Heterostructures based on g-C₃N₄ for the photocatalytic CO₂ reduction

A primary task of rational management and protection of the environment is to reduce the concentration and emissions of greenhouse gases. The development of methods for reducing concentrations of greenhouse gases is the subject of many studies[1-3], with much attention being paid to utilization of carbon dioxide, the major component of greenhouse gases[4-8]. Today, there are several industrial processes for converting СО₂ to valuable products such as urea, salicylic acid, ethylene carbonate and methanol. Due to the high thermodynamic stability of the CO₂ molecule, traditional processes of CO₂ conversion to products are carried out at high temperature and high pressure, e.g., synthesis of urea from СО₂ and NH₃ takes plates at a temperature of 185 °С and a pressure of 150 bar[9-11], which substantially increases the energy expenditure and decreases the economic attractiveness of these processes, despite the readily available feedstock.

For this reason, of particular interest are the ways to decrease the energy consumption of CO₂ conversion processes. One of the approaches to address this task is the photocatalytic reduction of CO₂, which can occur under ambient conditions[7, 12]. This process is based on the use of renewable resources, solar light and water, which makes photocatalytic reduction a promising method for CO₂ utilization, but its industrial implementation is primarily limited by the lack of efficient photocatalysts. Nevertheless, photocatalytic reduction of CO₂ complies with the principles of sustainable development and, in the future, it may become the most facile and inexpensive way for decreasing the CO₂ concentration in the atmosphere[12, 13]. Moreover, this process gives organic compounds such as CH₄, CH₃OH and HCOOH [equations (1),(2),(3),(4), (5)], standard electrode potentials vs. NHE at pH = 7 are given], which can be used as synthetic fuel and in chemical industy[12, 14-17].

In recent years, the attention of researchers has been attracted to a new polymer semiconductor, graphitic carbon nitride g-C₃N₄ (Figure 1), materials based on which can be used in various photocatalytic reactions[18, 19]. It is found that g-C₃N₄ has a planar structure based on heptazine or triazine units[20], in which carbon and nitrogen atoms are sp²-hybridized. This promising semiconductor has high chemical and thermal stability; in addition, g-C₃N₄ can be obtained from readily available precursors by simple methods[21, 22]. As compared with many semiconductors traditionally used in photocatalysis, for example TiO₂, g-C₃N₄ has a rather narrow band gap (2.7 eV); in combination with the highly negative position of the conduction band (–1.3 V vs. NHE), this results in light absorption over a broad range of wavelengths and a high reduction potential of photogenerated electrons[23, 24]. In addition, g-C₃N₄ has a number of properties inherent in two-dimensional materials such as high mobility of charge carriers, high surface area to volume ratio and the presence of quantum size effect[25, 26]. The drawbacks of g-C₃N₄ include fast recombination of photoinduced electron – hole pairs[27] and low adsorption ability, which hampers any heterogeneous catalytic process[28, 29]. A popular and efficient approach to overcome these difficulties is to form heterostructures with other semiconductors or materials, which would enhance the efficiency of charge carrier separation and the adsorption properties.

Figure 1

Number of publications dealing with \( \text{g-C}_{3}\text{N}_{4} \)-based photocatalysts in recent years according to the Scopus database.

It is known that photocatalysts are activated upon absorption of light quanta with energy equal to or exceeding the size of the band gap energy of the semiconductor. As this takes place, the electron migrates from the valence band to the conduction band, and an electron vacancy (hole) appears in the conduction band. Then the photogenerated electrons and holes can either migrate to the photocatalyst surface and be involved in redox reactions with adsorbed reagents or recombine in the photocatalyst bulk to release heat. The recombination of electron – hole pairs sharply decreases the efficiency of the photocatalytic reaction; therefore, an important task of photocatalysis is separation of the photogenerated charge carriers to increase their lifetime.

On contact of two semiconductors, the difference between the Fermi levels induces the electron transfer, giving rise to a built-in electric field, which, in turn, promotes the transfer of photoinduced charge carriers[30-32]. The charge carrier separation in composite structures can proceed by different mechanisms. The classical schemes of charge carrier transfer between two semiconductors include type I and II heterojunctions. Type I heterojunction appears when the conduction band (CB) of one semiconductor (SC 2 in Figure 2а) is located lower, while its valence band (VB) is located higher in energy than these bands of another semiconductor (SC 1 in Figure 2а). In such a structure, electrons and holes would migrate from the wide-band-gap semiconductor to the narrow-band-gap one and be accumulated in one semiconductor and subsequently recombine. In type II heterojunction, both the valence and conduction bands of one semiconductor are higher in energy than those of another semiconductor, and the transfer of electrons from a semiconductor with higher conduction band level (SC 1 in Figure 2b) to a semiconductor with lower conduction band level (SC 2 in Figure 2b) is formed, while holes migrate from lower valence band in SC 2 to higher valence band in SC 1 (Figure 2b). In this case, electrons and holes are accumulated in different semiconductors, and their lifetime is thus increased.

Figure 2

Schematic image of the electron transfer for heterojunctions of type I (a) and type II (b), direct S-scheme (c), direct Z-scheme (b) and indirect Z-scheme (c).

As a rule, charge carrier separation mechanisms in composite photocatalysts composed of g-C₃N₄ and another semiconductor are based on type II heterojunction or direct Z-scheme. Later, S-scheme heterojunction was proposed[33]. The direct Z-scheme transfer mechanism occurs between two semiconductors the energy structure of which is similar to that described above for type II heterojunction. However, in this case, an electron with a low reduction potential (in SC 2) recombines with a hole that has a low oxidation potential (in SC 1); in comparison with type II heterojunction, this gives rise to charge carriers with a higher redox potential. The separation of charge carriers between two semiconductors according to the S-scheme is similar to the direct Z-scheme; however, the S-scheme takes into account the semiconductor band bending caused by alignment of their Fermi levels and generation of the built-in electric field in the photocatalyst[34]. As a rule, Z-scheme is implemented by two n-type and p-type semiconductors, while S-scheme is implemented by n-type semiconductors, considered as an oxidation photocatalyst and reduction photocatalyst (see Figure 2)[3, 35, 36]. Both Z-scheme and S-scheme can be either direct (Figure 2d) or indirect (Figure 2e); in the latter case, compounds with high electrical conductivity, e.g., metals or some carbon materials such as graphene, serve as mediators[37]. It is important that implementation of either Z- or S-scheme in a composite photocatalyst not only increases the efficiency of separation of the electron – hole pairs, but also maximizes the redox capacity of charge carriers.

