Current Strategies to Improve the Properties of Graphitic Carbon Nitride for Effective and Scalable Wastewater Pollutant Removal: A Critical Review

Abstract

Heterogeneous photocatalysis has emerged in recent years as a promising and sustainable decontamination method. However, several drawbacks limit the effective usage of this process up to date, such as photocatalysts’ limited properties, difficulty in modifying and improving their properties, as well as the environmental impact and cost associated with the use of the metals on which conventional photocatalysts are based. Graphitic carbon nitride (gCN), a new carbon-based photocatalyst, offers the possibility of easy modification and improvement of their properties. There are several strategies to improve the properties of these derivatives, such as increasing the surface area (modifying morphology into 0D, 1D, 2D, or porous structures), increasing the absorption in the visible (doping), and improving the separation and mobility of photogenerated charges (introducing defects, vacancies, functional groups, and doping). In this review, a compilation of these modifications, the associated improvements in its properties, and its derivatives was carried out with focus on the degradation of emerging pollutants (EPs). The property modifications enhance their behavior and efficiency of degradation, all in a cheaper and more sustainable way. Thus, improved gCN derivatives offer real possibilities for the upscaling of heterogeneous photocatalytic processes as an effective alternative for decontaminating water bodies.

1. Introduction

Graphitic carbon nitride (gCN) has emerged as a highly promising photocatalytic material, particularly for applications under visible light irradiation. Its advantages include the use of low-cost, widely available, and environmentally friendly precursors, along with its straightforward synthesis and ease of structural and electronic modification. It offers suitable band gaps (valence band (VB) ≈ [(+1.6)–(+1.4)] eV; conduction band (CB) ≈ [(−1.1)–(−1.3)] eV) as well as appropriate redox potentials (suitable mainly for the production of superoxide, O₂(g)/O₂^•− ≈ −0.33 eV, or perhydroxyl radicals if the medium is acid, O₂/HO₂^• ≈ +0.07 eV, but not so much for the direct production of hydroxyl radicals, HO^•/H₂O ≈ +2.3 eV). Additionally, gCN shows high thermal stability (up to 600 °C) in diverse reaction media (acidic media, basic media, and organic solvents such as alcohols, DMF, THF, diethyl ether, toluene), and demonstrates excellent resistance to oxidation in air. In addition, gCN is inherently non-toxic, which makes it a compelling alternative to conventional metal-based photocatalysts [1,2,3,4,5,6,7].

The history of this compound is almost 200 years old. It was Jöns Jakob Berzelius who, in 1822, first described the synthesis of gCN (heptazines synthesized by pyrolysis of mercury thiocyanate, Hg(SCN)₂), and it was Justus von Liebig, in 1834, who ended up calling it ‘melon’. Its empirical composition was detailed by Franklin in 1922, who determined that it varied according to the conditions of synthesis. In 1937, Linus Pauling studied the crystal structure of heptazines; it was in 1940 that Redemann and Lucas determined ‘melon’ has a lamellar structure like that of graphite [8,9]; and it was in 1966 that Teter and Hemley elucidated the five different crystalline phases of C₃N₄ (graphitic, cubic, pseudo-cubic, α, and β phases) [10]. However, it was not until 2009 that Wang et al. [9,11] described the first documented use of gCN in photocatalysis (production of H₂ from H₂O).

The applications of bulk gCN and its modifications in photocatalysis have been documented extensively in the literature, with over 8619 publications (limited to articles, reviews, and book chapters with g-C₃N₄ (gCN) in the article title) in the last five years focusing on gCN and its derivatives, particularly in fields such as physical chemistry, materials science, and chemical engineering (Scopus, 8 April 2025). The application areas of these modified gCN materials include not only environmental remediation, but also hydrogen production, CO₂ reduction, and solar energy conversion, as inferred from the keyword co-occurrence map (Figure 1), based on the total referenced articles (≥250 rep.).

The main reported synthesis methods include thermal condensation and polymerization (in inert environment and in aerobic environment, respectively), hydrothermal condensation, reflux techniques, the ionothermal method, chemical and physical vapor deposition (CVD and PVD), and electrochemical deposition [1,6,7,12]. Among the methods mentioned, thermal condensation and polymerization are the most widely used [1,9,12,13], but this synthesis has poor yield and purity [1,5]. Additionally, some of the main precursors used in the synthesis of gCN include cyanamide, dicyandiamide, urea, thiourea, and melamine, but also other nitrogen-rich carbon compounds such as melamine, cyanurate, ammonium thiocyanate, hexamethylenetetramine, guanidine carbonate or guanidine hydrochloride, etc. [4,5,6,7,9,14].

Thus, two main phases are possible within the gCN family: those based on triazines, and those based on heptazines (Figure 2). It has been found that heptazine-based structures are more thermodynamically stable (30 kJ/mol more stable) [15,16]. For example, precursors such as guanidinium, cyanamide, and melamine; those used with eutectic molten salts; or those copolymerized with trimesic acid, as well as use of inert environments, favor these triazine structures [17,18,19,20].

Triazine-based gCN tends to have a wider band gap (2.8–3.1 eV) and tends to have a faster recombination of photogenerated charges due to the smaller extent of the π cloud (less conjugation) [21,22,23,24]. It is usually easy to differentiate between the two structures; the diffractogram of a triazine-based gCN (with A-B stacking) usually shows a peak corresponding to the (1,0,0) plane at 2θ = 20–21° (CCDC 2217016) [19], and heptazine-based gCN (with A-B stacking) at around 2θ = 15–17° (CCDC 1714809) [25], both for the hexagonal crystalline system (P-6m2 space group, Figure 3a). In the FTIR spectra, the peaks corresponding to the out-of-plane bending of the rings, around ≈810 cm⁻¹ in the heptazine-based gCN, are usually sharp and pronounced, while the triazine-based peaks are usually either attenuated or absent; also, heptazine-based gCNs show higher intensity in N-H stretches (3000–3500 cm⁻¹) due to the higher content of primary and secondary amines (Figure 3b) [10,17,26].

The choice of precursor and reaction conditions of the thermal treatment (Figure 4) directly influence the specific surface area, band gap, and overall photocatalytic efficiency of gCN [6,12,14,27,28,29,30].

Regarding reaction conditions, it is worth noting that higher condensation temperatures result in a larger surface area and a smaller band gap, although they also lead to reduced yield. When synthesized under the same conditions, the yield decreases according to the following preference order: melamine ≈ dicyandiamide > cyanamide > thiourea > urea [6,12,31]. Conversely, a lower heating rate enhances crystallinity and promotes charge mobility but significantly reduces the surface area [31,32].

Under the same thermal synthesis conditions, urea as a precursor yields a significantly larger surface area (especially in air, in a non-inert environment) although slightly wider band gap than with other precursors [27,29,33]. Among common precursors, dicyandiamide results in the smallest band gap [30].

The precursors urea and thiourea are particularly noteworthy. Urea is one of the most used precursors due to its low cost and ease of handling. During thermal treatment, urea undergoes a series of transformations (Figure 5), starting with the formation of isocyanic acid (≈175 °C), followed by condensation into melamine (>200 °C), which, after condensing as melam (390 °C), melem (>390 °C), and melon (500 °C), undergoes final condensation as gCN at 550 °C [1,5,34]. However, precursors such as cyanamide or dicyandiamide, apart from being expensive reagents, are difficult reagents to handle due to their explosive nature [1]. It is worth mentioning a common misconception found in the literature, namely, that the transformation of urea into cyanamide (by dehydration of the urea) then into dicyandiamide (by dimerization of two cyanamide units) gives rise to melamine: this route is impossible; it would only be possible in inert, anhydrous environments, and at low pressure [15,35,36,37,38].

Currently, advanced synthesis methods are being used to overcome the limitations of traditional thermal synthesis; alternative methods such as microwave-assisted synthesis and solvothermal methods have been explored. Microwave-assisted synthesis, for example, not only reduces synthesis time but also increases the number of active sites on the gCN surface, leading to improved photocatalytic performance [7,9]. In the case of solvothermal methods, although more complex and requiring toxic solvents, they offer better control over the morphology and crystallinity of the synthesized gCN, making them suitable for specific high-performance applications [13,39].

Despite the aforementioned advantages, pristine gCN suffers from several intrinsic limitations, such as a relatively large band gap (~2.75 eV, or 450 nm; only absorbs in the blue region of visible spectra), low specific surface area, rapid recombination of photogenerated electron–hole pairs, low conductivity (high internal resistance), low electronic mobility, and slow surface reactions [1,7,40]. These factors significantly limit their photocatalytic efficiency, particularly in environmental applications such as the degradation of emerging pollutants (EPs) in water bodies.

To address these challenges, various modification strategies have been developed to enhance the photocatalytic properties of gCN (Figure 6). These modifications can be broadly categorized as follows:

• Morphological modifications: Adjusting the morphology of gCN to create structures like 0D quantum dots (QDs); 1D nanorods, nanowires, or nanotubes (NRs, NWs, and NTs, respectively); 2D nanosheets (NSs); and 3D porous networks. These modifications mainly enhance the surface area as well as improve the mobility of the charge carriers (their diffusion to the surface) and may even reduce the band gap of the material by introducing defects in its morphology [4,6,13].
• Surface functionalization: Introducing functional groups like amino, imino, cyano, hydroxyl, and carboxyl groups to the gCN surface increases the specific surface area and thereby increases the number of active sites in the gCN [4,15,41,42].
• Modification of band structure:
  Vacancies and defects: The introduction of defects such as nitrogen and carbon vacancies in gCN can significantly influence its photocatalytic performance. Nitrogen vacancies (Nv) can enhance charge separation and extend the lifetime of photogenerated excitons [4,6], while carbon vacancies (Cv) can serve as electron reservoirs, facilitating the generation of superoxide radicals (O₂^•−) [43,44].
  Metal, non-metal, and self-doping: Doping with metal and non-metal elements influences the electronic structure of gCN by not only improving the absorption range (due to the introduction of intermediate energy levels in the band gap, as well as vacancies) but also by increasing the lifetime of the charge carriers and improving their separation [1,13]. This doping can also correct the increased gap produced in nanostructures (e.g., nanosheets) by the quantum confinement effect [1,13].

Figure 6. Methods of gCN modification and improvements in its properties. Grey: morphology modifications, blue: modifications by doping, beige: modifications by vacancies and defects, and green: surface modifications.

Figure 6. Methods of gCN modification and improvements in its properties. Grey: morphology modifications, blue: modifications by doping, beige: modifications by vacancies and defects, and green: surface modifications.

These modifications, applied individually or synergistically to gCN, improve its photocatalytic efficiency to degrade emerging pollutants in aquatic environments; however, applied individually, they can also lead to disadvantages such as possible increases in band gap (nanoarchitecture), the use of toxic reagents and reduced exciton mobility (vacancies), accelerated exciton recombination (non-metal doping), or secondary contamination due to leaching of metal ions (metal doping) (

Table 1. Advantages and drawbacks of the gCN different modifications.

Structure	Photocatalyst	Advantages	Refs.	Drawbacks	Refs.
0D, 1D, 2D and porous 3D	0D: gCNQDs; 1D: gCNNW, gCNNRs, gCNNF, or gCNNTs; 2D: nanosheets; or porous 3D	Favor the migration of charge carriers. Enlarge the active surface. Improve visible light absorption (3D porous). 0D, 1D, and 2D solubility in water. Non-toxic.	[4,6,13]	Quantum confinement effect by size reduction in 0D, 1D, and 2D structures. Gap broadening.	[6,13,14,45,46]
Functional groups	gCN with amino, imino, cyano, ureido, hydroxyl, carboxyl, and aromatic groups.	Amino and imino groups improve the anchoring of metal oxides during their support, as well as their dispersion on the surface. Improved visible light absorption. Improved separation of charges. Increased specific surface area and active sites number (-OH) may act as h+ trapping centers enhancing charge separation.	[4,41,42,47,48]	Cyano, ureido, carboxyl and aromatic groups often require toxic reagents that can cause contamination. Mobility of photogenerated charges restricted by the presence of these groups.	[1,42,49,50]
Vacancies	Cv-gCN	Serve as a reservoir for photogenerated electrons. Inhibit the recombination of h+ and e−. Serve as electron transfer centers for the adsorbed molecular oxygen favoring the production of superoxide radicals. Reduce the effective band gap.	[43,44]	Mobility of photogenerated loads is restricted by the presence of these vacancies.	[50]
	Nv-gCN	Improve the separation of charges. Increase the lifetime of excitons. Reduce the effective band gap.	[4,6,50,51]
Doping	Non-metallic doping	Introduction of intermediate energy levels in the gap. Reduction in effective gap. Increased exciton lifetime. Corrects for the increased gap due to the quantum confinement effect.	[1,2,13,40]	Non-metallic species do not participate in the transport of charges; they act as exciton recombination centers.	[12]
	Self-doping (C-gCN and N-gCN)	Introduction of intermediate energy levels in the gap; reduction in effective gap. Increased exciton lifetime. Corrects the increased gap due to quantum confinement effect. C-doping increases the number of delocalized π-bonds, improving conductivity, charge transfer, and their separation. Significant gap reduction.	[52,53,54]
	Metal doping	Introduction of intermediate energy levels in the gap; reduction in effective gap (especially with alkalis). Increased exciton lifetime. Correction for the increased gap due to the quantum confinement effect. With noble metals, the effect of surface plasmon resonance (SPR) is introduced; the excitation of electrons to the conduction band is enhanced (greater number).	[1,13]	Causes secondary contamination due to leaching of metal ions. Excess metal ions act as exciton recombination centers. Serious pollution is associated with the mining of metals and their refining.	[12]

).

The modification strategies may be helped by the use of computational tools. It is worth highlighting the versatility of ab initio computational methods, such as Hartree–Fock theory and density functional theory (DFT), in the design and optimization of nanomaterials. While Hartree–Fock provides high accuracy for small systems (such as triazine or heptazine units), DFT stands out for its balance between accuracy and computational cost, making it ideal for studying complex nanostructures. These techniques enable the prediction of properties like band gaps, electronic density of states, and adsorption energies. The choice of exchange–correlation functional in DFT (e.g., B3LYP, HSE06) is critical, as it directly affects key parameters such as the band gap and band positions in DOS calculations [55,56].

In recent years, machine learning has emerged as a powerful complement, allowing for the prediction of complex material properties based on theoretical or experimental data [57,58]. In the case of g-C₃N₄ and its derivatives, these in silico techniques have been essential to understanding their structure [59,60], electronic and optical properties [59,60,61,62], and the effects of modifications such as morphology, vacancies, functionalization, and doping in this photocatalyst [63,64,65,66,67,68,69].

