INTRODUCTION

The organic dyes released into water and wastewater by various industries have prompted widespread environmental concerns worldwide. The unchecked discharge of dye-containing wastewater into the environment damages human health, causes disease, and endangers the aquatic ecosystem, plants, and animals [1-3]. In the event of contact, dye pollutants permanently damage the eyes and irritate the skin [4]. They are carcinogenic and will cause burning mouth syndrome, dizziness, heavy sweating, nausea, and methemoglobinemia [5,6]. Dye-containing wastewater increases water turbidity after entering natural water sources, preventing light, which is essential for the synthesis and respiration of aquatic organisms, from entering water [7,8]. Therefore, these toxic and dangerous pollutants must be removed from wastewater before release into the environment.

Membrane filtration is an integral and continuous stage in advanced wastewater treatment, including dye-containing wastewater. It consists of microfiltration, ultrafiltration, nanofiltration, and reverse osmosis. The numerous advantages of these pressure-driven physical methods include the production of clean water, good efficiency, easy operation, lack of chemicals, and feasibility of large-scale production [9,10]. Although nanofiltration removes dye molecules at a high rejection rate with its particle size-based separation mechanism, it has disadvantages such as low water flux, fouling, and high energy consumption. Meanwhile, despite its higher water flux, the wider pores in ultrafiltration can allow some dye molecules to pass through, resulting in a lower rejection rate than nanofiltration [11,12]. Ultrafiltration is also constrained by the residual pollutant molecules in the feed solution, which should be removed in some way [13,14]. To address these constraints, ultrafiltration can be combined with advanced oxidation processes such as photocatalysis to create a photocatalytic membrane system with greater pollutant removal efficiency [2, 15-17]. Photocatalysis is an advanced oxidation technology in which photocatalyst material is excited to produce free radicals to degrade pollutant molecules under irradiation [18-21]. The photocatalytic membrane system addresses the problem of photocatalyst particle recovery from the solution, reduces membrane fouling by photocatalysis, and increases flux. Thus, the individual limitations of photocatalysis and membrane filtration are both addressed [22-24].

Photocatalytic dynamic membranes can be self-forming or pre-coated. The former initially mix wastewater with the photocatalyst and filter using the support membrane; whereas the latter put photocatalyst particles under pressure on the support membrane before wastewater filtration [2,25]. Habibollahi et al. [2] synthesized and used the Ag-ZnO/g-C₃N₄/GO nanocomposite photocatalyst in the photocatalytic dynamic membrane (self-forming and pre-coated) to remove methylene blue dye. The results showed that the synthesized photocatalyst improved the rejection rate, flux, and the hydrophilicity of membrane surface in the photocatalytic membrane system. Compared to the static mode, the photocatalytic dynamic membrane also provided better and faster dye degradation. Liu et al. [22] used the dead-end filtration system to deposit different amounts of the ZnIn₂S₄ photocatalyst on the PVDF membrane, fabricating a pre-coated photocatalytic dynamic membrane. The results illustrated that with 2.6 mg/cm² of photocatalyst loading, the photocatalytic dynamic membrane was more efficient for Fluvastatin (97.19%) and TOC (53.29%) removal than pure PVDF. Gao et al. [26] employed a photocatalytic dynamic membrane with the ZnIn₂S₄ photocatalyst for removing tetracycline. The results suggested that the dynamic ZnIn₂S₄ layer led to 57% tetracycline mineralization after 3 hr of filtration. Also, 92% tetracycline (100 ) removal efficiency was retained during 36 hr of reaction. 

Numerous water and wastewater treatment researchers have developed an interest in medium-band-gap photocatalyst material (active in the visible light spectrum). This can be attributed to the high percentage of visible light (45%) in sunlight, which is merely 7% for UV light. Therefore, photocatalysts capable of photocatalytic activity under sunlight are preferable [27-31]. Two-dimensional graphitic carbon nitride is a non-toxic semiconductor with easy synthesis and an average bandgap energy of about 2.7 eV. Hence, it provides adequate photocatalytic activity under visible light and is capable of photodegrading organic pollutants through oxidation-reduction (redox) reactions. To increase its photocatalytic activity and delay the recombination of electron-hole pairs, this semiconductor can be combined with another semiconductor with a larger band gap and conduction and valence band energies that closely emulate the conduction and valence band energies of graphitic carbon nitride to prepare the graphitic carbon nitride/semiconductor nanocomposite [32,33]. Despite studies on using graphitic carbon nitride and its composites for making photocatalytic membranes, it has limited use for producing photocatalytic dynamic membranes and removing dye pollutants. Therefore, this study aimed to use the graphitic carbon nitride/zirconium dioxide nanocomposite to fabricate a self-forming and pre-coated photocatalytic dynamic membrane and use it for the treatment of dye-contaminated wastewater (MB and CR solutions) under LED light exposure. The pure membrane and the photocatalytic dynamic membrane will be compared in terms of rejection rate and permeation flux in removing MB and CR dyes. Moreover, the photocatalytic membrane will be compared with the batch photocatalytic system in terms of dye removal efficiency under LED irradiation. To fit the experimental data with kinetic models, the zero-, first- and second-order models were used in batch photocatalytic systems and self-forming photocatalytic dynamic membranes to comprehend the system kinetics.

