The conversion of inexhaustible and clean sunlight into chemicals and fuels by means of a photocatalyst has been pointed out as an appealing approach to mitigate the energy crisis as well as the environmental concerns caused by the massive consumption of fossil fuels. In a typical photocatalytic process, the photocatalyst, typically a semiconductor material, must absorb photons with energy equal or larger than that of its band gap. After light-harvesting, electrons (e⁻) in the valence band (VB) are promoted to the conduction band (CB), leaving a positive charge, named hole (h⁺), in the VB. The photo-generated electron–hole pairs must migrate to an interface between the semiconductor surface and the reaction media before undesirable charge recombination takes place. Finally, the photo-generated electrons and holes at the interface are involved in red/ox reactions. Scheme 1 presents the fundamental steps of the photocatalytic process.

A large variety of semiconductor photocatalysts, such as TiO₂, ZnO, Fe₂O₃, BiVO₄, SrTiO₃, WO₃, CdS, MoS₂, MOFs, etc., have been widely investigated in different photocatalytic applications such as water splitting, artificial photosynthesis, N₂ fixation, organic transformations, and pollutant degradation (Yang and Wang, 2018). Among them, TiO₂ has been the most widely studied photocatalyst since the seminal paper of Fujishima and Honda (1972). However, TiO₂ is a wide-band gap semiconductor (3.2 eV for anatase), and therefore, only photons from the UV region of the solar spectrum, which accounts for less than 4% of the total solar irradiation, can be involved in photocatalytic reactions, rendering low production rates. Other narrow-band gap semiconductors, such as Cu₂O or Fe₂O₃, have shown large instability and/or suffering from high recombination rates (Zhou et al., 2016). Similarly, metal sulfides are also well known to decompose upon light irradiation (Iwashina et al., 2015; Ning and Lu, 2020). For these reasons, metal-free carbon-based semiconductor materials (doped graphenes, carbon nanotubes, carbon dots, carbon nitrides, etc.) have been recently proposed as good candidates in photocatalytic applications due to their robustness, large surface area, and optoelectronic properties, among others (Jin et al., 2018; Albero et al., 2019; Chu et al., 2020; Guo et al., 2020; Haiyan et al., 2020; Kandy, 2020; Mazzanti and Savateev, 2020; Reischauer and Pieber, 2021; Zou et al., 2021).

In this regard, the carbon nitride material family (C_xN_y) has attracted increasing interest in the last decades, and the number of both theoretical and experimental publications have grown exponentially in this field. Among them, the most widely studied so far has been the graphitic carbon nitride (g-C₃N₄), which was first reported in photocatalytic applications by Wang et al. (2009). This is a metal-free semiconductor material formed by the stacking of g-C₃N₄ layers consisting of tri-s-triazine units linked by amino groups. g-C₃N₄ can be prepared by thermal condensation of N-rich precursors, such as urea, thiourea, cyanamide, dicyanamide, and melamine, among others (Lakhi et al., 2017). Moreover, g-C₃N₄ has demonstrated suitable optoelectronic properties (2.7 eV band gap) as well as a high thermal, mechanical, and chemical stability. As a consequence, a wide variety of C₃N₄-based photocatalysts have been investigated to drive chemical reactions upon light irradiation (Wang L. et al., 2020), including H₂ evolution from water (Naseri et al., 2017), pollutants degradation (Ong et al., 2016), CO₂ reduction (Liu R. et al., 2020), and organic transformations, such as aromatic compounds oxidation, benzaldehyde and alcohol esterifications, oxidative cleavage of C–C bonds and allylic oxidations, among others (Wen et al., 2017). Nevertheless, the performance of C₃N₄ in photocatalytic applications is still limited since this is highly dependent on the crystal structure, porosity, charge transport, and the limited light harvesting in the visible (Vis) and near-infrared (NIR) regions. For that reason, crystalline carbon nitrides with different stoichiometry than that of C₃N₄ have recently emerged exhibiting very promising properties for optoelectronic applications (Tan et al., 2021).

Theoretical calculations have predicted different CxNy crystalline structures, beyond C₃N₄, with striking properties since 1989. The mechanical, optical, electronic, and structural properties of this new family of materials have been anticipated by DFT and first-principles calculations. Moreover, theoretical efforts are also being invested in the theoretical design of the materials (Kou et al., 2020). The wide range of compositional possibilities within these materials (CN, C₂N, C₃N, and C₃N₅, etc.), giving rise to the fine tune of their properties, have motivated not only theoretical but also experimental researchers to explore the performance of these materials in different fields, including electronics, sensing, mechanical, adsorbents, catalysis, photocatalysis, electrocatalysis and energy storage, among others (Tan et al., 2021).