These computational tools not only guide nanoarchitecture design (Section 2), the introduction of functional groups (Section 3), and defect engineering, but also allow for synergy prediction in multi-element doping (Section 4), thereby reducing costly experimental trials. They provide researchers with a series of advantages that not only help optimize experimental design but also contribute to aligning such designs with the 12 principles of green chemistry [70].

Without disregarding the potential of catalyst combinations to enhance their electronic and optical properties—such as homojunctions or heterojunctions (between gCN derivatives or between gCN and other non-gCN-based photocatalysts, respectively); their immobilization, which facilitates recovery from the reaction medium (both of which are also key for developing effective and scalable photocatalytic pollutant degradation systems in water bodies); or the stability and toxicity (both also key aspects for the viability of heterogeneous photocatalysis as a decontamination method) of gCN and its derivatives—it should be noted that each of these aspects requires extensive, stand-alone monographic analysis owing to the diversity and complexity in these fields. It is for all these reasons that the main objective of this bibliographic review is solely to provide a comprehensive overview of current experimental strategies used to improve the intrinsic properties of gCN (morphology, defects, functionalization, and doping) for effective and scalable photocatalytic water remediation.

Thus, in this review, the synthesis methods, advantages, and limitations of various gCN modifications are examined, focusing on their impact on photocatalytic efficiency and the degradation of emerging pollutants.

2. Nanoarchitecture Design of gCN

Regarding modifications related to nanoarchitecture (morphology), this review refers to those aimed at altering not only the state of aggregation (0D, 1D, and 2D), but also the porosity in 3D structures. As previously noted, these modifications primarily contribute to enhancing surface area, improving charge carrier mobility, and even reducing the band gap. Thus, there are two main routes to this target: bottom-up, like thermal treatment for 3D structures from precursors, solvothermal-assisted microwave and chemical or physical vapor deposition (CVD and FVD, respectively) from precursor for 0D structures, and supramolecular self-assembly or template methods from precursor for 1D structures; top-down strategies like chemical, ultrasound, microwave, and thermal exfoliation for 2D structures from 3D structures and hydrothermal and sonication treatments from 2D structures for 0D structures (Figure 7) [4,6,13].

2.1. Zero-Dimensional Structures (0D)

Zero-dimensional nanomaterials are structures confined to the nanoscale in all three spatial dimensions. They are particles smaller than 10 nanometers (<10 nm), typically consisting of a few thousand atoms such as quantum dots. These materials lack defined length, width, or height, and their properties are highly dependent on size and quantum effects. Typical 0D nanomaterials used in photocatalysis are metal oxides (TiO₂, ZnO), MXenes, chalcogenides (Cds), carbon quantum dots (CQDs), fullerenes (C₆₀), and gCNQDs, among others [71,72,73].

Due to their small size, gCNQDs exhibit a morphology that favors the migration of charge carriers [13], enlarging the active surface and enhance light absorption [13,45]. All these properties can be modulated by controlling their size. As their dimensions decrease, these structures display the so-called quantum confinement effect—a specific manifestation of quantum effects arising from size reduction—which, according to the Brus equation, leads to the band gap. In addition, gCNQDs also exhibit photoluminescence and water solubility and are considered non-toxic [6,13,14,45,74,75,76]. There are two different approaches to synthesize these types of nanostructures, as mentioned above, i.e., top-down and bottom-up techniques.

The most cited top-down methods include ultrasonic synthesis (sonication) and hydrothermal treatment, starting from exfoliated gCN (2D). Among these, it can be noted that sonication in a single step, although simple and easy to handle, requires a significant amount of time and ultimately results in a poor yield. However, sonication coupled with chemical oxidation, despite high yield and purity, is an expensive and complicated process. Similar observations apply to one-step and two-step hydrothermal treatments (coupled to oxidation): simple and cheap, with low yield for the one-step process, and complicated and expensive, but with good yield, for the two-step process [45,74,75,76].

Among the bottom-up processes, the microwave-assisted solvothermal method, the solid-phase process, and chemical vapor deposition (CVD) stand out. For the first one, there is an easy synthesis, with few steps, but with low yield; for the second one, there is high yield, good quantum yield of the material, and possibility to adjust the photoluminescence, but it is a long process, and post-treatments are necessary. As for chemical vapor deposition, it is only mentioned that it is an expensive process because of the high energy consumption, and it also requires post-treatment. Hydrothermal is the simplest, most sustainable, and most economical of the methods discussed [74,75,76].

Due to the main drawbacks of QDs, fast recombination of charge carriers, and wide band gap (owing to the effect of quantum confinement) [6,13,14,45,74,75,76], their main applications are in sensor fabrication, hydrogen production, and imaging.

The documented use of gCNQDs in decontamination involves both of the following:

• The fabrication of heterojunctions, mainly Z-type and II type (semiconductor/semiconductor);
• Heterojunction of the Schottky type (semiconductor/non-noble metal) [1,52].

As previously noted, since the aim of this work is not to provide an in-depth discussion of homojunctions or heterojunctions, only heterojunction types involving 0D gCN-based structures that demonstrate photocatalytic degradation of pollutants in aqueous environments are referenced in this subsection.

Examples of this type of compound in the degradation of pollutants in water bodies can be seen the work of Li et al. [77], in the degradation of Rhodamine B (RhB) (5 ppm) with a Z-type heterojunction of rTiO₂/gCNQDs (1:20; 1000 ppm) under visible light, prepared by heat treatment of TiO₂ (P25) and melamine at 500 °C for 4 h (2.3 °C·min⁻¹), which degraded 95% of the RhB in 240 min (the pristine gCN degraded only 35% in the same time) (

Table 2. Parameters of photodegradation of pollutants in water with different morphologies of gCN (0D, 1D, and 2D).

Morphology	Photocatalyst	Area (m2·g−1)	Gap (eV)	Pollutant (ppm)	Radiation	Efficiency (%)	Time (min)	k (min−1)	Refs.
0D	gCN (1000 ppm)	10.9	-	RhB: 5	500 W; Xe lamp (Vis.)	35	240	0.002	[77]
	rTiO2/gCNQD 15% (1000 ppm)	43.9	-			95		0.012
	gCN (600 ppm)	-	-	OTC: 40	500 W; Xe lamp (UV)	20	150	0.0017	[78]
	gCNQDs-CoTiO3/CoFe2O4 (600 ppm)	-	-			88		0.0141
	gCNQDs (600 ppm)	25.34	-	NOR: 30	1000 W; Hg lamp (UV)	35	120	0.0035	[79]
	gCNQDs/Ni5P4 8% (600 ppm)	83.61	-			92		0.022
	1D MCN (200 ppm)	29.07	-	RhB: 10	300 W; Xe lamp (Vis.)	99.5	30	0.12	[80]
	0D UCN (200 ppm)	44.27	-			70	30	0.036
	0D/1D MUCN (200 ppm)	57.24	-			99.96	20	0.26
1D	gCN (400 ppm)	9.9	2.65	SMX: 10	300 W; Xe lamp (Vis.)	15	140	0.001	[81]
	gCNNT (400 ppm)	100.4	2.31			100	120	0.035
	gCN (1000 ppm)	5.3	2.62	MO: 10	350 W; Xe lamp (Vis.)	71.1	120	0.009	[82]
	gCNNWs (1000 ppm)	74.25	1.52			98.5		0.02
	gCNNFs (1000 ppm)	60.16	1.61			90.9		0.01
	TGCN (1000 ppm)	8.52	2.8	PhOH: 20 MO: 20	SLB-300A, 300 W (Simulated sunlight)	PhOH: 50 MO: 62	PhOH: 180 MO: 150	0.004	[83]
								0.006
	SGCN (1000 ppm)	21.13	2.4			PhOH: 60 MO: 65		0.005
								0.007
	MGCN (1000 ppm)	31.84	2.45			PhOH: 85 MO: 92		0.01
								0.013
2D	Bulk (550 CN)	25	2.54	CIP: 5 RhB: 5	500 W; Xe lamp (Vis.)	CIP: 8	CIP: 60 RhB: 40	0.001	[84]
						RhB: 58		0.017
	Ultrasounds (550 ul CN)	12	2.56			CIP: -		_
						RhB: 75		0.027
	Thermic (475 CN)	29	2.42			CIP: -		_
						RhB: 50		0.009
	Acid (550 H CN)	114	2.73			CIP: 17		0.002
						RhB: 85		0.023
	Thermic + ultrasounds (475 ul CN)	21	2.46			CIP: 7		0.001
						RhB: 43		0.012
	Acid + ultrasounds (550 ul H CN)	74	2.74			CIP: -		_
						RhB: 80		0.03
	Thermic + acid (475 H CN)	71	284			CIP: -		_
						RhB: 80		0.018
	Thermic + acid + ultrasounds (475 ul H CN)	57	2.67			CIP: 57		0.006
						RhB: 92		0.037
	CN-B	18	2.42	Rh6G: 5	400 W; Xe lamp (Vis.)	67	30	0.031	[85]
	CN 500-4	31	2.56			78		0.046
	CN 550-4	107	2.68			92		0.085
	CN 550-5	165	2.73			96		0.102
	CN 550-6	295	2.89			98		0.139
	gCN	22.2	2.8	CIP: 20	Sunlight	45	60	0.009	[86]
	E-gCN	63.8	2.94			78		0.023

), attributing the improvements in degradation to a longer exciton lifetime (better charge separation; Figure 8a,b), and a slight increase in surface area, where the main species responsible for degradation are HO^● and h⁺ radicals.

Another example of a magnetically recoverable Z-type heterojunction based on gCNQDs is given by Feng et al. [78]. These researchers prepared a dual Z-type heterojunction of gCNQDs-CoTiO₃/CoFe₂O₄, by hydrothermal treatment and microwave-assisted impregnation, for the degradation of oxytetracycline (OTC). The authors were able to degrade OTC (40 ppm) with the compound gCNQDs-CoTiO₃/CoFe₂O₄ (0.8% in CoTiO₃; 600 ppm) by 88% in 150 min under UV light, where pristine gCN degraded only 20%, and gCNQDs degraded 30% within the same time and under the same reaction conditions (Table 2). The removal rate of OTC can remain above 85% after four cycles. They claim that the improvement in OTC photodegradation is mainly due to an improvement in the separation of the photogenerated charges (longer lifetime of the photogenerated charges; Figure 8c,d).

An example of a Schottky-type heterojunction could be the study reported by Feng et al. [79], in which gCNQDs/Ni₅P₄ were prepared by impregnation of gCNQDs on Ni₅P₄ nanoflowers, in EtOH and with ultrasound. With this Schottky-type gCNQDs/Ni₅P₄ (8% by mass of Ni₅P₄; 600 ppm), 92% of norfloxacin (NOR) (30 ppm) was degraded in 120 min under UV radiation compared to 35% achieved with the gCNQDs and 20% achieved by the unmodified gCN within the same time (Table 2). The NOR removal rate remained above 85% after four cycles. The improvement in degradation efficiency is attributed by the authors to better migration and separation of the photogenerated charges (Figure 8e,f).

Finally, an interesting homojunction (type II) between 0D and 1D structures of gCN (sample named as MUCN) was prepared by Zheng et al. [80] from melamine and cyanuric acid as supramolecular assemblies (gCNNTs; 1D) plus urea (gCN 0D nanoparticles) in a 2:1 ratio, and subsequent heat treatment at 520 °C 2 h, for RhB degradation (10 ppm). With this homojunction (MUCN; 200 ppm), 99.96% degradation was achieved in 20 min under visible radiation, where the urea-derived gCN (UCN) achieved only 70% and the melamine-derived gCN (MCN) 99.5% in 30 min under the same conditions (Table 2). As for the previous examples, the improvements in efficiency over the bulk materials are mainly due to improvements in the separation of the photogenerated charges (longer exciton lifetimes) and to a larger surface area (Figure 8g,h).

2.2. One-Dimensional Structures (1D)

As far as 1D materials are concerned, it can be noted that this category includes structures such as nanowires (NWs) [82,87], nanorods (NRs) [88,89], or the better-known nanotubes (NTs) [81,90], to cite only a few examples. These structures can be synthesized via hard-, soft-, self-templating, or template-free methods [13]. Examples include the synthesis of gCN nanorods with templates, e.g., by thermal condensation of cyanamide using an AAO oxide membrane as a template, which improved the orientation and crystallinity of the material with respect to the bulk, or without templates, e.g., by infrared-controlled heating of dicyandiamide, a synthesis that improves charge separation [1]. These structures show improvements both in surface area (more active sites) and in the increased migration speed of the charge carriers (shorter distances for diffusion of the photogenerated charges), but not only in [1,13,91].

An example of a synthesis of this type of morphology without templates (e.g., molecular assembly) is reported by Zhang et al. [81], on porous gCNNTs for sulfamethoxazole (SMX) degradation. In this work, the authors prepared the nanotubes, in two steps, by molecular assembly (mixture of aqueous solution of melamine and 1,5-naphthalene disulfonic acid, in different amounts, plus aqueous solution of cyanuric acid; both solutions prepared at 90 °C) and subsequent heat treatment of the assemblies (550 °C/4 h; 5 °C·min⁻¹). This material (sample MNCA-75; 75 ppm of 1,5-naphthalene disulfonic acid) degraded SMX (10 ppm), with 400 ppm of photocatalyst, to 100% in 120 min (unmodified gCN only reached 15% in 140 min) under visible light, 300 W Xe lamp (Table 2), attributing the improved photocatalytic behavior to the increased lifetime of the charges, the reduced gap, and a higher specific area (Figure 9a,b). The removal efficiency of the SMX was still above 82.5% in the fifth cycle.

Another example of 1D morphologies without templates (by molecular assembly) could be the work of Xie et al. [82], regarding gCN nanowires and nanofibers (gCNNWs and gCNNFs, respectively) evaluated in methylene blue (MB) degradation. These 1D nanostructures were synthesized by molecular assembly of cyanuric acid and melamine (in acetonitrile with stirring 12 h, washed also with acetonitrile, and subsequent drying at 80 °C 12 h), followed by two-step treatment: solvothermal (180 °C for 24 and 48 h) and heat treatment at 500 °C for 1 h (10 °C·min⁻¹). Two materials were obtained: gCNNWs with the 24 h solvothermal treatment, and gCNNFs with the 48 h solvothermal treatment. In this way, an MB degradation degree (10 ppm) of 98.5% was achieved with gCNNWs, and 90.9% with gCNNFs (g-CN, 71.1%) in 120 min, with 1000 ppm of photocatalyst under visible light (Table 2). These gCN-derived 1D structures showed improved degradation efficiency compared to pristine material mainly due to a considerable reduction in band gap, an increase in specific area (also higher microporosity), and longer photogenerated charge lifetime (Figure 9c,d).