EXPERIMENTAL

Materials

This part was presented in SI, section 1.

Synthesis of Graphitic carbon nitride/Zirconium dioxide nanocomposite

The synthesis of pure graphitic carbon nitride and pure zirconium dioxide is explained in our previous study [34]. To summarize the nanocomposite synthesis, graphitic carbon nitride, and zirconium dioxide nanosheets with an equal mass ratio were added to a specific amount of ethanol, dispersed in an ultrasonic bath for 1 hr, and finally stirred on a magnetic stirrer for 20 hr. The final mixture was oven-dried after centrifugation and multiple washes with water and ethanol. To link the two photocatalysts’ nanosheets, the resultant powder was calcined in a tube furnace at 400°C for 1 hr (Scheme 1). The graphitic carbon nitride was abbreviated as CNG and the resultant nanocomposite was called CNGZ.

Preparation of polyethersulfone membrane

The polyethersulfone (PES) membrane was synthesized through a phase inversion process based on the method proposed in our previous study [2]. To synthesize a 16 wt% membrane, PES was added to the DMF (dimethylformamide) solvent and stirred slowly at room temperature for 12 hr until obtaining a fully transparent solution. After remaining steady for 6 hr for degassing, the polymer solution was slowly poured on a piece of polyester glued on glass, where it was spread and molded using a casting knife. Then, to form a solid membrane through phase inversion, the membrane was immediately placed in a distilled water bath. To ensure complete phase inversion and removal of the impurities and residual solvent, the membrane was left in a water bath for 24 hr before it was removed and put between two pieces of filter paper to dry completely.

Batch photocatalytic system

To examine the MB and CR dye degradation capacity of the synthesized photocatalysts, a specific amount of the photocatalyst was added to 100 ml of the dye solution with a concentration of 10  and stirred. Before light irradiance, the mixture was stirred for 30 min in the dark to reach adsorption-desorption equilibrium between the photocatalyst and the pollutant molecules. Then, the photocatalytic reaction was completed under a 100 W household LED lightbulb (with a UV light filter) as the visible light source. During dye degradation, a specific amount of the dye solution was extracted within 20 min and its concentration was measured using the spectrophotometer (in maximum wavelengths of 668 nm and 496 nm for the MB and CR dyes, respectively). The dye removal efficiency was calculated using the following equation:

Where, C_i and C_t denote the initial dye concentration and dye concentration at time t, respectively.

Self-forming photocatalytic dynamic filtration

First, the synthesized PES membrane was cut down to specific dimensions and placed inside a plexiglass cell with 15.2 cm² of area and 150 ml of volume in a dead-end system. Then, a specific amount of the synthesized photocatalyst was mixed with 150 ml of 10  dye solution and poured into the cell and the system was put under 1 bar of nitrogen gas pressure. To activate the photocatalyst, the membrane setup was put under a 100 W household LED lamp. An output sample was taken every 20 min for concentration measurement. Dye molecule rejection rate and permeation flux were respectively obtained using equations 2 and 3:

Where,  and represent dye concentration in feed and permeate solutions, respectively. Also, V, A, and t are the volume of collected permeation (L), effective filtration area (m² ) and filtration time (hr), respectively.

Pre-coated photocatalytic dynamic filtration

At first, a specific amount of photocatalyst was added to a specific volume of distilled water, which is then stirred for a certain period. Then, the resultant solution was poured into the dead-end membrane system containing a piece of PES membrane at the end and applies pressure via nitrogen gas to remove all distilled water from the system output and create a layer of the photocatalyst on the PES membrane surface. After layer formation, the 150 mL dye solution with 10 ppm of concentration was added to the system. The system was put under 1 bar of nitrogen gas pressure in the presence of a household LED light for a specified time. Note that permeation concentration and volume were measured at 20-min intervals.

