Recent progresses in synthesis and modification of g-C₃N₄ for improving visible-light-driven photocatalytic degradation of antibiotics

Antibiotics are metabolites originally derived from organic sources, including microorganisms, animals, and plants, which exhibit antimicrobial activities (Chen et al. 2020a). They have been widely applied to treat bacterial diseases and prevent infection in animals and humans, or to stimulate growth in animals (Yi et al. 2019). These drugs are not completely utilized in the bodies of animals and humans. Approximately 10–90% of the administered antibiotics are released into the environment through their urine and feces (Kaur et al. 2021). Meanwhile, a significant amount of antibiotics is discharged into the ecosystem during the production and unfitting disposal of drugs. Wastewater from hospitals and pharmaceutical industries is another significant source. Antibiotics have now been widely seen in sewage, drinking water sources, groundwater, food, soil, and surface water. Antibiotics pollution is associated with the growth of bacteria-resistant organisms, noxiousness to aquariums, genotoxicity, and disruption of the endocrine system (Kaur et al. 2021). The presence of antibiotics in the aquatic systems results in the formation of antibiotic-resistant bacteria (ARB) and antibiotic-resistant genes (ARGs) (Xiong et al. 2019). There is a high possibility of transfer of ARB and ARGs to humans, which has been identified by the World Health Organization as the most significant public health distress of the century (Zarei-Baygi et al. 2019).

Currently, conventional centralized and decentralized wastewater treatment systems are being used to treat wastewater before discharge or reuse. However, they are not designed for effective and sufficient removal of antibiotics and their residues; thereby, their efficiency in removing antibiotics, which possess high solubility and stability, is far from satisfactory. Studies have reported high concentrations of antibiotics at the hot spots and surroundings of conventional centralized wastewater treatment plants (WWTPs) (Yi et al. 2019). Traces of antibiotics were even detected in waterways significantly far away from the plants' discharge points (Watkinson et al. 2009). Despite the low concentration of antibiotics observed, concerns remain on their antagonistic effect on the aquatic ecosystems and their long-term effect on human health. The effluent from WWTPs has been arguably identified as the main source of antibiotics in the environment (Yi et al. 2019; Shi et al. 2020). On the other side, the level of decentralized wastewater treatment normally decides the effluent disposal and use, such as soil absorption, spray or drip irrigation. The direct reuse and application of treated effluent from decentralized wastewater treatment systems with inadequate removal of antibiotics and residues may transfer those contaminants to surface water, groundwater, and soil.

There is globally a rise in the consumption of growth enhancers and antibiotics, which is attributed to two key factors namely the teeming world population and the increase in the demand for protein sourced from animals. The ineffective antibiotic removal using conventional centralized and decentralized wastewater treatment systems poses challenges and limitations on the discharge and reuse of treated effluent. Till now, researchers have devoted significant effort to developing a cost-effective, eco-friendly, highly efficient, and simple process for antibiotic removal from wastewater. Photocatalysis has been recognized as one of the appropriate methods, which could be potential enhancement to traditional methods in wastewater treatment and remediation, and then significantly reduce the release of antibiotics into the environment after wastewater disposal or reuse. So far, a wide range of photocatalyst materials have been investigated for use in reaction to remove antibiotics.

Graphitic carbon nitride (g-C₃N₄) has become the most researched non-metal photocatalyst for remediation of organic pollutants, due to its abundance in nature, visible light response, eco-friendliness, good stability, facile synthesis, and low cost (Martha et al. 2013; Wang et al. 2018a). Notably, the cost of g-C₃N₄ is mainly associated with the use of low-cost nitrogen-based raw materials, such as thiourea, urea, and melamine, along with the simple preparation (Wang et al. 2020). Table 1 compares the unit price of commercially available g-C₃N₄ with some of commercial photocatalysts in the market and several reported photocatalysts in the literature. g-C₃N₄ shows competitively low cost among various types of selected catalyst. Moreover, several studies have proven the low toxicity of pristine and modified g-C₃N₄-based photocatalysts (Duan et al. 2020; Li et al. 2020a; Abdel-Moniem et al. 2021; Ravichandran et al. 2024). For instance, Li and colleagues reported negligible toxicity of Er-doped g-C₃N₄ on pakchoi seeds and Escherichia coli (Li et al. 2020a). Similarly, the in vitro toxicity test of g-C₃N₄ and bismuth sulfide-modified g-C₃N₄ showed that they were non-toxic, environmentally safe, and could be adopted as green catalyst materials for wastewater treatment (Abdel-Moniem et al. 2021). More importantly, the photoactivity of g-C₃N₄ under visible light, which occupies around 45% of the solar spectrum, makes it more attractive than many traditional photocatalyst materials (e.g. TiO₂), which are only active under ultraviolet (UV) light. Though g-C₃N₄ has shown remarkable potential for the removal of organic pollutants, its performance is limited by high recombination of photogenerated electron-holes carriers, poor visible light absorption (<450 nm) and low specific surface area (Wu et al. 2012; Feng et al. 2017). To improve the performance of g-C₃N₄, various methods, i.e. elemental doping, morphology controlling, defect engineering, heterojunction construction, etc., have been utilized to modify g-C₃N₄. The development of visible-light-active modified g-C₃N₄ photocatalysts toward high-performance degradation of organics has become one of the research hotspots.