C_xN_y materials have already shown great potential in energy storage and electronic applications due to their semiconducting properties (Xu et al., 2017; Yang et al., 2017; Bafekry et al., 2019). However, the use of C_xN_y materials beyond g-C₃N₄ in photocatalysis is in its infancy, and further work must be carried out, especially experimentally, in order to confirm the high expectations raised by simulations and theoretical calculations. In this review, we aim to overview the scarce experimental and theoretical reports in the field of photocatalysis using novel C_xN_y materials, except g-C₃N_4, which has been widely reviewed elsewhere (Ong et al., 2016; Wen et al., 2017; Fang and Wang, 2018; Savateev et al., 2018; Nasir et al., 2019). We have paid special emphasis on the synthetic procedures and the crystalline structures of the obtained materials that have given rise to very promising performances as photocatalysts. Finally, we will look for the best candidates in this area based on the available experimental results and theoretical calculations, which could give rise to the next generation of metal-free carbon-based materials in photocatalysis.

The motivation in the preparation of nitrogen-rich carbon nitrides arise from the large band gap that the undoped g-C₃N₄ exhibits (2.7 eV, λ < 450 nm), which limits the application of this semiconductor in the photocatalytic field. A straightforward way to reduce its band gap goes through the addition of more N atoms in the carbon nitride framework. In this regard, Mane et al. reported in 2017, for the first time, a synthetic procedure, which enabled the formation of a well-ordered 3D porous carbon nitride with stoichiometry C₃N₅, exhibiting a band gap energy of 2.2 eV (Mane et al., 2017). This C₃N₅ semiconductor material was prepared using a single molecular precursor with a high nitrogen content (3-amino-1,2,4,-triazole) by means of a hard templating approach through precursor self-condensation inside the confined mesopores of KIT-6 silica at 500°C in Ar atmosphere. After carbonization, the silica template was dissolved in HF, and the obtained C₃N₅ was filtered and washed with EtOH. The sample exhibited ordered mesopores with a pore size of 3.42 nm and a specific surface area of 296.7 m²/g.

Synchrotron-based near-edge XRD fine structure (NEXAFS) showed three peaks characteristic of the carbon structure, very similar to the KIT-6 mesoporous silica, indicating that the obtained structure is very similar to that of the template. The as-prepared C₃N₅ presented a broad absorption band covering the whole UV-Vis spectrum (Figure 1), and its performance in photocatalytic H₂ evolution was investigated. Pt nanoparticles were photo-deposited on the semiconductor surface, and Pt-C₃N₅ aqueous dispersions (1 mg/ml) containing triethanolamine (10 vol.%) were irradiated under the outcome of a 300-W Xe lamp filtered with a long-pass cutoff filter (λ ≥ 420 nm). The photocatalytic H₂ production under these conditions was of 801 μmol after 3 h irradiation. The H₂ evolution rate of C₃N₅ was found much higher than that of C₃N₄ reported in the literature under similar conditions. It was determined that the reduced band gap not only enhanced the light harvesting of a large portion of the visible region of the solar spectrum but also provided more favorable thermodynamics for H⁺ reduction.

One of the most applied methods was developed by Zhang, Yan, and co-workers. They prepared a rod-like C₃N₅ material with a length up to 1–6 µm and a diameter of 100–500 nm through a KBr-guided approach using 3-amino-1,2,4-triazole as C₃N₅ precursor and KBr acting as guide (Wang et al., 2019). In this method, an aqueous solution containing KBr and the C₃N₅ precursor was heated at 80°C, and the water solvent was allowed to gently evaporate until the KBr crystals grew covered by the molecular precursor. The obtained solid was annealed at 500°C in air conditions. Finally, C₃N₅ was easily recovered by water washing. Opposite to the previous method, where HF was used to remove the silica template, these approach is more environmentally friendly and easy. However, the KBr-guided approach promoted rod-like shape C₃N₅ (Figure 2), exhibiting a 1.90 eV band gap. XRD patterns showed a diffraction peak at 13.4° attributed to the 100 facets, indicating the in-plane structural ordering of the obtained C₃N₅. In this report, the photocatalytic activity toward organic pollutant degradation using the model contaminant methylene blue (MB) was studied dispersing the C₃N₅-based photocatalyst (1 mg/ml) in an MB solution (20 mg/ml) and irradiated under visible light (λ ≥ 420 nm) with a 50 W- halogen tungsten lamp. For comparison purposes, pristine g-C₃N₅ was prepared, and the photocatalytic degradation of MB was studied too. It was demonstrated that the photocatalytic degradation ratios of pristine g-C₃N₅ and rod-like C₃N₅ reached 10% and 98% within 120 min, respectively. The authors ascribed the improved photocatalytic activity of the rod-like C₃N₅ to the higher surface area and smaller band gap than that of the pristine g-C₃N₅ (2.05 eV). Despite no comparison of the photocatalytic activity with other g-C₃N₄ has been provided, the reduced band gap that this material presents allows much broader photo-response in the visible range of the solar spectrum than any undoped g-C₃N₄.

