Overview of Photocatalytic Membrane Reactors in Organic Synthesis, Energy Storage and Environmental Applications

Abstract

This paper presents an overview of recent reports on photocatalytic membrane reactors (PMRs) in organic synthesis as well as water and wastewater treatment. A brief introduction to slurry PMRs and the systems equipped with photocatalytic membranes (PMs) is given. The methods of PM production are also presented. Moreover, the process parameters affecting the performance of PMRs are characterized. The applications of PMRs in organic synthesis are discussed, including photocatalytic conversion of CO₂, synthesis of KA oil by photocatalytic oxidation, conversion of acetophenone to phenylethanol, synthesis of vanillin and phenol, as well as hydrogen production. Furthermore, the configurations and applications of PMRs for removal of organic contaminants from model solutions, natural water and municipal or industrial wastewater are described. It was concluded that PMRs represent a promising green technology; however, before the application in industry, additional studies are still required. These should be aimed at improvement of process efficiency, mainly by development and application of visible light active photocatalysts and novel membranes resistant to the harsh conditions prevailing in these systems.

1. Introduction

Organic synthesis and environmental conversions can be carried out by heterogeneous photocatalysis (PC). This is an advanced oxidation process (AOP) which makes it possible to obtain the initiation of a chemical reaction or a change in its rate thanks to the action of ultraviolet (UV), visible (VIS) or infrared (IR) radiation in the presence of a photocatalyst [1,2].

Thanks to the absorption of radiation, photonic activation of the photocatalyst, instead of the traditional thermal activation, takes places [3,4]. The electronic structure of a photocatalyst is characterized by a valence band (VB) and a conduction band (CB). These bands are separated by a band gap of energy (Eg). Photons with energy (hν) at least equal to the Eg are able to excite the photocatalyst. Valence electrons (e⁻) are promoted from VB to CB, producing a hole (h⁺) in the VB. The so-formed electron/hole couples, which migrate to the photocatalyst surface, promote oxidation and reduction of the adsorbed substrate, usually by means of radical mechanisms [5].

PC has been extensively studied for about four decades, starting with the pioneering study of Fujishima and Honda [6] in 1972, which discovered the photocatalytic cleavage (splitting) of water into H₂ and O₂.

PC processes involve highly unselective reactions; therefore, they have been widely utilized in environmental applications, especially for the total oxidation of organic pollutants contained in water or in air to innocuous substances [7,8,9,10,11]. However, in recent years there has been a growing interest in the application of PC for synthetic purposes such as selective reduction and oxidation [12,13,14,15].

Potential advantages with respect to concurrent processes, making PC a green approach, are [16,17,18]: (i) the possibility to operate under mild conditions (ambient temperature and pressure); (ii) the use of greener and safer catalyst (mainly TiO₂) avoiding the use of more dangerous heavy metal catalysts; (iii) the use of mild oxidants, such as molecular oxygen; (iv) the possibility to obtain the real destruction of refractory and non-biodegradable contaminants with the formation of innocuous by-products; (v) the requirement of very few auxiliary additives; (vi) the applicability to a wide range of substrates in liquid, solid, and gaseous phases, even if diluted; (vii) the possibility to use renewable solar energy; and (viii) the possibility to couple PC with other physical and chemical technologies.

Despite these advantages, the costs related to the separation of the heterogeneous photocatalyst and the poor process selectivity represents the major drawbacks limiting PC application at industrial scale [19,20]. To overcome the separation problems, the deposition of a photocatalyst on various supports (e.g., conductive glass, stainless steel, porous metal mesh or titanium foil [21]) has been proposed. Another attempt is based on application of membrane technology.

A membrane separation process (MS) is a physical technique not involving a phase change, allowing one to obtain a desired separation by operating in continuous mode.

By synergistically coupling MS with PC, it is possible to operate continuously in systems in which the recovery and reuse of the photocatalyst (immobilized or in suspension), the separation of substances from the treated solution and/or the recovery of the reaction products occur simultaneously. This synergic coupling takes place in a Photocatalytic Membrane Reactor (PMR), which can be defined as a device existing in various configurations and synergistically combining a photocatalyst and a membrane to produce chemical transformations [19]. Then the implementation and use of PMRs represents a promising approach in view of large-scale application of PC.

PMRs can be designed in two main configurations: (i) PMRs with photocatalyst suspended in the reaction mixture (slurry systems) and (ii) PMRs with photocatalyst immobilized in/on a substrate material acting as a membrane (photocatalytic membrane). Both configurations present specific advantages and limitations depending on the specific application [1,22,23,24].

PMRs with suspended photocatalyst can be further classified in two categories [23]. The first one is represented by the so-called integrative-type PMRs, in which the photocatalytic reaction and MS take place in one apparatus, i.e., an inorganic or polymeric membrane submerged in the slurry photocatalytic reactor [25,26]. The second category comprises the so-called split-type PMRs. Here the photocatalytic reaction and MS take place in two separate apparatuses, the photocatalysis and membrane modules, which are appropriately coupled [27,28,29].

PMRs with immobilized photocatalyst, indicated as PMRs with photocatalytic membrane (PM) [30], are intrinsically integrative-type; the photocatalytic and the membrane separation processes take place in the same unit.

Additionally, on the basis of a different approach, the PMRs can also be classified considering the position of the light source, which can be: (i) above/inside the feed tank; (ii) above/inside the membrane unit; and, (iii) above/inside an additional vessel placed between the feed tank and the membrane unit [24].

Because of the larger active surface area, which guarantees a good contact between the photocatalyst and the pollutants, it has been reported that PMRs with suspended photocatalyst make it possible to achieve higher efficiency when compared to that of the immobilized system [1,23,31]. As a consequence, the configurations with suspended catalyst have been largely used in literature [32,33,34]. However, light scattering due to suspended photocatalysts nanoparticles (NPs) and membrane fouling, due to the deposition of photocatalyst NPs on the membrane surface with the consequent flux decline, still limits the performance of this kind of PMRs [33,35].

In the immobilized PMRs, despite the contact between the photocatalyst and the substrate being hindered by the mass transfer limitation over the immobilized photocatalyst, the recovery, reuse and regeneration of the photocatalyst is easier than the other ones. Additionally, systems with PMs generally show better performance with respect to conventional membranes in terms of reduced membrane fouling and improved permeate quality.

In the literature there are various critical overviews on the recent progresses and new perspectives of PMRs [1,22,23,24]. The present work gives a critical overview of the recent advantages of PMRs in both organic synthesis and pollutant degradation with a particular focus on the scientific literature of the last 5 years.

2. Process Parameters Affecting the Performance of PMRs

The selection of the appropriate process parameters is crucial to obtain good performance of PMRs when finalized to practical applications. A brief description of these parameters, including photocatalyst concentration, photocatalyst type, reaction rate, temperature, pressure, operating pH, irradiation of photocatalyst, light intensity, typologies of light sources and position, presence of other species, type of membrane, and type of membrane reactor, is reported in the following. The influence of PMRs operation variables on membrane fouling is also summarized. More details can be found in Molinari et al. [20] and Mozia [24].

The photocatalyst amount is one of the main operating parameters that influences PMR performance. The rate of the reaction is proportional to the amount of photocatalyst by working under a true heterogeneous catalytic regime, but above a certain level of photocatalyst mass, the rate of reaction reaches a plateau condition. Furthermore, photocatalyst aggregation and light scattering phenomena also increase with photocatalyst amount, causing lower system efficiency. Therefore, the optimum amount of photocatalyst must be chosen.

The photocatalyst type is another important parameter, since it significantly affects the PMRs performance. Key properties when choosing a photocatalyst are the band gap energy, the chemical and physical stability, the non-toxic nature, the availability, and the cost [22]. TiO₂ photocatalyst is the most utilized in PMRs because of its strong photocatalytic activity, relatively low electron-hole recombination rate, high (photo)chemical stability, low cost and toxicity. However, this material is capable to use only less than about 5% of the solar energy. Thus, the development of photocatalysts able to use visible light represents a challenge in view of large-scale application of PMR systems. Other metal oxide semiconductors, characterized by a different band gap, are also used in PMRs (e.g., ZnO, WO₃, CuO/ZnO) [17].

The reaction rate usually increases with substrate concentration until all the active sites of the photocatalyst are occupied by substrate molecules (saturation conditions). Thus, it is important to maintain high concentration of the substrate to favor its adsorption on the photocatalyst with respect to the other chemicals contained in the reacting environment, including the desired product [20,36,37]. However, in some cases, it is desirable to work with low amounts of substrate, below the solubility limit, to avoid the formation of micro-droplets of the substrate.

Photocatalytic systems do not require heating and in most cases are operated at room temperature. In general, the optimal temperature for photocatalytic reactions is in the range of 20–80 °C. Temperatures higher than 80 °C are unfavorable for photocatalytic processes, because of the promotion of electron/hole recombination. Additionally, at high temperatures the exothermic adsorption of the reactant is disfavored and becomes the rate limiting step. At very low temperatures (below 0 °C) the photocatalytic process is limited since desorption of products from the photocatalyst surface becomes the rate limiting step [23]. For the temperature range of 20–60 °C, an enhancement of the photodecomposition rate with increasing the reaction temperature was observed [24].

The operating pressure can affect the photocatalytic performance when the reactants are in gaseous phase. In the case of pressure-driven membrane separation, the flow rate across the membrane is influenced by the transmembrane pressure, which is the driving force. The flow rate determines the contact time between the photocatalyst and the reagents/products, influencing both system reactivity and selectivity.

The operating pH largely influences two fundamental steps of the photocatalytic process: adsorption of the substrate and desorption of the desired product. For example, TiO₂, widely used as photocatalyst, has positively charged surface under acidic conditions, whereas in alkaline media, it is negatively charged [38,39]. Depending on the type of substrate, the pH can favor adsorption or desorption of the product reducing or increasing its degradation, thus causing a positive or negative effect on reaction rate [39]. Moreover, the operating pH influences the substrates and products permeation across the membrane [40].

The step required to initiate the photocatalytic process is the irradiation of the photocatalyst surface with photons having energy (hν) equal to or higher than Eg. Photons that have a wavelength (λ) smaller than or equal to the absorption edge of the photocatalyst activate it efficiently, especially at low light intensity. Instead, at high light intensity, a recombination of electron/hole pairs can occur [3,41,42]. In general, the reaction rate increases by increasing the light intensity in the range of 0–20 mW cm⁻² (low light intensity), since the formation of electron-hole couples is predominant with respect to the electron-hole recombination. By operating at approximately 25 mW cm⁻² (medium light intensity), the reaction rate increases with the square root of light intensity, because of the competition between the electron-hole formation and their recombination. The light intensity does not influence the reaction rate by performing the photocatalytic process at values higher than 25 mW cm⁻², because of the achievement of mass transfer limit.

