3.1. Pristine Membranes Characteristic
The contact angle of membranes was in the range from 50 to 60° (
Table 6). Results should be identical due to the lack of a modifier. These values are comparable for those obtained by other authors and suggest the hydrophilic–hydrophobic properties of the membranes. For comparison, Guruvenket et al. [
16] set the contact angle of pure polystyrene of 66°. In addition, membranes made of PES had similar values (60.3–63.3°) [
17]. The results obtained in the porosity tests for rPS 18 and rPS 20 membranes assumed a value ranging from about 57 to 62% (
Table 6), for this reason, they exhibit a similar permeation characteristic (
Figure 3). The rPS 16 membrane had a different porosity, below 40%, which directly affected its low permeability at low pressures from 0.05 to 0.1 MPa. A slight difference in porosity between rPS 18 and 20 rPS membranes did not have a significant impact on the difference in the value of the streams. A slight increase in flux was noted for the rPS 18 membrane due to its lower angle of wettability and hence greater hydrophilicity. For rPS 16, the permeate stream, contrary to common assumptions, was definitely lower than expected. This phenomenon can be explained by the presence of a large number of closed pores, which were opened only when applying a higher pressure above 0.1 MPa. This is visible in
Figure 3, where the flow rate for the rPS 18 membrane increases rapidly at pressures of 0.15 and 0.2 MPa. The rPS 14 membrane was not tested due to its excessive permeability, over 100 [m
3 · m
−2 · h
−1]. For this membrane, only surface tests were performed, according to which it has a similar water contact angle and the highest porosity. There is a relationship between porosity and permeability confirmed by Abdel Aal et al. [
18], in which the most permeable membrane had the highest porosity.
3.2. Reduction Degree of Pollutants in Ultrafiltration with Unmodified Membranes
The treatment efficiency of Kłodnica water is shown in
Figure 4. Higher reduction degrees of all water quality parameters were observed for membranes with lower permeability. This trend has been repeatedly observed by Tiron et al. [
19]. Apart from the reduction of conductivity, rPS 16 membrane had the highest performance for treatment of Kłodnica water. Comparable results were obtained using the rPS 20 membrane, while the rPS 18 membrane had the lowest degree of reduction of all selected parameters. The highest reduction degree was noted for the color, which ranged from 63% to 86%. In the second place, it was a decrease in the absorbance and the phenol index, where the level of reduction ranged from 21% to 44%. The intermediate retention of phenolic compounds (expressed by phenolic index) requires explanation because these substances belong to low molecular weight organic micropollutants and their retention in the ultrafiltration process is usually low. However, sometimes these substances may adsorb or form complexes with other pollutants (e.g., colloids, polymers, natural organic matter) present in surface water. It results in the formation of larger agglomerates, which are retained due to the molecular sieve effect in the UF process [
20]. The least effective removal was observed for nitrates, which ranged from 0% for rPS 16% to 3.5% for rPS 20%. The conductivity was reduced by 1.7%, 3.5%, and 5.2% by membranes rPS 18, rPS 20, and rPS 16, respectively. The reduction of nitrates and conductivity was not possible with the use of ultrafiltration membranes because the conductivity is primarily responsible for dissolved salts in the form of ions, which are not retained in the UF process [
21]. Similarly for nitrates that are mainly removed in the reverse osmosis process [
13].
Comparing the obtained results with literature data, it can be noticed that the reduction of color and absorbance was at the level obtained for commercially available PES and cellulose ultrafiltration membranes. Kabsch-Korbutowicz and Urbanowska [
22] reported that color and absorbance of Odra river water were reduced by 40–100% and 40–90%, respectively. Importantly, the reduction of color depended on cut-off of membranes [
19]. However, in the Xia and Ni [
23] studies on membranes made of PVDF and graphene oxide, the absorbance was reduced by 23–28%. Similarly in Dudziak et al. studies [
24], turbidity was reduced by commercial ultrafiltration membranes by about 90%, the color by 80%, and the absorbance by less than 60%. The concentration of bisphenol A was reduced by only a few percent, which was a much lower value, however, not having a full ratio in relation to the phenol index. Said et al. [
25] indicated that using PES membranes and neutral pH, phenols removal was around 70%. Conductivity cannot be removed by ultrafiltration membranes; however, some trials had been undertaken. Wandera et al. [
26] reported a very low reduction degree of conductivity from 1108 μS/cm to 1038 μS/cm by means of a commercial unmodified membrane made of polypropylene and cellulose.