As an example of template-free synthesis, also by molecular assembly, but by microwave treatment, the work of Mohamed et al. [83] could be mentioned, in which they prepared gCN nanofibers by this method and evaluated their photocatalytic performance in the degradation of methyl orange (MO) and phenol under simulated sunlight. The nanofibers were synthesized by mixing melamine and cyanuric acid in acetonitrile, which after magnetic stirring was poured into a pressurized vial and introduced into a microwave reactor (CEM; Discover 2.0) at 300 W (270 °C and 16.55 bar) with stirring for 60 min. The material thus prepared (1000 ppm), MGCN, showed a gap of 2.45 eV (a thermally treated sample a gap of 2.8 eV for TGCN, and of 2.4 eV for the solvothermal-prepared material, SGCN) and was able to degrade phenol (20 ppm) by 85% in 180 min and MO (20 ppm) by 92% in 150 min (SGCN sample 60% phenol and 65% MO; TGCN sample 50% phenol and 62% MO, within the same time) under simulated sunlight (SLB-300A, 300 W) (Table 2). The authors attribute the efficiency improvements over the thermal- and solvothermal-treated materials to an improvement in surface area (31.84 m²·g⁻¹; 1.5 times the area of SGCN, and 3.7 times the area of TGCN), and a smaller gap than the TGCN sample (Figure 9e,f).

2.3. Two-Dimensional Structures (2D)

One of the most important morphology modifications to increase low surface area and improve mobility and charge separation in gCN is to prepare 2D structures by exfoliation. This flat, flexible, two-dimensional structure can improve compatibility with various modification strategies, such as heterojunction construction, doping, 0D and 1D materials, and the introduction of vacancies [13]. Among the different types of possible 2D structures are nanosheets, a few layered structures, nanoplates, etc. Several methods are possible for the exfoliation of this material, e.g., exfoliation with ultrasound (in a suitable solvent such as IPA, cheap and easy to remove; NMP, quite effective; DMF; 1,3-butanediol; EtOH; MeOH; or mixtures with water such as EtOH/H₂O, IPA/H₂O, and DMF/H₂O); thermal exfoliation; microwave-assisted exfoliation; mechanical exfoliation (e.g., ball milling); or by chemical exfoliation with mineral acids such as HCl, H₂SO₄, or HNO₃ [1,6,13]. Chemical exfoliation has great advantages over physical or thermal exfoliation, such as better performance, better exfoliation rate, and the possibility of functionalization [1]. Thus, treatments with different acids impart distinct properties to the photocatalytic material; however, chemical exfoliation has the disadvantage of not being a very environmentally friendly method [70]. With thermal oxidation, which is cheap, sustainable, and easily scalable, surface areas of up to [300–390] m²·g⁻¹ can be achieved, and in addition, for example, heat treatment in an H₂ atmosphere has led to improvements not only in the degree of exfoliation but also in the π-conjugated structure [6]. Microwave exfoliation leads to improvements in both the surface area and separation—and therefore lifetime—of photogenerated charges and is a more sustainable method than chemical exfoliation [92].

In a comparison between thermal, chemical, and ultrasonic exfoliation methods for the degradation of CIP and RhB, as carried out by Wang et al. [84], gCN (550 CN) was exfoliated in different ways (ultrasonic; thermal; acidic; thermal/ultrasonic; acidic/ultrasonic; acidic/thermal; acidic/thermal/ultrasonic). The smallest gap was obtained by the heat-treated exfoliated sample (475 CN). The sample with the largest specific area was the one obtained by acid exfoliation with HNO₃ (550 H-CN), and the best charge separation was achieved with thermal treatment + ultrasound (ul) + acid (475 ul-H-CN), also obtaining the best photocatalytic behavior, both in the photodegradation of CIP (5 ppm), with a degree of degradation of 57.2% in 60 min (for the unmodified gCN 8.3%), and in the degradation of RhB (5 ppm), with 91.52% in 40 min (gCN only 57.6%), with 1000 ppm of photocatalyst, under visible light (Table 2). After five cycles, the degradation rates of 475 ul-H-CN and 550 CN to RhB were 70% and 63%. The improvements achieved with the 475 ul-H-CN sample, with respect to gCN, and the other samples are mainly due to better photogenerated charge separation (Figure 10a).

Another comparison between chemical and thermal exfoliation can be found in the work of Papailias et al. [93], where it is concluded that chemical exfoliation produces a wider band gap than thermal exfoliation. It was also found that the specific surface area, porosity, and gap increased with the duration of the heat treatment, although not significantly.

It is worth mentioning that heat treatment is the cheapest and simplest treatment, as Li et al. [85] proposes in a work about the exfoliation of gCN, and it is also an effective method for obtaining ultra-thin gCN nanosheets. These authors started by thermally condensing (inert atmosphere) melamine at 600 °C (5 °C·min⁻¹) for 2 h. Thermal exfoliation was then carried out at different temperatures and different times. The highest surface area was obtained for the sample exfoliated at 550 °C 6 h (295 m²·g⁻¹ compared to 18 m²·g⁻¹ for the untreated gCN, CN-B). The band gap of this sample was 2.89 eV, showing that as the degree of exfoliation increases with T, the gap increases, but more so with condensation time. In addition, an increase in process temperature results in more specific area (more exfoliation) and more nitrogen vacancies (N_v), while an increase in process time leads to a higher degree of exfoliation, more carbon vacancies (C_v), and larger pore volume. Both the degree of exfoliation and the photocatalytic activity of Rh6G (5 ppm) improve with increasing heat treatment time and temperature under visible light with 500 ppm photocatalyst (Table 2). The CN 550-5 sample also shows a longer lifetime of photogenerated charges (compared to CN-B), mainly due to a reduction in migration distances (Figure 10b).

Zhang et al. [94] prepared, by thermal polymerization of a urea solution in water (in crucible at 550 °C for 3 h; 5 °C·min⁻¹), exfoliated gCN in one step, and documented, by comparison with other exfoliation methods (chemical and ultrasonic), that the one-step synthesis produces a material with higher photocatalytic efficiency (200 ppm photocatalyst) than the other methods studied in the degradation of BPA (20 ppm) under visible light. After five repeated cycles, the BPA degradation rate remains at more than 92% with this material.

A ‘green’ method for the synthesis of exfoliated gCN was documented by Pattnaik et al. [86] in the degradation of CIP under sunlight, by a so-called ‘bi-thermal’ method, in aqueous solution, starting from gCN powder (obtained from urea by thermal polymerization), by refluxing for 6 h. It was then allowed to cool, followed by freezing for another 6 h, and thus eight reflux–freezing steps were repeated. They achieved an improved specific area with respect to bulk, and up to 2.5 times more photocatalytic activity, than the unmodified gCN (despite a wider band gap, due to the quantum confinement effect). A CIP degradation rate (20 ppm) of 78% in 60 min was achieved with 1000 ppm photocatalyst under solar irradiation (untreated gCN 45% at the same time) (Table 2). The improvement in efficiency was attributed to an improvement in the surface area, as well as a slight improvement in the lifetime of the photogenerated loads.

2.4. Three-Dimensional Porous Structures (3D)

To conclude this subsection, the discussion focuses on the preparation of porous 3D materials. These porous materials have a larger surface area (a higher number of active sites) and better charge separation. The three main types of pores present in this type of structure are initially described: Micropores (<2 nm) allow for a greater specific surface area with an increased number of active sites. Mesopores (2–50 nm) are more conducive to the transport of reagents or solvents than the former due to a smaller distance between pores; in addition, the multiple reflection of light within mesopores improves the efficiency of light utilization. Lastly, macropores (>50 nm) have better adsorption capacity for organic macromolecules. Thus, good porous structure design achieves a good balance between light absorption and charge transport and separation, for timely optimization of quantum efficiency [95]. Among the most referenced synthesis methods are hard template with carbon nanotubes (CNNTs), soft template with ionic liquids or polymers, but also template-free methods by molecular assembly, acid treatment, or ammonium carbonate treatment, to name a few examples [1,6,45] (Figure 11).

2.4.1. Hard Template

For hard template methods, it can be mentioned that, in general, three synthesis steps are necessary:

i.

Coating the chosen hard template with the gCN precursor;

ii.

Treatment for conversion to gCN on the template;

iii.

Removal of the template.

Silica or alumina derivatives (SiO₂ or Al₂O₃, respectively) are the most used hard templates. This method provides a structure that will favor optimal migration of the photogenerated charges. The lengthy template removal process, which often requires toxic acids such as HF or NH₄HF₂, makes this method unattractive [6,13]. A more ecofriendly alternative could be the use of salts as a hard template, e.g., ClNa [13,95,96].

Another example of the use of hard templates in the synthesis of porous gCN, which are sustainable and easily removed and recovered, is proposed by Chen et al. [97] in a paper on the use of CNTs as an ecofriendly and recyclable hard template for the photodegradation of organic pollutants. These authors prepared the photocatalyst by mixing dicyandiamide and CNTs by grinding and heat treating at 550 °C 2 h in N₂ atmosphere (condensation). The obtained product was dispersed in deionized water (DW) and treated with ultrasound to separate the CNTs. The authors achieved a surface area in the material of 103.3 m²·g⁻¹, with a pore volume of 0.61 cm³·g⁻¹ (compared to 10.5 m²·g⁻¹ and 0.091 cm³·g⁻¹, respectively, for the unmodified gCN). The band gap of the material thus obtained (2.78 eV) hardly changed with respect to the unmodified gCN (2.79 eV). A degradation of RhB (10 ppm) of 63% in 90 min (25% for gCN within the same time) was achieved with 1000 ppm of photocatalyst under visible light (300 W; Xe lamp) (

Table 3. Parameters of photodegradation of different pollutants in water with porous 3D morphologies of gCN.

Morphology	Template	Photocatalyst	Area (m2·g−1)	Pore (cm3·g−1)	Gap (eV)	Pollutant (ppm)	Radiation	Efficiency (%)	Time (min)	k (min−1)	Refs.
3D	HARD	gCN (1000 ppm)	10.5	0.091	2.79	RhB: 10	300 W Xe lamp (Vis.)	25	90	0.004	[97]
		Porous 3D gCN (1000 ppm)	103.3	0.61	2.78			63		0.018
		gCN (500 ppm)	9.75	0.061	2.7	RhB: 10	Visible	25	40	0.007	[98]
	SOFT	Porous gCN (P123-6) (500 ppm)	73.29	0.27	2.75			98.7		0.1
	WITHOUT	gCN (1000 ppm)	8.36	0.02	2.6	RhB: 10	500 W Xe lamp (Vis.)	40	80	0.004	[99]
		Porous hexagonal gCN (1000 ppm)	67.3	0.32	2.43			100		0.053
High crystalline gCN		BCN (1000 ppm)	-	-	2.67	NPX: 8	350 W Xe lamp (Vis.)	20.9	70	0.013	[100]
		CCN (1000 ppm)	-	-	2.72			98.4		0.092

). The performance of porous g-C₃N₄ before the fourth cycle was still 92%. The better efficiency was attributed to the improved surface area and pore volume in addition to the increase in photogenerated charge lifetime (Figure 12a,b).

2.4.2. Soft Template

Soft template methods include the use of surfactants as template, for example, the use of ionic liquids such as 1-butyl-3-methylimidazolium dicyanamide (BmimDCN), 1-butyl-3-methylimidazolium chloride (BmimCl), or 1-butyl-3-methylimidazolium hexafluorophosphate (BmimPF6). However, the removal of this type of template can also result in gCN pores being resealed and a high amount of excess carbon being produced, thus decreasing photocatalytic activity [6,95].

Figure 12. (a,b) adsorption isotherms (left) and PL spectra (right) of porous gCN with CNNTs as hard template [97]; (c,d) adsorption isotherms (left) and PL spectra (right) of porous gCN with Pluronic-P123 as a soft template [98]; (e,f) adsorption isotherms (left) and PL spectra (right) of the porous gCN without template by supramolecular assembly [99]; and (g,h) adsorption isotherms (left) and PL spectra (right) of porous gCN, also without template, from melamine hydrochloride [101].

Figure 12. (a,b) adsorption isotherms (left) and PL spectra (right) of porous gCN with CNNTs as hard template [97]; (c,d) adsorption isotherms (left) and PL spectra (right) of porous gCN with Pluronic-P123 as a soft template [98]; (e,f) adsorption isotherms (left) and PL spectra (right) of the porous gCN without template by supramolecular assembly [99]; and (g,h) adsorption isotherms (left) and PL spectra (right) of porous gCN, also without template, from melamine hydrochloride [101].

An example of the use of soft templates to obtain porous gCN was proposed by Yan et al. [98], where they synthesized porous gCN using Pluronic-P123 (triblastic copolymer based on polypropylene oxide (PPO) and polyethylene oxide (PEO)) as a soft template in two steps: (i) first mixing melamine and Pluronic-P123 (in different mass ratios with respect to P123) in acidic medium and mixed with stirring for 24 h, and then solvothermal treatment in an autoclave at 130 °C/24 h; (ii) removing the P123 polymeric template by heat treatment at 550 °C for 6 h in an aerobic environment.

The photocatalytic efficiency of this material (gCN-P123-x; 500 ppm) was evaluated with RhB (10 ppm) under visible light. The gCN-P123-6 sample presented a surface area of 73.3 m²·g⁻¹ with a specific pore volume of 0.27 cm³·g⁻¹ (the solvothermal synthesized gCN, without template and without acid presence, presented a surface area of 9.75 m²·g⁻¹ with a specific pore volume of 0.061 cm³·g⁻¹). The band gap of the sample gCN-P123-6 was 2.75 eV (gCN without template, 2.70 eV). With the porous material (gCN-P123-6), they achieved a photocatalytic degradation efficiency of RhB of 98.7% in 40 min (gCN 25% within the same time) (Table 3). The efficiency of this material before five cycles was 90%. The authors attributed the increase in surface area and pore volume, together with improved exciton separation, to improvements in RhB degradation efficiency under visible light (Figure 12c,d).

2.4.3. Template Free

As template free methods for porous 3D structures, it could be said that the molecular assembly (or supramolecular pre-organization) is the most widely used method, e.g., the use of melamine as a precursor together with triazine derivatives, which gives rise to hydrogen-bonded assemblies, such as melamine–cyanuric acid or melamine–trithiocyanuric acid, with subsequent heat treatment (by changing the solvents used, different morphologies are possible) [1,6,96,101,102].

