Recent Strategies for Hydrogen Peroxide Production by Metal-Free Carbon Nitride Photocatalysts

Abstract

Hydrogen peroxide (H₂O₂) is a chemical which has gained wide importance in several industrial and research fields. Its mass production is mostly performed by the anthraquinone (AQ) oxidation reaction, leading to high energy consumption and significant generation of wastes. Other methods of synthesis found in the literature include the direct synthesis from oxygen and hydrogen. However, this H₂O₂ production process is prone to explosion hazard or undesirable by-product generation. With the growing demand of H₂O₂, the development of cleaner and economically viable processes has been under intense investigation. Heterogeneous photocatalysis for H₂O₂ production has appeared as a promising alternative since it requires only an optical semiconductor, water, oxygen, and ideally solar light irradiation. Moreover, employing a metal-free semiconductor minimizes possible toxicity consequences and reinforces the sustainability of the process. The most studied metal-free catalyst employed for H₂O₂ production is polymeric carbon nitride (CN). Several chemical and physical modifications over CN have been investigated together with the assessment of different sacrificial agents and light sources. This review shows the recent developments on CN materials design for enhancing the synthesis of H₂O₂, along with the proposed mechanisms of H₂O₂ production. Finally, the direct in situ generation of H₂O₂, when dealing with the photocatalytic synthesis of added-value organic compounds and water treatment, is discussed.

1. Introduction

H₂O₂ is considered an environmentally friendly, active, and safe chemical in a broad range of applications, acting as a powerful oxidizing agent. H₂O₂ has been reported in environmental remediation, textile whitening, paper bleaching, chemical synthesis, cleansing of electronic materials, processing of metals, energy storage, and electric energy generation in fuel cells [1,2,3].

H₂O₂ is frequently present in the textile industry, reacting with the colouring matter during the bleaching process [4,5]. In organosynthesis, H₂O₂ has the capability to accelerate oxidation reactions, being exploited for the production of many fine and bulk chemicals [6]. Another application for H₂O₂ is in the field of fuel cells as it is a promising alternative energy carrier [7,8,9,10], due to its high energy density and ease/safety of storage (contrasting with H₂ that presents some storage issues [11]). H₂O₂ can play a crucial role in the removal of pollutants from aqueous or gaseous effluents, being commonly combined with catalysts, ozone, or a light source. Wet peroxide oxidation (with and without catalysts), Fenton and photo-Fenton oxidations, and sono-, electro-, or photo-chemical reactions are some of the main processes where H₂O₂ is employed for water and wastewater treatment [12,13,14,15,16,17,18,19]. Particularly, H₂O₂ is widely used in photocatalytic water treatment because the addition or even in situ generation of H₂O₂ has been established as a promoter of the degradation of pollutants [20,21,22,23,24,25,26,27,28,29,30,31,32].

In response to an increasing H₂O₂ demand, different production processes have been developed and are currently operating at industry and laboratory scale [1]. However, many of these technologies often involve security issues regarding H₂/O₂ gas mixture, the use of organic solvents, metal catalysts, and the generation of undesirable by-products and solvent wastes [3]. Thus, to reduce these disadvantages, numerous approaches have been studied and developed for the synthesis of H₂O₂. These methods include: (i) chemical, sonochemical, and electrochemical processes, (ii) catalytic, and (iii) photocatalytic (metal-based or -free) systems.

H₂O₂ is produced mainly by the anthraquinone (AQ) route, commonly known as Riedl-Pfleiderer process [2,33,34,35]. This process is based on redox reactions [2,36,37], carried out in the presence of a mixture of organic solvents, e.g., ester/hydrocarbon or octanol/methyl-naphthalene. An alkylanthraquinone reacts with hydrogen, in the presence of a noble metal catalyst (e.g., Pt or Pd), and the corresponding alkylanthraquinol. The latter is oxidized under oxygenated conditions (air or O₂), generating H₂O₂ and the former alkylanthraquinone. The quinol species can be further hydrogenated and oxygenated under the same conditions producing H₂O₂ and tetrahydroanthraquinone. Afterwards, H₂O₂ is recovered from the organic solution with water to obtain a 30% H₂O₂ solution, which is then distilled under reduced pressure to remove impurities and increase its concentration. The anthraquinone process avoids implementing a H₂/O₂ gas mixture, which is known to be explosive, with this process being economically viable for large-scale production operations. However, there are still high operating costs associated with the acquisition of the raw materials, replacement of the catalysts, and the high energy requirement during the production procedure [2]. Another interesting technology for H₂O₂ generation consists on the application of sonochemistry which, by dissociation of oxygen and water, leads to the formation of H₂O₂ [38,39,40,41]. However, limitations of ultrasounds for chemical synthesis include long exposure time that can induce H₂O₂ decay and the difficulty in reactor design and process scale-up [42,43].

To reduce energy costs, the design of efficient and inexpensive catalytic processes has become imperative for the generation of H₂O₂. The use of metallic catalysts, such as Pd- and Au-based catalysts or metal oxides (e.g., Al₂O₃ or TiO₂), has been thoroughly documented due to their high selectivity [44,45,46,47,48,49,50,51,52,53,54,55,56]. Although several catalytic technologies have been discussed in the literature, the focus of this review is on the application of heterogeneous photocatalysis for H₂O₂ production. Briefly, in heterogeneous photocatalysis, an optical semiconductor is irradiated by an appropriate light source for its activation, leading to the formation of photogenerated electron/hole pairs which, under certain conditions, have the ability to produce H₂O₂ [57]. Depending on the reaction medium, the electronic properties of specific optical semiconductors, and the wavelength and intensity of the radiation source, heterogeneous photocatalysis can be adjusted to different processes (e.g., degradation of aqueous contaminants and production of high-value chemicals, among others [58,59]). Furthermore, photocatalytic processes may be employed under ambient temperature and pressure conditions and can be activated by free and inexhaustible solar light, further reducing the energy costs in comparison with traditional thermally activated routes. In a careful look at the literature, it was possible to retrieve more than 11,000 scientific reports on the various approaches for H₂O₂ synthesis (source: Scopus database, October 2018). Although photocatalytic production of H₂O₂ can be considered a recent topic, in recent years, the number of publications has been increasing significantly, which indicates that this technology could be considered an interesting alternative to traditional routes.