One more popular method for increasing the lifetime of photogenerated charge carriers is deposition of metals on the semiconductor surface. The Schottky junction is formed at the metal-semiconductor interface, and electrons from the conduction band of the semiconductor move to Fermi level of metal, which leads to alignment of the metal and semiconductor Fermi levels and bending of semiconductor bands near the interface.

Figure 3

Band structure of some reduction and oxidation photocatalysts[38].

Apart from the lifetime of photogenerated charge carriers, another important characteristic of semiconductor photocatalysts is the size of the optical band gap (below referred to as band gap), which determines the minimum energy of light quantum necessary for generation of electron – hole pairs (Figure 3)[38]. As a rule, the band gap energy is derived from the diffuse reflectance spectroscopy by the method proposed by Jan Tauc[39]. First, the absorption coefficient \( F(R) \) is found using the Kubelka – Munk equation:

where \( R \) is the reflection factor of the sample determined from the diffuse reflectance spectroscopy data.

Using the Tauc equation, the dependence of the absorption coefficient \( F(R) \) on the photon energy is found:

where \( h \) is the Planck constant, \( \nu \) is the photon frequency, \( E_g \) is the band gap energy, \( B \) is a constant, \( \gamma \) is the factor equal to 1/2 for direct junctions and to 2 for indirect junctions.

The intersection of the tangent to the Tauc plot with the abscissa gives the band gap energy of the semiconductor. It is noteworthy that this method is suitable for determining the band gap energy of unmodified semiconductors. Determining the band gap energy for composite materials, especially in the case of different types of electron transfer (direct/indirect) in semiconductors gives, most often, incorrect results. Nevertheless, in many publications, analysis of band gap changes based on diffuse reflectance spectroscopy is often performed for composite materials. In this case, the spectrum of the photocatalyst is usually a linear combination of the spectra of independet components, except the cases where quantum size effects or localized surface plasmon resonance influence the optical properties of materials. In some cases, the band gap of modified semiconductors can be estimated by drawing the tangent to the region called Urbach tail in the diffuse reflectance spectra. Then the intersection of this tangent with the tangent to the Tauc plot gives a correct band gap energy for a single component of a photocatalyst[40].

It is known that graphitic carbon nitride can behave as either p-type or n-type semiconductor depending on the external effects such as the applied voltage and redox potential of a reagent[41-45]. For example, g-C₃N₄ behaves as a p-type semiconductor when a negative voltage is applied and as an n-type semiconductor when the voltage is positive; therefore, it can be considered as an amphoteric semiconductor[42, 46, 47]. The amphoteric semiconductor properties also depend on the g-C₃N₄ bulk structure and surface functionalization[48]. Hence, the charge carrier transfer mechanism in composite photocatalysts composed of g-C₃N₄ and a semiconductor with a high potential of photogenerated holes can be considered as a Z-scheme or S-scheme, as described below.

Traditionally, g-C₃N₄ is obtained by heat treatment of various nitrogen-containing precursors such as melamine, dicyandiamide, and urea[49-51], but, most often, this method does not provide a material with a high specific surface area. In recent years, numerous new approaches to the synthesis of g-C₃N₄ have been proposed providing the possibility of varying the particle size, pore volume and specific surface area of the resulting material[52]. The methods of synthesis of g-C₃N₄ are comprehensively addressed in the literature[22, 53-57]; therefore, they are not discussed here.

This review is focused on g-C₃N₄ heterostructures with other materials used for photocatalytic CO₂ reduction. It is noteworthy that there are a few published reviews dealing with modification of g-C₃N₄ and fabrication of heterostructures for the photocatalytic reduction of CO₂[58-64] considering photocatalysts in the light of charge carrier transfer mechanisms or the effect of methods of synthesis on the photocatalytic properties. There are also reviews on using g-C₃N₄-based systems for other photocatalytic reactions (hydrogen evolution, decomposition of dyes, water treatment, etc.)[54, 65, 66]. A widely used photocatalyst is g-C₃N₄ with platinum deposited on its surface, which is obtained by photoreduction of H₂PtCl₆ in a solution of an electron donor, most often, triethanolamine or by reduction with a solution of NaBH₄[67, 68]. Modification of semiconductors with transition metals to enhance the photocatalytic activity has also been studied in detail and presented in the literature[69-71].

This review is devoted to the most recent advances in the field of g-C₃N₄-based heterostructures for photocatalytic CO₂ reduction. The attention is focused on the materials used and the most likely mechanism of charge carrier transfer between the semiconductors. In addition, a distinctive feature of this review is comparison of the photocatalyst activity in terms of the overall rate of consumption of photogenerated electrons, which makes it possible to compare the activities of photocatalysts with different selectivities to CO₂ reduction products. The review addresses the most widely used and promising systems containing metal oxides and sulfides, two-dimensional metal carbides (MXenes) and carbon materials. Since in some cases the literature provides information about different possible heterojunctions for the same type of heterostructure, the role of the type of charge carrier separation mechanism is also analyzed in each Section.

Titanium dioxide is one of the most popular photocatalysts used both for CO₂ reduction and for other photocatalytic processes, which is due to the low cost, stability and low toxicity of TiO₂[72-74]. However, the use of TiO₂-based photocatalysts is limited by the large band gap (~3.2 eV), since only UV photons have sufficient energy to induce the photoexcitation in TiO₂[12, 75]. As a consequence, TiO₂ has low activity under the sunlight, in which the fraction of ultraviolet radiation is much smaller than that of the visible radiation. To increase the response of TiO₂-based photocatalysts to visible light, TiO₂ is modified with narrow-band-gap semiconductors, for example, g-C₃N₄[76].

