1. Introduction

Membrane separation presents various advantages such as stability and high efficiency over traditional wastewater treatment methods. However, membrane separation has the disadvantage of membrane fouling, a phenomenon in which pollutants are accumulated on the surface of the membrane, reducing the effectiveness and lifespan of the membrane. Among several types of membranes, ceramic membranes are a potentially good candidate due to its hydrophilicity and chemical, thermal, and mechanical stability. The hydrophilic nature of ceramic membranes leads to antifouling properties. In addition, the chemical stability of ceramic membranes help it withstand the aggressive chemical treatment to remove pollutants on its surface [1-6].

Another way to combat membrane fouling is to make composite membranes such as graphene oxide (GO) composite membranes. Ma et al prepared composite GO and poly-dopamine (PDA) membranes which showed high water flux and effective dye removal [7-9].

Hybrid ceramic membranes with photocatalytic properties produce a synergetic effect of the advantages of photocatalysis and membrane separation while remedying the disadvantages. Photocatalytic oxidation is highly efficient and consumes little energy during operation but has a low reutilization rate, as well as a possibility of secondary pollution through the photocatalytic particles in the effluent. In a hybrid system, the membrane improves photocatalytic particle retention, improving efficiency and effluent quality. On the other hand, photocatalytic degradation can remediate fouling on the membrane surface, solving flux decline[10-13].

Under irradiation of light, pollutant molecules undergo photo-oxidation, thermal oxidation and photolysis. This review focus on the degradation in presence of photocatayst lie titanium dioxide (TiO₂) and zinc oxide (ZnO). Fig. 1 represents the schematic of classification of the review and Table 1 represent the summary of photocatalytic membrane.

2. Photocatalytic Ceramic Membrane

Fouling in membranes causes a variety of problems such as flux decline, reduction of membrane lifespan, and increase in operation cost. Therefore, photocatalytic particles have been added to alleviate this issue. Iron oxychloride (FeOCl)-coated ceramic membranes (FeOCl-M) were fabricated to investigate the photo-Fenton reaction and filtration performance[14]. Nitrobenzene (NB) was used as the model foulant. The surface of FeOCl-coated ceramic membranes was covered with rod-like catalysts, which was not the case in pristine ceramic membranes, which were porous. Different FeOCl loads (0, 0,28, 0.58, and 0.97 mg cm^-2) were tested to determine the optimal load. As FeOCl load increased, pure water flux dramatically decreased while NB rejection increased. Therefore, to prevent low pure water flux while maintaining high NB removal, 0.58 mg/cm^-2 was determined to be the optimum load. Compared to pristine ceramic membranes, FeOCl-M had smaller pure water flux, due to the silane coupling agent and FeOCl catalyst partly blocking the porous membrane surface. However, the blockage did not seriously interfere with water permeability, as water permeability with 0.14 μm pore was 6732.8 LMH MPa^-1 for pristine membranes and 2150.8 LMH MPa^-1 for FeOCl-M. Photodegradation of NB by UV irradiation, H₂O₂, and FeOCl-M was examined as well. The presence of FeOCl increased degradation rates for UV irradiation and H₂O₂, by 32.3% and 39.7% respectively. All three methods combined produced the fastest degradation waste, achieving complete elimination within 10 minutes.

2.1. Titanium dioxide

TiO₂ is a popular photocatalytic membrane support material because it is a highly effective catalyst, as well as having low environmental toxicity and chemical stability. TiO₂ composite membranes show low fouling rate, high water flux, and low energy consumption.

Nanostructured TiO₂/Al₂O₃ composite ceramic membranes were fabricated using a poly(oxyethylene methacrylate) (POEM) template using the sol-gel method to enhance photocatalytic performance in the treatment of textile wastewater[15].