Metal-metal oxide hybrid catalysts have also been widely studied as promising candidates in the context of photocatalytic production of H₂O₂, owing to their high chemical stability and low toxicity [60,61,62,63]. Among all the metal elements, Ru, Bi, Co, and Cd are the most commonly combined with optical semiconductors, such as TiO₂ (the standard photocatalyst) [36,64,65,66,67,68,69,70,71]. Despite their high efficiency for several photocatalytic applications, metal catalysts present viability problems and poor sustainability due to their high cost, difficulty of extraction from their ores, and possibility of leaching, which can contaminate the reaction medium and lead to the generation of hazardous wastes [72,73,74,75].

In this way, the development and optimization of metal-free structures as heterogeneous photocatalysts has been attracting wide interest. In the scope of metal-free photocatalytic H₂O₂ generation, the most employed material is polymeric carbon nitride (g-C₃N₄, here denoted as CN). Polymeric CN generally shows a relatively narrow band gap enabling visible light absorption, yet fast recombination of the photogenerated electron/hole pairs generally occurs. A sustainable process for photocatalytic H₂O₂ synthesis, besides requiring a visible-light activated metal-free material, requires the use of water instead of organic solvents (e.g., alcohols as sacrificial agents). CN can catalyse water splitting and selectively produce H₂O₂ under visible light irradiation though oxygen reduction [76,77,78].

Depending of the CN precursor and preparation method, the band potentials of CN (typically −1.12 and 1.58 eV, for the valence and conduction levels, respectively) are thermodynamically suitable for several applications, especially organic synthesis, H₂ production, and pollutants degradation [79,80,81,82,83]. Several reports have shown that pristine CN materials hold a small specific surface area and low chemisorption of oxygen, commonly leading to low photocatalytic efficiencies [79,84]. Thus, several modification strategies have been pursued to increase the efficiency of CN-based photocatalysts, such as soft templating approaches, exfoliation, elemental doping, or heterojunction formation [79,85]. However, some authors report that, besides material engineering, photochemical considerations are a rather important issue to discuss since proper charge separation and transfer kinetics are needed for efficient H₂O₂ production [86,87].

Recently, Haider et al. [88] briefly summarized some studies concerning the synthesis of hybrid CN-based photocatalysts for H₂O₂ production. Regardless of the several studies reported on the use of CN for H₂O₂ production by heterogeneous photocatalysis, a consolidated knowledge focused on the influence of both metal-free CN structure and the manipulation of the operating conditions for improving H₂O₂ productivity is still missing.

In the present review, the photocatalytic production of H₂O₂ using metal-free CN is explored. Herein, the influence of the catalyst tuning and process optimization is correlated and discussed. The capability of simultaneously producing H₂O₂ during the synthesis of value-added organic compounds and their ability to induce high photocatalytic degradation of organic pollutants present in waters is a crucial point taken into consideration in this study.

2. H₂O₂ Production by Carbon Nitride Photocatalysts

The photocatalytic production of H₂O₂ can occur through several pathways, which may differ depending on the use of metallic or non-metallic catalysts. The use of metal-based photocatalysts to generate H₂O₂ is well discussed in the literature [62,63,89,90,91,92,93]. However, using non-metals, such as CN materials, the reaction pathway for H₂O₂ generation is still under study. A general scheme is depicted in Figure 1, illustrating the main steps occurring after photoactivation of CN in the presence of molecular oxygen and water. After light absorption, the migration of electrons from the valence band (VB) to the conduction band (CB) occurs. Then, electrons (e⁻) in the CB and photogenerated holes (h⁺) migrate to the surface of the photocatalyst and participate in reduction and oxidation reactions, respectively.

To improve H₂O₂ synthesis with photoactive CN, it is necessary to understand the reactions that take place at the photocatalyst surface. The pathway for H₂O₂ synthesis using CN photocatalysts is generally ascribed to the capacity of this material to drive two-electron oxygen reduction. Figure 2 represents the photoactivation of CN with visible light and the mechanism for H₂O₂ production suggested by Shiraishi et al. [94]. The authors propose that the electron/hole pairs are localized in the 1, 4 and 2, 6 positions highlighted in Figure 2, with the negatively charged sites attracting oxygen and later reacting with the trapped protons in nearby N atoms. Then, a selected alcohol is used as sacrificial agent. Photoexcited electrons react with O₂, leading to the formation of the 1,4-endoperoxide species, which results in the liberation of H₂O₂. At the same time, a proton donor (e.g., an alcohol or water) undergoes oxidation and yields protons that contribute to generating H₂O₂ [76]. The efficient formation of the endoperoxide species suppresses one- and four-electron reduction of O₂ (Equation (1), (2), respectively), improving the selectivity of two-electron reduction of O₂ (Equation (3)). However, O₂ on pristine CN preferably undergoes reduction to O₂^•‒ via one-electron reduction (Equation (1)), while on modified CN materials with more surface defects, a more facile production of H₂O₂ is achieved (Equation (3)) [95].

O₂ + e^‒ → O₂^•–

(1)

O₂ + 4 e^‒ + 4 H⁺ → 2 H₂O

(2)

O₂ + 2 e^‒ + 2 H⁺ → H₂O₂

(3)

Shiraishi et al. [76] reported Raman spectroscopy and electron spin resonance (ESR) studies that in their work, rationalizing the mechanism behind the selective two-electron reduction of O₂ on photoexcited CN. After irradiation, the Raman spectrum of CN catalyst in an O₂-saturated solution showed a new broad peak at 891 cm⁻¹, ascribed to bond vibrations in the 1,4-endoperoxide species. Concerning the ESR analyses, products of one-electron reduction of O₂ were not found, proving that O₂ is being selectively reduced to H₂O₂ [76].