Titanium dioxide is known to have three stable crystalline phases — anatase, rutile, and brookite (Figure 4)[21, 77-79]. As a rule, the first two of them are used in photocatalytic studies; anatase is considered to be more active than rutile, because of the higher reduction potential of electrons and the ability to form hole traps[80, 81], which is attributable to different predominant orientations of the crystallite surfaces in rutile and anatase. In the case of anatase, the crystallite surface is enriched with {101} and {001} faces, while in the case of rutile, {110}, {100} and {101} faces predominate[82]. It is known that the {001} face is more reactive in photocatalytic reactions[83, 84]. Meanwhile, rutile has a narrower band gap than anatase and, hence, it can be used for reactions under irradiation at longer wavelengths[85]. The conduction band minimum in TiO₂ is approximately –0.2 V (vs. NHE at pH = 7)[21], which is insufficient for CO₂ reduction, since the formation potential of most products is more negative [see Eqns (1) – (5)]. However, the conduction band minimum in g-C₃N₄ is approximately –1.20 V (vs NHE at pH = 7), which is sufficient for the formation of various organic compounds from CO₂[21].

Figure 4

Crystal structure of \( \text{TiO}_2 \) phases: (a) anatase, (b) rutile, (c) brookite[79]. Published with permission from the Royal Society of Chemistry.

In the formation of g-C₃N₄-based composite photocatalysts, the particles of other components are usually distributed over the g-C3N4 surface, because it has a layered structure where the longitudinal particle size substantially exceeds the transverse size. These composite structures are commonly designated by X/g-C₃N₄ (X is a component of a g-C₃N₄-based photocatalyst), e.g., TiO₂/g-C₃N₄. This notation is used in this review, although authors of original publications may use other designations for heterostructures.

A fairly widespread method for the synthesis of TiO₂/g-C₃N₄ composites is physical mixing of components, according to which the components (TiO₂ and g-C₃N₄) are synthesized separately[19, 86]. TiO₂ is obtained, most often, by hydrolysis, hydrothermal or solvothermal treatment of titanium-containing precursors such as titanium alkoxides[19, 73].

Mehregan et al.[87] proposed a method for the fabrication of TiO₂/g-C₃N₄ composite by hydrothermal treatment [hereinafter, slash (/) is used in the designations of heterostructures, while a hyphen (–) designates semiconductors doped with metal or non-metal atoms]. Titanium dioxide was prepared by the sol – gel method using titanium tetrabutoxide as the precursor, while g-C₃N₄ was obtained by melamine calcination followed by exfoliation to give a layered structure (Figure 5). A suspension consisting of weighed portions of TiO₂ and g-C₃N₄ obtained in this way (in 2 : 1 w/w ratio of TiO₂ to g-C₃N₄) and aqueous ethanol was sonicated and then placed in an autoclave and kept at 120 °C for 3 h. The CO₂ reduction was carried out in the gas phase in the presence of water vapour on exposure to visible light. The authors studied the effect of light intensity on the reaction rate and showed a pronounced increase in the photocatalyst activity with as the light power density increased from 20 to 80 mW cm^–2. Thus, the light intensity of 80 mW cm^–2 provided the highest product formation rates, 33 and 1.4 mmol g^–1 h^–1 for CH₄ and CH₃OH, respectively. In some other publications, it is also shown that increase in the light power density increases the rates of electron and hole generation, which results in a higher photocatalyst activity[88, 89].

Figure 5

Schematic diagram of the approach to the fabrication of \( \text{TiO}_{2}/ \text{g-C}_{3}\text{N}_{4} \) heterostructures.

It should be emphasized that Evonik P25, a commercial TiO₂ powder, is used most often for the synthesis of various TiO₂-based photocatalysts[12]. This TiO₂ powder is composed of 80% anatase and 20% rutile with an average particle size of 25 nm and has a specific surface area of 40 – 60 m² g^–1[90]. Wang et al.[91] synthesized TiO₂/g-C₃N₄ composite photocatalysts with various component ratios by ball milling of TiO₂ (anatase) with g-C₃N₄ pretreated with a solution of HNO₃ to form a layered structure (Figure 6). Then the mixture was calcined at 400 °С for 1 h to form the composite material with TiO₂ particles deposited on the g-C₃N₄ surface. The authors demonstrated a change in the positions of the valence and conduction bands depending on the component ratio and increase in the band gap with increasing TiO₂ content. This study describes the liquid-phase СО₂ reduction in which the suspension with a photocatalyst in aqueous NaOH and triethanolamine solution was placed in a reactor, and then purged with СО₂ flow. The highest activity after 4 h of UV irradiation (8 W lamp) was found for the photocatalyst with TiO₂ to g-C₃N₄ ratio of 1 : 2. The rates of СО and СН₄ formation were 14 and 18 mmol g^–1 h^–1, respectively, which was twice as high as that for pristine g-C₃N₄.

Figure 6

Schematic diagram of the synthesis of the \( \text{TiO}_{2}/ \text{g-C}_{3}\text{N}_{4} \) photocatalyst[91].

Truc et al.[92] synthesized composite photocatalysts based on g-C₃N₄ and niobium-doped TiO₂ in which the direct Z-scheme heterojunction was formed (Figure 7)[92]. The introduction of Nb into TiO₂ structure leads to the formation of Ti³⁺ ions and the appearance of an additional energy level near the TiO₂ conduction band, thus reducing the band gap size from 3.2 to 2.9 eV. The change in the band gap by introducing metal and non-metal ions into a parent substance is a common method used to shift the absorption edge of photocatalysts[93-97]. The Nb – TiO₂/g-C₃N₄ composite photocatalysts were obtained by calcination of a mixture of Nb – TiO₂ with melamine. In a series of experiments on the gas-phase photocatalytic CO₂ reduction on exposure to light from two 30-W white lamps, the highest activity was detected for the photocatalyst with 1 : 1 ratio of (Nb – TiO₂) to g-C₃N₄. Apart from CO and CH₄, the reaction products were found to contain formic acid and oxygen resulting from water oxidation. The formation rates of CO, CH₄, HCOOH and O₂ were 420, 560, 700 and 1700 mmol g^–1 h^–1, respectively. The results obtained in this work demonstrate the high potential of the integrated approach to the synthesis of photocatalysts, including both doping and the formation of heterojunctions in composite materials.