In a similar approach, Jin et al. reported the synthesis of g-C₃N₅ following the previous approach based on the KBr-guided synthesis (Jin et al., 2020). The obtained g-C₃N₅ was mixed with Zn_0.5Cd_0.5S, sonicated, and filtered with the aim to obtain composites exhibiting Z-scheme heterojunctions. After loading optimization (15 wt.% of g-C₃N₅), a photocatalytic H₂ production of 142.8 mmol/g·h using Na₂S and Na₂SO₃ as sacrificial electron donors, a 300-W Xe lamp (100 mW/cm²) as light source, and a quantum efficiency of 33.7% under 420-nm monochromatic light, was obtained. In this case, the authors reported that g-C₃N₅ was used in the heterojunction instead of g-C₃N₄ because of the smaller band gap and the antecedents in the literature supporting the better photocatalytic performance of g-C₃N_5. Similarly, Vadivel et al. reported identical synthetic approach to prepare AgCl-loaded g-C₃N₅ composites for visible light photocatalytic pollutant degradation. In this work, Rhodamine B (RhB) was used as model pollutant, and the removal achieved 96% within 30-min efficiency under visible light irradiation with the output of a 300-W halogen lamp (Vadivel et al., 2020).

A variation of this method was reported by Zhang and collaborators (Li et al., 2020). In this, NaOH was used instead of KBr with the aim to introduce N-vacancies in the g-C₃N₅. The N-vacancies promoted a redshift of the absorption edge due to a band gap narrowing from 2.08 up to 1.50 eV. XRD measurements of pristine g-C₃N₅ and g-C₃N₅ with N-vacancies (NV-g-C₃N₅) revealed a peak at 27.1° related to the 002 crystal planes of C₃N₅, while the 002 peak of NV-g-C₃N₅ was shifted to a higher angle due to the repulsion between the electron-rich conjugated C₃N₅ layers. Besides, the lower intensity of the 002 peak suggests that the CN framework is distorted, and as consequence, the ordered structure of g-C₃N₅ is partially lost in NV-g-C₃N₅. The as-prepared NV-g-C₃N₅ formed a heterojunction with BiOBr, and the composite material was tested as photocathode for nitrogen reduction. Thus, NV-g-C₃N₅/BiOBr was deposited on FTO and used as a working electrode. The photoelectrochemical nitrogen reduction was carried out under a Sun-simulated light with a cut off filter (λ ≥ 420 nm) in a three-electrode H-type cell with a proton exchange membrane, a saturated Ag/AgCl (sat. KCl) reference electrode, and a Pt sheet as a counter electrode in HCl + Na₂SO₄ electrolyte. Under these conditions, a maximum NH₃ yield rate and Faraday efficiency of 29.4 μg/h·mg and 11%, respectively, at −0.2 V were obtained. The authors concluded that the N-vacancies were not only effective in narrowing the band gap, but also acting as catalytic sites for the activation and reduction of N₂.

Alternatively, ultrathin C₃N₅ nanosheets were prepared following a two-step method based on, first, a NH₄Cl-guided approach to obtain bulk g-C₃N₅ and subsequent protonation process consisting in bulk g-C₃N₅ treatment with HNO₃ and H₂SO₄ mixtures under continuous stirring for 12 h (Liu T. et al., 2020). The XRD pattern before and after protonation with the acids remained the same, exhibiting a peak at 27.3° attributed to the inter-layer stacking of the conjugated C₃N₅ framework. The obtained 2D g-C₃N₅ nanosheets with a surface area of 38 m²/g exhibited photocatalytic activity toward MB photodegradation as well as H₂ evolution under visible light irradiation (λ ≥ 400 nm) with a 300-W Xe lamp.

In a very appealing approach, Zhu and co-workers prepared g-C₃N₅/TiO₂ heterojunctions following the Zhan and Yan protocol, but replacing KBr by TiO₂ (Zhang et al., 2021). In this case, the guiding-metal oxide was not removed after annealing at 500°C, and instead, this is part of the photocatalytic system. XRD pattern showed a peak at 27.2° attributed to the 002 plane, indicating the interlayer stacking characteristic of this material. The composite exhibited excellent photocatalytic activity in visible light driven (λ ≥ 400 nm) from a 300-W Xe lamp, with NO photooxidation under relative humidity of 15%. It is worth commenting that this composite photocatalyst presented high stability, and after six consecutive runs, the NO removal efficiency only showed 7% loss, from 67% to 60%.