In another work, authors reported the formation of H₂O₂ by a two-step single-electron reduction of O₂ (Equations (1) and (4)), via the reduction of O₂^•‒ (reduction product of O₂) to H₂O₂ [96]. Additionally, under suitable conditions (i.e., with an appropriate VB energy), water oxidation may occur, generating H₂O₂ (Equation (5)) [96,97]. H₂O₂ can also evolve via the transformation of HO^• formed from the hole-oxidation of HO^‒ (Equation (6), (7)) [96]. These pathways enable the production of H₂O₂ by both oxidation and reduction routes (Figure 1). However, H₂O₂ production from water and O₂ is hard to facilitate using the bulk CN photocatalyst, owing to the small thermodynamic driving force between the VB energy (1.4 V) and the water oxidation potential (0.8 V) [76,98,99,100].

O₂^•– + e^‒ + 2 H⁺ → H₂O₂

(4)

2 H₂O + 2 h⁺ → H₂O₂ + 2 H⁺

(5)

OH^‒ + h⁺ → HO^•

(6)

2 HO^• → H₂O₂

(7)

H₂O₂ + e^‒ → OH^‒ + HO^•

(8)

H₂O₂ can drive the production of HO^• in the presence of light (Equation (8)), depending on the radiation wavelength and catalyst employed [20,22,23]. The degradation of H₂O₂ to form HO^• radicals has a potential of +0.39 V [101] and, therefore, is favourable to occur on the CB of the semiconductor. In this way, the presence of H₂O₂ and photoactivated CN can generate HO^• radicals. These radicals are desired for several applications in environmental remediation, such as in water treatment, e.g., abatement of phenols [102], dyes [103], and antibiotics [104].

3. Enhancing Photocatalytic Activity of Carbon Nitride

In the following sections, metal-free strategies to modify CN will be overviewed and discussed to understand their influence on the efficiency and productivity of H₂O₂ photosynthesis. Several approaches can be found in the literature for improving the efficiency of this material, including thermal treatments, chemical substitutions with other carbon materials, or with specific organic molecules [105,106,107]. All these approaches are summarized in the following sections along with the correspondent experimental conditions of the photocatalytic tests and their ensuing results. The H₂O₂ productivity is depicted in terms of the highest amount produced after a respective irradiation time and moles of H₂O₂ per catalyst load and time, i.e., production rate (µmol g_cat⁻¹ h⁻¹).

Multiple articles have reported the preparation of CN using distinct precursors and their posterior application using different experimental conditions [76,107,108]. In

Table 1. Metal-free CN-based materials, experimental conditions, and respective photocatalytic results.

