when compared to the performance of other membranes (see
Figure 9). The ORs for M1, M2, and M4 membranes were 48.71%, 65.86%, and 84.92%, respectively. The outstanding selectivity of the M3 membrane over those of M1, M2, and M4 membranes could be attributed to the better dispersion of CNTs in M3. The surface chemistry of the CNTs has been shown to be hydrophilic with the presence of hydroxyl groups, suggesting that CNT/Psf with uniformly dispersed hydrophilic CNTs in the polymer matrix will display a higher affinity for water as opposed to oil. Without the PVA layer on the membranes prepared by Maphutha et al. [
9], it is expected that the performance of their membrane prepared using Method 2 will be similar to that of M2 in this study. The use of PVA as a layer constitutes additional membrane thickness as well as additional operating costs to the membrane fabrication. However, adopting the dispersion method, M3, in this study could alleviate the aforementioned problems without jeopardizing the expected selectivity.
The kinetic diameter of C
8 is 7.5 Å (0.75 nm) [
24] and the kinetic diameter of water is 2.6 Å (0.26 nm) [
26]. Furthermore, kinetic diameter of hydrocarbons increases with increasing carbon number [
27], suggesting that the components in diesel will have bigger kinetic diameters than that of C
8. The competition between water and oil to occupy the porous sites of the membranes is clearly shown by the low OR displayed by M4, indicating that the surface chemistry of the polysulfone membranes favors the absorption of oil over water. The low OR displayed by M1 and M2 could be attributed to poor CNT dispersion and formation of interfacial defects between the CNT and polymer matrix. The increase in free fraction volume promotes permeation of oil through the membranes and hence a dramatic reduction in membrane selectivity.
Results of this study compared to literature are presented in
Table 4. Despite the difference in the preparation methods and operating conditions employed in the previous studies shown in
Table 4, the membrane prepared in this study and tested for the separation of oil–water emulsion displayed outstanding performance compared to that of the membranes reported.