An example of template-free porous gCN synthesis can be seen in the work of Chen et al. [99], about the synthesis of porous hexagonal prisms of gCN, wherein they were evaluated in terms of RhB degradation. The authors prepared the photocatalyst by supramolecular assembly of melamine and cyanuric acid in a solution of water and acetic acid (in different volumetric ratios of HAc:H₂O—5:1, 3:1, 1:1, 1:3, and 1:5), which, after being stirred, was first autoclaved at 180 °C/12 h, and then, after being freeze-dried, was heat-treated at 520 °C for 2 h (5 °C·min⁻¹). They obtained a photocatalytic material (the ACNH-3 sample) with a specific area de 67.3 m²·g⁻¹ and a specific pore volume of 0.37 cm³·g⁻¹ (the bulk gCN 8.36 m²·g⁻¹ and 0.02 cm³·g⁻¹, respectively). The band gap of this material was 2.43 eV (gCN 2.6 eV). They achieved a degradation efficiency of RhB (10 ppm) of 100% in 80 min (gCN 40% at the same time) with the thus-prepared photocatalyst (1000 ppm) under visible light (Table 3). After four cycles, this sample maintains nearly 90% of its original degradation capacity. The efficiency improvements over the gCN bulk were attributed to a larger specific pore area and volume, a reduction in band gap, and an improvement in photogenerated charge separation (Figure 12e,f).

To conclude this subsection, another example of porous gCN without template could be the work of Dou et al. [101] in the degradation of amoxicillin (AMX) and cefotaxime (CFX) under visible light. The authors prepared the photocatalyst by first mixing an aqueous solution of melamine and HCl with stirring, thus obtaining melamine hydrochloride after drying. It was then heat treated at 500 °C 2 h (20 °C·min⁻¹) in N₂ atmosphere and treated a second time at 520 °C 2 h (4 °C·min⁻¹). The material thus prepared (MCN) showed a surface area of 18.06 m²g⁻¹ (5.64 m²·g⁻¹ of the untreated gCN, BCN), with a gap almost the same as the untreated material (2.6 eV vs. 2.63 eV for the untreated material). They thus achieved a degree of degradation of 90% for AMX and 99% for CFX (for the unmodified gCN, 40% for AMX, and 80% for CFX, within the same time), both 2 ppm, in 60 min under visible light (300 W; Xe lamp) with 1000 ppm of photocatalyst (Table 3). After five cycles, the degradation of AMX still remains at 75%, also for the CFX 90% with this gCN derivative. The improvements in efficiency are mainly due to a slight improvement in surface area, and an improvement in exciton lifetime (Figure 12g,h).

2.5. High Crystalline gCN Structures

Another aspect that can be considered in this section is the crystallinity of the photocatalyst. A higher degree of crystallinity generally results in a faster migration rate of charge carriers and a reduced recombination rate of electron–hole pairs, though often at the cost of decreased surface area. The main methods to synthesize crystalline gCN derivatives are the molten salt method (the most widely used method), template-based method, solvothermal method, pure O₂-assisted method, and thermionic synthesis method [103,104]. One way to prepare highly crystalline gCN would be by post-thermal treatment of gCN with a mixture of KCl and LiCl (or NaCl/KCl) in a nitrogen atmosphere at 550 °C [1,49].

Few studies have reported the degradation of pollutants using highly crystalline gCN. One notable example is the work carried out by Wang et al., 2020 [100], which examined the degradation of pharmaceuticals and personal care products (eight different PPCPs; each at 8 ppm) using the thus-modified gCN. The authors prepared crystalline carbon nitride (CCN) in two steps: (i) by calcining 5 g of dicyandiamide at 550 °C for 3 h with a heating rate of 2.8 °C·min⁻¹ to obtain the bulk material (BCN); (ii) by grinding the BCN with KCl and LiCl, followed by thermal treatment under an inert atmosphere at 550 °C for 4 h using the same heating rate. The band gaps of BCN and CCN were measured to be 2.67 eV and 2.72 eV, respectively. Using this modified material, they achieved 98.4% degradation of naproxen (NPX) in 70 min, k = 0.092 min⁻¹ (the BCN, k = 0.013 min⁻¹) under visible light (350 W Xe lamp with a 420 nm cut-off filter) with 1000 ppm of CCN. The authors attributed the enhanced degradation efficiency compared to BCN to improved separation of photogenerated charges and increased O₂ adsorption on the photocatalyst surface, which leads to higher H₂O₂ production. The H₂O₂ subsequently decomposes into HO• radicals that effectively oxidize the PPCPs.

It is important to note, however, that the main application of this type of gCN modification is hydrogen (H₂) production via water splitting, or the reduction of CO₂, and less the degradation of EPs [104,105,106,107,108].

In gCN, the synthesis of 0D structures (quantum dots) improve charge carrier migration, enlarge the active surface area, and enhance light absorption, although their quantum confinement effect increases the band gap; their use is mainly restricted to their integration in heterojunctions (type II, Z, Schottky) or homojunctions mitigating charge recombination, optimizing applications in decontamination. One-dimensional morphologies (nanotubes, nanofibers) reduce charge diffusion distances, increase the surface area, and decrease the band gap, favoring efficient exciton separation and higher photocatalytic activity of the gCN structure. Two-dimensional structures (exfoliated nanosheets) maximize surface area, improve charge separation, and adapt the band gap by sustainable methods (thermal, ultrasound), while three-dimensional porous materials balance high surface area, reagent transport, and light exploitation by hierarchical design of micro/meso/macropores, achieving synergy between active sites and charge mobility. High crystalline gCN, mainly obtained by heat treatment with salts, optimizes carrier mobility, but may slightly reduce surface area. It is also worth mentioning that although not all the referenced studies report on the stability of these gCN derivatives after several cycles of use, those that do generally demonstrate good stability and maintain high efficiency after four or five cycles.

3. Surface Functionalization

These surface modifications related to the introduction of functional groups which, apart from enlarging the active surface (more active sites) can enhance charge separation and carrier mobility (the introduction of surface defects, such as vacancies, was considered more appropriate to deal with in the next section, dedicated to vacancies and defects). Thus, on the surface of the gCN itself, there are different groups of interest, including Brønsted and Lewis basic sites, a hydrogen bonding group, cyano groups, hydroxyl groups, etc. The nitrogen of the ring (imino) acts as Lewis basic site, while the amino groups at the edges act as Brönsted bases [15].

As an example, basic sites can be used to improve the anchoring of metal oxides during doping (as well as to improve their dispersion on the gCN surface), the adsorption of weak acids such as phenols, or cationic dyes. Such sites are favored by incomplete condensations, or by increasing the porosity. Protonation with inorganic acids not only improves charge separation but—as has been previously seen—also facilitate exfoliation. Thus, given the reactivity of the amino and imino groups in gCN, their covalent functionalization is also possible, which lowers the energy barrier for intramolecular charge transfer [4,42].

The main functional groups inserted into the gCN are as follows (Figure 13):

• Hydroxyls (introduced by hydrothermal treatment, for example). They can enhance adsorption of organic matter, and act as h⁺ trapping centers; however, they can slightly widen the band gap (they are electro-donators) [41,42,109].
• Cyano groups (by treatment of gCN with NaBH₄ and subsequent treatment at 150–350 °C, or with potassium thiocyanate (KSCN) and subsequent heat treatment at 500 °C, among other methods) [42,110]. These functional groups can considerably reduce the band gap (they are electron-accepting groups) [109]. The introduction of the cyano group into gCN structures increases O₂ adsorption and introduces lone pair electrons into the structure that participate in photoexcitation [111].
• Carboxyl groups (by oxidation with HNO₃, for example) improve the mobility and separation of the photogenerated charges as well as improve the surface hydrophilicity by being electron acceptor groups [42,48,112,113].
• Ureido groups (introduced in the same way as cyano groups, with the addition of a subsequent treatment with HCL and stirring for a long time) also improve charge separation [42].
• Aromatic groups (introduced by thermal copolymerization with aromatic precursors such as thiophene, aniline, benzonitrile, etc.) provide more active areas as well as extend the π-conjugated system and decrease the band gap [42,114].

Figure 13. Four examples of functional groups in heptazine-based gCN. Carbon atoms: grey sphere, Nitrogen atoms: blue spheres, Oxigen atom: red spheres, and Hydrogen atoms: white spheres.

Figure 13. Four examples of functional groups in heptazine-based gCN. Carbon atoms: grey sphere, Nitrogen atoms: blue spheres, Oxigen atom: red spheres, and Hydrogen atoms: white spheres.

These functional groups can enhance the absorption of visible light, improve charge separation, increase the specific surface area, and thereby increase the number of active sites on the gCN. Such surface modifications are mainly effective in CO₂ reduction, H₂ production, ammonia synthesis, H₂O₂ production, and disinfection [42,47].

An example of modifications of surface functional groups is proposed by Zhu et al. [48] in a paper on the surface carboxylation of gCN and its impact on the adsorption and photodegradation of MB and RhB (both 15 ppm). These authors modified gCN by first mixing melamine and urea, with subsequent heat treatment at 550 °C for 4 h. The synthesized gCN was then treated with HNO₃ (10 M) for 24 h at 80 °C. The product thus obtained was separated, washed, and freeze-dried. This material had a surface area of 88.5 m²·g⁻¹ and a pore volume of 0.48 cm³·g⁻¹ (38.17 m²·g⁻¹ and pore volume of 0.24 cm³·g⁻¹ for the untreated gCN). This material achieved a gap of 2.65 eV, a degree of degradation for MB of 79%, and 62% degradation for RhB, with the thus-prepared photocatalyst (gCN-HNO₃; 200 ppm) in 180 min under visible light (300 W Xe lamp) (

Table 4. Parameters of photodegradation of pollutants in water with gCN with -COOH, -C≡N, and thiophene groups.

Group	Photocatalyst	Area (m2·g−1)	Gap (eV)	Pollutant (ppm)	Radiation Xe Lamp (Vis.)	Efficiency (%)	Time (min)	k (min−1)	Refs.
-COOH	gCN (200 ppm)	38.2	-	MB and RhB: 15	300 W	-	180	-	[48]
	gCN-HNO3 (200 ppm)	88.6	2.65			MB: 79 RhB: 62		RhB: 0.026; MB: 0.033
-C≡N	gCN (1000 ppm)	21.49	2.72	RhB and TC: 15	300 W	100	30	RhB: 0.025; TC: 0.029	[115]
	Cyano-gCN (1000 ppm)	51.34	2.63			100		RhB and TC: 0.099
	gCN (500 ppm)	16.6	2.61	RhB: 5	500 W	45	90	0.0072	[116]
	Thiophen-gCN (500 ppm)	78.4	2.64			96		0.036

). The authors attributed the improvement in efficiency to a higher surface area and pore volume, and to a better separation of the photogenerated charges (Figure 14a,b).

Another example in which functional groups were introduced into gCN was the work of Razavi-Esfal M. et al. [115], in which they prepared porous gCN rich in cyano groups and evaluated the degradation of RhB and tetracycline (TC) under visible light. The authors prepared the photocatalyst by mixing melamine with SiO₂ as a hard template and then heat-treated it at 550 °C/3 h (2.3 °C·min⁻¹). After eliminating the SiO₂ template with HF, the material was washed and dried. The modified photocatalyst presented a specific surface area of 51.34 m²·g⁻¹ and a pore volume of 0.23 cm³·g⁻¹, much higher than those of gCN (21.49 m²·g⁻¹ and 0.126 cm³·g⁻¹, respectively). The effective band gap of the group-modified photocatalyst was 2.63 eV (2.72 eV for unmodified gCN). With this material, they were able to degrade RhB (15 ppm) by 100% in 30 min (gCN by 40% in 90 min) and TC (15 ppm) by 100% in 30 min (gCN by 27% in 90 min) with 1000 ppm of photocatalyst under visible light (300 W Xe lamp) (Table 4). According to the authors, the improvement in efficiency over the unmodified material is mainly due to improvements in specific area and porosity, reduction in band gap, and separation of photogenerated charges (Figure 14c,d).

A final example of gCN derivatives with functional groups is given by Xu et al. [116], with work on functionalization with aromatic groups, specifically thiophene, which evaluated the degradation of RhB, MO, MB, and CV (5 ppm) under visible light (500 W Xe lamp). Thiophene grafted in a gCN porous nanosheet photocatalyst was prepared employing a one-step method by mixing cyanuric acid and melamine in DW for 12 h, which after being separated and dried were mixed with different amounts of 3-thiophenecarboxylic acid and heat-treated at 550 °C/1 h (3 °C·min⁻¹). This material presented a surface area of 74.8 m²·g⁻¹ and a gap of 2.64 eV (gCN had a surface area of 16.6 m²·g⁻¹ and a gap of 2.61 eV). Thus, the sample with 30 mg of 3-thiophenecarboxylic acid and a heating ramp of 3 °C (3-CM-CN-Th A30) degraded RhB 96.1% in 90 min (gCN only 45% at the same time) under visible radiation with 500 ppm of photocatalyst. They also achieved a degradation efficiency of 53.2% of MO, 89.4% of MB, and 88.8% of CV (Table 4). The RhB removal efficiency remained at 89.22% after five cycles of use. This improvement in efficiency is attributed to the increase in specific pore area and volume, as well as to the enhanced separation of the photogenerated charges (Figure 14e,f).

Thus, the introduction of hydroxyl groups (-OH) improves the adsorption of organic matter and acts as a hole trap (h⁺), although it can slightly increase the band gap as they are electro-donating groups. The cyano groups (-C≡N) reduce the band gap (electron acceptor effect), favor the adsorption of O₂, and introduce free electron pairs, optimizing photoexcitation. Carboxyl (-COOH) increases mobility and charge separation, as well as improve surface hydrophilicity. Ureido groups enhance charge separation, while aromatic groups (e.g., thiophene) extend the π-conjugated system, reduce the band gap, and increase the number of active sites. These modifications increase surface area, tailor electronic properties, and improve pollutant degradation efficiency, as demonstrated by studies with RhB, MB, and TC, where functionalization with -COOH, -C≡N, and aryls achieved >90% degradations under visible light, outperforming unmodified gCN. Among the three references provided in this section, only one reports a stability study after several treatment cycles (RhB degradation using thiophene-functionalized gCN), showing good material stability after five cycles.

4. Electronic Structure Optimization

Modifications aimed at modifying the band structure of gCN are not only aimed at improving absorption in the visible spectrum, but also at optimizing the redox potentials to improve photocatalytic activity. Thus, several possibilities have been considered for this purpose, such as, for example, the design of vacancies, or the different types of doping (metallic, non-metallic), which are examined in the next section.

4.1. Defect Engineering

Vacancies refer to the absences in lattice positions, in this case of gCN. It can be noted that there are two main types of vacancies in gCN: nitrogen vacancies, N_v (anionic), and carbon vacancies, C_v (cationic).