Modification on CN	Preparation Method	Experimental Conditions	Photocatalytic Results		Ref.
			H2O2 Generated (µmol)	Production Rate (µmol gcat−1 h−1)
None	Thermal polymerization of cyanamide under N2 atmosphere	9/1 (v/v) propan-2-ol/water (5 mL); 4 g L−1; 2000 W Xe lamp (λ > 420 nm); O2	60 µmol (24 h)	125	[76]
None	Thermal polymerization of cyanamide under N2 atmosphere	9/1 (v/v) benzyl alcohol/water matrix (5 mL); 4 g L−1; 2000 W Xe lamp (λ > 420 nm); O2	109 µmol (24 h)	227	[76]
None	Thermal polymerization of cyanamide under N2 atmosphere	9/1 (v/v) ethanol/water (5 mL); 4 g L−1; 2000 W Xe lamp (λ > 420 nm); O2	30 µmol (24 h)	63	[76]
None	Thermal polymerization of cyanamide under N2 atmosphere	9/1 (v/v) butan−1-ol/water (5 mL); 4 g L−1; 2000 W Xe lamp (λ > 420 nm); O2	18 µmol (24 h)	38	[76]
None	Thermal polymerization of cyanamide under N2 atmosphere	9/1 (v/v) propan−1-ol/water (5 mL); 4 g L−1; 2000 W Xe lamp (λ > 420 nm); O2	6.3 µmol (24 h)	13	[76]
None	Thermal polymerization of cyanamide under N2 atmosphere	9/1 (v/v) propan-2-ol/water (5 mL); 4 g L−1; sunlight; O2	120 µmol (9 h)	667	[76]
None	Thermal polymerization of cyanamide under N2 atmosphere	9/1 (v/v) propan-2-ol/water (5 mL); 4 g L−1; sunlight (λ > 420 nm); O2	70 µmol (9 h)	389	[76]
None	Thermal polymerization of melamine	9/1 (v/v) propan-2-ol/water (30 mL); 1.7 g L−1; 2000 W Xe lamp (λ > 420 nm); O2	148 µmol (6 h)	493	[107]
Adding surface defects	Silica-templated thermal polymerization of cyanamide under N2 atmosphere	9/1 (v/v) ethanol/water (5 mL); 4 g L−1; 2000 W Xe lamp (λ > 420 nm); O2	90 µmol (24 h)	188	[94]
Adding C vacancies	Thermal polymerization of melamine and further treatment under Ar atmosphere	water (100 mL); 1.0 g L−1; 300 W Xe lamp (λ > 420 nm); O2	9 µmol (1 h)	90	[105]
Adding N vacancies	Thermal polymerization of melamine and further treatment under H2 atmosphere	water (100 mL); 1.0 g L−1; 300 W Xe lamp (λ > 420 nm); O2	1.5 µmol (1 h)	15	[105]
Adding N vacancies	Thermal polymerization of dicyandiamide and photo-assisted post-treatment with hydrazine	20% (v) propan-2-ol/water (60 mL); 0.83 g L−1; solar simulator (λ > 420 nm); O2	12.1 µmol (2.5 h)	97	[109]
Adding N vacancies	Thermal polymerization of melamine and further calcinated with sodium borohydride	water (100 mL); 1.0 g L−1; 300 W Xe lamp (λ ≥ 420 nm); O2	30.0 µmol (1 h)	300	[110]
Adding N vacancies	Thermal polymerization of melamine and further calcinated with sodium borohydride	water (100 mL); 1.0 g L−1; 300 W Xe lamp (λ ≥ 400 nm); air	17.0 µmol (1 h)	170	[110]
Adding N vacancies	Thermal polymerization of melamine and further calcinated with sodium borohydride	water (100 mL); 1.0 g L−1; 300 W Xe lamp (λ ≥ 400 nm); N2	2.5 µmol (1 h)	25	[110]
Adding N vacancies	Thermal polymerization of melamine and H2 plasma treatment	50% (v) ethanol/water (200 mL); 1.0 g L−1; 250 W high-pressure sodium lamp (λ > 400 nm); O2	26000 µmol (12 h)	2167	[111]
C doping	Thermal polymerization of melamine and sonication with glucose	5/95 (v/v) propan-2-ol/water; 1.0 g L−1; 300 W Xe lamp; O2	38.1 µmol (4 h)	318	[112]
Carbon nanotubes combination	Thermal polymerization of dicyandiamide and ammonium chloride and mixed with carbon nanotubes	5/95 (v/v) formic acid/water (100 mL); 1.0 g L−1; 300 W Xe lamp (λ ≥ 400 nm); O2	48.7 µmol (1 h)	487	[106]
Carbon nanotubes combination	Thermal polymerization of dicyandiamide and ammonium chloride and mixed with carbon nanotubes	5/95 (v/v) methanol/water (100 mL); 1.0 g L−1; 300 W Xe lamp (λ ≥ 400 nm); O2	23.1 µmol (1 h)	231	[106]
Carbon nanotubes combination	Thermal polymerization of dicyandiamide and ammonium chloride and mixed with carbon nanotubes	water (100 mL); 1.0 g L−1; 300 W Xe lamp (λ ≥ 400 nm); O2	1.3 µmol (1 h)	13	[106]
AQ-COOH coupling	Thermal polymerization of melamine and sonication with anthraquinone (AQ)-2-carboxylic acid	1/9 (v/v) propan-2-ol/water; 0.5 g L−1; 150 W Xe lamp (λ > 400 nm); O2	no data	361	[108]
AQ-COOH coupling	Thermal polymerization of melamine and sonication with anthraquinone-2-carboxylic acid	1/9 (v/v) propan-2-ol/water; 0.5 g L−1; 150 W Xe lamp (λ > 400 nm); air	no data	270	[108]
AQ-NH2 coupling	Thermal polymerization of melamine and sonication with 2-aminoanthraquinone	1/9 (v/v) propan-2-ol/water; 0.5 g L−1; 150 W Xe lamp (λ > 400 nm); O2	no data	233	[108]
AQ-SO3- coupling	Thermal polymerization of melamine and sonication with sodium anthraquinone-2-sulfonate	1/9 (v/v) propan-2-ol/water; 0.5 g L−1; 150 W Xe lamp (λ > 400 nm); O2	no data	131	[108]
AQ-OH coupling	Thermal polymerization of melamine and sonication with 2-hydroxymethylanthraquinone	1/9 (v/v) propan-2-ol/water; 0.5 g L−1; 150 W Xe lamp (λ > 400 nm); O2	no data	86.9	[108]
AQ-COOH coupling	Thermal polymerization of melamine and sonication with anthraquinone-2-carboxylic acid	water; 0.5 g L−1; 150 W Xe lamp (λ > 400 nm); O2	no data	24	[108]
Benzene substitution	Thermal polymerization of urea with trimesic acid	1/9 (v/v) ethanol/water (30 mL); 0.5 g L−1; 300 W Xe lamp (λ > 420 nm); O2	275 µmol (3 h)	300	[114]
N doping	Thermal polymerization of melamine, sonication with tetracycline hydrochloride and further thermal exfoliation	3/7 (v/v) propan-2-ol/water (100 mL); 0.5 g L−1; solar simulator (λ > 420 nm); O2	14 µmol (1 h)	279	[115]
O doping	Thermal polymerization of dicyandiamide with nitric acid and hydrothermal post-treatment	water (50 mL); 1.0 g L−1; 250 W high-pressure sodium lamp (λ > 400 nm); O2	760 µmol (6 h)	633	[116]
Phosphate doping	Thermal polymerization of melamine and hydrothermal treatment with H3PO4	2.6 mM EDTA aqueous solution (200 mL); 1.0 g L−1; 250 W high-pressure sodium lamp (λ > 400 nm); O2	1080 µmol (6 h)	900	[117]
CN-C60	Thermal polymerization of melamine and C60	1/9 (v/v) propan-2-ol/water; 0.5 g L−1; 150 W Xe lamp (λ > 400 nm); O2	no data	63.2	[108]
CN-GO	Thermal polymerization of melamine and sonication with GO	1/9 (v/v) propan-2-ol/water; 0.5 g L−1; 150 W Xe lamp (λ > 400 nm); O2	no data	62.3	[108]
CN-rGO	Thermal polymerization of melamine and sonication with hydrazine-reduced GO	1/9 (v/v) propan-2-ol/water; 0.5 g L−1; 150 W Xe lamp (λ > 400 nm); O2	no data	74.3	[108]
CN-PDI	Thermal polymerization of melamine and pyromellitic dianhydride (PMDA)	water (30 mL); 1.7 g L−1; 2000 W Xe lamp (λ > 420 nm); O2	50.6 µmol (48 h)	21	[107]
CN-PDI	Thermal polymerization of melamine and pyromellitic dianhydride (PMDA)	9/1 (v/v) propan-2-ol/water (30 mL); 1.7 g L−1; 2000 W Xe lamp (λ > 420 nm); O2	210 µmol (6 h)	700	[107]
CN-BDI	Thermal polymerization of melamine and biphenyl tetracarboxylic dianhydride (BTCDA)	9/1 (v/v) propan-2-ol/water (30 mL); 3.3 g L−1; solar simulator (λ > 400-500 nm); O2	22.2 µmol (2 h)	111	[77]
CN-BDI	Thermal polymerization of melamine and biphenyl tetracarboxylic dianhydride (BTCDA)	water (30 mL); 1.7 g L−1; solar simulator (λ > 420 nm); O2	11.6 µmol (24 h)	10	[77]
CN-MTI	Thermal polymerization of melem and mellitic acid trianhydride (MTA)	water (30 mL); 1.7 g L−1; Xe lamp (λ > 420 nm); O2	27.5 µmol (24 h)	23	[78]
CN-PDI-rGO	Hydrothermal treatment of melem and GO and thermal polymerization with PMDA	water (30 mL); 1.7 g L−1; 2000 W Xe lamp (λ > 420 nm); O2	60 µmol (48 h)	25	[119]
CN-PDI-rGO	Hydrothermal treatment of melem and GO and thermal polymerization with PMDA	9/1 (v/v) propan-2-ol/water (30 mL); 1.7 g L−1; 2000 W Xe lamp (λ > 420 nm); O2	550 µmol (9 h)	1222	[119]
CN-PDI-BN	Sonication of melem and CN and thermal polymerization with PMDA	water (30 mL); 1.7 g L−1; 200 W Xe lamp (λ > 420 nm); O2	28 µmol (24 h)	23	[122]
CN-PDI-BN	Sonication of melem and CN and thermal polymerization with PMDA	9/1 (v/v) propan-2-ol/water (30 mL); 1.7 g L−1; 200 W Xe lamp (λ > 420 nm); O2	370 µmol (6 h)	1233	[122]
CN-PDI-rGO-BN	Sonication of melem, GO and CN and thermal polymerization with PMDA	water (30 mL); 1.7 g L−1; 200 W Xe lamp (λ > 420 nm); O2	37 µmol (24 h)	31	[122]
CN-PDI-rGO-BN	Sonication of melem, GO and CN and thermal polymerization with PMDA	9/1 (v/v) propan-2-ol/water (30 mL); 1.7 g L−1; 200 W Xe lamp (λ > 420 nm); O2	550 µmol (6 h)	1833	[122]
CN-PI	Thermal polymerization of melamine and cyanuric acid and reflux condensation reaction with perylene tetracarboxylic dianyhdride (PTCDA) and imidazole	water (30 mL); 1.7 g L−1; 300 W Xe lamp (λ > 420 nm); no data	120 µmol (2 h)	1200	[96]
CN-BP	Thermal polymerization of urea followed by sonication with N-methyl-2-pyrrolidone and BP	1/9 (v/v) propan-2-ol/water (30 mL); 1.7 g L−1; 300 W Xe lamp (λ > 420 nm); O2	540 µmol (1 h)	540	[124]
CN-BN	Hydrothermal treatment and thermal polymerization of thiourea and melamine and sonication with BN dots	1/9 (v/v) propan-2-ol/water (50 mL); 1.0 g L−1; 300 W Xe lamp (λ > 420 nm); O2	72.3 µmol (1 h)	72.3	[120]
CN-BN	Thermal polymerization of urea and BN nanosheets	1/9 (v/v) methanol/water (40 mL); 0.5 g L−1; 300 W Xe lamp (λ > 305 nm); O2	112 µmol (4 h)	1400	[121]