This review reported recent progress in the synthesis and modification of g-C₃N₄ for the photocatalytic degradation of various antibiotic pollutants. In addition, the photocatalytic performances of g-C₃N₄-based materials were compared with other known photocatalysts in the literature. Photocatalytic reaction mechanisms were revealed to explain their enhanced performances. Limitations and challenges on scale-up synthesis, material design, intermediate analysis and economic viability were also discussed at the end to highlight future research opportunities.

Currently, there are two main approaches to manage water and wastewater treatment, centralized and decentralized treatment systems. The centralized systems are based on the collection and treatment of large volumes of wastewater and disposal of treated wastewater into the environment, typically through large-diameter pipes (Massoud et al. 2009). This method of wastewater treatment is widely practiced in developed and some of developing countries (Zhang et al. 2014). Significant advantages of such systems include less environmental impact, efficient treatment processes, and compliance with regulations for large residential areas (Massoud et al. 2009). However, they are not commonly practiced in many developing countries, especially with a low population density; due to the high cost of operation and management, reliance on treatment technologies, and need for skilled personnel in operation and maintenance (Massoud et al. 2009). On the other side, the decentralized systems are to collect, treat and dispose water near or at the point of generation; which allows flexibility in management and design by adopting onsite and cluster treatment technologies for specific sites (Battilani et al. 2010). Challenges associated with construction sites, including impervious soils, high groundwater table, and shallow bedrock, are avoided in the installation of decentralized systems (Massoud et al. 2009). Despite the fact that the decentralized systems show reliability and cost-effectiveness in providing a long-term wastewater treatment solution for small communities, there exist problems, e.g. selection of appropriate technologies, compliance with discharge standards and regulations, optimization of infrastructure setting and maintenance (Libralato et al. 2012). Treatments in neither conventional centralized nor decentralized systems effectively remove recalcitrant organic pollutants, including antibiotics. Therefore, there is a need to design and incorporate an effective advanced process.

There have been different treatment processes investigated by researchers to remove antibiotics from wastewater (in Table 2) (Ahmed et al. 2021). Each of these methods has its advantages and disadvantages. For instance, physical methods involving the use of adsorbents or flocculants can be relatively easy to incorporate into the existing wastewater treatment to increase the removal of antibiotics. However, the design, optimization and production of high-performance materials could limit the wide application of this approach (Dehghani et al. 2023). Bioremediation is known to be low-cost, but the efficiency is insufficient due to poor performance in the degradation of organic pollutants, which are toxic to the microorganisms employed (Shah & Shah 2020). The removal rate by microorganisms is relatively low and this process is time-consuming, such as when using constructed wetland method. Though chemical methods have been demonstrated to be effective in the elimination of pollutants, the formation of undesirable by products is a key setback to this method (Ahmed et al. 2021). Integration or combination of two different methods might potentially boost removal by taking advantages advantage of each; but increases complexity in operation and optimization (Köktaş et al. 2023).