4.1.1. Nitrogen Vacancies (N_v)

The formation of nitrogen vacancies (N_v) in gCN plays an important role both in charge separation (increase of exciton lifetime) and in influencing the band gap. The insertion of nitrogen vacancies can be carried out by controlling the condensation temperature, by reduction in H₂ atmosphere, by hydrothermal treatment in the presence of (NH₄)₂S₂O₃, or by heat treatment in alkaline medium (e.g., with KOH, NaOH, or Ba(OH)₂). By thermal synthesis in an alkaline medium, controlling the base/precursor ratio, the gap reduction can be modulated. The gap decreases with increasing base/precursor ratio, mainly due to the reduction in conduction band potential towards less negative values [4,6].

Similar conclusions were reached by Katsumata et al. [51] in a study on the degradation of BPA with N_v-gCN. These authors introduced the N_v, mixing in a first step KOH and urea (in different mass ratios) in distilled water with stirring, and then drying. The powder thus obtained was mixed with OA in a solution of H₂O/ethanol. The solvent was then evaporated, and the resulting solid was calcined at 600 °C for 2 h (2.5 °C·min⁻¹). This material (KOH-OA-gCN) presented a band gap of 2.6 eV, with a surface area of only 29 m²·g⁻¹ (the untreated gCN had a gap of 2.85 eV and a specific area of 85 m²·g⁻¹). This material (600 ppm) achieved 90% degradation of 10 ppm BPA in 150 min under visible radiation (300 W Xe lamp), while gCN only achieved 25% (

Table 5. Parameters of photodegradation of pollutants in water with gCN with vacancies.

Vacancy	Photocatalyst	Area (m2·g−1)	Gap (eV)	Pollutant (ppm)	Radiation	Efficiency (%)	Time (min)	k (min−1)	Refs.
Nv	gCN (600 ppm)	2.85	85	BPA:10	300 W Xe lamp (Vis.)	25	150	0.0021	[51]
	KOH-OA-gCN (600 ppm)	2.6	29			90		0.0147
	gCN (600 ppm)	75.7	2.79	OTC:20	24 W; LED (Vis.)	45.8	135	0.0046	[117]
	NvrCN (600 ppm)	64.3	2.73			92.5		0.018
	gCN (250 ppm)	85.4	2.75	TC: 15 SMX: 5	300 W Xe lamp (Vis.)	TC: 40 SMX: 52	TC: 60 SMX: 120	0.009 0.025	[118]
	ODH-CN-2 (250 ppm)	108.2	2.61			TC: 79.9 SMX: 91.5		0.007 0.020
Cv	gCN (300 ppm)	30.1	2.76	BPA: 10	350 W Xe lamp (Vis.)	78	120	0.003	[43]
	Cv-gCN (300 ppm)	14.7	2.65			90		0.006
	gCN (250 ppm)	272	2.94	TC; 16	300 W halogen lamp (Vis.) + ul (600 W/40 kHz)	30	60	0.0003	[44]
	Cv-gCN-20 (250 ppm)	331	2.9			96		0.0010
	gCN (1000 ppm)	79.7	2.73	4-clorophenol: 10	300 W Xe lamp (Vis.)	33.8	120	0.003	[119]
	Cv-gCN-575 (1000 ppm)	64.2	2.71			60.1		0.008

). After five usage cycles, the photocatalytic activity dropped significantly to 80%, which the authors attributed primarily to the loss of ultrafine particles during the recycling process. The authors also concluded that the better efficiency was due to the reduction in band gap and a better separation of generated charges (Figure 15a,b) and inferred that the addition of KOH leads to a substantial loss of specific area due to a decrease in mesoporosity.

Likewise, Lee et al. [117] carried out the synthesis of N_v-gCN by alkaline solvothermal method. They used a suspension of Mg-gCN in EG together with KOH, heated at 160 °C for 12 h, in a work on the degradation of OTC. The modified material had a band gap of 2.73 eV (gCN 2.79 eV and MgCN 2.89 eV), and a surface area of 64.3 m²·g⁻¹, with a pore volume of 0.16 cm³·g⁻¹ (75.7 m²·g⁻¹, 0.38 cm³·g⁻¹, and 87.3 m²·g⁻¹, 0.43 cm³·g⁻¹, for gCN and Mg-CN, respectively). This photocatalyst thus prepared achieved an OTC degradation (20 ppm) of 92.5% in 135 min (gCN 45.8% within the same time), under visible light with 600 ppm of photocatalyst (Table 5). After three cycles, the removal efficiency of OTC was still 82%. The better efficiency was attributed to the reduction in band gap and a better separation of generated charges (Figure 15c,d); also, the authors concluded that the reduction in surface area is mainly due to the decrease in pore volume in N_v-gCN.

And as a final example of this type of modification with N_v in the gCN structure, it is worth mentioning the research by Wang Y. et al. [118] on the degradation of tetracycline hydrochloride (TC-HCl) and sulfamethoxazole (SMX) with the thus-modified gCN. The authors synthesized the photocatalytic material by mixing urea and oxalyl dihydrazide (ODH) in different amounts and heat-treating them (polymerization in an aerobic environment) at 550 °C/4 h (5 °C·min⁻¹). The ODH-CN2 (Urea:ODH, 100:1) sample achieved a surface area of 108.2 m²·g⁻¹ (the unmodified gCN 85.4 m²·g⁻¹) and a band gap of 2.61 eV (the gCN 2.75 eV). This photocatalyst thus modified can degrade to 79.9% of TC-HCl (15 ppm) in 60 min and 91.5% of SMX (5 ppm) in 120 min under visible radiation with 250 ppm of photocatalyst (Table 5). The photocatalytic degradation efficiencies of TC-HCl and SMZ remained at 70.85% and 79.33%, respectively, after five cycles. The improvement in efficiency over the unmodified material is due (according to the authors)—as in the previous examples—to a higher surface area, to a narrowed band gap, and an improvement in photogenerated charge separation (Figure 15e,f).

4.1.2. Carbon Vacancies (C_v)

For C_v-modified gCN, it can be mentioned that these vacancies can serve as a reservoir of photogenerated electrons, thereby inhibiting the recombination of h⁺ and e⁻. These vacancies can also serve as electron transfer centers to the adsorbed molecular oxygen, favoring the production of superoxide radicals (O₂^●−), thus improving photocatalytic activity [43,44].

As proposed by Liang et al. [43], these vacancies may be introduced by thermal treatment of melamine in two steps:

i.

Calcination in an air muffle furnace at 550 °C 2 h;

ii.

Pyrolysis at 520 °C in a tube furnace in Ar atmosphere (2 h).

Under these conditions, the obtained material exhibited a surface area of 14.7 m²·g⁻¹ and a gap of 2.65 eV (the untreated gCN 30.1 m²·g⁻¹ and a band gap of 2.76 eV), and the degradation of BPA (10 ppm) was 1.65 times better with the C_v-gCN (90% degradation in 120 min) than with the untreated gCN (78% in the same time) using 300 ppm photocatalyst under visible light (350 W Xe lamp) (Table 5). After five cyclic experiments, the photocatalytic ability of C_v-gCN was almost unchanged. The authors concluded that, despite the decrease in surface area due to the Ar atmosphere treatment, after the introduction of the carbon vacancy, the recombination rate of photogenerated h⁺ and e⁻ is greatly decreased. In addition, the carbon vacancies, which serve as conversion centers, transferred most of the photogenerated electrons, trapped there, to the absorbed O₂ facilitating the generation of radicals (Figure 16a,b).

Another example of C_v-gCN would be, for example, the one proposed by Preeyanghaa et al. [44], in the TC degradation by sono-photocatalysis with C_v-gCN nanosheets. They prepared the photocatalyst in two steps: (i) they first melted the urea at 140 °C and added formaldehyde (different amounts) to the molten mixture, and then (ii) heat-treated both reagents at 550 °C for 3 h (5 °C·min⁻¹). Thus, they obtained material with a band gap of 2.9 eV for the C_v-gCN-20 sample (2.94 eV for gCN). They also achieved a specific area of 331 m²·g⁻¹ (272 m²·g⁻¹ for the gCN without formaldehyde with the same treatment), reaching a degree of TC degradation (16 ppm) of 96% in 60 min, under visible light and ultrasound, with 250 ppm of photocatalyst (Table 5). The sono-photocatalytic degradation efficiency of TC remains almost unchanged even after five consecutive cycles. The better efficiency of this photocatalyst thus modified was mainly due to an increased surface area and for an improvement in the separation of excitons (Figure 16c,d).

A final example of this type of vacancy in the gCN structure could be the work of Huang Z. et al. [119], on the introduction of C_v in gCN for the degradation of 4-chlorophenol under visible light. The authors synthesized the thus-modified gCN in three steps: (i) calcining urea at 550 °C/4 h (2.5 °C min⁻¹) in a flask; (ii) a second heat treatment at 500 °C/2 h (5 °C min⁻¹) for thermal exfoliation; and (iii) for the introduction of C_v, they mixed the obtained gCN (bulk and exfoliated) with powdered Mg and treated it in a quartz tube furnace in an inert environment at temperatures between 550 and 625 °C (10 °C min⁻¹). They obtained a sample (gCN-575) with a band gap of 2.71 eV (gCN a band gap of 2.73) and a surface area of 64.2 m²·g⁻¹ (gCN 79.7 m²·g⁻¹). The same sample achieved a 4-chlorophenol degradation efficiency (10 ppm) of 60.1% in 120 min (gCN 33.8% within the same time), with 1000 ppm photocatalyst under visible radiation (Table 5). After four consecutive cycles, GCN-575 exhibited a slight decrease in performance. The authors concluded that the improvement in efficiency, despite the decrease in surface area, is due to an improvement in the separation of photogenerated charges (Figure 16e,f).

In general, both defects (N_v and C_v vacancies) improve charge separation and introduce additional energy levels in the band gap that improve the absorption of the visible spectrum. Both vacancies promote the presence of unsaturated centers such as cyano groups (carbon vacancies) and carbonyl groups (nitrogen vacancies), to give just a couple of examples, which expand the number of active sites in the material. Optimization of the electronic structure by defect engineering allows for tuning both the visible light absorption and the redox potentials of gCN. The introduction of nitrogen vacancies helps to reduce the energy gap and improve charge separation. On the other hand, carbon vacancies act as electron reservoirs, decreasing recombination and favoring electron transfer to adsorbed oxygen to form radicals. Together, these defects generate additional energy levels that optimize carrier mobility and enhance photocatalytic efficiency. gCN with Nv and Cv vacancies remains stable after several degradation cycles, indicating that these modifications are stable in the reaction medium. These strategies are key to tailoring gCN for applications in pollutant degradation.

4.2. Element Doping Strategies

Generically, doping in gCN is defined as the process of replacing atoms in the lattice positions (non-metallic doping; anionic) or introducing them into lattice interstices (metallic doping; cationic). Both influence the electronic structure of the gCN, improving not only the absorption range (due to the introduction of intermediate energy levels in the band gap, as well as vacancies) but also enhancing the lifetime of the charge carriers and improving their separation. This doping can also compensate for the increased gap produced in nanostructures (e.g., nanosheets) by the quantum confinement effect [1,13].

4.2.1. Non-Metal Doping

Non-metal doping, as well as the introduction of vacancies, is a promising strategy to prepare metal-free photocatalysts. This type of doping (usually with B, P, S, C, N, O, and halogens, such as F, Cl, Br, and I) can modify the structure, improve conductivity and charge separation, and reduce the band gap. For electron-poor dopants, such as B, the intermediate energy levels in the band gap are closer to the valence band, increasing the p-type conductivity; for electron-rich dopants, such as O or S, increasing the n-type conductivity, the introduced levels are closer to the conduction band [2,13,40]. In addition, due to the high ionization energies and electronegativity of non-metals, they can quickly form covalent bonds with other compounds, facilitating the functionalization of the photocatalyst [13].

Doping with O

This subsection begins with a description of oxygen (O) doping. There are several possibilities for doping gCN with O, for example, Deng et al. [120] refer to their preparation in one step by thermal copolymerization (calcination at 550 °C 3 h with a ramp of 5 °C·min⁻¹) starting from urea (10 g) and different amounts of oxalic acid, H₂C₂O₄. The photocatalyst thus obtained has a considerably reduced band gap with respect to untreated gCN (1.93 eV for the doped material and 2.65 eV for the untreated gCN) (Figure 17a), which, according to the authors, ostensibly improves the degradation capacity of the antibiotic lincomycin (100 ppm), 99% in 180 min (with the gCN at 50% within the same time), with 300 ppm of photocatalyst under visible light (

Table 6. Parameters of photodegradation of pollutants in water with non-metal-doped gCN.

Dopant	Photocatalyst	Area (m2·g−1)	Gap (eV)	Pollutant (ppm)	Radiation	Efficiency (%)	Time (min)	k (min−1)	Refs.
O	gCN (300 ppm)	-	2.65	Lincomycin: 100	90 W LED (Vis.)	45	180	0.005	[120]
	O-gCN (300 ppm)	-	1.93			99		0.034
	gCN (200 ppm)	62.50	2.63	BPA: 10	300 W Xe lamp (Vis.)	14	120	0.001	[121]
	O-gCN (200 ppm)	70.32	2.60			99		0.032
P	gCN (1000 ppm)	26.86	2.69	RhB: 10	300 W, Xe lamp (Vis)	100	30	-	[122]
	P-gCN (1000 ppm)	40.5	2.84				10	-
	gCN (100 ppm)	73.8	-	RhB: 20	300 W Xe lamp (Vis.)	64.2	70	0.039	[123]
	P-gCN (100 ppm)	202.9	-			99.5		0.120
S	gCN (100 ppm for MB and 200 ppm for TC)	11	2.55	MB and TC: 20	300 W Xe lamp (Vis.)	MB: 4; TC: 10	MB: 300 TC: 240	MB: 0.00014 TC: 0.0003	[124]
	S-gCN (100 ppm for MB and 200 ppm for TC)	15	1.83			MB: 60; TC: 89		MB: 0.0014 TC: 0.037
	gCN (1000 ppm)	-	-	OTC: 10	300 W Xe lamp (Vis.)	57.1	40	-	[125]
	S-gCN (1000 ppm)	31.2	2.83			93.3		0.133
B	gCN (68 mg gCN/62 mg CFs)	-	2.71	RhB: 5	8 W LED lamp (Vis.)	82	120	0.015	[126]
	B-gCN (68 mg B-gCN/62 mg CFs)	-	2.69			95		0.024
	gCN (500 ppm)	17.5	2.73	RhB: 2	500 W Xe lamp (Vis.)	-	30	0.026	[127]
	B-gCN (500 ppm)	105.1	2.70			97		0.086
Halogen	gCN (500 ppm)	42.3	2.75	TC: 10	300 W Xe lamp (Vis.)	32	120	0.004	[128]
	Cl-gCN (500 ppm)	114.4	2.7			92		0.02
	gCN (1000 ppm)	-	2.78			30		0.004
	Br-gCN (1000 ppm)	-	2.75	OTC: 10	35 W LED lamp (Vis.)	75	150	0.018	[129]
	Cl-gCN (1000 ppm)	-	2.73			75		0.017
C	gCN (400 ppm)	-	2.77	TC: 30	35 W LED lamp (Vis.)	40	60	0.01	[130]
	C-gCN (400 ppm)	-	2.71			77		0.03
	gCN (1000 ppm)	43	2.88	BPA: 10	300 W Xe lamp (Vis.)	25	60	0.005	[131]
	C-gCN (1000 ppm)	85	2.19			96		0.053
N	gCN (500 ppm)	18.4	2.51	TC: 10	300 W Xe lamp (Vis.)	52.2	60	0.013	[132]
	N-gCN (500 ppm)	74.79	2.47			81.7		0.026
	gCN (1000 ppm)	76.69	2.51	PhOH: 10	300 W Xe lamp (Vis.)	70.1	180	0.002	[133]
	N-gCN (1000 ppm)	72.26	1.82			37.6		0.006

). This O-doped gCN derivative also maintains its efficiency in lincomycin degradation even after three consecutive degradation cycles.