Depending on the emission range of interest, different typologies of light sources are available, like xenon, mercury, and deuterium lamps. In recent years, laser induced photocatalysis has also been considered. Among the various types of lamps, LED ones, emitting in the UV or UV-VIS range, are gaining interest because of their efficiency and the possibility of also being powered by photovoltaic panels [22].

During the photocatalytic process, photons are consumed in the reactions as immaterial reagents [19]. Photons cannot be mechanically “mixed” as reactants. On the basis of this, the geometry of the reactor and irradiation system, as well as the optical properties of the reacting environment are of fundamental importance to obtain an efficient distribution of the radiation inside the reactor [19]. Split-type PMRs, where the photocatalytic reaction and membrane separation take place into two separate apparatuses, can be classified considering the position of the light source, which can be: (i) above/inside the feed tank; (ii) above/inside the membrane unit; and, (iii) above/inside an additional vessel placed between the feed tank and the membrane unit [14].

The rate of the photocatalytic process can also be influenced positively or negatively by the presence of other species in the reaction environment, e.g., they can prevent the charge carrier recombination or increase the by-products formation [43,44].

Another important parameter in the photocatalytic process is the membrane. The membrane can allow not only the easy recovery and reuse of the photocatalyst, immobilized on the membrane or just maintained in suspension, but also the selective separation of the desired product from the reaction mixture, contributing to limit side reactions [1]. In this last case it is important to choose a membrane characterized by a high permeability with respect to the desired product, permitting its prompt and efficient selective removal and recovery.

The performance of slurry PMRs, especially pressurized ones, is strictly associated with membrane fouling caused by photocatalyst particles. In case of PMRs equipped with photocatalytic membranes this phenomenon does not exist. Nevertheless, in every configuration fouling can be caused by contaminants present in feed. Factors influencing severity of fouling in PMRs include (i) type of membrane process (pressure driven or others); (ii) operating mode in case of pressure-driven membrane techniques (i.e., pressurized or depressurized); (iii) operating conditions (cross flow velocity, transmembrane pressure, aeration, back-flushing, etc.); (iv) photocatalyst type and loading (for slurry PMRs); and (v) feed composition [24,33,45,46,47,48,49,50]. The effect of these factors is usually determined experimentally in relation to PMR configuration and type of contaminants. The total effectiveness of PMR results from a synergistic effect of membrane separation and photocatalysis, regardless of the type of driving force of membrane process. A brief summary of the influence of various operation factors on the PMR performance is presented in

Table 1. The influence of various operation parameters on PMRs performance.

Operation Factors	Effects	Ref.
Type of membrane process	PMRs utilizing pressure driven membrane processes are more prone to fouling than systems with other driving forces (e.g., pervaporation, membrane distillation, dialysis).	[24]
Operating mode	PMRs utilizing pressure driven membrane techniques: (a) pressurized—tendency for membrane fouling caused by photocatalyst particles; (b) depressurized—less prone to fouling caused by photocatalyst particles.	[47] [45]
Operating conditions	Aeration: improvement of the process performance (fouling mitigation; increase of photodecomposition efficiency); however, at high aeration rates the bubble cloud can attenuate the UV/VIS light transmission in the photoreactor.	[45]
	Back-flushing: alleviation of membrane fouling caused by photocatalyst particles or in case of membranes with large pores no positive results.	[48] [47]
	Cross flow velocity (CFV): enhancement of membrane performance with increasing CFV as a result of higher shear rates prevents the deposition of photocatalyst cake layer.	[49]
	Transmembrane pressure (pressure driven membrane techniques): (a) Increase of transmembrane pressure (TMP) enhances deposition of photocatalyst particles on a membrane surface and leads to formation of a dense filtration cake; (b) Working below the critical flux value can prevent from the deposition of the filtration cake layer on the membrane surface.	[49] [47]
Photocatalyst	Under unfavorable process conditions, membrane fouling due to the presence of photocatalyst particles, especially in case of pressurized systems.	[47]
	Fouling mitigation due to decomposition of organic contaminants in case of both PMs and slurry PMRs.	[50]
	Higher photocatalyst loading can cause increase of permeate flux; however, at too high concentration the screening effect may attenuate the UV/VIS radiation transmission in the photoreactor.	[50]
Feed composition	Chemical nature of contaminants strictly affects their adsorption on photocatalyst particles thus influencing their removal efficiency and fouling severity.	[33] [46]
	Less severe membrane fouling at neutral and acidic pH compared to alkaline conditions.

. More details on this subject can be found elsewhere [24,33,45,46,47,48,49,50].

3. Types of PMRs

3.1. PMRs with Immobilized Photocatalyst (PMRs with PMs)

In this type of PMR configuration the photocatalyst is not solubilized/suspended in the reacting environment, but it is immobilized in/on the membrane, giving a photocatalytic membrane (PM). In such a system, the membrane acts both as a selective barrier for species involved in the reaction and as the support for the photocatalyst, thus the photocatalytic and the membrane separation processes take place in the same unit [22]. PMRs with immobilized photocatalyst can be operated in two different modes: dead-end and cross flow [51,52]. When the PMR is operated in dead-end mode and suspended particles are present in the aqueous solution, they are retained by the membrane and accumulate on its surface. The formed cake layer reduces the membrane permeability and the photocatalytic performance. Instead, in the cross-flow mode the feed flows tangentially to the membrane removing the particles deposited on its surface, thus resulting in a reduction of membrane fouling. The permeate flows perpendicularly across the membrane, while the retentate is usually recirculated in the feed tank. In the case of PMRs operated under cross-flow filtration mode, the photocatalytic efficiency increases by increasing the recirculation flow rate. This trend can be explained by considering the larger turbulence in the solution, which promotes the mass transfer from the bulk of the feed to the surface of the PM, while reducing the membrane fouling. Flow rate across the membrane depends on the applied driving force (e.g., transmembrane pressure in a pressure-driven membrane separation), membrane structure, and composition. It is a key parameter for the PMR because it determines the contact time between the photocatalyst and the reagents/products. Mass transfer of the reagents to the catalytic sites, and of the product away from them, should be fast enough in order to avoid reaction limitation, while the contact time of catalyst/reagent should be also appropriate to control reaction selectivity [22].

On the basis of the procedure of photocatalyst immobilization, PMs can be classified into three sub-categories, where the photocatalyst is: (i) coated on the PM, (ii) blended into the membrane matrix, and (iii) free-standing PM. When the photocatalyst is coated on the PM the membrane permeability can decrease. As reported by Horovitz et al. [53], coating of Al₂O₃ membrane with N-doped TiO₂ decreased its permeability by 50% and 12%, depending on membrane pore size. Other authors evidenced that membrane coating can increase the permeability by rendering the interface more hydrophilic [54]. In the second sub-category, the photocatalyst is entrapped in the polymeric matrix during the membrane preparation, reducing the possibility of photocatalyst leaching with respect to the PM prepared by photocatalyst coating. The PMs classified in the third sub-category are made of a pure photocatalyst. Thus, a further reduction of the possibility of photocatalyst leaching can be obtained.

PMs can be either made of polymeric (poly(vinylidene fluoride)—PVDF, polysulfone—PSF, polyethersulfone—PES, cellulose acetate—CA, etc.) or inorganic (TiO₂, Al₂O₃, stainless steel, carbon nanotubes (CNTs) or graphene oxide (GO)) material. In contrast to polymer membranes, ceramic membranes are more often used in PMRs with PMs due to the better resistance to UV light and oxidative species. Dzinun et al. [55] noticed that after 30 days of UV irradiation, a tensile strength of hollow fiber TiO₂/PVDF membranes significantly decreased, which shows a limited applicability of polymeric PMs.

Usually TiO₂ nanoparticles are used to form a photocatalytic layer on the membranes; however, recent literature shows that PMs modified with other photocatalysts, such as ZnO [56,57,58], Cu₂O [59], Fe₂O₃ [60], ZnIn₂S₄ [61], reduced graphene oxide with Bi₂WO₆ [62], Bi₁₂O₁₇Cl₂ with dopamine [63] and others are also effective in degradation of pollutants in water. Most of these photocatalysts increase the hydrophilicity of the membranes which leads to an increase of the permeate flux. However, modification with CeO₂ [64] was proved to increase membrane hydrophobicity and allow to achieve simultaneous pollutant degradation and oil-water separation.

There are numerous methods of PMs production, for example:

• Dip-coating [56,57,58,59,60,61,62,63,64,65];
• Phase inversion [30,59,66];
• Hydrothermal synthesis—filtration [58,59];
• Electrospinning and hydrothermal reaction [67,68];
• Physical deposition [69];
• Atomic layer deposition (ALD) [57,70,71] and chemical vapor deposition [58];
• Facile spray-deposition [64].

Figure 1 shows SEM images of PMs prepared by some of the above methods.

One of the most widely known techniques of the deposition of a photocatalyst layer onto the membrane is the dip-coating method (1). The method can be used for modification of both polymer and inorganic membranes. Alias et al. [65] used dip-coating for preparation of ceramic membranes (Figure 1a). First, they obtained ceramic membranes by phase inversion technique from a casting solution containing PES as a binder, N-methyl-2-pyrrolidone (NMP) as a solvent and kaolin powder as membrane forming component. The dry membranes were coated with a suspension containing poly(ethylene glycol) (PEG) and TiO₂ nanoparticles using a dip coater. The coated membranes were then dried and sintered. Alternatively, the membranes were first sintered and then dip-coated with TiO₂ suspension. Also, polymer membranes can be modified by the dip-coating approach. Kim et al. [56] functionalized polyurethane (PU) electrospun nanofiber membranes by dipping the nanofibers in an aqueous dopamine hydrochloride solution, washing with distilled water and then immersing in aqueous suspension of ZnO nanoparticles in a shaking incubator and drying.

Phase inversion technique (2) is used for the preparation of polymer membranes. This technique was applied, e.g., by Singh et al. [59] for the preparation of flat sheet PSF polymeric membranes. In this case, Cu₂O photocatalyst was dispersed in NMP, then PSF and PEG were added to the solution. After stirring and degassing the solution was casted over a glass plate using a casting knife and then immersed in a non-solvent (water) bath.

