PET was derived from waste plastic bottles. Phenol was used as a solvent (KGaA Merck, Germany). PVP (40,000 Da) was purchased from Sigma-Aldrich Co. Ltd (USA). Humic acid (HA) powder was also obtained from Sigma-Aldrich. The HA solution was synthesized by dissolving HA powder in 1 L of distilled water.

The membrane was prepared via thermally induced phase separation (TIPS). Firstly, phenol, as a solvent, was heated at 50°C until the liquid phase was reached. Fragments of PET bottles were dissolved in molten phenol at 100°C and stirred using a magnetic stirrer for 6 hours until homogeneous. In order to improve the performance of the membrane, 5wt of PVP was added to the solution. Four types of membranes were composed: the composition and the condition of each dope solution are summarized in Table 1. A minimum of four membranes were made of each type, and three membranes of each type were chosen for the filtration experiments (below).

After obtaining a homogeneous solution, the dope temperature was maintained at 100°C without any stirring to remove air bubbles. The homogeneous solutions were cast uniformly onto a glass plate using a Baker applicator (YBA-3, Yoshimitsu, Japan) at room temperature. The thickness of the membrane was set at 700 µm. The casting film was left in the air for 7 minutes to evaporate the solvent. The glass plate was then dipped into a coagulation bath containing water-propanol 1:12 as a non-solvent. Another casting film, for no evaporation treatment, was directly immersed into non-solvent. The glass plate was then dipped into a coagulation bath containing water-propanol 1:12 as a non-solvent. The membrane sheets formed were washed and stored in distilled water for 1 day to remove any residual solvent.

The morphologies of the membrane surface and cross-section were analyzed using a scanning electron microscope (model, JSM 6360LA; JEOL Ltd., Japan). The dried sheets of membrane were gold sputtered for eproducing electric conductivity. Photomicrographs of PET membranes were viewed in vacuum condition at 5 kV. The magnification image was taken at 10,000 x for the surface and 700 x for the cross-section.

Functional groups of the membrane were analyzed using a Shimadzu FTIR-8400 spectrometer (Japan). A wavelength of 4000–400 cm ⁻¹ was used, and the chemical groups of the membranes were identified by their peaks using IR solution 1.50 software (Shimadzu).

The ultrafiltration test was conducted using dead-end filtration and pressurized with nitrogen gas. The filtration area of the membranes was 15.2 cm ². The operating condition was set at 1 bar transmembrane pressure and room temperature. Filtration was carried out for 30 minutes, and the permeate was collected three times every 10 minutes. Three repeats for each membrane type was performed. The filtration experiment was evaluated in term of flux and rejection of HA solution. Flux is the total volume of permeate pass through the membrane in a determined filtration period calculated by Equation (1). Rejection is the amount of HA particle rejected by membrane, as analyzed by Equation (2).

Flux = V A . t ( 1 )

In which:         V = Volume of permeate (L)

                       A = Membrane surface area (m ²)

                       t = Filtration periode (hr)

The model for the ultrafiltration test in this study was HA solution at 10 mg/L concentration. The solution was prepared by dissolving HA powder in 1 L distilled water. The rejection value of the membrane in HA filtration was calculated as follows ^{11, 12} :

R % = ( 1 – C p C f ) × 100 ( 2 )

In which:         R = rejection (%)

                      Cf = HA concentration in feed

                      Cp = HA concentration in permeate

The HA concentration in the feed and permeate solution were measured using a UV-Vis spectrometer (model UV-1700; Shimadzu) at 490 nm wavelength. The HA filtration schematic using a PET membrane can be seen in Figure 1.

The PET membrane was prepared via TIPS. The constructed membranes were categorized as asymmetric ultrafiltration membrane. The top surface and cross-sectional images of the membranes are shown in Figure 2 and Figure 3, respectively. Figure 2 shows the changes in membrane morphology with the addition of PVP and evaporation on casting film. PET-2 and PET-4 were evaporated for 7 minutes before being immersed in a coagulation bath. The membranes formed showed a decrease in the membrane porosity. A high evaporation temperature (100°C) for 7 minutes increased the exchange rate of solvent from the surface of the casting film. This led to a higher polymer concentration near the top surface, and, therefore, the phase exchange of solvent and nonsolvent in the coagulation bath became lower. This is called delay demixing, which causes the membrane to have a less porous structure ¹³ .