Figure 7

Band structure of single materials \( \text{TiO}_2 \) (a) and \( \text{g-C}_{3}\text{N}_{4} \) (b) and schematic image of the processes that take place in doped \( \text{Nb-TiO}_{2} \) (c) and composite \( \text{Nb-TiO}_{2}/ \text{g-C}_{3}\text{N}_{4} \) (d) photocatalysts[92].

To compare the activity of photocatalysts with different selectivities to reaction products, it was proposed to use the total number of electrons absorbed in the photocatalytic process per unit time and per unit mass of the photocatalyst, We, which is calculated by the equation[98]:

where W_CO, W_CH4, W_CH3OH , and W_HCOOH are the rates of formation of the products per unit mass of the photocatalyst (as a rule, mmol g^–1 h^–1); coefficients are determined by the number of electrons needed for the formation of this product. Methane makes the greatest contribution to W_e, because its formation requires more electrons than the formation of other compounds [see Eqns (1) – (5)]. Equation (8) or similar relations are convenient for calculating the apparent quantum efficiency (AQE) [Equation (9)] and are widely used in the studies of the photocatalytic CO₂ reduction, although the terminology may differ[7, 98-101].

where W_e^* is the product formation rate with allowance for the electron balance (usually μmol h^–1), N_ph is the photon flux from the light source calculated from the data on the irradiation power and spectrum (mmol h^–1).

Unfortunately, AQE values for the photocatalytic CO₂ reduction (usually not exceeding 1%) are rarely reported in the literature; therefore, comparison of the photocatalyst activity will be based on calculation of the rate W_e[12]. However, it is noteworthy that AQE is an important characteristic of photocatalytic systems, as it reflects the photon utilization efficiency, while the rate W_e depends, among other factors, on the light irradiation power. The only outstanding AQE value was repotrted by Zhang et al.[102] A composite photocatalyst based on g-C₃N₄ and Ag-doped TiO₂ has an AQE of 2.4% in the liquid-phase CO₂ reduction. It is shown that the addition of Ag promotes implementation of S-scheme heterojunctions and electron transfer from TiO₂ to g-C₃N₄. In addition, high efficiency of CO₂ reduction is attained owing to high specific surface area and defects in the g-C₃N₄ structure.

The data on TiO₂/g-C₃N₄ composite systems that were reported in recent years are summarized in Table 1, which also indicates the reaction conditions and reagents added to the reactor. To carry out the photocatalytic reaction, a flow of ultrapure СО₂ is usually purged through the reactor; however, the СО₂ generation directly in the reactor, e.g., by the reaction of NaHCO₃ and H₂SO₄, can also be used. Most often, the source of protons in photocatalytic CO₂ reduction is water, but CH₄ can also serve for this purpose. Generally, TiO₂ is one of the most popular semiconductors for the synthesis of photocatalysts; therefore, there are numerous heterostructures based on TiO₂. A combination of TiO₂ with g-C₃N₄ in the TiO₂/g-C₃N₄ composite photocatalysts gives W_e values reaching 270 mmol g^–1 h^–1 under visible light irradiation[87], while modification of the TiO₂/g-C₃N₄ composite leads to even a more pronounced increase in the activity. As can be seen from Table 1, high W_e value was attained for composite system comprising Z-scheme heterojunction and synthesized from TiO₂ doped with metal ions; therefore, this method appears to be most efficient.

Table 1

\[ \]

Review of some publications on CO2 reduction in the presence of photocatalysts based on TiO2/g-C3N4 (GP means that СО2 reduction is carried out in the gas phase, while LP refers to liquid-phase reduction). Refs. [87, 91, 92, 102-112]

(1)

Zinc oxide ZnO is widely used in various fields owing to its mechanical, electrical, optical and photocatalytic properties[113, 114]. The most thermodynamically stable ZnO phase under ambient conditions is wurtzite (Figure 8)[115, 116]. ZnO has a band gap of approximately 3.4 eV and n-type conduction, which makes it similar to TiO₂ for photocatalytic applications[117]. Hence, combination of ZnO with narrow-band-gap semiconductors such as g-C₃N₄ is also a promising approach for increasing the photocatalytic activity similarly to TiO₂/g-C₃N₄ composites.

Figure 8

Crystal structure of wurtzite phase of ZnO[115].

A variety of methods for the preparation of ZnO/g-C₃N₄ composites such as hydrothermal and solvothermal synthesis, layer-by-layer deposition and other have been reported in the literature[19]. For example, Chen et al.[118] synthesized the ZnO/g-C₃N₄ composite photocatalyst by evaporation of a suspension consisting of a methanol solution, g-C₃N₄ and zinc acetate prepared in advance. The reduction involved CO₂ and water vapour, which were formed inside the reactor upon the reaction between NaHCO₃ and H₂SO₄. The W_e value found for the composite photocatalyst was 9.4 times higher than that for pristine g-C₃N₄. The use of W_e for comparison of the photocatalyst activities is especially important here, because the photocatalytic СО₂ reduction in the presence of g-C₃N₄ is dominated by the formation of CO, whereas in the case of ZnO/g-C₃N₄ composites, the reaction almost exclusively gives СН₄. The authors also studied the stability of the most active ZnO/g-C₃N₄ photocatalyst and demonstrated that the activity decreased by about 7% by the third cycle, which indicated a high stability of the synthesized photocatalyst.

Guo et al.[119] synthesized the ZnO@g-C₃N₄ composite photocatalyst with the core@shell structure by depositing g-C₃N₄ on porous ZnO nanosheets using two-step calcination and studied the effect of reaction temperature on the photocatalyst activity towards the CO₂ reduction. A temperature rise from 150 to 200 °C resulted in a 3-fold and 2-fold increase in the rate of CH₄ and CO formation, respectively. However, as the temperature was further increased to 250 °C, the formation rate of CO₂ reduction products increased by only 33% compared to the rate at 200 °C. This effect may be caused by diffusion processes and adsorption – desorption equilibrium, which are affected by temperature. Since impossibility of initiation of photocatalytic processes by thermal energy has been proved in the literature both experimentally and theoretically, the change in the activity can be due only to the change in the above mentioned dark stages of the reaction[12, 120-122].