During ultrafiltration of Kłodnica river, flux decline was observed suggesting occurrence of membrane fouling. This unfavorable phenomenon was caused by the macromolecular compounds present in the surface water. To better understand, the intensity of fouling was expressed by relative permeability (
Figure 5). The permeate flux decreased over time for each of the virgin membranes. The highest decrease was observed for the rPS 18 membrane with the highest permeate flow. The decrease was about 35% (for 240 min α = 0.65). This membrane should have the highest degree of contaminant retention according to its relative permeability but has the lowest. However, this phenomenon can be explained as follows: this membrane has the lowest water wettability and the highest porosity. The combination of these two features caused that large hydrophobic compounds more readily bind with the surface of the membrane, while high porosity promoted the simultaneous passing of small compounds through the membrane. A similar explanation can be used for the rPS 20 membrane where relative permeability was smaller and decreased only by 24% (for 240 min α = 0.76). This membrane had one of the highest treatment performances. The high porosity of this membrane caused a slower growth of the fouling layer. For the best membrane in contaminant removal, the decrease was about 31% (for 240 min α = 0.69), which corresponds to expectations. A similar tendency was received by Dudziak [
27], where the membrane with the lowest relative permeability had the highest removal performance.
3.3. Modified Membranes Characteristic
In the first part of the work, the membrane rPS 16 turned out to be the best and have the most interesting properties. Because of this reason, rPS16 membrane was chosen for second part of study aiming at modification by nanotubes and high removal of micropollutants from water. In these studies, Kłodnica water spiked with micropollutants was used.
Prepared membranes, despite the addition of carbon nanotubes, had quite similar surface properties in comparison to the unmodified membrane. The nanotubes did not have such a strong influence both on porosity and contact angle (
Table 7). Modification with hydroxyl groups was only aimed at an increase in the removal of micropollutants, while not changing the properties. The porosity of the membranes has been preserved as compared with pristine membrane rPS16. Contact angle was slightly lower, suggesting a more hydrophilic surface. This property could slightly influence the increase in flux through the membrane compared to an unmodified membrane (
Figure 6). The nanotubes positively influenced the mechanical stability of the membrane. A similar influence of nanotubes was confirmed by other authors [
28,
29]. Membranes made of PES with the addition of SWCNT-OH have similar surface properties as in the work of Kamińska et al. [
8].
The cross-sections of unmodified and modified membranes were shown in
Figure 7. It was found that all membranes had a typical asymmetrical structure, which was composed of a dense skin layer, a fingerlike structure below, and a macrovoid structure at the bottom. Macrovoids in rPS 16 were less visible than in rPS 16 0.1 membrane [
30]. Addition of nanotubes to the unmodified rPS membrane decreased the thickness of the membrane. A possible explanation is that the addition of nanocomposites slows down the solvent/anti-solvent exchange rate in the coagulation bath and this caused the formation of thinner membranes with a thicker skin layer due to the entrapment of more CNTs at the top surface during phase separation. These macrovoids and the developed finger-like structures enhance the permeation of water, thereby allowing for increased membrane flux [
31].
3.4. Reduction Degree of Pollutants and Micropollutants in Ultrafiltration Modified Membranes
Figure 8 presents the reduction degree of water quality parameters in ultrafiltration process with waste polystyrene membranes modified with carbon nanotubes. The biggest change in relation to
Figure 4 is the highest removal of phenols. The reduction of other coefficients is maintained at a similar level to unmodified membranes. The only perceived dependence is the increasing removability of phenols, along with the increase in the concentration of nanotubes in the membrane. There is also a slight increase in the retention of nitrates and substances responsible for the conductivity of the sample.