A very different approach was reported by Kumar et al. by thermal deammoniation of 2,5,8-trihydrazino-s-heptazine (melem) (Kumar et al., 2019). In this, melem hydrazine was first synthetized by hydrothermal reaction between melem and hydrazine hydrate solution at 140°C for 24 h. Then, the obtained solid was annealed at 450°C for 2 h. The obtained material presented C₃N₅, determined by XPS, and with a band gap of 1.76 eV. The chemical structure was elucidated by ss-¹³C NMR using the cross-polarization magic-angle spinning (CPMAS) and XRD. The C₃N₅ CPMAS-NMR spectrum showed two peaks at 164 and 156 ppm for N₂C–N=N– and CN₃ carbons, respectively. CRD pattern of C₃N₅ showed the typical broad 002 peak at 27.6° corresponding to the 0.33 nm interplanar layer distance. Photoelectrochemical water splitting experiments were carried out using Na₂S as hole scavenger and C₃N₅-sensitized TiO₂ as photoeletrode. This achieved a photocurrent density of 465 µA/cm² under AM 1.5 G irradiation, while under identical conditions, the value of photocurrent density using g-C₃N₄ sensitized TiO₂ was only of 373 µA/cm². The improved photoelectrochemical activity was attributed to the position of the valence band in C₃N₅ (+1.04 V vs. NHE) in comparison with g-C₃N₄ (+1.8 V vs. NHE) together with the photosensitizing behavior of C₃N₅ at longer wavelengths (505 nm) than that of g-C₃N₄ due to the reduced band gap. The same synthetic protocol was used by Yin et al. to prepare C₃N₅/Ag₃PO₄ heterojunctions, and the photocatalytic removal of tetracycline hydrochloride (TCH) was investigated (Yin et al., 2021).

Recently, P-doped C₃N₅ nanosheets have been synthetized using 3-amino-1H-1,2,4-triazole and phosphonitrilic chloride (Liu et al., 2021). These precursors were mixed and ground before thermal condensation at 500°C for 4 h. After natural cooling, the obtained solid was annealed again at 550°C for thermal copolymerization for 4 h. In this case, two diffraction peaks appeared in the XRD measurements, attributed to in-plane packing in the 001 plane and the interlayer stacking along the 002 direction at 13.2° and 27-7°, respectively. These peaks did not shift after P-doping, indicating that the structure of C₃N₅ is preserved. The obtained P-doped C₃N₅ photocatalyst was used in the photocatalytic RhB degradation. The photocatalytic efficiency for RhB degradation values obtained with the P-doped C₃N₅ was higher than the values reported in previous studies using doped g-C₃N₄ or its composites.

C₁N₁ is a porous polymeric carbon nitride comprising pores formed by three carbon atoms and three nitrogen atoms. The pores of ca. 3.4 Å of diameter are highly polar, which makes them excellent candidates to chelate metals and adsorb small molecules. Moreover, Srinivasu et al. reported in 2014 using first-principle calculations that a monolayer C₁N₁ structure have a band gap of 2.89 eV, which decreases to 2.75 eV in a multilayered structure, and studied the potential of the simulated C₁N₁ structure and different doped derivatives as photocatalysts (Srinivasu et al., 2014).

Though synthetic efforts to produce polymeric C₁N₁ can be found as early as in the 1800s when scientists condensed metal cyanides (i.e., production of paracyan), the advances in technology did not allow then to characterize whether the produced materials were crystalline or had a good photocatalytic response (Lopez-Salas et al., 2020). Surprisingly, despite the potential of C₁N₁ materials, the amount of reported cases of successful C₁N₁ syntheses is still very limited. In 2004, Gao et al. reported for the first time the preparation of tubular luminescent polymeric protonated C₃N₃ networks (i.e. [(C₃N₃)₂(NH)₃]_n) through a solvothermal route involving the reaction of C₃N₃Cl₃ with NH₄Cl in the presence of a little iron powder at 300°C (Guo et al., 2004). The tubular structures formed exhibited a characteristic of 002 basal plane diffraction peak in XRD diffactogram and comprised triazine rings according to FTIR. The luminescent and UV–vis absorption characteristics pointed at π–π* electronic transitions.

Soon afterward, the same group reported for the first time the synthesis of graphite-like lamellar C₁N₁ structures through the Wurtz reaction of C₃N₃Cl₃ in molten alkali metals (Na, K) (Guo et al., 2005; Yin et al., 2017). The obtained materials showed different compositions (i.e., g-CN_1.20, g-CN_0.96, and g-CN_0.25) depending on the synthetic temperature used (220°C, 300°C, and 380°C, respectively). All samples exhibited at 002 diffraction peak in their XRD patterns, but sample g-CN_0.96 showed poor crystallinity compared with samples prepared at 220°C and 380°C, and the authors suggested post-treatments (such as annealing) to improve its crystallization. However, such treatment would only lead to composition changes. Moreover, the structures showed weak SAED diffraction rings indicating the disorder of the lamellar microstructure of the samples.