Recently, to circumvent these limitations inherent to the above methods, researchers have been exploring other approaches. Photocatalysis, as one of the advanced oxidation processes (AOPs), has shown to be promising for the degradation of persistent or toxic organic pollutants, with advantageous features of high removal efficiency, mild reaction conditions, reduced use of harsh chemicals, potential complete mineralization of pollutants and eco-friendliness. Hence, it has been considered a potential enhancement to traditional wastewater treatment and remediation methods. Traditional semiconductors, such as TiO₂, have been studied for use as photocatalysts; however, many semiconductor photocatalysts can be activated only by UV (constituting about 5% of the solar spectrum). Consequently, the search for alternative photocatalysts with good photo-response in the visible region of the electromagnetic spectrum has received considerable interest. Graphitic carbon nitride (g-C₃N₄) is among the most researched non-metal semiconductors with good catalytic activity under visible light. Additionally, its easy synthesis, unique structure, good stability, and low cost have further attracted our research interests.

In our literature search, despite some attractive advantageous features as discussed above, we noted that most of photocatalytic reactions and associated photocatalysts (including g-C₃N₄) are still under study. It requires continuous effort to fully optimize material design and understand reaction mechanisms, toward potential optimization of industry-scale production, utilization, and associated costs. There is also a lack of toxicity analysis and quantification of intermediates during photocatalytic reactions.

Metal doping, using alkali, transition, or rare earth metals, has attained some popularity attributed to its ability to reduce band gap, increase light absorption, improve electron-hole separation and thereby enhance photocatalytic performance (Wang et al. 2018c; Xu et al. 2018b; Ma et al. 2019; Chi et al. 2022). Usually, metal doping was performed using a soluble salt of the metal combined with an amine-based precursor subjected to calcination temperature at around 550 °C (Wang et al. 2018c). Researchers have demonstrated that the doping of metal ions into g-C₃N₄ after high-temperature calcination led to the opening of heptazine rings and then the formation of defects (Hu et al. 2021). The doped metal defects formed energy levels below the conduction band of g-C₃N_4, and they were conceived to minimize charge carrier recombination by capturing the photogenerated electrons (Li et al. 2020a). Additionally, the doped metal species, such as Co, uniformly distributed and formed a stable coordination between the electron-rich N atoms cavities of g-C₃N₄. The obtained catalyst exhibited good stability and reusability along with high reactivity toward pollutants (Li et al. 2020c). K⁺ was reported to interact with the triazine rings of g-C₃N_4, causing electron redistribution. Such K doping in g-C₃N₄ improved free carrier density, enhanced transfer of photogenerated carriers, and minimized recombination of charge carriers (Yan et al. 2020).

The decoration of metal particles onto g-C₃N₄ forms metal-semiconductor interfaces, at which normally Schottky junction builds up (Rawool et al. 2023). The Schottky junction effectively improves the photocatalytic performance of g-C₃N₄ by minimizing charge recombination (Bai et al. 2022). It was normally carried out via mixing pre-formed g-C₃N₄ with metal-precursor-containing solution, followed by photo-reduction or chemical reduction (in NaBH₄ solution under N₂ gas) (Huang et al. 2020a). Noble metals that have been used for this purpose include Pt, Ag, Pd, and Au (Ong et al. 2014; Song et al. 2018b; Faisal et al. 2020). These metals are known to exhibit surface plasmon resonance (SPR) in the visible region. It assists in concentrating incident electromagnetic field, enhancing light absorption, promoting charge carrier generation and then increasing rate of photocatalytic reaction, especially in a nanostructure (Majeed et al. 2022). So far, a few studies have reported the use of noble metal nanoparticles (NPs) decorating g-C₃N₄, by taking advantages of their nano scale local surface plasmon resonance (LSPR) effect. Their results have proven that those noble metal NPs are effective captors for photogenerated electrons, thereby capturing free electrons, facilitating electron-hole separation, and improving photocatalytic activity of the resultant modified g-C₃N₄ (Zhang et al. 2018). For example, in Figure 4(c), the Au NPs (1%)-decorated g-C₃N₄ exhibited a high photodegradation of gemifloxacin mesylate (Faisal et al. 2020). The improved performance was ascribed to the even distribution of small Au NPs on the surface of g-C₃N₄, which served as captors to facilitate transfer of electrons to the catalyst surface and enhance charge separation. Likewise, Song et al. validated the LSPR effect of Ag NPs promoting the photocatalytic activity of g-C₃N₄ toward sulfamethoxazole removal (in Figure 4(d)) (Song et al. 2018b). However, one of the major limitations of noble metals is their high cost, which has hindered their commercial application. For this reason, cheaper non-noble metals, such as Bi and Cu, might be used as alternatives (Chen et al. 2017). The loading of bimetallic Ni-Cu NPs onto g-C₃N₄ resulted in enhanced light absorption, faster charge transfer and more efficient carrier separation; which were associated with the formation of Schottky barrier (Jin & Zhang 2020). Although non-noble metal NPs deposited on g-C₃N₄ have been demonstrated as an effective strategy minimizing charge carrier recombination and extending its light absorption toward improved photocatalytic activity for hydrogen evolution and dye removal, we noted there has been little study targeting at the photocatalytic degradation of antibiotics from water. Deposition of two or even more different metal particles may provide an opportunity to remarkably boost catalytic activity of resulting composites; however, their stability, durability and reusability should not be overlooked.