Zhu et al.[123] deposited copper nanoparticles on the ZnO/g-C3N4 materials with different contents of ZnO to be used in the photocatalytic CO₂ reduction in aqueous solution under mercury lamp irradiation. The highest rates of CO, CH₄ and CH₃OH formation equal to 64, 41 and 93 mmol g^–1 h^–1, respectively, were attained by using the 3% Cu/(30% ZnO/g-C₃N₄) photocatalyst (The contents of the components are given in mass percent.). These values are not only markedly higher than those for pristine g-C₃N₄, but they are also higher than those for the 30% ZnO/g-C₃N₄ composite photocatalyst. The authors proposed a possible microstructure of the 3% Cu/(30% ZnO/g-C₃N₄) photocatalyst and a mechanism of separation of the photogenerated charge carriers (Figure 9). Presumably, the charge carrier separation mechanism implemented in the photocatalysts is similar to type II heterojunction, but occurs via copper nanoparticles, which act as not only electron traps, but also as sources of electrons, causing a significant increase in the photocatalytic reaction rate.

Figure 9

Possible mechanism of electron transfer in the \( \text{Cu/ZnO/g-C}_{3}\text{N}_{4} \) heterostructures[123].

Data on some ZnO/g-C₃N₄ systems are summarized in Table 2. Currently, ZnO has been less explored for photocatalytic applications than TiO₂; hence, relatively few data on heterostructures based on ZnO composites with g-C₃N₄ have been reported in the literature. It is important that the activity of ZnO/g-C₃N₄ photocatalysts in the CO₂ reduction is at the level of TiO₂/g-C₃N₄ composite activity. Moreover, the W_e value of 1010 mmol g^–1 h^–1 for Cu/ZnO/g-C₃N₄ is especially notable, because three СО₂ reduction products, CO, CH₄ and CH₃OH, are formed at high rates in this case. Most types of heterojunctions present in the ZnO/g-C₃N₄ photocatalysts reported in the literature correspond to type II. However, the highest W_e value was attained for a three-component composite photocatalyst with a complex of heterojunctions.

Table 2

\[ \]

Review of some publications on CO2 reduction in the presence of ZnO/g-C3N4-based photocatalysts. Refs. [118, 119, 123-127]

(2)

Cerium dioxide CeO₂ is an n-type semiconductor with a wide band gap (2.8 – 3.1 eV) and the fluorite structure[128, 129]. An important feature of this compound is the high proneness of Ce⁴⁺ cations to be reduced to Ce³⁺[130]. This change in the oxidation state results in a change in the stoichiometry and formation of oxygen vacancies, which are known to enhance the absorption of visible light and also act as photoinduced charge carrier traps or as adsorption sites (Figure 10)[131, 132].

Figure 10

Crystal structure of the \( \text{CeO}_2 \) fluorite phase[132].

Liang et al.[133] investigated hollow g-C3N₄@CeO₂ photocatalysts for CO₂ reduction under irradiation with a xenon lamp with a 420 nm cut-off filter (l > 420 nm). The heat treatment of the composite photocatalyst in a hydrogen atmosphere resulted in the formation of oxygen vacancies and partial Ce⁴⁺ reduction to Ce³⁺. The highest rates of CH₄, CH₃OH and CO formation were attained in the presence of the g-C₃N₄@49.7%CeO₂ photocatalyst and amounted to 1.2, 1.7 and 5.6 mmol g^–1 h^–1, respectively, which exceeds these values for single g-C₃N₄ and CeO₂ catalysts. The synergistic effect is caused by the formation of type II heterojunction between the two semiconductors and by the large number of oxygen vacancies in CeO₂. The Ce⁴⁺ cations can trap photoinduced electrons, whereas the Ce³⁺ cations apparently provide the formation of the CO₂ – radical anion, which is an intermediate of CO₂ reduction[134]. Moreover, the hollow structure of the photoсatalyst increases the light utilization efficiency due to multiple reflections[135].

Wang et al.[136] synthesized a series of CeO₂/g-C3N4 composite photocatalysts with a built-in electric field. The electric field formation was confirmed by density functional theory (DFT) calculations. The calculation results indicate that electrons are accumulated on the g-C₃N₄ surface, while holes are concentrated on CeO₂. This gives rise to a built-in electric field directed from CeO₂ to g-C₃N₄, which promotes the transfer of photogenerated charge carriers according to the S-scheme. The highest product formation rates attained on the 1.75% CeO₂/g-C₃N₄ photocatalyst were 0.56 and 15 mmol g^–1 h^–1 for CO and CH₄, respectively, which is almost 20 times higher than those attained with single g-C₃N₄ or CeO₂.

Li et al.[137] reported a multistep hydrothermal synthesis of the CeO₂/g-C₃N₄ composite photocatalyst; then partially reduced graphene oxide (rGO) was deposited on the composite surface. Partially reduced graphene oxide is a graphene-like material with specific structural defects and oxidized groups on the surface[138]. The combination of these characteristics gives rise to a two-dimensional material with a large specific surface area, high electron mobility and chemical stability, which makes it a promising component for the synthesis of composite photocatalysts[139]. The photocatalytic CO₂ reduction was carried out in a suspension consisting of an alkaline solution of TEOA and a photocatalyst under xenon lamp irradiation. The CO and CH₄ formation rates were 63 and 33 mmol g^–1 h^–1 for rGO/CeO₂/g-C₃N₄ vs 15 and 5.2 mmol g^–1 h^–1 for g-C₃N₄. The results of DFT calculations suggest the formation of a built-in electric field in the composite photocatalyst caused by changes in the Fermi levels, similarly to what was described by Li et al.[137]. In the resulting multicomponent heterostructure, the formation of S-scheme heterojunction for electrons was suggested.

Data on the CeO₂/g-C₃N₄ systems are summarized in Table 3. The use of CeO₂ in photocatalytic studies is primarily due to the variable stoichiometry. The highest W_e values for CeO₂-based composite photocatalysts were obtained for heterostructures in which the charge carrier separation mechanism corresponds to the S-scheme and which are used for liquid-phase CO₂ reduction. As can be seen from Table 3, the formation of S-scheme hetero-junction provides the highest activity of CeO₂/g-C₃N₄-based photocatalysts. However, the activity of photocatalysts of this type is usually lower than that of TiO₂/g-C₃N₄ or ZnO/g-C₃N₄. Apparently, CeO₂ may be a promising material for the synthesis of photocatalysts for СО₂ reduction; however, additional studies along this line are needed.