Hydrothermal method with filtration (3) was used by Rao et al. [60]. They prepared a photocatalytic membrane interconnecting TiO₂ nanowires, Fe₂O₃ nanoparticles and graphene oxide sheets. The flat sheet composite membrane was obtained by one-step blending method. TiO₂, Fe₂O₃ and GO were sonicated together in deionized water and then vacuum-filtered onto a glass-fiber filter paper. The formed membrane (with filter paper being a support) was dried in air and then hot pressed.

Another approach was based on a combination of electrospinning, DBD cold plasma treatment and hydrothermal treatment (4) to produce TiO₂-PVDF membranes. First, PVDF was dissolved in N, N-dimethylformamide (DMF) and acetone, then tetrabutyl titanate (TBT) was added into the prepared gel. The next step was electrospinning of the PVDF-TBT gel through a syringe with a gauge needle connected to a high voltage supply. After that, the dry membranes were treated with cold plasma and then placed into stainless steel autoclave and immersed in HCl and TBT. The hydrothermal synthesis was carried out at 90 °C for 9 h [67]. Other authors combined the electrospinning technique with the hydrothermal method (Figure 1b). Polyacrylonitrile (PAN) nanofibers were prepared by electrospinning technique and then carbonized at 500 °C to get CNFs. The CNFs were then immersed in the solution containing titanium isopropoxide, cetyltrimethyl ammonium bromide and water, ethylene glycol and urea and then transferred into autoclave and heated at 150 °C [68].

Physical deposition (5) was used by Kovács et al. [69] (Figure 1c). In this method, the TiO₂ suspension was simply filtered through the membrane in a dead-end stirring cell. The TiO₂ coating did not wash off during the washing with distilled water; however, the photocatalyst was not distributed uniformly.

Atomic layer deposition (6) was applied by Feng et al. [57] for preparation of inorganic PM. Pristine carbon nanotube membranes were put into ALD reactor chamber at 150 °C for 30 min in vacuum before deposition. The precursors (TiCl₄ for TiO₂ and (C₂H₅)₂Zn for ZnO deposition) were alternatively pulsed into the ALD reactor by the carrier gas. After that the samples were calcined at 600 °C.

De Filpo et al. [58] deposited ZnO and TiO₂ on PVDF membrane via chemical vapor deposition (Figure 1d). ZnO coating was formed by argon sputtering of ZnO target. The deposition of nanostructured TiO₂ was made by reactive sputtering using a Ti target in a gaseous mixture of argon and oxygen. The gas mixture reacted with the substrate and sputtered atoms forming a thin layer of TiO₂ onto the substrate.

Baig et al. [64] used membranes fabricated by facile spray-deposition (7) of CeO₂ nanoparticles dispersion in tetrahydrofuran (THF). The dispersion was sprayed onto stainless steel mesh membrane with spray gun inside a fume hood and then annealed at 200 °C. The prepared membrane was highly hydrophobic.

3.2. Slurry PMRs

Slurry PMRs combine the advantages of classical photoreactors (PRs) and membrane processes with a synergy of both technologies (e.g., chemical reactions and separation are performed in one step), thus minimizing the environmental and economic impact [1,15,20]. The membrane allows not only the easy recovery and reuse of the photocatalyst, but also the selective separation of the desired product from the reaction mixture. The function of the membrane is to set the interface for mass exchange between the phases [72]. Possible cases of slurry PMRs are: gas–liquid, liquid–liquid, and liquid–gas membrane contactors, where the first term indicates the phase of the feed and the second one indicates the phase of the permeate.

In a PMR with a solubilized or suspended photocatalyst, a membrane with appropriate molecular weight cut off (MWCO) can be used for the retention of the catalyst in the reactor [22]. Usually globular proteins in aqueous solution are used for MWCO measurements; however, in membrane selection it is necessary to consider that the MWCO depends on the solvent and the solution composition and polarity might change in relevant way during the reaction [73]. The catalyst retention can be optimized by enlargement of the catalyst, such as: dendrimers, hyperbrached polymers or catalyst bound to a soluble polymer [74]. Moreover, the membrane can have the multiple roles of confining, the catalyst, the substrate, and the intermediates into the reaction environment.

4. Photocatalytic Membrane Reactors (PMRs) in Organic Syntheses and Energy Storage

4.1. Photocatalytic Reductions

4.1.1. Applications of PMRs with PMs

Photocatalytic Conversion of CO₂

Today, global warming is considered one of the main environmental problems. For this reason, the scientific community dedicates massive efforts towards CO₂ reduction and/or valorization through a sustainable process because CO₂ largely contributes to the global climate change [75].

In industry, CO₂ is widely used for Sabatier reaction [76,77,78] and reverse water–gas shift (RWGS) reaction. Regarding the Sabatier reaction, CO₂ reacts with H₂ to form CH₄ and H₂O at 300–400 °C. As for the RWGS reaction, CO₂ firstly reacts with H₂ to form CO and H₂O, then CO can be easily converted into hydrocarbons by the Fischer-Tropsch reaction. However, both reactions require high temperature and high pressure to reduce CO₂ into hydrocarbons. Thus, another method to conduct CO₂ reduction is expected to be developed. Photo-hydrogenation of carbon dioxide (CO₂) is a green and promising technology and has received much attention recently. This technique could convert solar energy under ambient temperature and pressure into desirable and sustainable solar fuels, such as methanol (CH₃OH), methane (CH₄), and formic acid (HCOOH) [76].

In recent years the photocatalytic reduction of CO₂ has been widely studied by using different photocatalysts [79,80,81,82,83,84,85], despite the main drawbacks of this process, which include the low photoconversion efficiency due to thermodynamic limitations, and the occurrence of decomposition of the obtained products (like methanol, formic acid, formaldehyde) due to the higher reactivity of these compounds with respect to CO₂. The use of a membrane reactor where the photocatalyst is immobilized into a polymeric membrane can limit the above-mentioned problems because it allows the recovery of the catalyst and the control of the contact time between species in the solution and the catalyst.

Titanium dioxide (TiO₂) is one of the most widely studied photocatalyst material due to its wide availability, low cost, nontoxicity, and long-term stability [79,80,81,82,83,84]. However, the rapid recombination of the photogenerated electron-holes and low adsorption capacity for CO₂ of TiO₂-based photocatalysts, limits their catalytic efficiency toward CO₂ conversion. Consequently, there is an intense effort to develop alternative catalytic materials, with improved charge separation kinetics while providing higher adsorption capacity for CO₂, to facilitate the conversion process. Metal organic frameworks (MOFs) have recently gained attention as potential alternatives to TiO₂ catalysts, due to their high porosity and surface area, and good adsorption capacities for CO₂. [86,87,88,89]. Maina et al. [86] studied a membrane reactor utilizing semiconductor nanoparticles and MOF for CO₂ conversion. Semiconductor nanoparticles-doped ZIF-8 membrane reactors were fabricated using rapid thermal deposition (RTD), and their photocatalytic efficiency toward CO₂ conversion was investigated under UV light. The fabrication process involved the deposition of a primary pristine ZIF-8 membrane layer across a GO coated stainless steel substrate, after which a secondary thin film layer consisting of semiconductor nanoparticles dispersed within the ZIF-8 matrix was deposited. A thin film composed of catalytic semiconductor nanoparticles dispersed within the MOF matrix can be deposited on a preformed MOF membrane, in just about 15 min. As a proof of concept, they demonstrated that the TiO₂ and Cu−TiO₂ doped ZIF-8 membrane exhibited higher catalytic efficiency toward CO₂ conversion as compared to the pristine ZIF-8 membrane, and the product yield can be controlled depending on the composition and the dosage of the semiconductor nanoparticles.

Another type of material that can be used to synthesize membrane is a natural biopolymer such as chitosan. Zhao et al. [90] selected chitosan to prepare the membrane matrix due to its abundant amine and carboxyl groups, which are favorable for CO₂ adsorption. They prepared a CdS/NH₂-UiO-66 hybrid membrane and tested it in the photocatalytic conversion of CO₂. NH₂-UiO-66 was chosen as the auxiliary catalyst due to its good adsorption of CO₂ and photocatalytic activity for CO₂ reduction under visible light irradiation. The semiconductor/MOF hybrid membrane showed excellent photocatalytic activity compared to the CdS membrane, NH₂-UiO-66 membrane, and CdS/NH₂-UiO-66/ chitosan mixed powder, which was attributed to the higher adsorption of CO₂. Additionally, by incorporating MOFs and semiconductors into membranes, an enhancement in the activity for the CO₂ photocatalytic reduction reaction was obtained. This trend was due to the synergistic reaction between the semiconductor and MOF, which accelerated the transfer of electrons and inhibited the recombination of electron–hole pairs in the hybrid membrane.

Chen et al. [76] studied the photo-reduction of CO₂ to C1-feedstock in the gas, liquid, and gas-liquid phases, respectively, over 1 wt. % Pt/CuAlGaO₄ photocatalyst under different H₂ partial pressures. Interesting results were observed in the gas-liquid phase. The authors reported that the increase of the partial pressure of H₂ could improve the yield of products. However, they observed that H₂ might compete with CO₂ for occupying the active sites when an extra amount of H₂ was provided, giving an adverse effect on CO₂ photo-hydrogenation. CH₃OH is the major product in the case of liquid and gas-liquid phases, while CH₄ is the only product for the gas phase. The quantum efficiencies in gas, liquid, and gas-liquid phases under 0.01 atm of H₂ were about 0.0001%, 0.0005%, and 0.0011%, respectively. The maximum total hydrocarbons amount (8.302 µmol g⁻¹) was achieved by operating in the gas-liquid phase.

The reduction of CO₂ emissions is a desirable solution but not achievable in the short term. Another approach is the reuse of emitted CO₂ as a raw feedstock to promote value-added products such as formic acid or fuels. Very recently, Pomilla et al. [75] studied the CO₂ conversion into liquid fuels such as methanol and ethanol by using a catalytic membrane. Exfoliated C₃N₄ was incorporated into a Nafion matrix. The such obtained photocatalytic membrane was used in a continuous photocatalytic reactor for converting CO₂ into liquid fuels by using H₂O as reducing agent. The photocatalytic tests were conducted in a PMR (Figure 2) by irradiating the membrane with a medium−high mercury vapor pressure lamp (Zs lamp, Helios Ital quartz, Milan, Italy) emitting in the range from 360 nm (UV-VIS) to 600 nm. The flat sheet PM was assembled in a stainless-steel module equipped with a quartz window allowing UV−VIS irradiation. The membrane module was continuously fed with CO₂ and H₂O by means of a mass flow controller and an HPLC pump, respectively.