, the results in terms of H₂O₂ production using several neat CN materials are listed. Metal-free CN materials modified by several approaches are also shown in this table and discussed below.

Shiraishi et al. [76] reported the application of a metal-free CN material for H₂O₂ production using various alcohols for improving H₂O₂ selectivity. In this study, high H₂O₂ selectivity (>90%) was achieved by using aqueous solutions of aliphatic or aromatic alcohols, namely ethanol, propan-2-ol, butan-2-ol, and benzyl alcohol. Concerning the proton donor, benzyl alcohol and propan-2-ol yielded the higher amounts of H₂O₂. Then, testing the system using solar irradiation with or without a light filter (λ > 420 nm), higher selectivity for H₂O₂ formation is obtained when CN is activated by visible light rather than using the total spectrum range. This is due to ultraviolet light leading to the unwanted decomposition of H₂O₂.

Comparing the same matrix, light source, and gas but changing the precursor used in the catalyst synthesis, a much higher rate is achieved when using melamine [107] instead of cyanamide precursor [76].

The same group investigated the use of silica templates for changing the textural properties of CN [94]. The increase of surface defects leads to the formation of a large number of primary amines, which act as active sites for four-electron reduction of O₂. A decrease on the selectivity towards H₂O₂ formation was observed in this system with the highest production rate of 188 µmol g_cat⁻¹ h⁻¹ and 60% selectivity.

3.1. Surface Chemistry Modulation

Li et al. [105] reported that H₂O₂ production can be improved up to 14 times in the absence of an organic electron scavenger in the presence of carbon-vacancies (Cv). This study revealed that a post-treatment with argon led to the destruction of the crystallinity of CN and, therefore, produce defects, i.e., carbon vacancies (Figure 3). The Cv enhanced the trapping of the photogenerated electrons. Moreover, amino groups were formed and promoted electron transfer and changed the H₂O₂ production pathway from two-electron direct reduction of O₂ to sequential one-electron reduction of O₂ (Figure 3). In addition, it was also found that Cv decreased the band gap energy but did not interfere with the CB potential. This study showed that the chemisorption of O₂ on the catalyst was enhanced with this modification. The effect of nitrogen vacancies (Nv) was also assessed, with the H₂O₂ production being much lower than using materials with Cv. The highest H₂O₂ production rate obtained for CN-Cv and CN-Nv was 900 and 150 µmol g_cat⁻¹ h⁻¹, respectively, under visible light (λ > 420 nm) irradiation.

In another study, the effect of the presence of nitrogen defects on a CN catalyst has been discussed in terms of their ability for reducing electron-hole recombination and improving the contact between the reactant and active sites on the catalyst surface [109].

Thermal post-treatment of bulk CN drives the cleavage between tri-s-triazine moieties forming nitride vacancies and increasing C≡N groups on the matrix (Figure 4) [110]. This is accomplished by the incorporation of electron-deficient or π-conjugated monomers that falls for low bands energy potentials and inhibit the recombination. Moreover, the incorporation of strong electron acceptor groups can reduce the band gap energy and positively shift both CB and VB. The same authors investigated the impact of the saturated gas by testing the catalyst activity in the presence of O₂, air, and N₂ at the same conditions. It was observed that the H₂O₂ production rate was enhanced by saturating the suspensions with these gases, following the order O₂ > air > N₂. These results seems to indicate that most H₂O₂ production derives from O₂ reduction with a small contribution from water oxidation.