The doping of g-C₃N₄ with non-metal species, such as C, N, O, S, P, I, and F, has been proven to reduce photogenerated electron-hole recombination and induce a red shift in visible light absorption (Xing et al. 2018; Starukh & Praus 2020). The conduction band and valence band of g-C₃N₄ can be tuned to match the redox potential of a photocatalytic reaction. Various methods have been reported for non-metal doping on g-C₃N₄, including hydrothermal/solvothermal reaction, thermal copolymerization, precipitation, oxidation, thermal condensation, and gas-templating methods (Starukh & Praus 2020). Non-metal doping is to chemically attach or introduce non-metal elements to the heptazine units of g-C₃N_4. The dopants can form strong covalent bonds to the g-C₃N₄ backbones by gaining electrons based on their electronegativity (Salim et al. 2019). The framework of non-metal-modified g-C₃N₄ has been reported to be more easily activated, due to the high polarity of covalent bonds formed between g-C₃N₄ backbones and non-metal species (Tahereh Mahvelati-Shamsabadi 2020). Some of non-metals, e.g. O, S, and I, could substitute the N atoms in the aromatic triazine ring of g-C₃N₄ (Chu et al. 2020; Hu et al. 2020a); whilst some, e.g. P and B, selectively replaced the C atoms in the triazine ring (Ran et al. 2015; Chu et al. 2020). Such substitution of C or N led to the generation of defects, which served as electron captors and hence inhibited charge recombination (Chen et al. 2018; Patnaik et al. 2021). The N doping into the g-C₃N₄ framework was suggested as an effective method to adjust the band gap by introducing defects and forming midgap states (Patnaik et al. 2021). Non-metal dopants would also contribute to greater delocalization of π-conjugated electrons and in turn improve charge mobility (Tahereh Mahvelati-Shamsabadi 2020). It should not be overlooked that the strategy of non-metal doping circumvents the limitations of metal doping, such as photocorrosion and element leaching. The incorporation of non-metal dopants has been seen to alter the physicochemical properties of g-C₃N₄, including structural, morphological, and optical characteristics (Ismael & Wu 2019).

Penneri et al. reported a remarkable increase in both adsorption and photocatalysis toward efficient degradation of tetracycline using the porous C-doped g-C₃N₄ granules, which were synthesized via spray drying, as compared to its bulk form (Panneri et al. 2016). The greater photocatalytic activity of porous C-doped g-C₃N₄ was explained by the C doping creating delocalized π bonds, enlarging surface area, increasing interconnected porous network, enhancing visible light absorption, and reducing exciton recombination. Figure 4(e) shows the work on the utilization of S-doped g-C₃N₄/carbon dot (S-1–S-8) for photocatalytic degradation of sulfamethoxazole, in comparison with the pure S-doped g-C₃N₄ (S-0) (Xu et al. 2019). Upon optimization of doping, the observed improvement in the photocatalytic performance was associated with exclusive pore structure, large surface area, improved visible light absorption and reduced charge carrier recombination.