The membrane reactor mainly consists of three parts: the feed/retentate and the permeate chambers and the catalyst-loaded membrane. The two reactor chambers can be considered as systems with lumped parameters. The reaction takes place in the membrane layer, and no concentration gradient of any chemical species is expected, also because of the low conversion. Thus, the retentate and permeate streams possess the same composition of these two volumes. The reactor was placed in a vertical position to facilitate permeate and retentate sampling, as well as to avoid the formation of stagnant zones. The authors reported that the photocatalytic membrane reactor converted at least 10 times more carbon than the batch system, because of the better dispersion of the photocatalyst which was embedded in the Nafion matrix. Alcohol production was promoted by the low contact time obtained thanks to the fast removal of the reaction mixture from the reacting volume, which limited oxidation and/or secondary reactions. On the contrary, the slow removal caused a partial oxidation of MeOH and EtOH, favoring HCHO production. In all cases, a water defect corresponded to a larger HCHO production, reaching a flow rate of 27 μmol g_catalyst⁻¹ h⁻¹ at a H₂O/CO₂ feed molar ratio equal to 0.5. The highest MeOH and EtOH selectivity were 54.6% and 45.4%, respectively, at H₂O/CO₂ feed molar ratio equal to 5 and contact time of 2 s. Alcohol production rate of 32.8 μmol g_catalyst⁻¹ h⁻¹ was obtained at the best operating conditions. It has to be emphasized that liquid fuel production by CO₂ reduction may be a useful way to create valuable products, but it is not an impactful strategy for reducing emissions.

Although significant works using pure and modified C₃N₄ as an efficient potential photocatalyst for CO₂ reduction have been reported, tests on photocatalyst stability and controls in the absence of CO₂ are usually not presented [91,92,93]. Some authors report that the photocatalysts maintained a percentage of stability after several photocatalytic runs [94]. When the g-C₃N₄ stability is evaluated, the changes in product yield with time are observed, wherein the activity is reported to be constant with reuse [95,96,97]. In a recent work, Xiao et al. [98] studied the g-C₃N₄ stability in the presence of water pollutants. An accumulation of carbon species (by total organic carbon analysis) and NO₃⁻ ions in water under photocatalytic reactions (sunlight/O₂/nanostructured C₃N₄ conditions) was found, suggesting chemical instability of g-C₃N₄ in the aqueous phase. Pomilla et al. [99] studied the changes in photocatalytic activity as a function of time and reuse cycle during the photocatalytic reduction of CO₂ in the gas phase by using water as the reducing agent and g-C₃N₄ or an oxygen-doped g-C₃N₄ as photocatalysts. The main detected product was CO. Formation of other products was not observed. The authors found that the pure material was more active than the oxygen-modified analogue for CO production. A reduction in the rate of CO generation under irradiation was observed during both single runs and repeated test cycles with the same photocatalyst. During control tests in the absence of CO₂, the production of CO was observed under irradiation, raising questions about the origin of CO and the stability of g-C₃N₄. The authors reported that the photoexcitation led to oxidation of C₃N₄ rather than water splitting, resulting in a loss of photocatalytic activity, i.e., photocorrosion. Since the valence band edge of g-C₃N₄ is close to the water oxidation potential, this material has limited overpotential to drive water oxidation and, as such, may accumulate intermediate Reactive Oxygen Species (ROS) that can oxidize the catalyst, changing its activity as observed. C₃N₄ may still be useful as a photocatalyst but the use of a Z-scheme or a second co-catalyst material on which the oxidation reaction performs may be necessary to avoid g-C₃N₄ oxidation. Many researchers use a Z-scheme consisting of g-C₃N₄ in composite materials such as g-C₃N₄−TiO₂ where the latter material realizes the water oxidations, thus decreasing the probability of g-C₃N₄ oxidation.

Hydrogen Production

The conversion of solar energy into a fuel such us hydrogen by photocatalytic water splitting is considered a promising approach to meet the growing global energy demands in a sustainable way [5,100]. Recently, the attention to develop PMRs combining photocatalytic water splitting with membrane separation [5] has been increased. Bipolar membranes (BPMs) have been used for this purpose since they are efficient separators in water-splitting cells [100,101]. BPMs consist of an anion-exchange membrane (AM) and a cation-exchange membrane (CM), that are permselective to anions and to cations, respectively, with an interfacial layer between them [102]. Catalysts have been included in the BPM interlayer to facilitate water dissociation, to reduce overpotential, and improve energy efficiency [103]. Inorganic nanomaterials are considered important catalysts to facilitate water dissociation. It is foreseen that these systems can help in planning new configurations of PMRs.

In recent years, graphene quantum dots (GQDs) have attracted increasing attention because of their high quantum effect and large specific surface areas [104]. Liu et al. [100] reported the fabrication of a sandwich GQDs–Cu₂O/BPM with GQDs–Cu₂O catalyst inside the interlayer. The authors observed that GQDs–Cu₂O decreased membrane impedances under sunlight irradiation and that GQDs–Cu₂O/BPM minimized pH gradient formation.

The most widely used semiconductor for photocatalysis is TiO₂. However, it is well-known that the application of TiO₂ as photocatalyst is strongly influenced by the transport and recombination of photogenerated electrons and holes. Indeed, it is photoactive only in the UV range [1]. To address these problems and to improve the photocatalytic activity, several strategies have been adopted such as the introduction of doping ions. Recently, there is a growing attention in fabricating nanostructured semiconductors for photocatalysis, including nanosheet, nanotube, and nanorod, because their special geometries provide large surface area, small lateral diffusion resistance, and low reflectivity compared to their bulk phase [105]. A large variety of fabrication routes for TiO₂ nanotubes such as hydrothermal, anodization, and template methods have been proposed by several authors [106]. Recently, Hattori et al. developed an all-electrochemical technique for fabricating a bilayer structure made of a titanium dioxide (TiO₂) nanotube array (TNA) and a palladium film (TNA/Pd membrane). This system was effective for the photocatalytic production of high-purity hydrogen [106]. Electroless plating was used for depositing the Pd film on the TNA surface prepared by anodizing a titanium foil. This ultrathin membrane with sufficient mechanical robustness showed photocatalytic H₂ production from water/methanol mixture under ultraviolet illumination on the TNA side, immediately followed by the purification of the generated H₂ gas through the Pd layer.

Very recently, Su et al. [107] have reported the fabrication of Al- and Zn-doped TiO₂ nanotubes by atomic layer deposition combined with polycarbonate (PC) membrane as the template [108,109]. This method allows control of the pore size and doping level. The bilayers were alternately deposited on the PC membrane template by ALD with various cyclic sequences. Increase of charge recombination and decrease of hydroxide radical formation, which reduced the photocatalytic activity of TiO₂, were observed by increasing the Al doping. In contrast, the photocatalytic activity was enhanced at Zn doping ratio of 0.01, obtaining a hydrogen production rate from water splitting 6 times higher than that of commercial P25 TiO₂. The energy-band diagram of Zn-doped TiO₂ determined by ultraviolet photoelectron spectroscopy revealed shift up of the Fermi level to provide more electrons to the conduction band. Photoinduced trapped electrons and holes were detected in Zn-doped TiO₂ by in situ electron paramagnetic resonance spectroscopy, revealing that Ti³⁺ sites on the surface and surface oxygen vacancies played a key role in promoting the photocatalytic process.

4.1.2. Applications of Slurry PMRs

Conversion of Acetophenone to Phenylethanol

Only few works have appeared till now on the use of a PMR in reduction reactions such as photocatalytic transfer hydrogenation of acetophenone (AP) [37,110]. In recent years [37], exploitation of photocatalytic reactions in the design of efficient and environment-friendly technologies has attracted great attention. Among the main chemical transformations, currently applied in industrial practices, reduction processes, such as the reduction of carbonyl compounds in the corresponding alcohols, play an important role in organic synthesis [16,111,112]. In particular, acetophenone hydrogenation has been widely studied since the resultant reduction product, 1-phenylethyl alcohol named also phenyl ethanol (PE), is a common precursor for the preparation of analgesic and anti-inflammatory drugs as well as fragrances and perfumes [113]. Recently, Molinari et al. [37] studied the photocatalytic membrane reactor to perform the photocatalytic hydrogenation of acetophenone to phenylethanol using water as solvent, formic acid as hydrogen and electron donor to develop a green and sustainable process, using commercial TiO₂ and homemade Pd/TiO₂ photocatalysts under UV and visible light. Photocatalytic tests were performed in the PMR schematized in Figure 3, consisting of a biphasic membrane contactor coupled with the batch photoreactor [37]. The volume of recirculating solution was 725 mL. The aqueous solution was delivered from the photoreactor to the membrane contactor with a peristaltic pump. Then, the aqueous phase returned in the photoreactor under the action of gravity. The membrane contactor, immersed in a thermostatic bath maintained at the same temperature as the photoreactor, constituted of two compartment cells (each one with a volume of 125 mL) separated by a flat sheet polypropylene membrane with an exposed membrane surface area of 28.3 cm². The first compartment contained the aqueous phase coming from the photoreactor, while the second one contained acetophenone, acting as both the organic extracting phase and the substrate reservoir, maintained under mechanical stirring. The phenylethanol produced in the aqueous phase diffused through the membrane and then dissolved into the organic extracting phase, where it was preserved from successive over-hydrogenation, with an enhancement of process selectivity. The use of a membrane extractor instead of a traditional liquid/liquid extraction unit avoided the physical mixing between the organic and the aqueous phases, and allowed a continuous extraction process. Samples were taken from the organic phases every hour for analyses.

Different methods for the substrate addition in the membrane photoreactor were tested, finding the best performance when the acetophenone was used as both solvent and reactant (substrate). An improvement of the efficiency of the photocatalytic reaction compared to a batch reactor (productivity 4.44 mg g ⁻¹ h ⁻¹ vs. 2.96 mg g ⁻¹ h ⁻¹) was obtained thanks to the extraction of phenylethanol in the organic extracting phase, simultaneously to the reaction, thus enhancing the process selectivity. The photocatalytic activity in the PMR under visible light irradiation was improved five times by a simple doping of the commercial TiO₂ photocatalyst with Pd, using deposition precipitation method (productivity 22.0 mg g⁻¹ h⁻¹). This PMR can be classified as an L-L membrane contactor because the separation is given by the distribution of a component between two immiscible phases, where the solubility of that component (phenylethanol) is different.