The cross-sectional image in Figure 3 shows the asymmetric structure of the sub-layer membrane. The evaporated membranes had a thicker, dense top layer due to the delay demixing. In the sub-layer of the modified membrane, the porous structure changed with the addition of PVP. The effect of PVP in the casting solution caused the formation of pores and sponge structure in the sub-layer ⁹ . In the membrane modification process, enhancing the pore density with uniform pore size is essential. A sponge structure, like the one formed in the modified membrane, affects the filtration quality and mechanical properties of the membrane ² .

The FTIR spectrum was analyzed to determine the changes of the chemical groups on the membrane surface ³ . Regarding polymer composition in this research, FTIR analysis was carried out for PET-1 and PET-3 membranes only. Polymer composition of PET-2 and PET-4 membranes was similar with PET-1 and PET-3, respectively; the IR spectra of PET-2 and PET-4 are equal to the IR spectra of PET-1 and PET-3, respectively. Figure 4 shows the FTIR spectrum of the PET-1 membrane. A peak of 3630-3300 cm-1 indicated the presence of the alcohol functional group (OH). In the range of 3200-3000 cm ^-1 and 1700 cm ^-1, bands of OH and CO were derived from the carboxylic acid functional group. An aromatic functional group of C = CC band was located at 1650-1600 cm ^-1. At 1400 cm ^-1 and 860 cm ^-1, a C-H band of alkanes and aromatics were observed. A very weak peak at 1250 cm ^-1 indicated a C-O band of the phenol functional group; phenol is composed of the aromatic ring and OH groups ¹⁴ . The identification of the membrane chemical groups is presented in Table 2. The chemical structure of PET, composed of several chemical groups, is shown in Figure 5. According to the data shown in Table 2, the membrane was composed of PET material.

The comparison of the FTIR spectrum analysis between PET-1 and modified membranes (PET-3) can be seen in Figure 6. Generally, the FTIR spectrums of PET-1 and modified PET/PVP membranes were similar, because both membranes were made with PET as the basic material. However, a lower peak at 1820 cm ^-1 and 1180 cm ^-1 of the modified membrane (PET-3) indicated a carbonyl functional group and CN band. The existence of these bands showed that the membrane was composed of PVP. The molecular structure of PVP is shown in Figure 7.

Water filtration is related to membrane characteristics, such as hydrophilicity and pore size. Furthermore, the addition of membrane modifying agent and the evaporation process also affects water permeability ¹³ . In this study, a feed solution of humic acid (HA) was tested at 10 mg/L. The concentration of HA solution in the feed and permeate were measured using a UV-Vis spectrometer. The comparison of the original and modified PET membranes in the HA flux and rejection is given in Figure 8. According to Figure 8, the PET/PVP modified membrane with evaporation (PET-4) had the highest HA rejection of up to 76%, followed by PET-3, PET-2, and PET-1 which are 67.3, 50.07, and 25.02%, respectively.

The differences in HA rejection in Figure 8 show the influence of PVP as an additive material and the evaporation time on the performance of the membrane. Figure 8 also shows the flux of the HA filtration. The membrane with smaller and uniform pores (PET-4) was better in HA rejection, but produced less permeates (the addition of PVP in the membrane solution increased the total concentration of the polymer and led the membrane to have smaller pores). Additionally, PVP improved the hydrophilic nature of the membrane surface. This prevented the hydrophobic HA molecules from getting closer to the membrane surface ³ . Therefore, the PET/PVP- modified membrane followed by evaporation (PET-4) was the best at HA rejection compared to the original PET membrane without evaporation (PET-1), or original PET membrane with evaporation process (PET-2).