Finally, another approach was performed by a plasma treatment with a power input of high voltage under H₂ atmosphere with the main aim of the inclusion of N vacancies on the CN matrix [111]. These vacancies act as active sites for O₂ adsorption and also have the capability to promote electron transfer, thus accelerating the reduction step. Furthermore, this catalyst suppressed the decomposition of H₂O₂, which resulted in a very high production rate of 2167 µmol g_cat⁻¹ h⁻¹ in the presence of ethanol and pure O₂.

3.2. Functionalization

The most recent studies found in the scope of metal-free photocatalytic H₂O₂ production consisted on the combination of CN with carbon, via doping or bonding with carbon nanotubes (CNT). Using C-doped CN, a positive shift of the band potentials was found (Figure 5). This change on the VB accelerates water oxidation, and on the CB enhances oxygen reduction, improving H₂O₂ production owing to reduced kinetic barriers of the corresponding reactions. It was verified that H₂O₂ production is strongly dependent on the carbon content and the electronic structure of CN. The synthesized material with highest H₂O₂ production simultaneously presented the highest formation and lowest decomposition rate constants, yielding a maximum H₂O₂ rate of 365 µmol g_cat⁻¹ h⁻¹ in a 5% propan-2-ol solution with O₂ saturation [112]. This study strengthened the claim of the formation and decomposition of H₂O₂ being two competitive reactions, with the formation following a zero-order kinetics due to continuous O₂ saturation, and the decomposition following a first-order kinetics.

The covalent combination between CNT and CN (CN-CNT) promotes the sequential two-step pathway (Figure 2), using formic acid or methanol as electron donors. However, CN-CNT catalyses both water oxidation and O₂ reduction, forming H₂O₂ without the need of an electron donor. This was proved by the presence of benzoquinone, which depresses H₂O₂ production by inhibiting the sequential two-step O₂ reduction. This material reached a maximum production rate of 487 µmol g_cat⁻¹ h⁻¹ using a 5:95 formic acid:water solution in O₂-saturated conditions [106].

Another strategy to improve the CN photocatalytic activity is the anchoring of organic compounds, such as AQ. Moon et al. [113] showed that H₂O₂ production was dependent on the concentration of AQ and that, for higher AQ loads, there is a light block effect. When AQ is physisorbed on CN, the H₂O₂ production is improved. The authors explained the results based on the reutilization studies in which it was found that AQ remain at the CN surface, thus achieving a highly stable photocatalyst. H₂O₂ production was enhanced not only due to the higher H₂O₂ formation but also as a consequence of lower H₂O₂ decomposition. The authors also tested several AQ sources which lead to the functionalization with different groups. The material with COOH groups showed the highest formation and lowest decomposition rates, enabling continuous production over extended irradiation time in the presence of an electron donor. Furthermore, the apparent quantum yield profile resembles its absorption spectrum, meaning that efficient optical absorption and charge collection are achieved; however, with some useless recombination remains.

Benzene doping was tested by Kim et al. [114], achieving an H₂O₂ production rate of 300 µmol g_cat⁻¹ h⁻¹ that was obtained in O₂-saturated conditions and using a 10% ethanol aqueous solution. The increase of photoactivity compared to the bulk material was mainly ascribed to the structure distortion (Figure 6), which was promoted by the substitution of the N atoms in the matrix by benzene with a much higher molecule size. In addition, the presence of benzene leads to an easier charge transfer and hinders the recombination of electron/hole pairs.

Several authors have reported doping CN with nitrogen, oxygen and phosphate as an efficient technique for increasing the efficiency for H₂O₂ production [115,116,117]. In the case of N-doping, it was described that it decreases O₂ adsorption energy and enhances charge transfer [115]. Doping with oxygen promoted higher light absorption and efficient charge separation with a lower recombination rate [116]. The anchoring of phosphate on CN led to enhanced O₂ adsorption, which was ascribed as the main reason for the increased H₂O₂ productivity [117].

3.3. Construction of Heterostructures

The combination of carbon materials, like fullerene (C₆₀), graphene oxide (GO), and reduced graphene oxide (rGO), with CN have shown to promote a negative impact for H₂O₂ production owing to the higher affinity to one-electron O₂ reduction route [108]. Even with O₂ saturation and in the presence of a propan-2-ol solution (regarded as one of the best proton donors for this process [76]), the yield of H₂O₂ achieved with these hybrid materials was relatively low.

Aromatic diimides are n-type semiconductors with high electron mobility and stability. Therefore, their incorporation on the CN structure can lead to a positive shift on both VB and CB bands, owing to the high electron affinity [107].

Reports have been shown that pyromellitic diimide (PDI) units increase the rates of H₂O₂ formation as the valence band shifts promoting water oxidation to O₂, facilitating H₂O₂ production (Figure 7). Shiraishi et al. [107] reported the use of a CN-PDI material using water and propan-2-ol as solvents. With this study, the authors obtained a much higher rate (573 µmol g_cat⁻¹ h⁻¹) for H₂O₂ production when the alcohol was present, due its capacity of acting as a strong proton donor [107]. In another work, the CN material was modified with biphenyl diimide (BDI) [77], and the effect of polymerization temperature was evaluated. The authors found an optimal temperature of 653 K, which yielded 6.8 µmol of H₂O₂ after 24 h of irradiation corresponding to a rate of 5.7 µmol g_cat⁻¹ h⁻¹. By increasing the polymerization temperature, the authors found a significant catalyst weight loss, followed by a decrease on the photocatalytic activity of the resulting materials. Moreover, the amount of BDI on the CN was studied, and among all the resulting materials, the best photocatalytic activity was achieved with a molar ratio of 50% BDI in the catalyst, yielding 41 µmol after 48 h of irradiation and a 9.7 µmol g_cat⁻¹ h⁻¹ rate. According to the calculated apparent quantum yield, BDI doping is more effective than PDI. BDI doping leads to a positive shift on the VB and CB, enabling water oxidation and promoting H₂O₂ formation (Figure 7). Additionally, ab initio calculations suggest that there is a significant spatial charge separation (h⁺ in BDI and e^‒ on melem; on PDI both h⁺ and e^‒ are on melem) which hinders their recombination improving H₂O₂ formation.