Very recently, Lavorato et al. [110] evaluated the photocatalytic properties of TiO₂-loaded faujasite (FAU) zeolite and Pd/TiO₂/FAU in the heterogeneous transfer hydrogenation of AP under UV and visible light in batch tests and in a membrane reactor. Various samples of FAU crystals were prepared by loading different amount of TiO₂ and using bare and suspended FAU crystals. The FAU TiO₂ samples were also loaded with Pd and then they were characterized by transmission electron microscopy (TEM), N₂ adsorption-desorption at 77 K for specific surface (BET) area determination and atomic absorption spectroscopy (AAS). The activity of synthesized photocatalysts was screened under visible and UV light in batch tests and then the best samples were tested in the photocatalytic membrane reactor evaluating productivity, produced phenyl ethanol and extracted phenyl ethanol in the organic phase. The authors reported that the incorporation of Pd into a TiO₂-functionalized faujasite framework was performed with the double aim of extending the catalyst activity in the visible light region and enhancing the absorption properties of the catalyst. While the transition metal improves visible light utilization, the zeolite matrix contributes to delay charges recombination by a mechanism of electron hopping within the framework [114]. In addition, the zeolite acts as support for both TiO₂ and Pd metal particles, which were homogeneously dispersed in the matrix. The final properties of the catalyst were strictly related both to the initial amount of TiO₂ precursor used in the synthetic procedure and the presence of solvent during the ion exchange process. A highly homogeneous dispersion of TiO₂ catalyst particles on the zeolite surface was obtained in the sample TF10P, prepared by using the lowest amount of TiO₂ precursor and isopropanol as solvent. The highest BET area and porous volume values were also obtained for this sample. Accordingly, the TF10P sample showed the best performance in the batch photoreactor under UV light. It is noteworthy that the synergic coupling of TiO₂ and FAU allowed to obtain a homogeneous and stable dispersion of the photocatalyst on the support FAU particles, a more effective utilization of light and the efficient charge-transfer to the substrate molecules. The results showed that Pd doping of TF10P sample gave the photocatalyst Pd_TF10P active in the visible light.

The productivity obtained in the membrane reactor was higher in this work [110] compared to a previous work [35] of the same authors (productivity 99.6 mg g_TiO2 ⁻¹ h⁻¹ vs. 22 mg g_TiO2 ⁻¹ h⁻¹). The photocatalytic membrane reactor had an extraction percentage of phenylethanol from the reactive phase to the organic phase of ca. 25%.

Hydrogen Production

Very recently, Marino et al. [115] described the one-step hydrogen and oxygen evolution through a PMR which mimics the Z-scheme mechanism by using Au/CeO₂ as photocatalyst leading to O₂ formation via water oxidation, and Au/TiO₂ as photocatalyst which promotes H₂ formation via water reduction (Figure 4). The aqueous suspensions containing the two photocatalysts were separated by a membrane able to transport electrons via a redox couple (Fe³⁺/Fe²⁺) acting as an electron redox mediator.

The best results in terms of hydrogen and oxygen evolution by using different amount of initial Fe³⁺ were obtained at 5 mM concentration. Under irradiation with visible light, hydrogen and oxygen were produced in stoichiometric amounts. The optimal percentage of Au-loading on titanium dioxide for hydrogen and oxygen generation was 0.25 wt. %.

4.2. Photocatalytic Oxidations

4.2.1. Applications of PMRs with PMs

A universal industrial reaction to obtain important intermediates to fabricate nylon-6 and nylon-66 polymers is the selective oxidation of cyclohexane to cyclohexanol (A) and cyclohexanone (K) (the mixture of the above is called KA oil) [37,110]. Despite more than 90% KA oil having been produced in the way of cyclohexane selective oxidation, the conversion efficiency of this process is lower than 10% when the selectivity of KA oil is higher than 80% [37,111]. Photocatalytic cyclohexane oxidation with the use of clean and cheap molecular oxygen as the oxidant and solar light as the driving force is especially attractive in green chemical reaction [16,112,113,114,115]. Zhao et al. studied [116] the C-doped Cr₂O₃/NaY composite membrane supported on stainless steel mesh for photocatalytic activity for cyclohexane oxidation. They prepared C-doped Cr₂O₃ photocatalyst on the upper surface of NaY zeolite membrane. This composite membrane included three layers with different functions: (i) the stainless-steel mesh, used as supported carrier, was at the bottom, (ii) the NaY membrane, used as adsorbent, was in the middle, (iii) and C-doped Cr₂O₃ used as photocatalyst was on the top. The C-doped Cr₂O₃ photocatalyst was prepared using a chromium-containing MOF as precursor. The results showed that comparing the photocatalytic activity of C-doped Cr₂O₃ and C-doped Cr₂O₃/NaY powders over the composite membrane, the production rate of KA oil was substantially enhanced. This was caused by the adsorption of the products on the NaY membrane with high surface areas and polarity. It was found that 5 h was the optimum catalytic time for the reaction system. Under the optimum conditions, the conversion efficiency of cyclohexane was 0.93%, and the selectivity to KA oil was up to 99.73%. Based on the above photocatalytic results the authors proposed a mechanism of cyclohexane photocatalytic oxidation on the layered composite membrane showing a cooperation of the adsorbent (NaY membrane) and the photocatalyst (C-doped Cr₂O₃). In the photocatalytic cyclohexane oxidation process, the conduction band of C-doped Cr₂O₃ that is −0.57 V versus NHE [117] is more negative than the O₂/O₂^−• (−0.13 V vs. NHE) [118]. So the dissolved O₂ can be captured and superoxide anion (O₂^−•) is obtained. The superoxide anion (O₂^−•) can react with cyclohexyl radical (C₆H₁₁^•) to obtain cyclohexyl hydroperoxide (C₆H₁₁OOH), which is an intermediate for cyclohexanone production. NaY zeolite membrane with high specific surface area, suitable pore size, uniform pore size distribution and polarity is used to capture the product and avoid its over-oxidation improving the yield of KA oil.

4.2.2. Applications of Slurry PMRs

In the following, a broad description on the conversion of ferulic acid to vanillin is reported. For another, less recent system, the conversion of benzene to phenol, only a brief description is given.

The exceptionally widespread utilization of vanillin in the food, cosmetic, pharmaceutical, nutraceutical and fine chemical industries makes it one of the most important aromas and justifies the very large volume of research on the improvement of the production processes [119]. Currently this aromatic aldehyde is mostly synthesized by chemical routes, [1,117] but in the last 10 years, some authors [119,120,121,122], proposed a photocatalytic synthesis of vanillin by the reaction of natural substances. However, vanillin is easily degraded into other chemicals, thus, to limit this problem and enhancing the yield of the aromatic aldehyde, some authors [40,120,121,122,123,124,125] used a pervaporation photocatalytic reactor (PVPR), coupling the reaction with a membrane separation. Pervaporation (PV) is a membrane separation process in which the target compounds are transported from a liquid retentate to a vapor permeate selectively through a non-porous membrane [126]. When photocatalysis is coupled with pervaporation or, also, dialysis, the recovery of the product of the partial oxidation of the aromatic alcohol prevents its degradation, thus enhancing the rate, the selectivity and the yield.

Böddeker and coworkers [124,125] demonstrated that membranes from polyether-block-amide (PEBA) are very suitable for pervaporating low volatile aromatics [124] and, in particular, vanillin [125]. Brazinha et al. [127] analyzed the pervaporation of vanillin and ferulic acid with polyoctylmethylsiloxane (POMS) membranes, giving an explanation to the high separation factor of vanillin with respect to ferulic acid. Augugliaro et al. [119] reported the photocatalytic oxidation of trans-ferulic acid, isoeugenol, eugenol or vanillyl alcohol to produce vanillin in aqueous medium by using different TiO₂ samples as photocatalysts. Photocatalytic tests were carried out in two different batch systems: a cylindrical and an annular photoreactor. The authors reported a selectivity to vanillin ranging from 1.4 to 21 mol % with a transmembrane flux about 3.31 g h⁻¹ m⁻² during the photo-oxidation tests conducted at room temperature and the complete removal of the heterogeneous photocatalyst using a non-porous PEBAX^® 2533 membrane. Vanillin vapors were recovered as crystals with a high degree of purity (≥99.8%) by downstream freezing in a liquid nitrogen trap. Camera-Roda et al. [122] studied the AROMA (Advanced Recovery and Oxidation Method for Aldehydes) process, demonstrating that the yield of the production of aromatic aldehydes and, specifically, of vanillin [121] can be enhanced by the coupling of pervaporation with the photocatalytic reaction [128]. The results obtained in the integrated photocatalytic/PV system, compared with the ones obtained operating in batch without the membrane, evidenced that coupling the photocatalytic oxidation with the PV unit permitted to obtain enhanced yield (3.9% vs. 6%) and conversion (22% vs. 35%). The main advantages obtained by the authors are the higher purity of the recovered aldehyde and the complete removal of the heterogeneous photocatalyst. In a more recent work, Camera-Roda et al. [40] improved the membrane performances in the pervaporation reactor to enhance the vanillin yield. On the basis of the results obtained by process simulation they investigated the methods to enhance the pervaporation performances (enrichment factors and permeate flux). The apparatus used for the photocatalytic synthesis of vanillin is reported in Figure 5. The vanillin pervaporation was carried out using PEBAX membranes.

The authors reported that increasing the membrane thickness enhances the enrichment factor of vanillin, since the resistance to vanillin permeation remains low, while the resistance to water permeation increases. An improvement of the enrichment factor can also be obtained by raising the temperature, with the additional positive effect of increasing the vanillin flux. The pH has a minor influence on the rejection of the substrate, which remains high also at low pH, when the substrate in solution is not dissociated.

Another important oxidation reaction widely studied in photocatalysis is the direct benzene oxidation to phenol. Phenol is an important chemical intermediate for the synthesis of petrochemicals, agrochemicals, and plastics [129,130]. Phenol production is obtained mainly by the three-step cumene process. Among the various routes, phenol production by one-step direct benzene oxidation represents an attractive alternative pathway, and many studies have been performed in recent years with the aim to develop more efficient and environmentally benign processes [1,19,34]. The photocatalytic approach is very interesting because it is a “green process” where light and a photocatalyst are used to generate OH^• radicals to oxidize benzene [20,131], but this reaction is a little selective, because phenol is more reactive than benzene, and by-products can be formed [1,20,132]. To limit this problem, some authors have used a PMR with the suspended photocatalyst to perform phenol separation (with high phenol permeability and complete rejection of the photocatalyst) simultaneously to the photocatalytic reaction [36].