A time gap was given from these last results until the next C₁N₁-like structures were reported. In 2014, Liu et al. reported the preparation of similar materials by the trimerization of 2,4,6-tricyano-1,3,5-triazine in molten ZnCl₂ (to avoid evaporation) at 400°C (Liu et al., 2014). The produced polymers had a multilayered 2D shape and virtually no accessible surface area as shown by nitrogen adsorption–desorption isotherms at 77 K. In 2017, H. Yin et al. used the method of Guo to prepare C₃N₃ from C₃N₃Cl₃ in the presence of NaCl in autoclaves. The materials showed the typical 002 basal plane diffraction, comprised triazine rings, and had a C:N ratio of 1. Again, the materials exhibited poor N₂ adsorption uptake according to N₂ adsorption–desorption isotherm at 77 K. Interestingly, the material Coulombic efficiency was enhanced at 600°C and the C₃N₃ products were used in Li-ion batteries. Probably due to the large heat treatment temperature, Yin and coworkers obtained black products and did not report on the color of the obtained material or their absorption properties. Also, Liu et al. did not investigate further in this regard.

In 2019, C₃N₃ frameworks obtained from 2,4,6-tricyano-1,3,5-triazine also using an ionothermal approach were also reported by J. Zheng et al. (Scheme 2) (Zeng et al., 2019). They used a pre-polymerization step in triflic acid and then a heat treatment in ZnCl₂ at 400°C for 6 days. The resulting material showed two broad peaks at 14.10° and 27.25° corresponding to 100 and 002 reflections of A−B stacking of layered C₃N₃ ordered structures (with unit parameters a = b = 7.25 Å and c = 6.64 Å). The materials were then coordinated with FeCl₃ and used as oxygen reduction electrocatalysts. Also in 2019, S. Fenetti et al. reported the preparation under high-pressure conditions of C₆N₆-like materials using 2,4,6-tricyano-1,3,5-triazine as a precursor (Fanetti et al., 2019). The behavior of the precursor under different temperature and pressure conditions was analyzed. Synchrotron x-ray diffraction suggested the presence of layered carbon nitride domains or structurally disordered nanocrystalline phases within the sample. More recently, a synthetic route to produce high-quality C₃N₃ materials in a relatively larger scale was reported by J. Feng and M. Li (Feng and Li, 2020). They used the Ullman reaction to also polymerize cyanuric chloride (C₃N₃Cl₃) on top of a copper foil. In the paper, the authors describe how using copper films is much advantageous than the previously reported synthetic methods because copper promotes the formation of radical instead of anionic intermediates, which favor the formation of large amounts of crystalline products, can migrate into the material matrix, and allow layer by layer growth of C₃N₃ (Figure 3). The reaction was optimized at 260°C, and a gray layer on top of the copper film was produced, which already pointed at the formation of a photoactive material. Before acid washing, the sample is crystalline, and its XRD pattern corresponds to (C₃N₃)·(CuCl). However, acid washing leads to the loss of crystallinity of the sample.

Copper-anchored C₃N₃ frameworks exhibited a much lower electrochemical overpotential for the hydrogen evolution reaction (306 mV vs. RHE at 10 mA cm⁻²) than the acid washed sample (533 mV vs. RHE at 10 mA cm⁻²). Due to its calculated bandgap of 2.2 eV, the authors also ran experiments using light excitation with a 300-W Xe lamp with a 420-nm filter in 1 M KOH electrolyte, with Ag/AgCl as reference electrode and a graphite rod as counter electrode. Photoelectrochemical reduction of hydrogen over copper-anchored C₃N₃ occurred with even lower overpotential (256 mV vs. RHE at 10 mA cm⁻²) probing the photoactivity of the material. Moreover, chopping light experiments showed that current increased sharply when light was on indicating that charge recombination was negligible. All in all, the results of copper-loaded C₃N₃ indicated that these materials are very promising photocatalysts. However, more research efforts are necessary to further produce pure C₃N₃ networks and explore the difference in the photoelectrochemical response that different dopants induce in the frameworks.

C₂N is probably the member of the carbon nitride family that has attracted more attention in the last years. C₂N is also a porous material ,and as for C₁N₁, its structure has electron-rich 6 N-pores that are predicted to anchor metals and adsorb small molecules (Du et al., 2016; He et al., 2016; Bhattacharyya et al., 2018; Qin et al., 2018). In 2015, Mahmood et al. reported for the first time the preparation of layered two-dimensional C₂N via a simple wet-chemical reaction (Mahmood et al., 2015). The material was prepared by simply mixing hexaaminobenzene (HAB) trihydrochloride and hexaketocyclohexane (HKH) octahydrate in N-methyl-2-pyrrolidone and using a few drops of acid as catalyst. Moreover, the resulting material was dark black in color and exhibited calculated and experimental bandgaps of ca. 1.70 and 1.96 eV, respectively (Figure 4) which points at this material as an excellent photocatalyst. The material was used to prepare a field-effect transistor device fabricated that exhibited an on/off ratio of 10⁷.