Kinetic Study

Zero-, first- and second-order kinetic models were used for investigating the degradation kinetics of MB and CR dyes in the batch system and the self-forming photocatalytic dynamic membrane using the following equations [2,35]:

 Where,  and  are the initial (t=0) and secondary dye concentrations (t=t), (mg/L), respectively. Also, ,  and  are the rate constants of the zero-, first- and second-order kinetic models, respectively.

RESULTS AND DISCUSSION

TEM, FESEM, PL, and XPS analysis

Fig. 1 shows the morphology and particle size of the CNG photocatalyst and its nanocomposite with zirconium dioxide using TEM analysis. Fig. 1a shows that pure zirconium dioxide has a thin, transparent, and planar structure with a length of several hundred nanometers. However, CNG has much larger transparent sheets exceeding 1 μm in length (Fig. 1b). Figs.1c and d illustrate that zirconium dioxide fits well in the CNG structure with adequate proximity and overlap. The proper integration of the two materials’ nanostructures leads to a high electron transfer rate between materials, reducing electron-hole pair recombination and PL intensity (Fig. 1e) and increasing the photocatalytic activity of the nanocomposite [36]. According to the results of DRS (SI, section 2), the band gap energies of CNG, CNGZ nanocomposite, and zirconium dioxide are 2.76 eV, 2.67 eV, and 4.01 eV, respectively, indicating the visible-light-driven photocatalytic performance of CNG and CNGZ, and UV-driven photocatalytic performance of zirconium dioxide.

Moreover, Fig. 1f shows the FESEM image of the support membrane surface (PES). The support membrane has smaller pores than the nanocomposite photocatalyst structure (SI, section 3), suggesting that in the pre-coated mode, the photocatalyst tends to be deposited on the membrane surface than within its pores. The photocatalyst particles placed on the PES membrane surface prevent pollutants from contacting the support membrane surface and permeating into membrane pores.

Data about the elemental composition and graphitic carbon nitride and zirconium dioxide bonds were obtained using XPS analysis and results are presented in Fig. 2. Fig. 2a illustrates the XPS spectra (full scan) of the pure CNG and CNGZ nanocomposite. The N and C elements are observed in CNG-associated spectra, while the XPS spectra of the CNGZ nanocomposite show Zr, O, N, and C elements, which are related to the presence of both semiconductors in the nanocomposite structure [37,38]. Also, the high-resolution spectra of elements (C 1s and N 1s) are presented in Figs. 2b and c, respectively. According to Fig. 2b, CNG-associated C 1s spectra display two peaks at 284.6 and 288.1 eV, which are related to the sp² C-C and sp³ N=C-N binding energies, respectively. It is noteworthy, the new diffraction peak at 286.7 eV (in the CNGZ-associated C 1s spectra) corresponds to C-O bond energy in the interface of CNG and zirconium dioxide [38,39]. Furthermore, the three peaks located at 398.85, 399.9, and 401 eV in the CNG-associated N 1s spectra are related to C=N-C, N-C_3, and N-H bonds, respectively. As can be seen, these peaks in CNGZ nanocomposite have a left-shift (398.6, 399.1, and 400.6 eV) compared to pure CNG, which illustrates that N atoms of CNG gain electrons from zirconium dioxide [38,39].

Effect of CNGZ nanocomposite dose

To examine the effect of the CNGZ nanocomposite dose on the dye photocatalytic degradation, 5, 15, and 30 mg of the nanocomposite were used for MB dye removal. Fig. 3 exhibits MB dye removal efficiency with various doses of CNGZ photocatalyst. Increasing the photocatalyst dose from 5 to 30 mg increases dye degradation efficiency, which can be attributed to greater access to surface adsorption sites and the creation of more electron-hole pairs that increase the degradation rate of pollutant molecules. Since the highest dye degradation efficiency was achieved with 30 mg of nanocomposite, this was selected as the optimal photocatalyst dose.

Self-Forming and pre-coated photocatalytic dynamic membrane

Fig. 4 shows MB and CR dye removal efficiency and rejection rate using batch processes and self-forming photocatalytic dynamic membranes, respectively. In both systems, the photocatalytic activity and removal of both dyes gradually increase. However, the self-forming dynamic membrane system has a greater degradation efficiency than the batch system. Note that the rejection rates of the pure PES membrane were respectively 29% and 47% for the MB and CR dyes. The results indicate that the highest rejection rate of MB and CR dyes corresponds to the self-forming system with the CNGZ nanocomposite at respectively 89% in 120 min and 100% in 80 min. The greater dye degradation rate of the CNGZ membrane relative to the CNG membrane can be attributed to the greater electron-hole pair separation rate and recombination prevention due to the zirconium dioxide presence in the nanocomposite sample. Zirconium dioxide traps the electrons transferred from the valence band to the conduction band of graphitic carbon nitride, preventing electron recombination. This produces more free radicals during photocatalysis, which, when contributing to the remaining pores in the valence band, degrade dye molecules (Scheme 2). Overall, the dye removal results point to the synergistic effect of the membrane process and photocatalysis, which improves the dynamic membrane’s removal efficiency compared to the batch system (standalone photocatalysis).