Table 3

\[ \]

Review of some studies on CO2 reduction in the presence of CeO2/g-C3N4-based photocatalysts. Refs. [133, 136, 137, 140-145]

(3)

The iron oxide α-Fe₂O₃ (hematite) is a readily available, thermodynamically stable and environmentally benign narrow-band-gap n-type semiconductor[146-149]. However, the narrow band gap (2.2 eV) not only enhances the visible light absorption, but also markedly decreases the lifetime of photogenerated electron – hole pairs[150]. Moreover, the energy level of the bottom of the conduction band hampers the use of this material in reduction reactions due to the low potential of photogenerated electrons[19, 151]. Thus, α-Fe₂O₃ can be used in the photocatalytic CO₂ reduction only upon the formation of heterostructures with other semiconductors. In this respect, g-C₃N₄ can act as a semiconductor with an appropriate level of the conduction band minimum and with wider band gap compared to that of α-Fe₂O₃.

Guo et al.[152] used the hydrothermal method to prepare the α-Fe₂O₃/g-C₃N₄ composite for the photocatalytic CO₂ reduction to CH₃OH in water under xenon lamp irradiation with a 420 nm cut-off filter. The highest CH₃OH formation rate is 5.6 mmol g^–1 h^–1, which is almost three times higher than that for pristine g-C₃N₄ (1.9 mmol g^–1 h^–1). This increase is attributable to the formation of direct Z-scheme, which prevents recombination of photogenerated charge carriers and promotes the generation of electrons with a high reduction potential.

Duan and Mei[153] synthesized the α-Fe₂O₃/g-C₃N₄ photocatalyst by hydrothermal method from a colloidal solution containing both components (Figure 11). The CO₂ reduction reaction was conducted in an aqueous solution of DMF and TEOA under irradiation with a 60 W white light emitting diode (LED). The major reaction product was CH₃OH formed at a high rate of 74 mmol g^–1 h^–1. The solvent effect on the reaction rate was investigated by replacing DMF with acetonitrile, but the photocatalyst activity significantly decreased, indicating that the use of DMF in the photocatalytic CO₂ reduction may be promising for α-Fe₂O₃/g-C₃N₄ photocatalysts (but not necessarily for other ones[154]). It is worth noting that under the light irradiation, TEOA can be converted in the reaction medium to give carbon-containing products, which can affect the photocatalytic reaction rate[155-158]. This effect was ruled out by conducting an experiment with ¹³CO₂; this confirmed that CO₂ was the only source of CH₃OH.

Figure 11

Schematic image of the synthesis of the \( \alpha \text{-Fe}_{2}\text{O}_{3} /\text{g-C}_{3}\text{N}_{4} \) photocatalyst[153].

Padervand et al.[159] synthesized a complex composite photocatalyst, K₄Nb₆O₁₇/Fe₃N/α-Fe₂O₃/C₃N₄, by one-step thermal pyrolysis of the precursor. The photocatalyst combines the properties of three semiconductors with different positions of bands and magnetic properties of Fe₃N, which produces a highly efficient system for photocatalytic reactions. The highest product formation rates in the gas-phase CO₂ reduction under visible light irradiation were 7.0 and 1.3 mmol g^–1 h^–1 for CO and CH₄, respectively. The authors also proposed an electron transfer mechanism: since K₄Nb₆O₁₇ has a wider band gap with a very low top of valence band, type I heterojunction with α-Fe₂O₃ can be formed simultaneously with S-scheme heterojunction between α-Fe₂O₃ and g-C₃N₄. Moreover, Fe₃N acts as an electron trap, similarly to metals. These features explain the high CO₂ conversion upon photocatalyst irradiation with visible light.

Data on α-Fe₂O₃/g-C₃N₄ systems reported recently are summarized in Table 4. The highest W_e value of 440 mmol g^–1 h^–1 was found for the reaction in an aqueous solution of DMF + TEOA, whereas standard conditions of CO₂ reduction in aqueous solutions or in the presence of water vapour resulted in a lower rate, even when complex heterostructures were used as photocatalysts. Thus, the formation of composite photocatalysts consisting of g-C₃N₄ and narrow-band-gap semiconductors such as α-Fe₂O₃ is not a promising trend. In this case, high rates of product formation are not attained, probably because of the low lifetime of photogenerated charge carriers caused by the narrow band gaps of both components.

Table 4

\[ \]

Review of some studies on CO2 reduction in the presence of α-Fe2O3/g-C3N4-based photocatalysts. Refs. [152, 153, 159-161]

(4)

Cadmium sulfide CdS is a semiconductor material widely used for photocatalytic CO₂ reduction and many other photocatalytic reactions, because it has a narrow band gap (~2.4 eV) and appropriate arrangement of the valence and conduction bands[162-165]. Meanwhile, the photocatalytic application of CdS is limited not only by recombination of electron – hole pairs, but also by photocorrosion caused by the oxidation of sulfide ions by photogenerated holes[166-169]. An effective method to overcome these drawbacks is to fabricate composite photocatalysts based on CdS and other semiconductors.

Vu et al.[170] synthesized the CdS/g-C₃N₄ composite and studied it in the photocatalytic CO₂ reduction under irradiation with a solar simulator (100 mW cm^–2); the photocatalyst was suspended in a solution containing acetonitrile, TEOA, water and [Co(bpy)₃]Cl₂. An aqueous solution of acetonitrile was used as a solvent, TEOA served as an electron donor and the organometallic complex acted as a co-catalyst. It is noteworthy that no CO₂ reduction products were detected in the absence of TEOA and the Co complex. Carbon monoxide was formed as the major product of CO₂ reduction, CH₄ was detected in a small amount, and H₂ was formed as a by-product. The rate of СО evolution was 240 mmol g^–1 h^–1, which was almost four times higher than that with pristine g-C₃N₄. The selectivity to СО was 73%. The photocorrosion of CdS was inhibited via migration of photogenerated holes to g-C₃N₄: study of the photocatalyst stability indicated no loss of activity after four reaction cycles. The synergistic effect of CdS/g-C₃N₄ heterostructure is attributable to implementation of the direct Z-scheme with a bridging C – S – Cd bond at the interface, which results in increasing rate of transfer and separation of photogenerated charge carriers (Figure 12).