Ever since it was synthesized for the first time, reports on using C₂N materials have flourished (Tian et al., 2020), and the synthetic routes to produce C₂N have been also described elsewhere (Tian et al., 2020; Tan et al., 2021). However, experimental reports describing the photocatalytic activity of C₂N are still rare. In 2018, Lei Wang et al. reported the preparation of Z-scheme overall water splitting C₂N and aza-fused microporous polymeric (aza-CMP) ultrathin nanosheet photocatalysts (Wang L. et al., 2018). Simultaneous evolution of H₂ and O₂ with a molar ratio of 2:1 was observed with a maximum H₂ production rate of 5 µmol/h under irradiation of a 300-W Xe lamp filtered at 420 nm, and a solar-to-hydrogen conversion efficiency of 0.23% with quantum efficiency of 4.3% at 600 nm. C₂N was prepared as described by Mahmood, and none of the two polymers was able to catalyze the overall water splitting under visible light when used independently. However, the aza-CMP/C₂N heterojunctions were obtained by mixing and annealing the two types of nanosheets. Comparison of the N K-edge XANES spectra of aza-CMP, C₂N, and aza-CMP/C₂N indicated the formation of the junctions. Different aza-CMP/C2N ratios were used and the final heterojunctions were able to produce H₂ and O₂ in a 2:1 stoichiometric ratio (Figure 5).

Interestingly, based on the already reported results on aza-CMP-C₂N Z-schemes, Y. Fan et al. reported that the same heterojunctions hold a great potential as photocatalysts for the CO₂ reduction reaction as calculated by first-principle computations (Fan et al., 2021). However, still there are no more experimental evidences of the performance of C₂N or C₂N heterojunctions as photocatalysts.

In 2019, X. Zhang et al. reported 41%–56% exfoliation yields for C₃N₄, C₂N, and aza-CMP in acid media (Zhang et al., 2019). In their work, C₂N was also synthesized using HAB and HKH in NMP. However, sonication of the materials in 1 M HCl or 12 M HCl solutions yielded 5-nm-thick C₂N layers that remain suspended stably for months. No peak shift was observed by PXRD before and after exfoliation, and thus, the authors hypothesized that the exfoliation took place through a delamination process facilitated by water molecules and not induced via Brønsted acid intercalation. Moreover, all the samples exhibited similar N K-edge XANES and N1s XPS spectra and unchanged ¹H MAS solid-state NMR, indicating that the structures remain intact after the treatment. The photo-response of the exfoliated C₂N was fivefold larger than g-C₃N₄ in water/triethanolamine solution under visible light (λ ≥ 420 nm) with a 300-W Xe lamp (100 mW/cm²), as chopping light experiments indicated. Moreover, the sample response was very stable after cycling (Figure 6).

The crystalline structure of C₃N, as well as its electronic configuration, was first predicted in 2012 (Hu et al., 2012), and this attracted immediately the attention of the scientific community as a consequence of its semiconductor behavior, their optoelectronic properties, high specific surface area, among other chemical, and physical characteristics. Since then, a considerable number of theoretical studies have investigated its mechanical, electronic, optical, structural, etc., characteristics in different configurations, such as monolayer, bilayer, bulk, or alternatively, modifying the crystalline structure with defects in the form of vacant, voids, or heteroatom doped (Wang X. et al., 2018; Shi et al., 2018; Shirazi et al., 2018; Tagani, 2018; Wu et al., 2018). However, only a few papers have reported synthetic procedures to obtain crystalline C₃N, and as consequence, the direct application of this material in photocatalysis has not been explored so far to the best of our knowledge, despite theoretical calculations having predicted its suitability.

Encouraged by the striking properties predicted by theoretical calculations, bulk C₃N was first prepared from the thermolysis of m-phenylenediamine at 800°C under static vacuum for 7–9 days by King et al. in 2013 (King et al., 2013). In this preparation method, hollow carbon microspheres in the absence of N atoms were also obtained. XRD measurements showed two broad signals at 26° and 44° corresponding to the 002 and 100 facets, respectively, indicating that the C₃N possess multi-layered turbostratic graphitic arrangement without periodic ordering between each C₃N layer. XPS analysis confirmed that the flakes contained a substantial amount of nitrogen, obtaining an empirical formula of C_3.33N. The excess of C atoms from the ideal composition was attributed to contamination from the microspheres.