Much more interesting relationships were obtained in micropollutants removal studies presented in
Figure 9. A slight or complete lack of caffeine retention was caused by the specific properties of this substance. At the same time, it was the smallest substance with the lowest octanol–water partition (log Kow) coefficient, indicating low hydrophobic property, thus low affinity to adsorption. The same property was also responsible for low CBZ removal rates. Substances with Log Kow below value 2.5 are mostly soluble in water. Substances with log Kow higher than 2.5 can interact with membranes by hydrophobic interactions [
32]. This is the reason BPA and END were always removed by all three membranes. In particular, the rPS 0.1 membrane had high removal rates of these two micropollutants (CAF and CBZ) compared to the others. Modified membranes were not able to remove compounds with low log Kow. In many works, polystyrene acts as a phenol adsorbent in technological processes. For example, Siyal et al. [
33] achieved removal of 75% to almost 90% of phenols from industrial wastes by the use of modified polystyrene. This demonstrates the high capacity of this polymer to remove these types of compounds as well as the adsorption potential of phenols themselves. BPA is able to create a hydrogen bonding between compound and membrane surface [
34]. In addition, the highest degree of adsorption was observed at neutral pH; for Siyal et al. [
33], it was pH 7, and for Adamczak et al., it was pH 6.5 [
17].
As already mentioned, the high BPA and END retention were due to the properties of the substances themselves. The highest degree of BPA and END removal was observed for rPS 0.02 and rPS 0.1 membranes. In studies of Adamczak et al. [
35], identical relationship was observed: much better separation properties had membranes with 0.02 wt.% and 0.1 wt.% than the membrane with the concentration of 0.05 wt.% of carbon nanotubes functionalized with carboxyl groups. The introduction of nanotubes to the structure of membranes is only beneficial in a well-defined range of concentration. According to many authors, finding the optimum nanotubes concentration is a crucial factor [
36]. It is a small number of nanoparticles that make the membrane then have the most favorable separation properties [
37,
38]. A very large number of nanotubes can block the pores and promote a strong separation effect [
39]. On the other hand, this type of pore-blocking mechanism leads to a reduction in the permeate flow and a reduction in membrane performance, as can be seen in
Figure 4. All modified membranes had similar porosity, while the contact angle indicates indirect hydrophobic–hydrophilic properties, which may limit the adsorption of compounds with a hydrophilic character. The porosity of the membrane was sufficient to remove macromolecular compounds responsible for turbidity or color but promoted the penetration of hydrophilic micropollutants through the membrane.
Figure 10 presents relative permeability in time for modified membranes. Comparing this result to unmodified membranes (
Figure 5), modified membranes had higher values of relative permeability, indicating better antifouling properties. Among rPS 0.02, rPS 0.1, and rPS 0.05 membranes, the rPS 0.05 membrane had the best resistance to fouling. This membrane had the lowest flow as well as the weakest retention properties. However, it showed the highest stability with a stream decrease of only 10% (for 240 min α = 0.90). The other modified membranes had a decrease of 20% (α = 0.80) and 25% (α = 0.75) for rPS 0.02 and rPS 0.1, respectively. These results also indicate slightly better properties than for unmodified membranes. We can also observe a similar phenomenon as in the first part of the study, where the best retention membrane had the lowest α coefficient [
27].
However, ultrafiltration is not the absolute method for wastewater treatment. The addition of nanotubes highly improved contaminants retention, but nitrates and salts removal were still low. This inconvenience could be improved by using posttreatment processes. Usually, complementary treatment is achieved by the addition of nanofiltration (NF) or reverse osmosis (RO). UF-NF and UF-RO systems are a combination of low-and high-pressure processes and removal of salts and nitrates is much higher in comparison to only UF process. For example, Zhou et al. [
40] reduction of conductivity in UF-only process was unnoticeable, after the use of UF-NF it was higher for 90–95%. Mielczarek et al. [
41] were using UF-RO system for coke plant wastewater treatment. Reduction of conductivity after UF and after RO process was around 7% and 90%, respectively. Removal of ammonia values was 25.5% and 84.0%, respectively. Ultrafiltration process can be also combined with other processes like electrocatalysis or electrooxidation for removal of the smallest substances: ammonia and salts [
42,
43].