5. Photocatalytic Membrane Reactors (PMRs) in Environmental Applications

5.1. Applications of PMRs with PMs

In PMRs with PMs the degradation of the pollutants takes place on the surface or inside the pores of a membrane, while the reagents pass through the membrane. Due to this fact, the membrane is the irradiated element in this configuration. Therefore, it is important to use a membrane material which is resistant to destruction by irradiation and the action of the oxidative species (mainly hydroxyl radicals) [24].

There are two main configurations of PMRs with PMs. In the first one, the photoactive layer is the separation layer, while the porous support is inactive. In the second configuration, the photocatalyst is placed in the porous support layer, and the separation layer is inactive. In both of these configurations the light source must be placed on the side of the photoactive layer, which means, respectively, on the side of the feed, and the permeate [133].

In the PMR configuration, where a PM has a photoactive support layer and the light source is placed on the permeate side the most common membrane material is ceramics. In reference [53] the authors used a PM made of α-Al₂O₃ and TiO₂ photocatalyst modified with nitrogen for the degradation of carbamazepine. Figure 6 shows the PMR used in this study. A quartz cover was placed above the photocatalytic membrane to seal the membrane chamber. The feed solution was filtered from the bottom up to the TiO₂-coated side of the membrane, which means that only permeate was exposed to the irradiation from the solar simulator (300 W ozone-free xenon arc lamp). Treated water was collected on the permeate side and pumped to the feed bottle. The main advantage of this configuration was that the photocatalytic layer of the membrane was not exposed to fouling, and thus, the effectiveness of the degradation did not decrease in time. It should be also mentioned that since the treated stream is permeate, the substances exposed to the degradation are smaller than the membrane pores.

A more common configuration is the one utilizing a PM with photoactive separation layer. This configuration was tested by Pastrana-Martínez et al. [134]. The authors prepared flat sheet cellulose membranes modified with TiO₂ and graphene oxide. The PMR installation shown in Figure 7 consisted of cylindrical glass reactor with the photocatalytic membrane and a medium-pressure mercury vapor lamp. Before the UV irradiation was started the solution was passing through the reactor for the time needed to saturate the membrane with tested pollutant. When the experiments were conducted in the dark, the permeate flux was lower than in case of the presence of UV light, which was linked with better hydrophilicity of the membranes and effective degradation of foulants.

The mitigation of membrane fouling in the case of PMs results from their self-cleaning properties. The group of Lv et al. [135] fabricated a self-cleaning membrane using a mussel-inspired method. The membrane consisted of a polydopamine (PDA)/polyethyleneimine (PEI) intermediate layer casted on a UF membrane support and a photocatalytic layer made of β-FeOOH nanorods. The prepared membranes showed high photocatalytic efficiency towards dye degradation under visible light and in the presence of hydrogen peroxide. Also, the dynamic water contact angle decreased from 60° to 20°, indicating enhanced wettability. During experiments realized in the presence of irradiation, the authors observed a slight decrease of permeate flux in the initial 2 h, while it recovered to nearly the original value after 6 h of filtration. In contrast, in the absence of VIS light, a 35% flux decline was reported after 6 h of the process due to membrane fouling by model dye.

The literature reports on PMs refer mainly to the removal of organic contaminants from water and wastewater, although some other environmental applications have been also proposed. They include disinfection of water [136,137,138,139], removal of heavy metals from water [140,141] and purification of gases [68,142].

There are several reports on the removal of dyes, such as methylene blue (MB) [57,64,143], rhodamine B (RhB) [144] and eosin yellow [30]; pharmaceuticals, including ibuprofen [59], diclofenac [58], carbamazepine [53] and tetracycline [61]; or other pollutants, e.g., phenol [66], humic acids [60,65], and pesticides such as diuron and chlorfenvinphos [145]. There are also reports on using PMs for purification of simulated brackish water and seawater from methyl orange and diphenhydramine [134], as well as secondary effluents from pharmaceuticals [146]. The summary of the mentioned photocatalytic degradation experiments is shown in

Table 2. Selected examples of application of PMs in removal of organic contaminants from water.

Pollutant	Membrane	Parameters	Photodegradation Efficiency	Ref.
Methylene blue	atomic-layer-deposited carbon nanotube ZnO-TiO2;	C0 = 20 mg L−1, UV irradiation	94% in 100 min	[57]
	hydrophobic CeO2 coated stainless steel membrane;	C0 = 10 mg L−1 in oil-water mixture, UV irradiation	>99% in 80 min	[64]
	PVDF/GO/ZnO nanocomposites membrane	C0 = 10 mg L−1, xenon light irradiation	86.84% in 100 min	[143]
Rhodamine B	g-C3N4 NT/rGO NF membrane	C0 = 10 mg L−1, VIS irradiation	>98% in 300 min	[144]
Eosin yellow	PSF/N,Pd co-doped TiO2 composite membrane	C0 = 100 mg L−1, VIS irradiation	>97% in 4 h	[30]
Ibuprofen	Cu2O modified PSF UF membrane	C0 = 10 mg L−1, VIS irradiation	up to 86%	[59]
Diclofenac	ZnO and TiO2 sputtered membranes	C0 = 9.3×10−5 M, UV irradiation	almost complete in 6 h	[58]
Carbamazepine	N-doped TiO2 coated PM	C0 = 1 g L−1, UV and VIS light	90% higher reaction rate	[53]
Tetracycline	ZnIn2S4 coated PVDF membrane	C0 = 100 µg L−1, VIS irradiation	92% in 36 h	[61]
Phenol	PES/O-g-C3N4 hybrid membrane	C0 = 10 mg L−1, UV and VIS irradiation	35.8% in 300 min	[66]
Humic acids	TiO2 nanowires/Fe2O3 nanoparticles/GO sheets membrane;	C0 = 25 mg L−1, solar radiation	92% in 12 h	[60]
	ceramic membranes coated with TiO2	C0 = 0.2 g L−1, UV irradiation	98.56% removal	[65]
Diuron Chlorfenvinphos	magnetron sputtered TiO2 on ceramic membrane	C0 = 1 mg L−1,	75% in 3 h	[145]
		solar radiation	78% in 3 h

.

The disinfection of water by PMs was examined by Guo et al. [147]. They described virus bacteriophage P22 removal in a PMR system with commercial ceramic tubular membrane (nominal pore size of 0.8 μm). The permeate side of the membrane was coated with TiO₂ P25 photocatalyst by dip-coating method. In the presence of the PM, the virus log removal of 5.0 ± 0.7 was found, which was significantly higher than observed for the simple UV disinfection and MF realized in series (2.0 ± 0.5) or together but in the absence of the TiO₂ coating on the membrane (2.4 ± 0.2). Due to the fact that the permeate side was irradiated, the proposed configuration could be possibly applied for treatment of highly turbid water which cannot be disinfected in other ways [147].

Rodríguez-Chueca et al. [136] applied porous stainless-steel MF membranes (pore size of 0.2 and 0.5 µm) coated with TiO₂ for removal of Escherichia coli, Enterococcus faecalis, and Candida albicans from synthetic wastewater. The highest microbial removal in the presence of UVC irradiation was observed for the lowest transmembrane pressure (0.05 MPa) applied. The immobilization of TiO₂ on the membrane improved both the inactivation and separation efficiency. The photocatalytic disinfection was also examined by Horovitz et al. [137] using N-doped TiO₂-coated Al₂O₃ PM for removal of MS2 bacteriophage. Virus removal from natural surface water under solar UV-VIS irradiation was 4.9 ± 0.1 log (>99.99%) The authors concluded that removal of MS2 was driven by electrostatic forces in addition to photocatalytic inactivation.

Photocatalytic membranes were also used for removal of heavy metals from water. Kazemi et al. [140] examined the photocatalytic reduction of Cr(VI) with commercial thin film composite (TFC) membrane coated using layer by layer (LBL) approach with chitosan and photocatalytic nanoparticles composed of nano-zerovalent iron (nZVI) and TiO₂. A significant decrease of Cr(VI) concentration in both retentate and permeate was observed. Moreover, the membrane exhibited high water flux and good antifouling properties. The removal of Cr(VI) was also studied by Wang et al. [141], who used algae decorated TiO₂/Ag chitosan hybrid nanofiber membrane. The addition of algae, which underwent photodegradation, significantly improved the photoreduction of Cr(VI) under visible light irradiation.

Another direction of PM environmental applications refers to reactions conducted in gaseous phase. Zhang et al. [68] applied TiO₂ nanowire/metal organic framework (MOF)/carbon nanofiber (CNF) membranes for the degradation of H₂S in wet air. The photocatalytic oxidation of H₂S was conducted under UV irradiation. The authors found that MOF coated on the surface of the TiO₂ nanowires significantly improved the photocatalytic activity of the hybrid membrane by increasing its adsorption ability towards H₂S. When the membrane without MOF addition was applied the H₂S conversion ratio was 91.2%, while in the presence of MOF it reached 93.5%. Moreover, the removal of the model compound was faster in the presence of MOF coating.

The degradation of volatile organic compounds, such as acetaldehyde and methanol in the presence of PVDF electrospun membranes modified with TiO₂, Ag₂CO₃ and GO was investigated by Roso et al. [142]. A complete degradation of acetaldehyde was obtained within 15 min (for an initial concentration of 1600 ppm) and methanol within 40 min (for 600 ppm) of visible light irradiation using PVDF/TiO₂–Ag₂CO₃–GO system. The authors explained the high photoactivity by heterojunction formed between Ag₂CO₂ and TiO₂, which effectively separated the charge carriers.

One of advantages of PMRs with photocatalytic membranes is that there is no need to separate the photocatalyst from the reaction environment. Moreover, the photocatalyst usually increases the hydrophilic properties of the membrane, thus improving the fouling resistance. The photodegradation of the pollutants can be made both on the feed and on the permeate side of the membrane, which gives a flexibility in process design. However, the main drawback of the application of photocatalytic membranes is that the effectiveness of the pollutant degradation is usually lower than in the case of the suspended photocatalyst. It is also not possible to adjust the quantity of the photocatalyst to the specific reaction. When the photocatalyst loses its activity, it is necessary to exchange the whole membrane. Furthermore, polymer membranes might be damaged by UV light or the action of oxidative species. Thus, more stable but also more expensive ceramic membranes should be applied.