The photoactivity of catalysts, such as CN, which are π-conjugated semiconductors, depends on the density and mobility of the photoformed charge carriers [118]. The same authors that used PDI also reported the combination of CN with mellitic triimide (MTI) [78]. The incorporation of MTI units and the subsequent stacking of melem layers can lead to efficient inter and intralayer charge transfer. Therefore, the CN-MTI catalyst showed improvements on the conductivity and charge transport, as well as a higher photoactivity, compared with the pristine CN towards H₂O₂ production.

PDI-, BDI-, and MTI-modified CN has also been combined with reduced graphene oxide (rGO). In general, rGO has the ability to trap photogenerated electrons from the CB of the CN-PDI material, acting as active sites for the two-electron reduction of O₂. On the other hand, CN-PDI-rGO photocatalyst promoted slight decomposition of H₂O₂. However, using a physical mixture of CN-PDI and rGO, no significant effect was observed, which can be ascribed to the low interaction between CN-PDI and rGO materials [119].

Structures of CN coupled with boron nitride (BN) were prepared to further enhance electron transfer [120,121]. The composite made with BN quantum dots favours the acceleration of charge transfer and the decrease of recombination [120]. The material with BN nanosheets resulted in an elevated production (1400 µmol g_cat⁻¹ h⁻¹) since the authors managed to decrease H₂O₂ decomposition while maintaining very high formation rates [121]. The addition of BN seems to promote the separation of holes hindering recombination. Kofuji et al. have explored the combination of a CN-PDI material with BN [122]. The photogenerated holes in the VB on the CN/PDI moiety move to the VB of BN leading to their entrapment, which enhances charge separation and promotes H₂O₂ production. Moreover, addition of rGO further inhibits the recombination of electrons and holes. The material consisting of CN-PDI combined with both BN and rGO resulted in the highest H₂O₂ production rates due to promoting both water oxidation and oxygen reduction (Figure 8), while increasing the charge transfer. Moreover, these studies also supported the impact of the presence of an alcohol as proton donor. The use of propan-2-ol as solvent increased the H₂O₂ production rate by a factor of 50, comparing with water.

Wang et al. [96] investigated the use of perylene imide (PI), to create a Z-scheme heterojunction with CN nanosheets for enhancing the oxidative power of photogenerated holes [123]. The CN-PI structures led to higher H₂O₂ production; however, for excessive PI amounts, light absorption by the nanosheets was compromised. PI seems to favour the separation of charge carriers and leads to faster interfacial charge transfer (Figure 9). The addition of PI changes the production of H₂O₂ from a direct two-electron oxygen reduction to a two-channel pathway (Equations (3) and (4)). Moreover, the presence of PI inhibits the subsequent decomposition of H₂O₂ [96].

The combination of CN and black phosphorus (BP) was reported by Zheng et al. [124]. This composite allowed for a higher H₂O₂ productivity than the lone CN, owing to BP being highly reactive to oxygen.

To compare the different structures and experimental conditions by an unbiased parameter, the apparent quantum yield (AQY) and the solar-to-chemical conversion (SCC) of H₂O₂ production are typically applied. These coefficients were calculated by the respective authors and were collected in

Table 2. Solar-to-chemical conversion (SCC) and apparent quantum yield (AQY) efficiencies for some studies. Nv = nitrogen vacancies.

Material	AQY/%	SCC/%	Ref.
CN-PDI	2.6	0.10	[77]
CN-PI	3.2	no data	[96]
CN-Nv	4.3	0.26	[110]
CN-BDI	4.6	0.13	[77]
CN-PDI-BN	4.8	0.19	[122]
CN-MTI	6.0	0.18	[78]
CN-PDI-rGO	6.1	0.20	[119]
CN-PDI-BN-rGO	7.3	0.27	[122]

. The AQY gives the information about the formation of H₂O₂ relative to the number of incident photons and relates the stoichiometric amount of H₂O₂ formed with a specific light intensity and emission wavelength [125]. The SCC is related to the performance of a catalyst to yield H₂O₂, relating the free energy of H₂O₂ formation and the total incident energy [125]. Therefore, higher AQY and SCC values can demonstrate the photoactivity efficiency and proneness to selectively generate H₂O₂. For instance, Kofuji et al. [122] reported the hybrid CN-PDI-BN-rGO, which yielded the highest reported values of AQY and SCC compared to other works, as well as the highest rates for H₂O₂ production.

4. Photocatalytic Application with In Situ H₂O₂ Generation

H₂O₂ is used in numerous applications and, due to its oxidizing power, is commonly added in many systems, namely in the abatement of organic contaminants and in fine chemistry, to improve conversion and accelerate the reaction. For instance, the application of sonochemistry has been reported for the simultaneous in situ generation of H₂O₂ and degradation of organics [39,40]. The coupling of ultrasounds and photocatalysts (sonophotocatalysis) has been employed for the degradation of phenol and 4-chlorophenol with titania-based materials, where the presence of in situ evolved H₂O₂ markedly improved the degradation process [126]. Particularly, in photocatalysis, the presence of H₂O₂ is reported to increase the mineralization or removal rates of several organics [12,13,14,15,16,17,18,19] and enhance the selectivity of photochemical synthesis [6]. Therefore, the in situ generation of H₂O₂ in applications in which it is used as reactant is not only an advantage in terms of process design but also in terms of cost reduction, as described in the next sections.