The photocatalytic degradation mechanisms using non-metal-modified g-C₃N₄ have been explored in several studies. Figure 4(f) shows the most probable location of anions in non-metal doped g-C₃N₄ (denoted NM g-C₃N₄). The introduction of dopants formed defects within the structure of g-C₃N₄ (Ran et al. 2015; Guo et al. 2021). Consequently, the band gap of g-C₃N₄, which is typically 2.7 eV, was narrowed after non-metal doping. Upon illumination, electrons were excited from the VB to the CB of g-C₃N₄ composites. If the photogenerated electrons and holes were not properly separated, they recombined again. In the CB, the photogenerated electrons were utilized in a photoreduction reaction, converting O₂ to oxygen radical (⁠⁠). In the presence of H₂O, the was converted to hydrogen peroxide (H₂O₂) and a highly oxidizing hydroxyl radical was produced, which was used to decompose antibiotics. On the other hand, the photogenerated holes in the VB decomposed antibiotics into smaller organic compounds and sometimes mineralized them into CO₂ and H₂O. The improved performances of non-metal doped g-C₃N₄ composites relative to its pristine form were based on two main factors: (i) the absorption of longer wavelength of visible light due to the presence midgap states and (ii) midgap states as temporary harbor for electrons, thus aiding charge separation (Jiang et al. 2019). Similar mechanisms have also been utilized to explain the enhanced degradation over the g-C₃N₄-based catalysts after metal doping (metal-doped g-C₃N₄ denoted as M g-C₃N₄ in Figure 4(f)).

Recently, the co-doping of two elements into g-C₃N₄ has gained increasing research interest, relative to mono-doping (Hasija et al. 2019), due to the potential to integrate advantages of various components and hence boost photocatalytic activity of resultant catalyst under their synergistic effect. The co-doping combination can vary from non-metal/non-metal, metal/metal to metal/non-metal. Co-doped g-C₃N₄ composites benefit from the synergistic effect of two dissimilar dopants, thus improving electron-hole separation, increasing specific surface area, improving charge mobility, creating more active sites and extending light absorption range (Wu et al. 2018; Li et al. 2020d).

It should be noted that, some researchers recently considered tri-doping of elements into g-C₃N₄. For instance, the Fe/Co/O tri-doped g-C₃N₄ exhibited a rate constant of about 2.67 times that of pristine g-C₃N₄, for the photocatalytic degradation of tetracycline (Wu et al. 2022). However, limited effort has been contributed to tri-doping of g-C₃N₄ for photocatalytic degradation of antibiotics. It is necessary to develop more effective and sustainable routes to synthesize co-doped or tri-doped g-C₃N₄ and then optimize their removal efficiency.

Defect tuning is effective to regulate the photoelectronic properties of g-C₃N₄ without introducing new chemical species during synthesis. Defective g-C₃N₄ was synthesized under high temperature treatment in an inert gas or air. The disruption of periodic arrangement of aromatic triazine unit in g-C₃N₄ resulted in the formation of carbon or nitrogen defects (Liao et al. 2020). In most cases, defect tuning did not alter the main conjugate structure of photocatalysts (Niu et al. 2014a). The N or C defects (vacancies) can enhance visible light absorption, improve charge separation, and increase active sites on the modified g-C₃N₄ toward improved photocatalytic performance (Liang et al. 2015, 2020). For instance, Ding and co-workers explored the effect of N vacancies from the uncondensed terminal of NH_x in the modified g-C₃N₄ on its photocatalytic oxidation performance (Ding et al. 2018). The enhanced oxidation potential of N-defective g-C₃N₄ over its pristine counterpart was credited to the improved visible light absorption and faster charge transfer. However, the relationship between defects and performances of modified g-C₃N₄ is still yet to be fully understood, especially in antibiotic removal.

Another promising strategy to boost the photocatalytic performance of g-C₃N₄ is tuning its morphological properties and textural structures by controlling sizes, shapes, dimensions, and porosity. The tuned g-C₃N₄ photocatalysts with large specific areas usually exhibited enhanced reaction performance due to the presence of more active sites, increase of electron-hole separation, and decrease of charge carrier recombination, as compared to the bulk counterparts (Kong et al. 2019; Zhang et al. 2019b). Other benefits of morphology tuning include an increase in optical activity and an improvement in pollutant adsorption performance. In addition to the greater surface area, the nanosized g-C₃N₄ might endow other advantageous features to achieve superior photocatalytic performance, including shorter charge migration length, better solubility, and tunable electronic properties (Chen et al. 2012). For instance, as compared with the bulk counterpart, ultrathin g-C₃N₄ nanosheets enhanced charge separation and migration to their surfaces, because of their quantum confinement effect (Niu et al. 2014b; Zhang et al. 2015). To date, g-C₃N₄ with various morphologies have been synthesized and employed in photocatalysis; these include 0D quantum dots (Wang et al. 2014; Wang et al. 2016a), 1D nanowires/nanorods/nanotubes (Gao et al. 2012; Bai et al. 2013; Zhao et al. 2014), 2D nanosheets (Yang et al. 2013; Zhang et al. 2013), and 3D hierarchical structures (Shen et al. 2014).