Lately, Baek and co-workers reported in 2016 the synthesis of 2D C₃N crystals from the direct pyrolysis of hexaaminobenzane trihydrochloride (HAB) single crystals at 500°C for 2 h (Mahmood et al., 2016b). They studied in detail the C₃N formation mechanism, which involved the evolution of ammonia and ammonium chloride, following Eq. 1. The C₃N chemical composition was confirmed by XPS, which was in good agreement with the theoretical calculations performed so far. They also characterized the structure by solid state ¹³C NMR (ss-¹³C NMR), obtaining only two characteristic peaks of 150.80 and 126.02 ppm, attributed to carbon atoms bonded to tertiary nitrogen and carbon atoms linked to primary or secondary amines, respectively. In this case, XRD revealed three major peaks at 12.8°, 26.1°, and 44.7° corresponding to d spacing of 6.87 Å, 3.40 Å, and 2.02 Å, respectively. The peak at 26.1° was attributed to the interlayer d spacing (3.40 Å) of the obtained material. This crystallographic result was in good agreement with the simulated XRD patterns after Pawley refinement. The electrical properties of the as-prepared C₃N were also investigated, and a conductivity of 0.72 S/cm was determined, which makes this material an ideal candidate for opto-electronic applications. C 6 N 6 H 15 C l 3 → C 6 N 2 + 3 N H 4 C l + N H 3 (1)

In the same year, Lazauskas and coworkers reported the synthesis of C₃N thin films from polymeric C_xN_y powder thermal evaporation (Lazauskas et al., 2016). The C_xN_y precursor was obtained through a two-step synthetic procedure by direct urea thermal treatment at 300°C for 3 h, and after fine grinding and milling, the obtained material was heated again at 520°C for 4 h. XRD analysis determined that during the first thermal treatment, urea decomposed to biuret, which subsequently also decomposed, evolving to cyanuric acid. In the second thermal treatment at 520°C, cyanuric acid decomposes to ammelide, ammeline, and melamine at temperatures below 400°C. At larger temperatures, these compounds evolved to polymeric C_xN_y. The obtained material was characterized by an orthorombic crystal structure and an A–B layers staking. Finally, C₃N thin films were deposited on arbitrary substrates by thermal vacuum deposition of the polymeric C_xN_y at 10⁻⁶ mbar and 1 Å/s rate. The chemical composition and structure were determined by XRD, XPS, FTIR, and Raman spectroscopy. XRD pattern showed a broad peak at 23.7°, indicating the presence of dispersed polymeric C_xN_y fragments. The authors suggested that during thermal vacuum deposition, the polymeric C_xN_y is fragmented, and the previously observed crystallinity is no longer observed in the thin films. The composition was determined by XPS, obtaining a surface atomic concentration of C_2.94N. UV-Vis measurements showed an absorption edge at approximately 423.2 nm, corresponding to an optical band gap of 2.93 eV.

One year later, Yang et al. reported the controllable large-scale synthesis of C₃N QDs through 2,3-diaminophenazine polymerization by hydrothermal treatment (Yang et al., 2017). The polymerization steps were characterized by matrix-assisted laser desorption ionization time-of-flight mass spectroscopy (MALDI-TOFSMS). Vertical and horizontal polymerization was advanced by the addition of precursor monomers leading to the formation of large C₃N flakes. The vertical polymerization consisted first in H abstraction from C–H and N–H bonds, providing C and N atoms with dangling bonds, subsequent formation of C–N bonds between the growing polymer fragment and the added monomer, and finally the release of the excess of hydrogen. The reaction for horizontal polymerization of two monomers and the whole polymerization process is depicted in Scheme 3. HRTEM and SAED (Figure 7) exhibited a well-crystallized graphene-like honeycomb structure and the two-fold hexagonal single crystal structure of C₃N indicating D_6h-symmetry of the N and C atoms in agreement with theoretical calculations.

More recently, Ma et al. reported the synthesis of C₃N from the imidazolate-based MOF ZIF-8 and urea (Ma et al., 2019). These two precursors (ZIF-8 and urea) were dispersed in water for 2 h, filtered and annealed at temperatures between 650 and 950°C under N₂ atmosphere for 2 h. The obtained carbonaceous material was, subsequently, HCl washed, filtered, and dried prior to characterization. The successful C₃N composition formation was confirmed by XPS for the sample annealed at 850°C. XRD measurement showed two characteristic peaks at 25° and 43° attributed to the 002 and 100/101 diffractions of carbon, indicating the existence of disorderly oriented graphitic type fragments in the carbon matrices.