The hydrophilic TiO₂ precursor TTIP has a strong interaction with the hydrophilic POEM chains and forms a POEM/TTIP hybrid. The low viscosity of the POEM/TTIP solution will allow it to penetrate the AlO3 membrane, influencing nanostructure morphology. Thus, PVP was coated on the membrane to prevent the penetration. Crystalline TiO₂ was formed from TTIP and the organic components (POEM, PVP, and solvents) decomposed during calcination at 450°C. Compared to bare Al₂O₃ membranes, TiO₂/Al₂O₃ composite membranes showed a drastic decline in membrane permeability, from 65.2 to 9.9 L m^-2 h^-1 bar^-1 without UV illumination and from 71.5 to 14.2 m^-2 h^-1 bar^-1 with UV illumination. The improvement of permeability under UV illumination was caused by the enhancement of membrane wettability or the increase in water temperature due to UV irradiation. The effect of temperature on permeability was less prominent in composite membranes compared to bare membranes. Membrane fouling after a filtration time of 24h was higher in bare membranes, with or without UV illumination. Without UV illumination, the composite membranes showed approximately 89% lower flux decline than the bare membrane. With UV illumination, composite membrane showed no flux decline, while bare membranes showed a permeate flux decay of around 51.2% due to a cake layer forming on the surface of the membrane under UV illumination. The TiO₂/Al₂O₃ composite membrane showed dye removal efficiency of around 96% after filtration time of 24h, due to the catalytic active sites induced by the POEM template encouraging membrane photocatalytic ability.

A photocatalytic membrane based on diethylene glycol (DEG)-assisted TiO₂ nanostructure on a porous alumina support was fabricated using hydrothermal method, with potassium titanium oxide oxalate dehydrate (PTO) as the TiO₂ precursor and DEG as the capping agent and investigated for antifouling properties[16].

Three samples were fabricated with differing DEG content, labelled T1, T2, and T3, with a ratio of water to DEG 9:31, 11:29, 13:27 respectively. The morphology of the membranes changed as DEG amount changed in synthesis. In lower amounts, the morphology of the membrane had a layered structure as opposed to a nanorod structure, which was prevalent in high DEG content membranes, due to DEG suppressing the hydrolysis of PTO. The non-porous layered structure of the low content DEG membranes led to a decrease in photocatalytic reactivity. TiO₂ amount varied with DEG content as well, with TiO₂ content increasing as DEG content decreased. Pure water permeability of the three membranes were lower than membranes with bare alumina support, due to the TiO₂ layer resisting water flow. The pure water flux drastically decreased in the highest content DEG membrane (T3), due to its shorter and thicker layered clusters. Antifouling properties were best in the T1 membrane, achieving an almost full recovery of permeate flux within 1 h of UV irradiation. Figure 4 compares the effect of UV irradiation on membrane surfaces of T1, T2, T3, and bare membranes after membrane filtration. The red/black fouling layer on membranes before UV irradiation was eliminated in the T-series membranes, with the T1 membrane showing the most effective elimination.

Graphene oxide (GO) and nitrogenated reduced graphene oxide (NrGO) membranes supported over ceramic porous materials were prepared using the vacuum filtration technique[17]. GO membranes showed a homogeneous porous structure with a rougher surface area while NrGO membranes showed a heterogeneous consonance distribution with lower porosity. In NrGO-ZrO₂ membranes, ultra-thin coatings were stable, but not as homogenous as GO-ZrO₂ membranes due to the hydrophobic nature of GO-ZrO₂ membranes. Electroconductivity was much higher in NrGO membranes than GO membranes. Water permeability dramatically decreased after the GO and NrGO layer was deposited, due to the 3D structure of GO, which is much thicker than graphene sheets and hydrophobic properties of NrGO. Ceramic support pore size (SPS) was also investigated for its effects on membrane flux. Membrane resistance differences were not significant in different SPS, with the GO layer mainly affecting pure water permeability.

A slurry photocatalytic membrane reactor (PMR) was prepared and tested to assess its performance in degrading anticancer drugs spiked in real secondary wastewater effluent[18].

Parachlorobenzoic acid (pCBA) was used as the catalyst. Laboratory grade water (LGW) was used to test the permeability of the PMR system excluding the catalyst to assess the system’s behavior in secondary wastewater effluent (WWeff). The water transmembrane fluxes did not change after the suspension of the catalyst, which indicates the catalyst does not hinder water transfer through the membrane. However, at a lower cross-flow velocity (CFV), water permeation resistance increased, due to the concentration of suspended particles decreasing and a catalyst cake layer forming. The formation of a catalyst cake layer was more prominent in WWeff and had a higher porosity, which led to less resistance of water transfer. The cake layer did not influence membrane separation properties significantly. When tested with dextran 150 kDa (1g/L), the rejection rate remained similar to membranes without a catalyst cake layer, indicating that the catalyst cake layer had an exclusion size that was not smaller than the average membrane pore. The optimal pCBA load was determined to be 1.5 g/L, as it had high degradation rates without forming high amount of catalyst particles that would obstruct photon penetration lengths. On three anticancer drugs, 5-fluorouracil, cyclophosphamide, and capecitabine, adding a TiO₂ catalyst led to an increase in removal rate of +90%, +250%, and +130% respectively.