Long et al. [121], in a work on the degradation of BPA with O-doped gCN, point out the synthesis by thermal polymerization (at 550 °C/4 h, 5 °C·min⁻¹) of this material without adding any precursor other than urea, by removing the gCN at room temperature after 300 °C (on cooling) until reaching room temperature. With heat treatment (at 550 °C) for different times (2, 4, 6, and 10 h), different degrees of doping were achieved by air cooling to room temperature. The sample treated at 550 °C/6 h, OCN-6, reached a surface area of 70.32 m²·g⁻¹ (gCN 62.5 m²·g⁻¹) and a band gap of 2.6 eV (gCN 2.63 eV). The best result in BPA degradation (10 ppm) was obtained with OCN6 (200 ppm), 99% in 120 min (gCN 14% within the same time) under visible light (Table 6). The degradation efficiency of BPA remained above 96% after four cycles. The authors also concluded that the improved efficiency of the O-doped material with respect to bulk is mainly due to an improvement in surface area, but above all to a substantial improvement in the mobility and separation of the photogenerated charges (Figure 17b).

Doping with P

Doping gCN with P improves the delocalization of the π cloud, improving the transfer of photogenerated electrons [40], improving visible light absorption, the separation of charge carriers, and overall photocatalytic activity [1].

A good way to introduce P into the gCN structure is by thermal copolymerization (in air atmosphere) at different temperatures (550–650 °C) with guanidinium hydrochloride (CH₅N₃-HCl; GndHCl) as a precursor of gCN, and hexachlorocyclotriphosphazene (HCCP), in different percentages with respect to gCN, as a cheap and sustainable source of P, with excellent photocatalytic activity in the degradation of organic dyes (RhB), as Zhou et al. [122] concluded in a paper on the degradation of this dye with P-gCN under visible light. These authors controlled the degree of doping by adjusting the HCCP amounts and the calcination temperature. This material shows a specific area for the P10-600 sample of 40.5 m²·g⁻¹ (26.86 m²·g⁻¹ for gCN) and a band gap of 2.84 eV (2.69 eV for gCN). The photocatalyst thus modified achieved 100% degradation of RhB (10 ppm) in 10 min with the same sample (gCN 100% in 40 min) at a concentration of 1000 ppm under visible light (Table 6). After four cycles of degradation, this photocatalyst still maintains 75% degradation efficiency. Overall, the authors attributed this improvement in efficiency over gCN to the effective suppression of electron and hole recombination photogenerated and a better redox capacity (Figure 18a).

In other study, Li et al. [123] prepared P-doped gCN in three steps, such that (i) synthesis of gCN by thermal polymerization of urea at 500 °C/4 h (heating rate: 2 °C·min⁻¹); (ii) porous gCN was synthetized with a new thermal treatment at 500 °C/2 h (5 °C·min⁻¹); (iii) and the subsequent doping of the gCN was made with sodium hypophosphite, SHP, prepared with different mass ratio for SHP to gCN, in a tube furnace at 350 °C/1 h. The resulting material exhibited a surface area of 202.9 m²·g⁻¹ (compared for gCN 73.8 m²·g⁻¹). The photocatalyst thus doped degraded 99.5% of the RhB (20 ppm) in 70 min (gCN 64.2% within the same time) with 100 ppm of photocatalyst under visible light (Table 6). The improvement in efficiency of the sample P-gCN (40:1) over undoped porous gCN is mainly due to its higher surface area and improved charge separation (Figure 18b).

P-doping can also be achieved by thermal condensation of gCN precursors with adenosine phosphates, sodium hypophosphite (NaH₂PO₂), etc. [54] Co-doping with other elements such as O or S is also interesting and gives good results in the degradation of pollutants in water bodies [134,135].

Doping with S

Another widely used non-metallic dopant is sulfur, S, which, like the non-metallic dopants mentioned above, enhances light absorption and improves charge mobility and separation [52,53,54]. The insertion of S into the gCN structure can be carried out by different techniques, such as by heat treatment of the gCN precursor in an SH₂ atmosphere, thermal polymerization (in air atmosphere) with S-containing precursors (thiourea, trithiocyanuric acid, thioacetamide, etc.), or by copolymerization of the precursors with S-containing compounds (benzyl disulfide, H₂SO₄, etc.) [54].

A representative example of S-gCN synthesis was reported by Guan et al. [124], who used dicyandiamide as the main precursor and trithiocyanuric acid (and thiourea) as S sources. The thermal treatment was carried out in a molten salts mixture (LiBr/KCl) at 500 °C for 3 h. By controlling the percentage ratio of trithiocyanuric acid to dicyandiamide, they adjusted the degree of doping. Thus, for a trithiocyanuric acid/dithiodiamide ratio equal to 5%, the material shows a gap in the material of 1.83 eV (gCN a gap of 2.55 eV) and a surface area of 15 m²·g⁻¹ (gCN 11 m²·g⁻¹). The material thus prepared achieved an improvement in photocatalytic activity in the degradation of MB and TC of 10 and 20 times, respectively; the effectiveness of the undoped gCN, i.e., 60% in 300 min for MB and 89% in 240 min for TC (gCN without any additive 4% for MB and 10% for TC, within the same time); with 100 ppm photocatalyst for MB (20 ppm) and 200 ppm photocatalyst for TC solution (20 ppm) under visible light (Table 6). The improvements in MB and TC degradation efficiency are mainly due to a slight improvement in surface area (at the expense of an improvement in microporosity), an improvement in charge separation, and an effective reduction in band gap (Figure 19a,b).

A ‘green’ option to prepare S-gCN was proposed by Dou et al. [125] in a study on the degradation of OTC under visible light with the nanosheets of the thus-doped gCN. These authors proposed a synthesis by thermal polymerization of thiourea in three steps: (i) the precursor was heated from room temperature to 259 °C and held for 30 min; (ii) then to 426 °C and held for 30 min; and (iii) finally to 550 °C and held for 4 h. The material thus prepared had a gap of 2.83 eV and a surface area of 31.17 m²·g⁻¹. The authors succeeded in 40 min to degrade almost completely (93.3%) the OTC (10 ppm) with 1000 ppm photocatalyst under visible light (Table 6). This modified gCN derivative retained 79% of the OTC degradation efficiency after four consecutive degradation cycles. The improvement in degradation compared to the unmodified material is mainly due to enhanced charge separation.

Doping with B

Doping with boron (B), like its non-metallic predecessors, improves both the surface area and the band gap. The doped gCN changes from an n-type semiconductor to a p-type semiconductor. As before, it is the introduction of intermediate energy states that reduces the gap [1]. Thermal condensation is the most common means used for this purpose. Thus, among the most frequent reagents as a source of boron are boron oxide, B₂O₃, borazane (BNH₆), boric acid (B(OH)₃), or sodium or potassium borohydrides (NaBH₄ or KBH₄) [1,2].

Among the most reported syntheses, in the different works consulted on the subject, mention should be made of doping by thermal synthesis in a muffle furnace, starting from melamine and B(OH)₃, as carried out by Lei et al. [126]. They prepared B-gCN supported on textile carbon fibers on RhB degradation, starting from melamine (5 g) and different quantities of boric acid in ethanol (EtOH), and coating with this solution, by immersion, carbon textile fibers (CFs) with subsequent thermal polymerization (calcination) at 570 °C for 2 h. The band gap for this sample (B-gCN-0.4) only reached a value of 2.69 eV (2.71 eV for the undoped and equally supported material). With the photocatalyst thus doped (B-gCN-0.4), a degradation of RhB (5 ppm) of 95% was achieved in 120 min (the undoped gCN 82%) with 68 mg of B-gCN on 62 mg of CFs under visible light (8 W LED lamp) (Table 6). This material not only achieves good degradability but also maintains the degree of degradation after 10 cycles of use. The improvement in activity is mainly due, according to the authors, to an improvement in charge mobility, charge separation, and porosity (Figure 20).

A final example of B-doping may be the work of Zou, Jingye et al. [127], in a paper on the degradation of RhB under visible light with the gCN thus doped. These authors synthesized the B-doped gCN by first dissolving melamine and boric acid (B(OH)₃) in EtOH. They then removed the EtOH by heating at 80 °C, followed by drying at 100 °C/6 h. The material was then treated in a vacuum microwave oven (4.0 kW) at 560 °C/10 min. They obtained one material (BCNNs) with a surface area and a pore volume of 105.06 m²·g⁻¹ and 0.229 cm³·g⁻¹, respectively (PCN, 17.45 m²·g⁻¹, 0.08 cm³·g⁻¹), and a band gap of 2.73 eV (2.7 eV for the PCN sample). This thus-doped material reached an RhB degradation percentage (2 ppm) of 97% (50% for the undoped material) in 30 min under visible light with 500 ppm of photocatalyst (Table 6). The modified gCN showed no loss in RhB degradation efficiency after four degradation cycles. The authors attributed the improvement in photodegradation to a longer lifetime of the photogenerated charges and an increase in the surface area and pore volume due to the formation of nanosheets (despite the increase in the band gap due to the quantum confinement effect).

Doping with Halogens

It is now time to mention halogen doping (F, Cl, Br, and I). Like their preceding non-metallic partners, they can enhance absorption and improve separation and charge mobility. Of the halogens, Cl seems to improve semiconductor properties the most. Possible syntheses are the thermal (pyrolytic) condensation of melamine (and/or urea) and ammonium chloride (NH₄Cl). These materials thus prepared have good efficiency in both CO₂ reduction and H₂ production [52,53].

A similar synthesis was carried out by Guo et al. [128], in the degradation of TC. The authors prepared the photocatalyst by mixing melamine with various amounts of NH₄Cl in deionized water. This solution was dried and then calcined at 550 °C for 3 h (0.5 °C·min⁻¹). The thus-modified gCN (CN-Cl-2 sample) showed a gap of 2.69 eV and the CN-Cl-1 sample of 2.7 eV, where the pristine gCN showed a gap of 2.75 eV (the gap decreases with Cl content), with a surface area of 114.4 m²·g⁻¹ and a pore volume of 0.34 cm³·g⁻¹ (42.3 m²·g⁻¹ and 0.11 cm³·g⁻¹ for the undoped gCN). The CN-Cl-1 sample achieved a TC degradation (10 ppm) of 92% in 120 min, k = 0.02 min⁻¹ (32%, k = 0.004 min⁻¹, for pristine gCN), under visible light (300 W; Xe lamp) with 500 ppm of photocatalyst (Table 6). After three consecutive cycles, this photocatalyst showed a scarce decrease in TC degradation. According to the authors, the improvement in degradation is mainly due to the improvement in separation and mobility of the photogenerated charges, reduction in the gap, and increase in specific area (Figure 21a).

A similar strategy (thermal copolymerization of melamine and the corresponding ammonium halide (NH₄Cl and NH₄Br), at 550 °C, 3h (15 °C min⁻¹)) for Br and Cl doping is provided by Hong et al. [129] in a paper on the degradation of OTC with Br-gCN and Cl-gCN nanosheets under visible light. They obtained a band gap of 2.78 eV for gCN, 2.73 eV for Cl-gCN, and 2.75 eV for Br-gCN. The authors concluded that with less Cl doping (5 g NH₄Cl/5 g melamine; CNN-Cl5, 1000 ppm), it is possible achieve the same percentage of OTC degradation (10 ppm) within the same time (150 min) as with more Br doping (15 g NH₄Br/5 g melamine; CCN-Br15, 1000 ppm), i.e., 75% degradation (gCN 30%) under visible light (35 W; LED), k = 0.004, 0.017, and 0.0179 min⁻¹ for the BCN, CNN-Cl, and CNN-Br samples, respectively (Table 6). In addition, they achieved a facile exfoliation route for this doped material compared to the conventional two-step techniques. During the calcination (polymerization) of melamine, these ammonium halides decompose into their respective hydrogen chloride, or fluoride, plus ammonia, which act as exfoliating agents. These small gaseous and polar molecules can easily intercalate in the gCN, facilitating its exfoliation, obtaining a yield for the exfoliated material of close to 50%. The photodegradation efficiency of the CNN-Br sample remained unchanged after the three cycles. The better efficiency of this type of doping was due to improvement in exciton lifetime (mainly) and a reduction in the band gap (Figure 21b).

Self-Doping

This list of non-metallic doping will end with self-doping; carbon doping, C-gCN; and nitrogen doping, N-gCN. This is a convenient method to improve photocatalytic activity without the introduction of defects, and/or impurities.

Carbon doping increases the number of delocalized π-bonds, which improves conductivity and thus charge transfer and separation, as well as adsorption capacity, and greatly broadens the absorption spectrum in the visible range [53,54].