4.1. Pollutant Degradation

The photodegradation of organic molecules and removal of biological contaminants using metal-free CN materials has been extensively investigated. Some works report the addition of H₂O₂, resulting in the enhancement of the degradation process [22,127,128,129,130,131]. In most cases, the CN photocatalyst acts as a Fenton-mimic since it turns H₂O₂ into HO^• which attack the pollutants, leading to increased mineralization [20,132,133]. This interesting duality of CN is worth of being explored, and several authors have already verified the in situ evolution of H₂O₂ to enhance the oxidative abatement of contaminants. As previously discussed, metal-free CN, under visible-light, leads to the formation of H₂O₂, and is used for the removal of several organic compounds by photocatalysis. Many studies report the formation of H₂O₂ simultaneously to the degradation of the contaminant molecules [20,23,120,134,135,136,137,138,139,140,141]. The degradation is accompanied by the formation of reactive oxygen species (ROS), H₂O₂ being detected during the photocatalytic experiments. Two studies recently followed the time-dependent H₂O₂ concentration along the photocatalytic degradation reaction of phenol [95,142]. Zhang et al. detected H₂O₂ using CN nanosheets under visible light irradiation with the production being markedly dependent in the structure of CN, yielding larger amounts for more exfoliated materials which promote selective two-electron O₂ reduction [95]. Additionally, the formation of highly reactive oxygen species promoted phenol degradation, such as HO^• originated from H₂O₂ decomposition. The presence of O₂ in an aqueous solution can lead to the formation of H₂O₂ and other reactive oxygen species which aid the oxidation of organic molecules. H₂O₂ is a fast reacting molecule; however, its stabilization and production is very dependent of the medium [1]. For instance, the pH dependency of H₂O₂ is known and, in more acidic media, H₂O₂ is more stable than in an alkaline environment [143]. However, CN has been proven to act efficiently in all pH range since this material presents amphoteric properties [144,145]. Relative to H₂O₂ formation, it is observed that many other factors have to be taken into consideration, namely the content in dissolved oxygen, pollutant initial concentration, catalyst load and light source. The work developed by Svoboda et al. [142] somewhat showed the impact of these parameters on H₂O₂ production since quenching experiments lead to changes on the degradation of phenol (sacrificial agent for H₂O₂ formation). However, the use of scavenging species, to study H₂O₂ formation and decomposition, may suffer interference owing to their degradation by different reactive oxy-species. This can hinder or facilitate the generation of H₂O₂ leading to ambiguous results. Svoboda et al. [142] investigated the degradation of phenol using CN nanosheets obtained from the thermal post-treatment of melamine-derived bulk CN. This allowed for an increase of the surface area of the material, leading to much higher photoactivity. In this work, an impressive H₂O₂ production rate of 3300 µmol g_cat⁻¹ h⁻¹ was reported, using visible-LEDs with a maximum emission wavelength of λ = 416 nm, continuous air purging and a phenol initial concentration of 20 mg L⁻¹.

Furthermore, using differently-substituted phenolic compounds and exfoliated CN it is possible to obtain high production rates between 633 and 3103 µmol g_cat⁻¹ h⁻¹ [146], using visible-light emitting diodes and continuous air saturation. This study contemplates the relation between O₂ concentration and H₂O₂ production, establishing that as the dissolved oxygen content increases up to 21% there is an increase in H₂O₂ formation. Moreover, the exfoliated CN in this work was tested in a propan-2-ol aqueous solution, achieving a H₂O₂ production rate of 19,200 µmol g_cat⁻¹ h⁻¹.

To date, the combination of oxidation driven by in situ evolved H₂O₂ in CN photocatalysts with other AOPs for water treatment has been reported for ozonation and persulfate activation [147,148]. These two studies investigate the in situ evolution of H₂O₂ during the reaction and discuss the synergic effect that promoted the removal of the contaminant molecules. However, it is interesting to point out that H₂O₂ is fundamental in the Fenton reaction. In addition, in a homogeneous Fenton system, the treated water matrix remains with dissolved iron which has to be separated. Iron-doped CN photocatalysts result on the combination of traditional Fenton and CN photocatalysis which enhances oxidation by promoting a two-channel pathway of H₂O₂ reduction to generate HO^•. In this way, many authors have synthesized iron-doped CN to try combat the disadvantage of dissolved iron in the mineralized waters [149,150,151,152].

CN has been applied as metal-free photocatalyst for the degradation of several organic pollutants, but the generation of H₂O₂ was not monitored in the publications [153,154]; thus, they are not considered in this review.

4.2. Fine Chemistry

In the case of selective organic synthesis, there are reports of H₂O₂ addition while employing metal-free CN photocatalysts [155,156]. The presence of H₂O₂ improves the oxidation of selected molecules, such as the conversion of toluene to benzaldehyde [155] or of cyclic olefins to the respective epoxides [156]. However, there have been reports where H₂O₂ formation was observed in the presence of both visible light and a CN catalyst. Lopes et al. [157] achieved very high production rates of ca. 5000 µmol g_cat⁻¹ h⁻¹ using nanosheets of CN in an anisyl alcohol solution. In that work, H₂O₂ was formed as a by-product of the oxidation of aromatic alcohols into the corresponding aldehydes. The simultaneous formation of H₂O₂ is a further advantage to the selective organic synthesis owing to the oxidizing power of H₂O₂, such as Zhang et al. [158] describes with an oxygen-enriched CN material employed for the transformation of amines into imines.

5. Conclusions and Future Prospects

This review article summarizes the state-of-the-art on modified metal-free carbon nitride photocatalysts for the selective evolution of H₂O₂, a high-value and multi-faceted chemical. Conventional processes for H₂O₂ synthesis are generally characterized by high energy consumption and waste generation. The latest studies, employing sustainable metal-free carbon nitride materials and clean aqueous matrices as solvents, show an emergent photocatalytic technology for H₂O₂ generation, including the smart tailoring of these materials for optimal conversion. Moreover, the ambivalence of the photocatalytic process, with simultaneous production and direct application of H₂O₂, has already been explored for pollutant degradation and fine chemical synthesis. The referred studies provide a favourable starting point to achieve sustainability in the industry of H₂O₂ production, albeit more research has to be performed to develop the necessary scale up operation, productivity enhancement, and overall process optimization.