Different synthetic methods have been adopted for tuning the morphology and structure of g-C₃N₄ and their related materials, including templating, chemical exfoliation, ultrasonic-assisted liquid exfoliation, and thermal exfoliation; which may be categorized into ‘top-down exfoliation’ and ‘bottom-up assembly of molecular units’ (Dong & Cheng 2015). For example, Guo and co-workers employed a facile bottom-up approach to prepare Cl-doped porous g-C₃N₄ nanosheets, which displayed about 2.4 times the tetracycline removal compared to the bulk g-C₃N₄ (Guo et al. 2019). The authors ascribed the enhanced performance to the doping of Cl element, leading to narrower band gap, large specific surface area, more active sites, and greater electron-hole separation. Yang's group developed a two-stage bottom-up method via combining hydrothermal reaction with secondary calcination under N₂ gas in a closed environment to synthesize highly crystalline g-C₃N₄ nanosheets (HCCNNS) with dramatically improved photocatalytic degradation of ciprofloxacin and sulfamethazine (Yang et al. 2021). The increase in photogenerated charge separation and transfer was associated with the thinner and larger lamellar morphology and high crystallinity.

Adhikari et al. synthesized a mediator-containing Z-scheme heterostructure, MoO₃/Ag/g-C₃N_4,via simple hydrothermal and borohydride reduction methods, in which the interfacial contact among the three components, Ag, MoO₃ and g-C₃N₄, was displayed in Figure 7(b) (Adhikari et al. 2019). The fast electron transfer between MoO₃ and g-C₃N₄ was mediated by the Ag NPs which located between both. This stable and recyclable photocatalyst removed 89% of tetracycline and 96% of ofloxacin after 100 min visible light photocatalytic reaction. The remarkably enhanced removal as compared with Ag/g-C₃N₄ and MoO₃/Ag/g-C₃N₄ affirmed the co-operative effects between Ag and MoO₃. Yang's group constructed a new direct Z-scheme photocatalytic system – g-C₃N₄-decorated ZrO_2−x nanotubes by using anodic oxidation and subsequent thermal vapor deposition (Chen et al. 2020b). This heterojunction was proven to enhance visible light absorption, faster separation of photogenerated carriers and lower their recombination. Figure 7(c) shows the photocurrent density of optimized heterostructure was about 3.5 and 15 times that of ZrO_2−x nanotubes and g-C₃N₄, respectively. It removed 90.6% tetracycline hydrochloride after 1 h visible light photocatalytic reaction, as compared with only ∼40 and 50% when using g-C₃N₄ and ZrO_2−x nanotubes, respectively. In addition, a novel design of catalyst comprising Bi₂₄O₃₁Cl₁₀ (BOC), MoS₂ (MS) and g-C₃N₄ (CN) was synthesized using impregnation-calcination. The enhancement of tetracycline photocatalytic degradation on CN/MS/BOC was due to the obtained dual Z-scheme ternary heterostructure, which improved visible light absorption, accelerated carrier transfer, hindered electron-hole recombination, and strengthened redox ability (Kang et al. 2020). The tuning of catalyst structural and morphological properties, as well as improvement of interfacial contact between semiconductors in heterojunctions could also result in highly efficient photocatalytic reactions (Yu et al. 2013; Yang et al. 2015).

As can be seen from the above sections, various modification strategies have been explored to improve the properties of g-C₃N_4, including visible light absorption, generation and separation of charge carriers, and then photocatalytic performances in antibiotic removal. By comparing Tables 3 and 4, the most effective strategies to modify g-C₃N₄ are heterostructure formation, co-doping, and defect tuning. In particular, the heterostructure formation has attracted greatest research interests as a promising route to modify g-C₃N₄ for high performance antibiotic removal.