Thus, regarding the doping of gCN with C, the work of Shi et al. [130] about the degradation of TC under visible light with C-gCN nanosheets can be mentioned. These authors prepared the photocatalyst by mixing 10 g of urea with different amounts of n-octanol (0.1; 0.5; 3; 5; 10 mL). It was stirred for 30 min and then heat-treated in a muffle furnace at 550 °C for 2 h (with a heating ramp of 5 °C·min⁻¹), achieving a slight improvement in the band gap of the material (from 2.77 eV for gCN to 2.71 eV for C-gCN-5), and also reaching with the sample treated with 5 mL of n-octanol (C-CN-5) a 77% degradation of tetracycline (TC; 30 ppm), at pH = 7; gCN was 40% at the same pH, in 60 min under visible light (30 W; LED) with 400 ppm of photocatalyst (Table 6). After three cycles, the degradation efficiency of TC was reduced by only 5%, indicating the good stability of this material. The authors point out that the degradation of TC in a basic medium is improved; at higher pH, the production of hydroxyl radicals, HO^•, is favored. The better efficiency was mainly due to the improvement in the charge separation (Figure 22a).

The synthesis proposed by Li et al. [131] can also be added, regarding the synthesis of porous C-gCN, in a work on the production of H₂O₂ with this photocatalyst and the subsequent degradation of Bisphenol A (BPA). They used dried and ground dead leaves as a carbon source and morphology regulator. Different materials were obtained by mixing different amounts of dead leaf powder with urea and then calcined at 550 °C for 2 h. The sample with 10 mg of leaves (PCCN10) had an effective band gap of 2.19 eV (2.88 gCN) and a specific area of 85 m²·g⁻¹ (43 m²·g⁻¹ gCN). This sample achieved a 96% degradation of BPA (25% the gCN), 10 ppm, in 60 min under visible light with 1000 ppm of photocatalyst (Table 6). The improvement in BPA degradation efficiency over undoped gCN was mainly due to an improvement in surface area, a reduction in the band gap, and an improvement in the separation of photogenerated charges. (Figure 22b).

Other reagents as a carbon source are DMF, agar gel and melamine, sucrose [54], and ammonium citrate [136].

As for nitrogen doping (N-gCN), it can be said to provide similar improvements to C. Thus, the transfer and separation of charge carriers are improved, apart from extending the absorption range. Nitrogen doping is less widely used than its non-metallic precedents, perhaps due to synthesis processes with not very ‘ecofriendly’ reagents such as hydroxylammonium chloride (NH₃OHCl) or hydrazine (N₂H₄). Other friendlier syntheses could consist of using trichloromelamine (C₃H₃Cl₃N₆) as the only precursor together with heat treatment at 600 °C, but also with precursors such as 3,6-di(azido)-1,2,4,5-tetrazine, 2,4,6-triazido-1,3,5-triazine, etc., precursors with which excellent results have been achieved both in the production of H₂ and in the degradation of pollutants [52,53].

An example of N-gCN synthesis is that proposed by Jiang et al. [132] on the degradation of TC with N-gCN nanosheets, with dicyandiamide dissolved in DMF, stirred (12 h), dried (100 °C), thermally condensed at 550 °C for 4 h, and with subsequent heat treatment at 500 °C (4 h) for exfoliation. The authors thus achieved a material with a specific surface area of 74.79 m²·g⁻¹ (18.4 m²·g⁻¹ for unmodified gCN); they reduced the band gap from 2.51 eV of the pristine material (DCN) to 2.47 eV of the N-doped material (NDCN-4), but due to the quantum confinement effect of exfoliation, the band gap increased to 2.54 eV for the N-gCNNSs (NDCN-4S). With the NDCN-4S sample (the doped and exfoliated material), they achieved TC degradation (10 ppm) of 81.7% in 60 min (52.2% for the unmodified gCN) with 500 ppm of photocatalyst under visible light (Table 6). After the fifth consecutive degradation cycle, this material showed a slight decrease in TC degradation efficiency (79.55% compared to 81.7% in the first cycle). The authors were thus able to verify improvements in the BET area, light absorption, and separation of the photoexcited charges, and thus in efficiency (Figure 22c).

A final example of N-doped gCN could be the work carried out by Zhu et al. [133], wherein they evaluated in the degradation of phenol (PhOH). The authors prepared the photocatalytic material by mixing urea and citric acid monohydrate (HOC(COOH)(CH₂COOH)₂-H₂O) and heat treating them (polymerization in a non-inert environment) at 550 °C/4 h (2 °C·min⁻¹). Different degrees of doping were achieved by changing the urea/citric acid ratio. They obtained an N-gCN sample (NCN-(2:2); 2 g urea/2 mg citric acid) with a slightly lower surface area than the undoped material, i.e., 72.26 m²·g⁻¹ versus 76.69 m²·g⁻¹ for undoped material. The band gap was reduced from 2.51 eV for the pristine material to 1.82 eV for the doped material (NCN-(2:2)). This sample showed a 70.1% PhOH degradation efficiency (37.6% degradation efficiency of pristine gCN), 10 ppm, within 180 min with 1000 ppm photocatalyst under visible light (300 W Xe lamp; Vis.) (Table 6). After three consecutive cycles of degradation, the efficiency decreases up to 50%, according to the authors, due to a significant loss of photocatalyst during the process. According to researchers, both the reduction in band gap and an improvement in the separation and lifetime of the excitons are responsible for the better efficiency compared to the undoped material (Figure 22d).

Non-metal doping in gCN allows for tuning its electronic structure by introducing intermediate levels, which improves conductivity and charge separation and reduces the gap. Oxygen doping significantly decreases the energy gap and optimizes carrier mobility, enhancing the degradation of pollutants. Phosphorus and sulfur doping enhances absorption in the visible range by favoring the delocalization of the π-system and improving charge separation, while boron doping transforms the material from an n-type to a p-type semiconductor, increasing electron mobility. Halogen doping not only reduces the gap and improves charge separation, but also facilitates exfoliation, increasing the surface area. Finally, self-doping with carbon or nitrogen extends the absorption in the visible spectrum and improves charge transfer without introducing additional defects, which together increase photocatalytic efficiency. As a final remark, it can be noted that this type of doping has minimal impact on the stability of the photocatalysts after several cycles (and if there is any decrease in efficiency, it is due to the loss of photocatalytic material during the process), indicating good stability of these derivatives in the reaction medium.

4.2.2. Metal Doping

The photocatalytic activity of gCN doped with metal nanoparticles from surface plasmon resonance (SPR)—by resonating with the incident radiation—enhances the excitation of electrons to the BC (more free electrons are introduced into the system) while introducing intermediate energy levels into the band gap that broaden the absorption spectrum and reduce the effective gap (especially with alkali metals) [1,13]. Nitrogen atoms located in the interstitial sites (6 N in heptazine- and 3 N in triazine-based structures) are responsible for interacting with metals via electrostatic interactions.

Several strategies have been adopted for this type of doping, such as deposition–precipitation techniques, thermal polymerization, the solvothermal method, photodeposition, etc., as pointed out by Thomas et al. [1] in a review article. Thus, these authors refer to the doping of gCN with Pt nanoparticles by solvothermal method (after thermal condensation of the precursor) with chloroplatinic acid hexahydrate (H₂PtCl₆-6H₂O), mixing the gCN obtained with this Pt source in ethanol, with ethanol being the reducing agent that will reduce the Pt⁺⁴ nanoparticles. Another similar example, for Ni doping, reported in the same literature review [1], is also carried out by solvothermal method with Ni acetylacetonate (Ni(CH₃COCHCOCH₃)₂) dispersed in DMF, which, after being mixed with the gCN, was autoclaved. Other salts used as metal sources, for the same (solvothermal) method, would be, for example, CoCl₂ and HAuCl₄ for Co and Au, respectively [1], or potassium fluoride KF and NaBH₄ for K and Na insertion, respectively [53]. Another way to carry out this type of doping would be by thermal condensation, for example, of H₂PdCl₄ and dicyandiamide, dispersed in deionized water and mixed at 80 °C, and heat-treated in an Ar atmosphere. Doping with alkali and alkaline earth metals by heat treatment with the corresponding halides is also possible (Figure 23a) [137]. To illustrate the diversity in metal doping strategies, the following section describes example syntheses for each type of metal dopant.

Alkaline and Alkaline Earth Metal Doping

For alkali metal doping, Zhang et al. [138] studied the optical properties of alkali metal-doped gCN (not focused on EP degradation button CO₂ reduction, though the synthesis remains of interest). They synthesized doped gCN by mixing 3 g of urea with 0.1 g of different alkali metal nitrates (LiNO₃, NaNO₃, KNO₃, and RbNO₃), mixed in DW, and treated with ultrasound for 5 h. The water content was then evaporated at 70 °C, followed by heat treatment (calcination) at 500 °C 2 h. (2.3 °C·min⁻¹), and then the powder thus obtained was heat treated at 350 °C 3 h. They obtained reduced gaps for all the doped materials, e.g., 2.2 eV for K-gCN, and only 2 eV for Rb-gCN.

In another synthesis, Yan et al. [137] dissolved urea in an aqueous solution of alkali (Na and K) and alkaline earth (Ca and Mg; 3% by weight) bromides, mixed them overnight, and then heat-treated them at 550 °C for 2 h (15 °C·min⁻¹). Gaps ranging from 2.46 eV for Mg-gCN to 2.29 eV for Ca-gCN, 2.41 eV for K-gCN, and 2.43 eV for Na-gCN (undoped gCN had a gap of 2.57 eV) were obtained, in addition to a considerable increase in the half-life of the charge carriers. The ENR degradation kinetic coefficient (5 ppm) was multiplied by 3.3 for Ca-gCN (50%), by 4.4 for Mg-gCN (70%), by 4.8 for Na-gCN (80%), and by 5.4 for K-gCN (95%), with respect to gCN (22%) in 120 min under visible radiation, with 400 ppm of photocatalyst (

Table 7. Parameters of photodegradation of pollutants in water with metal doped gCN.

Dopant	Photocatalyst	Area (m2·g−1)	Gap (eV)	Pollutant (ppm)	Radiation	Efficiency (%)	Time (min)	k (min−1)	Refs.
Alkaline and alkaline earth metals	gCN (400 ppm)	177	2.57	ENR, SMX, and TC: 5	300 W Xe lamp (Vis.)	ENR: 22	120	0.014	[137]
						TC: 42		0.036
						SMX: 19		0.001
	Mg-gCN (400 ppm)	107	2.46			ENR: 78		0.062
						TC: 79		0.067
						SMX: 40		0.003
	Ca-gCN (400 ppm)	100	2.29			ENR: 58		0.046
						TC: 78		0.067
						SMX: 35		0.003
	K-gCN (400 ppm)	81	2.41			ENR: 82		0.075
						TC: 80		0.072
						SMX: 67		0.006
	Na-gCN (400 ppm)	90	2.43			ENR: 81		0.067
						TC: 78		0.070
						SMX: 65		0.006
Rare earths (Ln)	gCN	46.33	2.73	RhB, MB, and PR: 15	300 W Xe lamp (Vis.)	RhB: 54	150	0.0062	[139]
						MB: 52		0.0061
						PR: 56		0.007
	Yb-gCN	52.56	2.5			RhB: 88		0.017
						MB: 91		0.019
						PR: 88		0.018
	Nd-gCN	46.38	2.56			RhB: 74		0.012
						MB: 87		0.018
						PR: 74		0.015
	Ce-gCN	58.09	2.47			RhB: 91		0.019
						MB: 93		0.023
						PR: 91		0.020
Noble metals	gCN (1000 ppm) Ag-gCN (1000 ppm)	59.5	2.66	TC, OTC, and CTC: 20	300 W Xe lamp (Vis.)	TC: 35	120	0.0037	[140]
						OTC: 50		0.0053
						CTC: 43		0.0039
		58.4	2.55			TC: 83		0.0142
						OTC: 81		0.0130
						CTC: 85		0.0139
	gCN (1000 ppm)	-	2.7	BZF: 3	500 W Xe lamp (Vis.)	27	90	0.012	[141]
	Pd-gCN (1000 ppm)	-	2.62			100		0.036
Transition metals	gCNNSs (600 ppm)	91	2.7	RhB: 20	300 W Xe lamp (Vis.)	60	60	0.012	[142]
	Fe-gCNNs (600 ppm)	132	2.5			100	30	0.117
	gCN (1000 ppm)	95.95	2.73	MB; 10CIP; 15	300 W Xe lamp (Vis.)	MB: 67	150	0.013	[143]
						CIP: 30		0.002
	gCNNSs (1000 ppm)	70.17	2.76			MB: 81		0.014
						CIP: 37		0.003
	Ni-gCNNSs (1000 ppm)	8.23	2.68			MB: 93		0.017
						CIP: 52		0.003
	gCN (600 ppm)	11.4	2.7	TC: 30	300 W Xe lamp (Vis.)	47	120	-	[144]
	Cu-gCN (600 ppm)	142.8	2.45			98		-

). For SMX and TC degradation, the degradation efficiency followed the same order as for ENR (K-gCN > Na-gCN > Mg-gCN > Ca-gCN > gCN). The improvements in respect to the efficiency of undoped gCN are mainly due to the reduction in band gap and an improvement in exciton lifetime (Figure 23a).

Rare Earths (Ln) Doping

For rare earth (lanthanide; Ln) doping, the available literature is scarce, and only the research by Yu et al. [139] is discussed. The study focused on the degradation of organic dyes (RhB, MB, and PR) using gCN doped with rare earths (Yb, Nd, and Ce). They used 10 g of urea and 2% molar Yb(NO₃)₃·5H₂O, Nd(NO₃)₃·6H₂O, and Ce(NO₃)₃·6H₂O, dissolved in 15 mL of deionized water with stirring 30 min, then dried it, followed by heat treatment at 550 °C for 2 h (2 °C·min⁻¹). They found that photocatalytic activity improved for all three metals. Ce-gCN is the material with the largest surface area of the three, with 58.09 m²·g⁻¹ (46.33 m²·g⁻¹ for undoped gCN). The calculated band gap was 2.47 eV for Ce-gCN (2.73 eV for gCN). The most efficient doping was Ce doping, reaching 91.23% for RhB, 93.45% for MB, and 91.25 for PR, all 15 ppm, in 150 min (54%, 53%, and 56% with the undoped gCN for RhB, Mb, and PR, respectively) with 300 ppm photocatalyst under visible light (Table 7). The authors studied the stability of this material only in the degradation of RhB and MB. Thus, after five degradation cycles, this sample (Ce-gCN) showed a slight decrease in the degradation efficiency of RhB, and no loss in MB degradation. The enhanced efficiency was mainly attributed to band gap narrowing and improved separation of photogenerated charges.