Figure 12

Mechanism of the photoinduced electron transfer in the \( \text{CdS/g-C}_{3}\text{N}_{4} \) photocatalysts in a solution with \( [\text{Co(bpy)}_{3}]\text{Cl}_{2} \)[170]. Published with permission from the American Chemical Society.

Guo et al.[171] synthesized a photocatalyst based on g-C₃N₄ and Zn_0.2Cd_0.8S nanoparticles using a combination of ultrasonic treatment and hydrothermal method (Figure 13). The Zn_xCd_1–xS solid solutions are known as photocatalysts that are activated under visible light and have a tunable band structure, which can be controlled by varying the Zn : Cd ratio, as in other solid solutions[172-175]. The photocatalytic activity in the CO₂ reduction was measured in an aqueous suspension of the photocatalyst at 80 °C under irradiation with a xenon lamp with a 420 nm cut-off filter. Methanol was the only detected product, which formed at a rate of 12 mmol g^–1 h^–1; this was higher than the methanol formation rate in the reaction performed using single components of the photocatalyst. Stability testing showed a decrease in the activity by only 7% after 28 h of the reaction, which indicated successful inhibition of photocorrosion. The authors suggested that separation of the photogenerated charge carriers in this photocatalyst corresponds to type II heterojunction, which promotes increase in both stability and activity.

Figure 13

Schematic image of the synthesis of the \( \text{Zn}_{0.2}\text{Cd}_{0.8}\text{S/g-C}_{3}\text{N}_{4} \) photocatalyst[171].

Data on some CdS/g-C₃N₄-based systems are summarized in Table 5. Since CdS is also a narrow-band-gap semiconductor, its combination with g-C₃N₄ usually does not provide high rates of CO₂ reduction. To attain high activity, additional modification or more complex reaction system is required. In particular, the two highest W_e values were found for systems using m-CdS/g-C₃N₄ photocatalyst dispersed in a solution, whereas W_e observed for the gas-phase СО₂ reduction in the presence of (Au/Zn_xCd_1–xS)@g-C₃N₄ was only 7.9 mmol g^–1 h^–1.

Table 5

\[ \]

Review of some studies on CO2 reduction in the presence of CdS/g-C3N4-based photocatalysts. Refs. [170, 171, 176-178]

(5)

Tin disulfide SnS₂ is a non-toxic semiconductor with a narrow band gap (~2.2 eV); therefore, it may be of interest for photocatalytic studies as a combination with wider-band-gap semiconductors[179-184].

Wang et al.[185] synthesized SnS₂/g-C₃N₄ composite photocatalysts with type II heterojunction by using self-assembly based on the electrostatic interactions between the components in an ethanol solution. The prepared photocatalysts with different SnS₂ contents were tested in the photocatalytic CO₂ reduction under xenon lamp irradiation. The highest rate of formation of the major product (CO) was 0.64 mmol g^–1 h^–1 in the presence of 60%SnS₂/g-C₃N₄, which is much higher than the rates attained using single g-C₃N₄ or SnS₂. The stability testing carried out for the most active sample showed that the activity remained almost unchanged after four cycles of the photocatalytic reaction. Since the results of X-ray diffraction and X-ray photoelectron spectroscopy did not show any significant changes either, it can be assumed that the synthesized photocatalyst has a high photocatalytic stability. Study of the charge carrier transfer mechanism provided the conclusion that type II heterojunction is formed between g-C₃N₄ and SnS₂. This increases the charge carrier separation rate, but, simultaneously, it also leads to a decrease in the redox potentials of photogenerated electrons and holes.

Yin et al.[186] obtained the SnS₂/g-C₃N₄ composite photocatalyst by a two-step method comprising sonication and hydrothermal synthesis followed by photodeposition of Au nanoparticles. The resulting photocatalyst was studied in CO₂ reduction in an aqueous solution of TEOA under irradiation with a xenon lamp. The major reaction products were CO and CH₄, with their formation rates being 94 and 75 mmol g^–1 h^–1, respectively, while g-C₃N₄ and SnS₂ taken separately had lower activity. The synergistic effect is due to two factors. First, both the hydrophilicity and CO₂ adsorption increase in the series g-C₃N₄ < SnS₂ < SnS₂/g-C₃N₄ < SnS₂/Au/g-C₃N₄, which leads to increasing adsorption of the reactants on the photocatalyst surface. Second, it is assumed that gold nanoparticles act as electron mediators for the Z-scheme implemented in the composite photocatalyst, which facilitates the electron transfer and increases the rate of charge carrier separation.

SnS2 is not the only tin sulfide that is used as a photocatalyst. For example, Huo et al.[187] reported hydrothermal synthesis of a composite photocatalyst consisting of porous g-C₃N₄ and Sn₂S₃ modified with diethylenetriamine (DETA) (Figure 14). The CO₂ reduction experiment was carried out in the gas phase in which CO₂ and water were formed upon the reaction of NaHCO₃ with HCl. A xenon lamp with a >420 nm cut-off filter was used as the light source. The major reaction products were CH₄ and CH₃OH, with the rates of their formation being 4.9 and 1.5 mmol g^–1 h^–1, respectively; this is higher than the activities of the initial g-C₃N₄ and Sn₂S₃-DETA. The replacement of H₂SO₄ by HCl did not result in any noticeable change in the reaction rate, which implies that the acid is not involved in the photocatalytic reaction. It is assumed that the synergistic effect is due to implementation of direct Z-scheme in the synthesized composite photocatalyst with a built-in electric field.

Figure 14

Schematic image of the synthesis of the \( \text{Sn}_{2}\text{S}_{3}\text{-DETA/g-C}_{3}\text{N}_{4} \) photocatalyst[187].