In addition to g-C₃N₄, there have been some other attractive photocatalyst materials used for antibiotic removal, such as TiO₂, ZnO, and WO₃. WO₃ is a n-type visible-light-driven semiconductor with a bandgap of 2.4–2.8 eV (Fu et al. 2019). TiO₂ (E_g = 3.2 eV) and ZnO (E_g = 3.1 eV), which are traditional photocatalyst materials and commercially available in the market, have large bandgaps and thus restrain their uses to UV light (Zeinali Heris et al. 2023). As such, they are commonly modified with other visible-light-active semiconductors to obtain activity in the visible region. Table 5 shows the comparison of g-C₃N₄-based heterostructures to other composites consisting of TiO₂, ZnO, or WO₃ in photocatalytic removal of different antibiotics. Taking consideration the different light sources and pollutant concentrations, g-C₃N₄-based heterostructures are seen to be competitive for removal of different antibiotics, including tetracycline, amoxicillin, ciprofloxacin and sulfamethoxazole.

It is known that photo-induced electrons, holes, and reactive oxygen species are generated and initiate the redox reaction, when a photocatalyst is exposed to light illumination. One of the advantageous features and prime objectives using photocatalytic reaction in the removal of antibiotics is their potential complete mineralization. However, partial mineralization of antibiotics produces various reaction intermediates and by-products, which might be more harmful and toxic than the original pollutants. It should be noted that most of current studies, which are related to photocatalytic degradation of antibiotics, reported degradation efficiency based on the ratio of concentrations of target antibiotic before and after treatment on a UV–vis spectrometer. So far, some effort has been contributed to examination and understanding on the formation and toxicity of intermediates during photocatalysis, which is essential to explore and confirm the practical applicability of modified g-C₃N₄ in wastewater treatment.

Despite the promising improvement in photocatalytic performances of the modified g-C₃N₄, most of the relevant photocatalysts and photocatalytic reactions are still under study and limited to the laboratory. The underlying mechanisms and toxicity of intermediates in those photocatalytic reactions remain unclear, which in turn limit their practical application. Therefore, future research in this field is necessary to address the following research gaps and explore potential for commercialization.

• (1) Economical and environmentally friendly synthesis of photocatalysts

Most synthesis of g-C₃N₄-based photocatalysts requires high temperature – over 500 °C and long pyrolysis reaction time – several hours (e.g. 4 h) (Yang et al. 2018; Li et al. 2023b). Meanwhile, many of the amine-based precursors employed to prepare g-C₃N₄ can generate toxic substances (e.g. sulfur-containing by-products) released into the environment (Shamilov et al. 2023). Therefore, there is a need to explore more energy-efficient techniques and more environmentally friendly chemicals for use during synthesis.

So far, various high performance photocatalysts, including modified g-C₃N₄, have been developed, but their synthesis is only limited to the laboratory scale. It is necessary to validate that their synthesis processes are stable, simple, and reliable for industry-scale fabrication, while their unique structural, chemical, and morphological characteristics can be largely controlled and retained. Moreover, other factors should be investigated in their synthesis when transiting from laboratory to full-scale production of photocatalysts, including its reliability, scalability, production yield, process sustainability and safety.

• (2) Improvement of photocatalyst performance, recovery, and reuse

Some of the modified g-C₃N₄ photocatalysts are prone to photocorrosion, due to the defects formed and/or doping of foreign elements, such as metal ions or particles. Leaching of dopants into water environment may cause another contamination and also affect reusability of catalysts. Some, such as peroxides generated in photocatalytic reactions, have been identified to deactivate and limit the performance of g-C₃N₄ by binding to its surface, thereby restraining its industrial applications (Vasilchenko et al. 2022). The chemical stability of modified g-C₃N₄ needs to be largely retained to achieve superior recyclability and reusability. Future research would need to ascertain their long-term stability, especially for specific applications and extreme operation conditions.

The recovery of modified g-C₃N₄ NPs after photocatalytic removal of antibiotics from water is a key challenge to its industrial application. The introduction of magnetic property can aid the efficient recovery of g-C₃N₄-based materials in a short period of time simply with the help of an external magnetic field. Other forms of catalysts, such as membranes or films, would also benefit catalyst recovery and practical application.

• (3) Continuous exploration of mechanisms and analysis of intermediates

Although significant projections on photocatalytic reactions have been outlined by researchers, there is a need to explore the relationship between heterostructure and performance of modified g-C₃N₄, especially with respect to the charge transfer and separation. Continuous effort should be devoted to providing insight on photocatalytic mechanisms of modified g-C₃N₄ based on a complete analysis of physicochemical characteristics, electrochemical properties, and theoretical modelling (such as DFT calculations). The use of machine learning models alongside experimental research will expedite optimization of photocatalytic degradation over a selected catalyst, minimization of chemical use and waste disposal, and aid better understanding of that photocatalytic reaction, as well as benefit development of other photocatalytic catalysts and systems.