Noble Metals Doping

As an example of noble metal doping, the work of Ren et al. [140] on the degradation of different tetracyclines (OTC; CTC; and TC) with Ag-gCN under visible light can be highlighted. These authors prepared the semiconductor thus doped by dissolving gCN (obtained by thermal treatment, calcination, at 550 °C/4 h., 2.3 °C·min⁻¹) in deionized water, and mixing this solution with another aqueous solution of AgNO₃ (Ag/gCN: 4%, 6%, 8%, 10%, and 12% by weight), with subsequent photodeposition (with a 300 W Xe lamp for 2 h). It was then centrifuged, washed, and dried. The material obtained (Ag-CN-8) had a surface area of 58.4 m²·g⁻¹ (gCN of 59.5 m²·g⁻¹) and a band gap of 2.55 eV (2.66 eV for gCN). They obtained a photocatalyst with an improved band gap, and with a longer half-life of the charge carriers. The best degree of degradation was obtained with the Ag-gCN-8 sample (8% Ag): 83% for TC; 81% for OTC; and 85% for CTC (all three antibiotics 20 ppm) in 120 min, and with 1000 ppm of the photocatalyst under visible light (Table 7). After four consecutive photocatalytic cycles, a reduction in Ag content from 7.47% to 6.20% was observed, indicating that the diminished catalytic efficiency could be attributed to silver leaching during the recovery process. They (the authors) concluded that better efficiency was due to the better separation of generated charges and a reduced band gap, and that by increasing the pH, the tetracyclines are negatively charged, and although this may decrease the adsorption on gCN, at the same time, the production of hydroxyl radicals is enhanced, thus improving degradation (Figure 23b).

Another example of noble metal doping is provided by Yin et al. [141] in the degradation of bezafibrate (BZF; a lipid regulator) under simulated sunlight, using Pd-gCN as a photocatalyst. These authors prepared the Pd-doped material by mixing gCN powder synthesized from melamine by thermal condensation in two steps (first at 500 °C for 2 h, at 2 °C·min⁻¹; and then, after cooling and grinding, a subsequent treatment at 550 for 3 h, also at 2 °C·min⁻¹). They then added Pd(NO₃)₂ (different percentage ratios) to 0.2 g gCN and treated with ultrasound 1 h. It was then stirred for 12 h at 25 °C. NaBH₄ (0.1 M) was added and stirred magnetically for another 2 h. The product obtained was then washed and dried overnight. The narrowest band gap was for the sample with Pd/gCN ratio of 1%, with 2.62 eV (2.7 eV for undoped g-CN). The material thus obtained (Pd-gCN-1%) had a BZF degradation efficiency (3 ppm) of 100% in 90 min (27% for gCN at the same time) with 1000 ppm photocatalyst (Table 7). After five consecutive degradation cycles, this doped g-C₃N₄ derivative exhibited only a 10% loss in BZF removal efficiency. In this case, the better efficiency was due to the reduced band gap and due to the improvement in charge separation (Figure 23c).

Figure 23. (a) Transient photocurrent of alkaline- and alkaline earth-doped gCN [137]; (b) transient photocurrent of gCN doped with Ag in different amounts [140]; (c) PL spectra of Pd-gCN [141]; (d) the transient photocurrent of gCN doped with Cu [144].

Figure 23. (a) Transient photocurrent of alkaline- and alkaline earth-doped gCN [137]; (b) transient photocurrent of gCN doped with Ag in different amounts [140]; (c) PL spectra of Pd-gCN [141]; (d) the transient photocurrent of gCN doped with Cu [144].

Finally, this section discusses the Au (Au-gCN) doping reported by Faisal et al. [145], in a study focused on the degradation of MB and gemifloxacin mesylate (an antibacterial compound) under visible light. Thus, these authors report a synthesis (like the previous method with Pd) by treating gCN (synthesized by thermal condensation of urea at 550 °C/3 h.), in aqueous solution, with different amounts of HAuCl₄ (0.2%, 0.5%, 1%, 2%, and 5%), agitated by ultrasound for 1 h. It was then filtered by centrifugation, washed, and dried. The band gap of these doped materials was almost the same for all of them (2.86 eV, compared to 2.85 eV for undoped gCN). These authors concluded that the sample with the best BZF degradation efficiency was the Au-gCN-1% sample with 95.13% dye degradation in 90 min (69% for gCN at the same time) with 400 ppm photocatalyst under visible light (400 W; lamp metal halide). This photocatalyst showed no loss in degradation efficiency after five consecutive degradation cycles. The authors attributed the increase in efficiency for Au-gCN-1%, with respect to gCN, to a substantial increase in the lifetime of the excited charge pairs.

Transition Metal (Non-Noble) Doping

To conclude this subsection, transition metal doping is mentioned. It was precisely because of the high cost of noble metals (and their scarcity) that transition metals became an alternative to noble metals as dopants [4]. Like their metallic partners, transition metals can also improve the absorption range by decreasing the gap, as well as improving mobility and charge separation [2,13,52].

To begin with, as an example of Fe doping, the work carried out by Nguyen Van et al. [142] on the degradation of RhB with Fe-gCN is worth mentioning. They carried out the synthesis by dissolving 0.5 g of gCN nanosheets (obtained from urea by thermal polymerization at 550 °C, 2 h) in distilled water with magnetic stirring for 30 min, followed by ultrasound treatment for 1 h. To this solution, different amounts of FeCl₃-6H₂O (3, 5, 7, 8, and 10 molar) were added, followed by heat treatment at 90 °C for 12 h with magnetic stirring. This solution was centrifuged, and the residue was rinsed with EtOH and dried. This sample had a surface area of 132 m²·g⁻¹ (91 m²·g⁻¹ for gCN nanosheets) and a band gap of 2.5 eV (2.7 eV for gCN). The material thus prepared (Fe-gCN-7 sample) showed a degree of degradation of RhB (20 ppm) of 100% in 30 min, with a kinetic coefficient 10 times higher than that of gCN with 6000 ppm of photocatalyst under visible light (Table 7). The photodegradation percentage of RhB was still up to 95% after three cycles. They also concluded that, on the one hand, Fe⁺³ traps the photoexcited electrons, allowing for a longer lifetime of the excited pairs, which, together with the increase in specific area, improves the degradation of RhB.

Another doping procedure with this type of metal, namely, Ni, was carried out by Zhou et al. [143], where they managed to prepare Ni-gCN nanosheets, with good degradation of MB and CIP under visible light. The photocatalyst was prepared in several steps: (i) synthesis of gCN by calcination at 520 °C for 4 h; (ii) hydrothermal exfoliation of the gCN (after ultrasonic treatment in HCl solution for 30 min) at 150 °C for 5 h in an autoclave; (iii) additional calcination at 500 °C for 2 h; (iv) mixing of the exfoliated nanosheets with the Ni precursor (Ni(CH₃COCHCOCH₃)₂; 5, 10, 15, and 20 wt.% Ni) in DMF under ultrasonic stirring; (v) the resulting solution was solvothermal treated at 200 °C for 10 h. The authors were able to verify that the surface area of the nanosheets decreased with increasing Ni %; for Ni-CN-10%, this was 70.17 m²·g⁻¹ (95.95 m²·g⁻¹ for undoped nanosheets, CN-500). This photocatalyst thus doped showed a band gap for Ni-gCN-10% equal to 2.68 eV (nanosheets 2.76 eV and unmodified gCN 2.73 eV). The degradation of MB (10 ppm) and CIP (15 ppm) with 1000 ppm photocatalyst improved with increasing Ni content up to 10% with 93% degradation (67% untreated gCN) for MB, and 52% (30% untreated gCN) for CIP, in 150 min under visible light (Table 7). The Ni-doped gCN, after three degradation cycles, showed no loss in efficiency for either MB or CIP degradation. According to the authors, the improvement in the efficiency of this material (Ni-CN-10%) compared to gCN nanosheets (CN-500) and bulk gCN is mainly due to a reduction in the band gap and an improvement in photogenerated charge separation. The material is also magnetically recoverable.

Finally, Cu doping is added to this subsection, mentioning the work of Bao et al. [144] on the degradation of TC with Cu-doped gCN. The synthesis was carried out by calcination at 550 °C 6 h (1.5 °C·min⁻¹) from a mixture of melamine, cyanuric acid, and copper acetate (Cu(OAc)₂), with 0, 4, 8, 12, and 16 wt.% copper acetate, dissolved in EtOH (after evaporation). The 3Cu-pCN sample had a surface area d 142.8 m²·g⁻¹ and a cumulative pore volume of 1.15 cm³·g⁻¹ (11.4 m²·g⁻¹ and 0.11 cm³·g⁻¹ for the gCN). The best gap was achieved with a 3Cu-pCN sample (12% Cu) equal to 2.45 eV (2.7 eV for gCN), reaching the best TC degradation (30 ppm), 98% in 2 h (47% for gCN within the same time), with 600 ppm of photocatalyst under visible light (Table 7). This photocatalyst (3Cu-pCN sample) showed a decrease in efficiency from 98% to 80.6% after seven cycles of TC degradation, attributed to catalyst loss during washing and centrifugation steps. The improved performance of degradation was mainly attributed to the narrowed band gap, enhanced visible light absorption, and increased exciton lifetime (Figure 23d).

Metal doping in g-C₃N₄ takes advantage of plasmon resonance to increase electronic excitation and reduce the energy gap, which broadens the absorption of visible light. This modification introduces intermediate energy levels that facilitate more efficient electron transfer and improve charge separation. Doping with alkali, alkaline earth, and rare earth metals decreases the gap and extends the lifetime of excitons, boosting the degradation of pollutants. In addition, the incorporation of noble and transition metals further optimizes carrier mobility, resulting in higher photocatalytic efficiency. Taken together, these strategies enhance adsorption, with charge transfer notably strengthening the performance of gCN in environmental applications. Finally, it should be noted that, regarding the stability of metal-doped photocatalysts, their efficiency is generally maintained over multiple reuse cycles, with observed declines typically attributable to physical losses of material during handling rather than to structural degradation.

5. Conclusions

This review provides a comprehensive and detailed overview of the different methodologies for gCN synthesis and modification and a set of reference data for photocatalyst design strategies. These contributions pursue the development of more efficient, sustainable, and economically competitive water decontamination systems.

gCN is a highly promising and sustainable alternative to conventional catalysts based on metal oxides, chalcogenides, nitrides, or carbides. Its use minimizes the environmental impact associated with the extraction and processing of traditional metal-based materials while reducing energy and operating costs, since its ability to absorb light in the visible range enables the use of renewable energy sources. Moreover, the synthesis of gCN using low-cost and environmentally benign precursors, such as urea, presents a competitive advantage that paves the way for large-scale production.

In this context, surface functionalization, defect introduction, and doping strategies (both with non-metals and metals) are of fundamental relevance, as they allow for fine tuning the electronic and structural properties of gCN, improving light absorption, charge separation, surface area, and, consequently, its performance in the degradation of a wide variety of pollutants. The extensive body of literature reviewed here demonstrates the high efficiency of these materials in the degradation of EPs in aqueous media, and their stability in reaction media without significant losses in photocatalytic efficiency, placing them at the epicenter of research in heterogeneous photocatalysis. However, several limitations have been identified in the materials and their evaluation:

• The documented leaching of metals when doping gCN and taking place during degradation cycles could be an environmental and/or public health issue, and further research into doping techniques to overcome this effect would be desirable.
• The stability of the synthesized materials is a matter of concern and there is no consensus in this respect. Not all the studies present these stability studies, whether carried out with only 2 or 3 or up to 10 consecutive degradation cycles, which makes it difficult to have a clear perception of this. Given the operational requirements, i.e., the high flow rates to be treated (both in WWTPs and DWTPs), and the need for the photocatalysts to be economically viable, more than five cycles would be necessary to get an idea of the operational stability of these materials in real situations. It is also noteworthy that 2D and porous 3D structures frequently exhibit superior recyclability, which positions them as particularly promising for long-term applications. It would also be advisable to accompany with stability studies—under the same conditions—of unmodified gCN in order to obtain a clear idea of the effect of the modifications on the stability of these gCN derivatives. Also, accompanying these stability studies with characterizations (FTIR, XRD, etc.) of the modified photocatalyst would reinforce this perception.
• The lack of a standardized protocols for both the study and preliminary optimization of catalysts (generally with organic dyes) and for the degradation conditions of model pollutants—as exists for the evaluation of photocatalytic ceramic materials in aqueous media (ISO 10678) [146], for the evaluation of methods for air-purification performance of semiconductor photocatalytic materials (ISO 22197) [147], or for water purification performance of semiconductor photocatalytic materials by measurement of forming ability of active oxygen (ISO 10676) [148,149]—is a drawback. Standardization of these processes would be desirable, as it would facilitate comparability between materials.
• For the degradation of model EPs, experimental studies should also include assessments of the potential toxicity of both the materials and the degradation products of the target EPs. This approach would improve the positive perception of heterogeneous photocatalysis within industrial sectors.
• Finally, in the 21st century, and amid the planet’s environmental crisis, there is requirement for a commitment by researchers to align their work with Green Chemistry principles (the 12 tenets established in 1998 by Paul Anastas and John Warner).

Other aspects that can be considered as future directions or opportunities identified in this review are as follows:

• Recent advances in DFT and machine learning offer a powerful means to pre-screen and predict the performance of modified gCN, allowing for researchers to optimize band structure, charge dynamics, and catalytic behavior before experimental validation. Integrating these computational tools could greatly accelerate materials discovery while aligning with the principles of green chemistry. Few studies incorporate prior in silico analysis to guide and justify the selection of gCN modification methodologies, which would enhance the sustainability of experimental designs and processes.
• The improvement of photocatalytic capabilities by the combination of photocatalysts in the form of heterojunctions or homojunctions is also an important aspect for the improvement of the photodegradation efficiency of PEs that must be assessed by photocatalysis researchers. Although this review deliberately focuses on modifications of gCN itself, it is essential to highlight that the integration of gCN into heterojunction systems (e.g., type II, Z-scheme, Schottky) offers exciting avenues for synergistic performance improvements, meriting dedicated and systematic exploration in future studies.
• Immobilization of photocatalytic materials, although addressed in numerous studies, requires solutions and more investigations that do not compromise the inherent photocatalytic efficiency of gCN. Even though it is not an aim of this review, the fine and dispersed nature of gCN nanoarchitecture complicates its recovery after the photocatalytic process, as was detected in the reuse of the materials relying mostly on the loss of catalytic material. It is a challenge shared with other nano-photocatalysts, both metallic and those based on carbonaceous compounds.

The integration of these approaches, combined with innovative support or immobilization techniques, is essential to advance the scalability and economic viability of heterogeneous photocatalysis as a large-scale decontamination solution.

Finally, while gCN represents a powerful tool for the degradation of pollutants in water bodies, it is essential to remember that the ultimate solution to water pollution lies in prevention. The integration of advanced technologies with prevention and environmental management strategies will undoubtedly be the key to achieving effective and long-lasting decontamination outcomes.