5.2. PMRs with Photocatalyst in a Slurry: System Configurations and Applications

In recent years, significant attention has been directed towards PMRs coupling photocatalysis in suspension and micro-(MF) or ultrafiltration (UF) processes [45,148], while only a few reports refer to PMRs with nanofiltration (NF) [32], reverse osmosis (RO) [149] or membrane distillation (MD) [150].

In PMRs utilizing MF/UF, both pressurized and depressurized (i.e., with submerged membranes) operation modes are applied. The main role of the membrane in these systems is separation of a photocatalyst from the treated solution, while the membrane is usually not able to reject low molecular weight compounds and photodegradation by-products, causing the permeate quality to be rather low [24]. Moreover, these systems, especially pressurized ones, are prone to membrane fouling. Hence, the research is mainly focused on the examination of influence of process parameters in relation to fouling alleviation and improvement of treatment efficiency. The PMRs utilizing pressure driven membrane techniques have been applied for removal of different organic compounds, such as natural organic matter (NOM) [46,151,152,153], pharmaceuticals [50,154,155,156,157], dyes [158,159,160], phenanthrene [33], and viruses [25] or Cr(VI) [161]. Moreover, application of PMRs for treatment of textile and wood processing industry wastewater [162,163], primary and secondary effluents from municipal wastewater treatment plants [26,148,164,165,166], and surface waters [167] has been reported.

5.2.1. PMRs Utilizing Pressure-Driven Membrane Techniques

Depressurized PMRs: Systems with Submerged Membranes

The most promising configuration in terms of fouling mitigation seems to be depressurized PMR. Thus, this type of hybrid system has been especially widely investigated in recent years. A PMR equipped with submerged hollow fiber UF membrane made of polyvinyl chloride (PVC) for the degradation of humic substances (HS) in water was proposed by Du et al. [45]. They reported that aeration with air flow rate ranging from 0 to 3.2 L min⁻¹ was an efficient approach to mitigate membrane fouling, leading to a rapidly decreasing fouling rate [45]. However, the scouring effect of air bubbles is not the only important factor in PMRs. In these systems, aeration is also a source of oxygen participating in the photocatalytic reactions. In general, oxygen acts as an electron trap, decreasing the electron-hole recombination. Ong et al. [168] found that air bubbling under UV irradiation was crucial for improvement of TOC degradation due to formation of more ROS at higher air supply. However, it was also reported that at high oxygen concentrations, the rate of photocatalytic reaction might be reduced due to competition with the reacting substrates for the adsorption sites [19]. Moreover, at high aeration rates, the bubbles can attenuate the UV/VIS light transmission in the photoreactor and decrease the reaction rate [169]. Therefore, the aeration parameters should be carefully selected with reference to, e.g., reactor scale, type of treated contaminants and UV light transmission [24].

Another attempt for minimization of membrane fouling in a submerged PMR equipped with hydrophilized PVDF UF membrane with a nominal pore size of 0.04 μm, was presented by Patsios et al. [151]. During their investigations on HA and sodium alginate (SA) removal they found that application of backwashing and maintaining of moderate permeate flux (14 L m⁻² h⁻¹) allowed to control membrane fouling under the conditions of experiment. Depending on the process parameters, mineralization rates for HA and SA were in the range of 7.11–14.13 mgTOC h⁻¹ and 3.33–10.37 mgTOC h⁻¹, respectively, which corresponded to mineralization efficiencies of 49.6–73.9% and 30.9–74.4% [151].

A depressurized PMR equipped with PVDF hollow fiber MF membranes with nominal pore size of 0.1 µm for HA removal was described by Khan et al. [152]. Laboratory synthesized TiO₂-ZrO₂ binary oxide was applied as a photocatalyst. The authors revealed that the composite photocatalyst in loading of 150 mg L⁻¹ exhibited higher inhibition of membrane fouling compared to TiO₂, while maintaining the similar quality of permeate in terms of TOC removal [152].

High attention has been directed recently to the removal of pharmaceuticals, which are contaminants of emerging concern. For this application, a submerged PMR equipped with hollow fiber membranes, presented in Figure 8, was proposed [155]. The PMR was applied for removal of diclofenac (DCF) from ultrapure water (UW), tap water (TW), ground water (GW) and real water (RW). The most efficient DCF degradation (∼99.5%) and mineralization (∼69%) in groundwater was obtained at pH of ca. 6 and TiO₂ loading of ca. 0.5 g L⁻¹ when PMR was operated under constant permeate flux of 15 L m⁻² h⁻¹ [155]. Irrespective of matrices, membrane fouling was found to be negligible as a result of backwashing application.

Three different drugs—progesterone (PGS), ibuprofen (IBU) and naproxen (NAP)—were decomposed in a submerged PMR equipped with oscillatory PVDF membrane (Figure 9) [50]. The role of oscillatory motion of the membrane was mixing of the photocatalyst suspension and prevention from the deposition of photocatalyst particles on the membrane surface. Researchers proved a very effective degradation of organic contaminants using ZnO as a photocatalyst, reaching 92.3, 94.5, and 98.7% for PSG, IBU and NAP after 120 min of UVA irradiation, respectively [50].

Recent studies have been directed towards application of visible light active photocatalysts in PMRs. Such approach exhibits two important advantages: possibility of utilization of solar light and elongation of polymeric membranes lifetime due to exclusion of destructive action of UV radiation. Visible irradiation was applied in a submerged PMR equipped with flat sheet PVDF MF membranes operating in the presence of N–TiO₂ photocatalyst [158]. The decolorization efficiency during Reactive Orange 29 (RO29) decomposition reached 84.2% after 240 min of experiment.

Application of hybrid photocatalysis–membrane systems also provides the possibility to remove or inactivate microorganisms present in feed. A submerged PMR was applied for removal of bacteriophage f2, representing human enteric virus (mean size ca. 25 nm), from synthetic water [25]. The PMR was equipped with a flat-sheet PVDF membrane with nominal pore size of 0.15 µm. The most efficient reactor performance was reported for intermittent suction mode with a flux of 40 L m⁻² h⁻¹ or higher and TiO₂ P25 photocatalyst loading ranging from 10 to 25 mg L⁻¹. After 24 h of continuous operation, 5 log of bacteriophage f2 removal was observed (for f2 in feed tank equal to 5.22 log). The process responsible for inactivation was photocatalysis while the membrane served as a physical separation barrier only [25].

Although submerged PMR systems have been widely investigated in recent years, the literature reports on their application for treatment of real wastewater are rather limited. Doruk et al. [162] used a two-step hybrid system combining the PMR equipped with a flat sheet polyethersulfone UF membrane (pore size of 0.04 μm, cut off ~150 kDa) and reverse osmosis post-treatment unit for treatment of raw textile and wood processing (from middle density fiberboard (MDF) production) industry wastewater (Figure 10). ZnO photocatalyst particles deposited on the membrane surface reduced membrane fouling due to photocatalysis, while the membrane efficiently retained photocatalyst in the reaction solution. The authors reported that the initial chemical oxygen demand (COD) of wastewater affected significantly the efficiency of photocatalytic treatment. The highest removal rate was obtained for initial COD value of 150 mgO₂ L⁻¹ for each type of wastewater. After 3 h of PMR operation COD decreased to 70 and 80 mgO₂ L⁻¹ for MDF and textile wastewater, respectively. Application of RO allowed to further decrease COD to 20 mgO₂ L⁻¹. The authors suggested that RO permeate could possibly be applied as process water [162].

A novel integrated treatment process composed of fungal biodegradation in fungal membrane bioreactor (FMBR) followed by photodegradation in a submerged PMR was applied for industrial textile wastewater treatment by Deveci et al. [163]. In both reactors, flat sheet PES membranes with pore size of 0.05 µm were used. During PMR experiments, two different kind of photocatalyst (i.e., ZnO and TiO₂) in three loadings (1.0, 1.5, 2.0 g L⁻¹) were applied. Researchers proved that the integrated FMBR-PMR system was characterized by higher treatment effectiveness in terms of COD and color removal, than the two processes applied separately. They also emphasized that the configuration in which PMR was operated after FMBR was more favorable than when PMR was applied as a pretreatment technique [163].

Investigations on removal of secondary effluent organic matter (SEOM) in a submerged MF PMR with application of TiO₂ as a photocatalyst were carried out by Jiang and Choo [26]. The system was equipped with hollow fiber hydrophilized polyethylene membranes with a nominal pore size of 0.4 μm and a surface area of 60 cm². In the examined system, application of air bubbles and bubbleless backpulsing for 30 s every 1 h at a pressure 0.05 MPa was effective in membrane fouling mitigation and permeate flux restoration, as well as high SEOM degradation (more than 60%) [26]. Jiang et al. [169] used the same system for inactivation of bacteria in secondary effluents obtained from the biologically activated sludge process. The authors achieved the highest efficiency of microorganism removal at a TiO₂ P25 concentration of 1.0 g L⁻¹. In these investigations, aeration and backpulsing had a negative influence on bacteria removal. However, the authors concluded that the effect of intermittent backpulsing was less significant than that of continuous aeration. The dynamic TiO₂ cake layer formed on the membrane was found to be responsible for the bactericidal performance of the system. Furthermore, an increase of permeate flux caused decrease of bacteria inactivation due to shorter reaction times. More than 2.5 log removal in total bacterial count was noted in the PMR, while during MF alone it was only ca. 0.5 log. It was stressed that removal of bacteria was due to rejection by membrane, inactivation by direct UV radiation, adsorption onto photocatalysts particles and oxidation by reactive oxygen species [169].

Pressurized PMRs with Polymeric Membranes

A pressurized PMR system with flat sheet PES UF membrane was applied for HA removal [153]. It was reported that high concentration of inorganic anions such as SO₄²⁻, HPO₄²⁻ and HCO₃⁻ had a negative effect on the permeate flux. The highest membrane fouling was observed in the presence of HCO₃⁻, which was explained in terms of alkaline pH and formation of a dense fouling layer. In the presence of HA and inorganic ions a very thick filtration cake was deposited on the membrane surface, which was attributed to bridging effect between HA molecules and photocatalyst particles under high ionic strength conditions. Furthermore, the inorganic salts caused decrease of HA decomposition rate due to the hole and hydroxyl radicals scavenging effect. Monitoring of membrane separation properties revealed that the membrane was damaged by abrasive action of photocatalysts particles; however, the loss of separation properties was less severe when the process was performed under acidic conditions compared to alkaline ones [153].