An optimized photocatalytic ceramic membrane was synthesized and assessed for its applications in wastewater treatment[19]. The membrane consists of a polymeric TiO₂ mesoporous top layer, a colloidal titania (TiO₂) mesoporous intermediate layer, and an alumina (Al₂O₃) microporous support. To determine the optimal number of the intermediate layer, one, two, and three colloidal layers were tested for Rhodamine B (RhB) degradation. RhB degradation increased as the number of layer increased, with 12.8%, 29.6%, and 33.7% for one, two, and three layers respectively. For the top polymeric layer, the degradation also increased as the number of layers increased, but the increase was not as dramatic as the intermediate layer. Therefore, the optimal number of coatings was two intermediate colloidal layers and one top polymeric layer with RhB degradation rate of 24.7%. The membrane performance was tested on oily wastewater. As pressure increased, the permeation flux increased while COD rejection decreased. A pressure of 5 bar was found to be the optimal operating pressure. The optimal temperature was determined to be 35°C, at which the membrane had a high permeation flux. At the optimal conditions, the membrane showed a permeation flux of 29.1 kg m^-2 h^-1 and COD rejection of 78.4%.

A photocatalytic membrane with a thin TiO₂ coating was fabricated and assessed on its performance under a solar simulator using real, nonpotable surface water [20].

Light transmitting sintered glass membrane substrates were used as a solution to light attenuation. The sintered glass substrate has a high transparency in shorter wavelengths under 355 nm but decreases to 10% at higher wavelengths over 370 nm, which was enough to facilitate photocatalytic activity on the surface of the membrane. The photocatalytic response was effective in reducing the irreversible fouling of the membrane by 8-folds (from 0.0001 m²/L to 0.0008 m²/L) in hydraulically irreversible fouling index (HIFI) values. In addition, the time to chemical clean (TtCC) improved by 4.5 folds, from 2.0 days to 8.8 days under solar irradiation. This led to an improvement in filtration energy saving, as energy consumption was calculated to be 0.01 kWh/m³ while no solar irradiation had 0.02 kWh/m³ energy consumption. Turbidity removal and UV₂₅₄ absorbance was also improved by the TiO₂ coating. Turbidity removal increased from 39% to 94% with the added TiO₂ coating. High performance size exclusion chromatography (HPSEC) for total organic carbon (TOC) revealed that the main rejec in TOC removal was membrane rejection, not photocatalysis.

A combined photocatalysis (TiO₂/UV) and membrane filtration system was studied for its applications in the disinfected secondary effluent treatment[21].

The secondary effluent consisted of dissolved organic carbon (DOC) and was characterized by liquid chromatography- organic carbon detector (LC-OCD) technique. DOC removal was tested at different TiO₂ concentrations (0.5, 1, 2, and 4 g/L). There were no significant changes in adsorption in high TiO₂ concentrations, as the removal rate ranged from 9 to 12% in 0.5 and 4 g/L concentrations. This indicated that the TiO₂ surface could only adsorb a portion of organics. With the addition of UV exposure, after 20 minutes, DOC concentration slightly increased due to intermediate UV oxidation products being released from the TiO₂ and then continued to decrease for 1,2, and 4 g/L TiO₂ concentrations. DOC removal was between 3% to 23% in 0.5 and 4 g/L concentrations, indicating that the effective organic matter removal was not achieved in the secondary effluent. UV absorbance at 254 nm decreased by 8%, 20%, 33%, and 52% for concentrations of 0.5, 1, 2, and 4 g/L respectively. These values were twice as big as DOC removal. In a control test with humic acid (HA) system, the DOC and UV₂₅₄ removal rate was > 80% and > 90% respectively. Permeate flux in the ceramic ultrafiltration membranes were improved as the TiO₂ concentration increased. Figure 8 represents permeate flux after photocatalysis of wastewater containing different TiO₂ slurry concentrations. DOC and UV₂₅₄ removal improved with the combination of photocatalysis and ceramic filtration, from 23% to 60% and 52% to 54% in 4 g/L TiO₂ concentration. LC-OCD analysis revealed that the amin foulants in secondary effluent were substances eluting in the biopolymer peak.