As shown in Figs. 4 c and d, the self-forming dynamic membrane is more efficient and faster for removing CR dye than MB dye. Ultrafiltration membranes have wider pores than pollutant dye molecules and could let dye molecules through. Electrostatic repulsion between dye molecules and membrane surface somewhat compensates for the adverse effect of pore size and prevents dye molecules from passing. Given the anionic CR dye and the negatively-charged PES surface, the electrostatic repulsion between CR molecules and membrane surface blocks dye molecules from passing through the membrane, keeping dye molecules in the feed solution and degrading them by photocatalyst particles under light [12,40].

Fig. 5 shows the MB dye rejection rate of the CNGZ pre-coated photocatalytic dynamic membrane over time. In the first 20 min of filtration, the pre-coated membrane shows a high dye rejection rate of 98.6%. However, the value gradually decreases over time, which differs from the rejection rate changes of the self-forming membrane system. This can be attributed to the difference in the accessibility of photocatalyst particles to pollutant molecules in pre-coated and self-forming systems over time. The pre-coated system provides the highest dye removal efficiency at the beginning of filtration when most of the photocatalytic layer contacts a small number of dye molecules. The rejection ratio gradually decreases due to the reduction in untouched photocatalyst materials that contact pollutant molecules [2,41,42].

Fig. 6 shows the permeation flux of MB and CR dyes in the self-forming photocatalytic dynamic membrane. The pure PES membrane has the lowest permeation flux of 4.7   for MB and 7.6 .  for CR, which grows by adding the photocatalyst to the membrane system, which can be attrited to the high hydrophilicity of CNG and CNGZ photocatalysts.

The flux of the self-forming dynamic membrane containing CNGZ nanocomposite was obtained at 15.7  and 18.5  for MB and CR, respectively. Moreover, the dynamic membrane with the CNGZ nanocomposite achieves the highest flux due to the presence of hydrophilic zirconium dioxide nanosheets in the composite sample and its greater hydrophilicity than pure CNG [34].

Also, Fig. 6 shows that the pre-coated dynamic membrane containing the CNGZ photocatalyst achieves greater permeation flux for MB dye filtration than the self-forming dynamic membrane, which is in agreement with the results of several studies [2,43-46]. In the pre-coated membrane, the dynamic layer formed on the support membrane blocks dye molecules from passing through membrane pores. Before they reach the support membrane surface, the pollutant molecules are trapped by the upper photocatalytic layer and destroyed under light.

Kinetic Study

Based on zero-order, first-order, and second-order kinetic models, Figs. 7 and 8 show the fitness of experimental data for the batch system and the self-forming photocatalytic dynamic membrane for the MB and CR dyes. Moreover, Table 1 lists the kinetic parameters of mentioned models. As the table and the figures show, the experimental data of MB and CR removal using the batch system and the self-forming photocatalytic dynamic membrane follow the second-order model. Table 1 indicates that the highest constant rate for the MB and CR dyes corresponds to the self-forming photocatalytic dynamic membrane system with the CNGZ nanocomposite.

Comparison of the rejection efficiency with reported publications

The rejection efficiency of MB and CR dyes by various composite membranes is presented in Table 2. Compared with the listed membranes, the photocatalytic dynamic membrane constructed in this study illustrated good performance for both dyes.

CONCLUSION

In summary, the photocatalytic dynamic membrane was fabricated in self-forming and pre-coated modes with the CNGZ nanocomposite and used for removing organic pollutants from water under a household LED light. Both photocatalytic membrane systems showed exceptional dye degradation efficiency. Moreover, the dynamic membrane system outperformed standalone photocatalysis (batch system) in removing MB and CR dyes. The rejection rates of MB and CR dyes in the self-forming photocatalytic dynamic membrane with CNGZ were 89% in 120 min and 100% in 80 min, respectively. Furthermore, photocatalytic particles increased the permeation flux of the membrane system compared to the pure PES membrane. Moreover, the experimental data for the batch system and the self-forming dynamic membrane for both pollutant dyes were consistent with the second-order kinetic model. The results suggest that the self-forming photocatalytic dynamic membrane with the CNGZ nanocomposite is promising for increasing rejection rate and membrane flux in dye-containing wastewater treatment.

CONFLICT OF INTEREST

The authors hereby declare that there is no conflict of interest.