In spite of the synthetic efforts carried out to prepare crystalline C₃N materials with different morphologies, only a few examples of their application in photocatalysis have been reported and are mainly based on theoretical calculations. In 2016, Baek et al. reported the band gap of C₃N to be 2.6 eV (as calculated electrochemically and using scanning tunneling spectroscopy) (Mahmood et al., 2016a). Thus, it was predicted that heterostructures composed of the semiconductor materials BC₃ and C₃N can form p–n heterojunctions with strong interlayer electron coupling and enhanced photocurrent (Zhang et al., 2018). With this in mind, Wang et al. studied the formation of BC₃/C₃N and BC₃/BC₆N van der Waals heterostructures as Z-scheme heterojunctions for photocatalytic overall water splitting by DFT calculations (Wang Z. et al., 2020). The theoretical investigations revealed that the delocalized nature of π-conjugated electrons in the monolayer components allows for high carrier mobility between the different layers, promoting efficient photocatalytic overall water splitting. Thus, these heterojunctions were proposed as good candidates for photocatalytic water splitting. However, experimental studies confirming the simulations carried out with these heterojunctions as well as with other photocatalytic systems containing C₃N have not been reported yet. Hence, the use of C₃N is this area remains unexplored despite the predicted favorable optoelectronic properties.

Boron carbonitride (BCN) materials are inorganic semiconductors, which exhibit a lower binding energy of excitons and faster charge migration properties than other polymeric photocatalysts, such as covalent organic frameworks (COFs), conjugated microporous polymers (CMPs), and graphitic carbon nitrides (g-CN). BCN are visible light-responsive semiconductors, and the band gap energy and energy levels can be easily tuned depending on the amount of carbon incorporated into the boron nitride. Recently, Yuan et al. reported the preparation of CBN and its application in photocatalytic organic transformations (Yuan et al., 2021). CBN was prepared by mixing and grinding urea, boric acid, and glucose in different proportions. Then KCl was added and ground again. The mixed precursors were heated at 1,250°C for 5 h under ammonia flow. The obtained BCN was used as visible light-responsive photocatalyst (LED light source at 420 nm, 15 W) for the coupling of halides with different synthons. It was demonstrated that a wide range of organic halides, including aryl-, alkyl-, bromides, and chlorides, could be reduced to carbon-centered radicals at remarkable yields.

Alternatively, Rao and coworkers reported the synthesis of BCN with two different compositions, (BN)_0.75C_0.25 and (BN)_0.3C_0.7, by varying the ratio of urea, boric acid, and activated charcoal, and heating the resultant mixture at 900°C in N₂ atmosphere (Singh et al., 2018). The photocatalytic activity of these materials forming composites with MoS₂ (1:2) was investigated for H₂ evolution. Aqueous suspensions of the two different composites containing triethanolamine (20 vol.%) were irradiated with UV-vis light with a halogen lamp (100 W). (BN)_0.75C_0.25/MoS₂ composite showed the highest H₂ production rate (6,965 µmol/g·h).

Barrio et al. reported the preparation of C₃N₃O and its photocatalytic activity toward RhB degradation (Barrio et al., 2019). C₃N₃O photocatalyst was synthetized by means of the preparation of tri-aminopyrimidine-cyanuric acid supramolecular assemblies by mixing both monomers in a 1:1 molar ratio and subsequent calcination at low temperatures (300–400°C). C₃N₃O calcined at 400°C exhibited a 2.17 eV band gap with a conduction band maximum (CBM) suitable for CO₂ reduction, while the C₃N₃O calcined at 300°C presented a 1.41 eV band gap with CBM above the H⁺ reduction potential, but not enough for CO₂ reduction. Thus, the photocatalytic activity for these two materials for RhB degradation was tested under visible light illumination (λ ≥ 410 nm), and C₃N₃O calcined at 400°C showed the highest photoactivity, while C₃N₃O calcined at 300°C presented negligible activity.

The different experimental procedures developed for the synthesis of C_xN_y materials are in its infancy. For instance, the majority of the synthetic routes are based in either expensive or toxic compounds, and the material processability is still very limited. Thus, one of the main current challenges for researchers working on the field is to find new and scalable strategies to produce these materials.

Regarding their application as photocatalysts, some of the performance of the materials have been theoretically predicted and simulated but, not fully explored neither experimentally nor theoretically yet. For instance, majority of the reports on the photocatalytic performance of C_xN_ys focus on water splitting. However, theoretical calculations predict that heteroatom-doped C_xN_y, as well as heterojunctions with other 2D materials, promote important differences in the final material band gap, which forecast a great potential of the C_xN_y family in a broader spectrum of photocatalytic reactions. Moreover, the porous nature of some of the materials (i.e., C₁N₁ or C₂N) can facilitate the adsorption of certain molecules extending the limits of the photocatalytic activity of the materials from only water splitting to more complex water purification systems. In this regard, we also expect material processability (for example, in the form of nanoribbons, thin films, powders, and quantum dots) to have a great impact in the overall performance as photocatalysts (i.e., facilitating the creation of defects or heteroatom doping itself).

Overall, we envision that C_xN_y materials will be extensively used as photocatalysts for a variety of reactions in the near future. However, strong efforts are still needed so the possibilities of these materials are fully explored, their limits understood, and they are produced through scalable synthetic routes.