A TiO₂ nanofiber membrane (TNM) loaded on porous fly ash ceramic support (PFACS) was fabricated using semi-dry press and hydrothermal process to study its applications in the removal of organic pollutants Rhodamine B (RhB) and heavy metal ions (Cu(II), Cd(II), Cr(VI))[22]. TNM-PFACS showed high efficiency in the removal of divalent heavy metal ions, with 96.38% and 89.33% removal efficiency for Cd(II) and Cu(II) respectively, over a period of 240 minutes. This is due to the electrostatic force of the negatively charged surface and high surface area of TNM-PFACS. With the RhB solution, TNM-PFACS showed a 40.84% removal rate. When all pollutants were existent in the solution, cationic species showed a competition adsorption, which was shown by the decrease in the Cd(II) and RhB removal from 96.38 to 41.53% and 40.84 to 22.44% respectively. However, Cr(VI) removal increased due to the existence of RhB enhancing Cr(IV) adsorption and electrostatic attraction. In the photodegradation process, Cu(II) and Cd(II) removal was not significantly affected, while the removal of RhB and Cr(VI) removal significantly increased, achieving maximum removal rate of 97.09% for Cr(VI).

2.2. Zinc Oxide

ZnO has numerous benefits such as high degradation rates for organic pollutants compared to other metal oxides such as TiO₂. In addition, ZnO exhibits high photocatalytic reactivity, and is cheap to source. ZnO can be modified to absorb visible light as well as UV light by nitrogen doping.

The Ultrasonic Spray Pyrolysis (USP) method was used to generate particles with high oxidation characteristics to use in organic pollutant treatment via oxidation[23]. The same parameters were applied in the generation of Au/ZnO particles. Photocatalytic activity was higher in Au/ZnO particles than pure ZnO particles in RhB aqueous solution photoysis under UV irradiation. This is due to Au nanoparticles trapping electrons excited from ZnO, causing the recombination of excitation electrons to decrease while promoting hydroxyl radicals (⋅OH) and superoxide radical anions (O₂⋅^-) to form. Thus, the photolysis rate constant increased following the Au loading increase from 0.025 mass% to 0.1 mass%. However, photolysis rate decreased when Au loading further increased to 0.2 mass%, due to the surplus Au loading accumulating excess exciton electrons. As photocatalyst dose increased, at a range of 2 mg to 10 mg, photocatalytic activity improved, whereas further increase of dose above 10 mg did not dramatically improve photocatalytic activity, due to turbidity blocking UV light.

A hybrid system of ZnO photocatalyst and ceramic nanoporous membranes was prepared and investigated for its degradation abilities of Rhodamine-B dye in various conditions[24]. As inlet pressure to the membrane increased from 2 to 4 bars, the permeate flux increased but showed a decline as time progresses due to organic pollutants polarizing on the surface of the membrane. Different ZnO loading quantities (0.25, 0.5, 0.75, and 1 g L^-1) were studied to determine the optimal quantity. ZnO catalyst loading of 0.5 g L^-1 was the optimal loading quantity, as it had the highest dye decolorization rate of 86%. The decline of dye decolorization rate in catalyst loading over 0.5 g L^-1 was potentially due to the scattering of light and reduction of light penetration in the solution. Dye decolorization was also affected by the pH of the dye solution, increasing as pH values increased. Neutral pH showed the highest decolorization rate while acidic and alkaline medium pH showed decreased rates. This is due to the dissolution and photodissolution of ZnO in acidic and basic pH causing low reactivity of ZnO. Initial dye concentration was tested from 500-2000 mg L^-1 to assess its effect on decolorization. As initial dye concentration increased from 500 to 2000 mg L^-1, decolorization decreased from 86% to 54% due to the amount of dye absorbed on the photocatalyst increasing while other parameters remained constant, reducing efficiency. Ceramic nanofiltration and photocatalysis worked synergistically in the system, achieving 96% of dye decolorization and 70% TOC removal over 90 minutes of operation.

3. Conclusions

Wastewater treatment through membrane separation shows high efficiency and cost effectiveness. These advantages can be improved by preparing a hybrid photocatalytic membrane separation system, such as composite graphene oxide membranes. Various photocatalyst particles can be used for the preparation of a hybrid system such as TiO₂, graphene oxide, and ZnO. Photocatalytic particles can help prevent membrane fouling under UV illumination. Current practical challenges of this technique arise from the supply of artificial light sources. This review discusses the preparation and wastewater treatment characteristics of hybrid photocatalytic membrane separation systems according to the photocatalytic materials: TiO₂ and ZnO.