A PMR with a with flat sheet PES UF membrane was also used for treatment of primary (PE) and secondary effluents from a municipal wastewater treatment plant [164]. In the absence of photocatalysis, the permeate flux decreased for 80% when UF of PE was performed and for 60% in case of SE. Application of the hybrid photocatalysis–UF made it possible to improve the permeate flux compared to UF alone. At TMP of 0.2 MPa and TiO₂ concentration of 1.5 g L⁻¹ the permeate flux increased compared to UF alone for 25–38% in case of PE and for 33% when SE were used [164].

A polypiperazine amide NF (MWCO 200 Da) membrane was applied in a PMR for removal of congo red (CR) dye in the presence of various types of ZnO photocatalyst [32]. The highest photodegradation efficiency (ca. 65%) and the lowest decrease of permeate flux (25%) were achieved for the photocatalyst fabricated under vigorous stirring in the presence of PVP (ZnO-PVP-St), applied at a loading of 0.3 g L⁻¹. This behavior was attributed to the large surface of ZnO-PVP-St and the effective absorption of UV radiation by this photocatalyst [32]. The schematic diagram of laboratory installation is presented in Figure 11.

Abid et al. [149] examined the degradation of methyl violet 6B (MV) dye in water using combined photocatalysis/low pressure reverse osmosis (LPRO) system. The investigations were focused on optimization of variable process parameters such as initial dye concentration (10–50 mg L⁻¹), TiO₂ photocatalyst loading (200–800 mg L⁻¹), suspension flow rate (Q_L = 0.3–1.5 L min⁻¹), pH (5–10), and addition of H₂O₂ (200–1000 mg L⁻¹). The installation, presented in Figure 12, contained a spiral wound module made of composite polyamide with an effective membrane area of 0.7 m². The authors concluded that optimal parameters of the system were photocatalyst loading of 400 mg L⁻¹, Q_L = 0.5 L min⁻¹, pH 5 and H₂O₂ concentration of 400 mg L⁻¹. They also stated that the product of the hybrid process could be recycled and reused [149].

Pressurized PMRs with Ceramic Membranes

Tubular ceramic UF membranes made of TiO₂, with MWCO of 100 kDa were applied in a PMR (Figure 13) utilized for HA removal from water [46]. The treatment efficiency strictly depended on feed cross flow velocity (i.e., hydraulic conditions in membrane module), TiO₂ P25 photocatalyst loading, feed pH and composition. Alkaline conditions had a negative influence on the permeate flux and HA removal, whereas at low and neutral pH, no membrane fouling was noticed. Ca²⁺ and Mg²⁺ cations present in feed containing SO₄²⁻, HPO₄²⁻ and HCO₃⁻ anions caused an improvement of the efficiency of HA removal and contributed to membrane fouling mitigation. However, similarly to in the case of PES membranes described earlier, after 400 h of operation, a deterioration of membrane separation properties was noted as a result of the abrasive action of photocatalyst particles [46].

The PMR presented in Figure 13 was also applied for treatment of secondary effluents from a municipal wastewater treatment plant [165]. The effectiveness of PMR operation was compared with that of hybrid UV/H₂O₂ system. In the PMR, the TiO₂ P25 photocatalyst loading was changed from 0.5 to 2 g L⁻¹. A tubular ceramic membrane with ZrO₂ separation layer deposited on a TiO₂ support (MWCO 5000 kDa) was used. No significant influence of photocatalyst concentration on permeate flux was noticed; however, lower membrane fouling was observed in case of PMR than in the photolysis–UF and UV/H₂O₂–UF systems. A significant contribution of adsorption to the overall removal of organic contaminants in the PMR was found. The highest total organic carbon (TOC) removal after 5 h of PMR operation was measured for 1.5 g L⁻¹ of TiO₂ (61%). Comparable treatment efficiency was found in the case of the UV/H₂O₂–UF process when 0.15 g H₂O₂ L⁻¹ was used (65%) [165].

Similar configurations were employed for surface water treatment from a lake [166]. An asymmetric, single-channel, ceramic membrane with ZrO₂ separation layer (5 kDa) was used. Application of photocatalysis led to enhancement of permeate flux compared to photolysis–UF for about 16–35%, depending on TiO₂ P25 loading. In this system the adsorption stage played a significant role in total process effectiveness as well. After 5 h of operation at 1 gTiO₂ L⁻¹ the total efficiency of TOC removal (i.e., due to adsorption, photocatalysis and membrane separation) reached 58% [166].

The efficiency of three different hybrid systems coupling UF (MWCO of membrane: 10 kDa) and advanced oxidation processes (including TiO₂ photocatalysis) during treatment of SE from a municipal wastewater treatment plant and surface water (SW) spiked with antibiotic, oxytetracycline (OTC), was investigated by Espíndola et al. [157]. In the case of SE treated at a TiO₂ loading of 1.0 g L⁻¹, the OTC was removed completely, and total mineralization was equal to 49%. In the case of experiments performed with SW as a matrix, higher OTC degradation rates and mineralization were reached due to the presence of a lower concentration of inorganic salts. Adsorption on the photocatalyst particles and/or on the ceramic membrane surface was found to contribute significantly for the removal of antibiotic and dissolved organic carbon (DOC). Regardless of feed type, application of photocatalysis led to an increase of permeate flux of about 20% in comparison to the photolysis–UF process [157].

A PMR with tubular ceramic MF membrane (nominal pore size 0.1 µm) was proposed by Song et al. [148] for treatment of SE in a PMR with TiO₂ as a photocatalyst. The permeate flux decline during 60 min of PMR operation was observed; however, the decrease was lower (30%) in the presence of TiO₂/UV compared to MF alone (50%). A significantly higher adsorption of organic compounds on photocatalyst particles at low pH was also found [148].

PMRs Utilizing Non-Pressure Driven Membrane Techniques

In recent years, only a few reports on PMRs for water and wastewater treatment employing other than pressure driven membrane techniques have been published. They refer to hybrid systems coupling photocatalysis with various configurations of membrane distillation. A PMR utilizing direct contact membrane distillation (DCMD) and UVC radiation was applied for diclofenac, ibuprofen and naproxen sodium salts removal from ultrapure water and tap water, as well as primary and secondary effluents of municipal wastewater treatment plant [166]. The efficiency of drugs removal was dependent on feed matrix and followed the order: UW > TW > SE > PE. In the case of PE and TiO₂ P25 photocatalyst loading of 0.5 g L⁻¹, DCF was completely removed, IBU concentration decreased by 73% and NAP for 90%. When SE was applied as a feed matrix and TiO₂ concentration was set at 1.5 g L⁻¹, DCF, IBU and NAP concentrations decreased by 100%, 93% and 94%, respectively. Although pharmaceutical degradation was high, the efficiency of mineralization did not exceed 14% for PE and 23% for SE. The drugs were not detected in distillate; however, the product contained high amount of N-NH₄⁺, especially in case of PE. The overall efficiency of removal of DOC exceeded 99% for both PE and SE. No permeate flux decline was observed for TW and SE; however, during PE treatment it decreased significantly (50–60%) [166].

A system based on microwave assisted photocatalysis coupled with vacuum membrane distillation (VMD) for treatment of (i) HA solution containing Ca²⁺ ions and (ii) coal gasification wastewater was investigated by Wang et al. [170]. A schematic diagram of the installation is presented in Figure 14.

Application of microwave irradiation led to a decrease of adsorption of TiO₂ particles and HA molecules on the polypropylene (PP) membrane surface; hence, the membrane fouling was alleviated. The authors reported that the system was efficient in treatment of the coal gasification wastewater and the removal rate exceeded 96% for COD and 98% for N-NH₄⁺ after over 120 h of operation [170].

6. Summary and Future Perspective

The results described in the present paper evidence that PMRs represent a very promising technology of great research and industrial interest.

PMRs, depending on their configurations, can be divided into two main categories: systems with suspended photocatalyst (slurry PMRs) or systems with the photocatalyst immobilized in/on a photocatalytic membrane. Slurry PMRs can be further classified in integrative-type PMR, where the photocatalytic reaction and the membrane separation take place in one unit, and in split-type PMR, where the photocatalytic reaction and the membrane separation take place in separate units.

PMRs realize functional synergies between a membrane separation and a photocatalytic conversion in agreement with the fundamentals of the process intensification strategy, thus making it possible to obtain many advantages in terms of output and costs. PMRs, thanks to the selectivity of the membrane separation unit, make it possible to achieve continuous operation in a system that allows not only the recycle and reuse of the photocatalyst but also the selective separation of the product(s). Such benefits cannot be obtained in conventional photoreactors. In the slurry systems the photocatalytic reaction must be followed by separation of the photocatalyst. Application of conventional coagulation–flocculation–sedimentation makes the reusing of the photocatalyst practically impossible. Moreover, in any type of classical photoreactor, the product (e.g., synthesized compound, purified water, etc.) cannot be collected from the very beginning of the process or needs further purification. In the case of environmental applications, this is especially important in terms of the presence of by-products of the photodegradation which might be more (eco)toxic than the initial compounds. When organic synthesis is considered, separation of products, which can be realized in PMRs, allows for prevention from secondary reactions and improvement of process selectivity compared to classical photoreactors.

In the PMRs with photocatalytic membranes, the problem of membrane fouling that can be caused by the TiO₂ particles is avoided. However, the photocatalytic efficiency was found to be lower than that of systems with the photocatalyst in suspension. Moreover, in the case of polymeric membranes, there is a risk of membrane degradation by UV light or hydroxyl radicals. In slurry PMRs, the membrane allows effective separation of the photocatalyst particles from the treated solution but, for pressure driven membrane processes, membrane fouling is observed. This unfavorable phenomenon might be overcome or reduced by application of submerged membranes.

Considering the harsh conditions present in the PMRs, significant attention should be paid to membrane selection. In the case of PMs, the use of polymeric membranes is dangerous because of the possibility of membrane degradation by both UV irradiation and ROS. This risk can be minimized by using slurry PMR in split-type configuration, thanks to the separation of reaction from membrane zones. Moreover, polymeric and ceramic membranes can be damaged by abrasive action of photocatalyst particles in the slurry-type PMRs. On these bases, the long-term investigation of the stability of membranes before their selection is recommended.

The results summarized in the present overview demonstrated that PMRs can be considered as useful green systems for water/wastewater purification, as well as organic synthesis, although additional studies are still needed before taking advantage of their potentiality at industrial level.

The integration of PMRs with other techniques such as electrochemistry or biotechnology can widen the application area and prompt further development [171].

A sustainable process can be obtained when a PMR is used to exploit the sun as a cheap and clean source of light. On this aspect, the development of new photocatalysts with high activity under visible light and their application in the various research fields is a mandatory task.