Research effort on degradation pathways and toxicity changes of intermediates over the modified g-C₃N₄ is still limited. There is a lack of direct evidence proving the complete mineralization of target pollutants and understanding the potential effects of residues on the environment, including microorganisms.

• (4) Assessment of economic viability of photocatalytic reaction and operation

It is imperative to assess the economic viability of a photocatalytic reaction using a specific catalyst material as a potential wastewater treatment process for antibiotic removal. Photocatalysis relies on a photocatalyst to drive a chemical reaction using light energy. Hence, promoting the economic viability and commercial application of photocatalysis requires consideration of selecting efficient light source and photocatalyst, as well as operation optimization.

Commercial UV lamps are costly with a short life expectancy, and are usually ineffective for generating UV light after one year of service (Xiao et al. 2015). In general, a photocatalytic process involving the use of UV lamp and UV-sensitive photocatalysts, such as TiO₂ and ZnO, among others, would be expensive and less economically viable, due to the high energy demand and high cost of UV lamps. On the other hand, the use of visible light or solar-active photocatalysts, such as g-C₃N₄, WO₃, etc., could be more cost-effective and economically viable. Sunlight is a low-cost, renewable, and abundant energy resource that can be used to initiate and promote photocatalytic reactions; however, the treatment by sunlight is severely restricted by sunshine duration. A few solar photocatalytic water treatment reactors have been reported in recent years (Al-Nuaim et al. 2023). There are also reports on the optimization of photoreactors by using internal illumination, e.g. LEDs and optical fibers (O'Neal Tugaoen et al. 2018). Energy consumption is one of important characteristics to evaluate economic viability of photocatalytic degradation; electricity energy per order (EE/O) has been used to study it in tetracycline degradation (Xia et al. 2020b). However, no effort has been found in the research on modified g-C₃N₄ for photocatalytic removal of antibiotics.

On the other side, the photocatalytic activity, reusability, and stability of a catalyst are crucial to ensure economic feasibility of the reaction, which have been discussed above. The use of suspended photocatalyst particles in the treatment of wastewater might be seen with a limited recycling and recovery, and in turn make the treatment laborious and less economical. Along with the development of an energy efficient photocatalytic reactor, small g-C₃N₄-based photocatalyst particles might be immobilized onto a recyclable, reusable, environmental-friendly solid supports, such as clay particles and polymer matrices, to increase the sizes of resulting catalysts. However, it might induce reduced photocatalytic activity (Sraw et al. 2018), which needs to be minimized. As per the experimental conditions, the optimization of various operating parameters, including light source, catalyst dosage, and water chemistry, need to be well investigated on photocatalytic efficiency. The combination between photocatalysis and other AOPs, such as ozonation and Fenton process, could assist in minimizing reaction time and thereby enhance economic viability of pollutant removal (Saritha et al. 2007; Mena et al. 2012).

Antibiotic pollution has been a menace to the environment for decades and this constitutes a major challenge to public health. The processes employed in centralized or decentralized water treatment systems are not sufficient for their effective removal. Visible-light-driven photocatalysis has been regarded as a potential approach to address this challenge; in which g-C₃N₄, having features of chemical and thermal stability, low cost, easy synthesis, and narrow band gap (∼2.7 eV), has been established as a promising material for practical photocatalytic applications with use of solar energy. Modification of g-C₃N_4,via metal doping, non-metal doping, co-doping, metal decoration, defect tuning, structural engineering, or heterostructure formation, has been proven to further enhance photocatalytic activity of the pristine semiconductor and then improve efficiency for removal of antibiotic pollutants. It was found that the g-C₃N₄ modified via heterostructure formation has attracted greatest research attention and showed promising photocatalytic degradation performances. It should be noted that the integration of two or more modification strategies would provide great potential to boost photocatalytic activity of resulting composites. Future research effort is suggested to addressing challenges and limitations which have been identified, including economical and environmentally friendly synthesis of photocatalysts, economic viability of photocatalytic reaction and operation, and mechanism exploration and intermediate analysis.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.