Table 6

\[ \]

Review of some studies on CO2 reduction in the presence of photocatalysts based on tin sulfides and g-C3N4. Refs. [185-191]

(6)

The data on the systems based on tin sulfides and g-C₃N₄ are summarized in Table 6. Despite the fact that the tin sulfide band gap is narrower than that of g-C₃N₄, tin sulfides (II, III, IV) are also used to form heterostructures. Apparently, SnS is potentially the most appropriate photocatalyst in this series for the photocatalytic CO₂ reduction, because the highest W_e value of 980 mmol g^–1 h^–1 was observed for SnS/g-C₃N₄ system in the gas-phase reduction of CO₂ under irradiation with a solar simulator. For comparison, the SnS₂/g-C₃N₄ system modified by gold nanoparticles, which was tested in the liquid-phase reduction of СО₂ under irradiation with full-spectrum xenon lamp, provided W_e of 790 mmol g^–1 h^–1. Note that tin sulfides are not widely used for the synthesis of heterostructures with g-C₃N₄, and there are few publications describing the use of tin sulfides as photocatalytic materials for CO₂ reduction, although SnS-based photocatalysts can potentially exhibit a fairly high activity in other reactions, for example, decomposition of dyes[192].

MXenes are a new class of 2D metal carbides, nitrides or carbonitrides discovered in 2011[193]. As a rule, MXenes are obtained from MAX phases by removing layers of A element (usually Al) with a potent etching reagent containing F^– anions (Figure 15)[193-198]. Different methods of synthesis result in the formation of different surface functional groups (e.g., –O, –F, –OH and –Cl). The general formula of MXenes can be written as M_n+1X_nT_x, where M is transition metal, X is C or N, and T is a functional group. MXenes have a number of remarkable properties such as high electrical conductivity, hydrophilicity and ordered layered structure, also, the composition of functional groups on their surface can be controlled[197, 198]. MXenes have already proved to possess a high potential for many applications including photocatalysis and are considered as promising materials for photocatalytic CO₂ reduction, especially in combination with other semiconductors forming heterostructures. The electronic structure of MXenes is characterized, most often, by the absence of band gap and high electron work function; therefore, they can be used as co-catalysts to form Schottky junction at the semiconductor – MXene interface, similarly to heterostructures based on transition metals and semiconductors.[199]

Figure 15

Periodic Table of Elements with highlighted elements present in other MAX phases (a) and basic diagram of MXene synthesis (b)[198].

Li et al.[200] developed mesoporous Ti₃C₂T_x/g-C₃N₄ photocatalysts for CO₂ reduction under irradiation with a xenon lamp (Figure 16). The MXene phase was prepared by a widely used method involving etching of Ti₃AlC₂ with hydrofluoric acid to remove Al layers. The reduction of CO₂ in the presence of Ti₃C₂T_x/g-C₃N₄ gave CO and CH₄ as the major products, with the formation rates being 4.0 mmol g^–1 h^–1 for CO and 2.2 mmol g^–1 h^–1 for CH₄; these values attained with mesoporous g-C₃N₄ were 3.1 and 0.88 mmol g^–1 h^–1, respectively. This effect is attributable to higher specific surface area, larger number of defects, higher electron transfer rate and, consequently, lower recombination rate of photogenerated charge carriers.

Figure 16

Schematic image of processes that take place in the \( \text{Ti}_{3}\text{C}_{2}\text{T}_{x}/\text{g-C}_{3}\text{N}_{4} \) photocatalyst[200].

М.Tahir and B.Tahir[201] synthesized a composite material consisting of g-C₃N₄ and layered bentonite clay (Bt), on which Ti₃C₂ particles were then deposited using the ultrasonic self-assembly technique. Bentonite mainly consists of smectite minerals, particularly montmorillonite (usually Ca montmorillonite and Na montmorillonite)[202], has a columnar multilayer structure and can significantly enhance CO₂ adsorption and charge carrier separation rate owing to bentinite surface properties and the presence of metal cations, which increase the photocatalytic activity (Figure 17)[202, 203]. Indeed, the photocatalytic experiments on CO₂ reduction demonstrate that the Ti₃C₂/g-C₃N₄/Bt ternary composite provides a much higher rates of formation of the major products (CO and CH₄), especially CH₄. It is presumed that the increase in the selectivity to CH₄ is caused by the efficient heterotransfer of charge carriers between the phases, which promotes the eight-electron reduction of CO₂ to СН₄. The effect of a sacrificial agent is also investigated, and it is found that the addition of acetic acid increases the rate of CH₄ formation by a factor of 4.2. CH₃COOH acts as an electron donor, decreases the charge carrier recombination rate and serves as a source of H₂, which is a more thermodynamically favourable reagent for CO₂ reduction than water[203-207].

Figure 17

(а) Structure of bentonite; (b) schematic image of processes taking place in the Ti3C2/g-C3N4/Bt photocatalyst[202, 203].

Studies dealing with composite photocatalysts for СО₂ reduction based on other MXenes, apart from Ti₃C₂, have also been reported. For example, Madi et al.[208] synthesized the V₂C/g-C₃N₄ photocatalyst by physical mixing of the components and sonication. The activity of the resulting photocatalyst in the gas-phase reduction of CO₂ under irradiation with a 35 W xenon lamp was considerably higher than the activity of g-C₃N₄ and somewhat higher than the activity of V₂AlC/g-C₃N₄. The authors attributed this synergistic effect to high electronic conductivity, which is one of the main benefits of V₂C and to the formation of Schottky junction between g-C₃N₄ and V₂C, which promotes the separation of photogenerated charge carriers[209-213]. A drawback of the composite is low stability: the photocatalyst activity decreases with every irradiation cycle, with the decrease reaching 40% by the end of the third cycle.

Data on the MXene/g-C₃N₄-based systems are summarized in Table 7. Analysis shows that, unfortunately, modification of g-C₃N₄ with Ti₃C₂ does not lead to a significant increase in the activity. It is noteworthy that data on the gas-phase CO₂ reduction giving CO and CH₄ as the major products are mainly presented in the literature. An outstanding W_e value of 8400 mmol g^–1 h^–1 was attained for the system containing bentonite clay under irradiation with a xenon lamp and with addition of acetic acid to the reaction system. A relatively simple V₂C/g-C₃N₄ heterostructure had W_e of 490 mmol g^–1 h^–1, which far exceeded the activity of most Ti₃C₂/g-C₃N₄ systems. It can be concluded that MXenes have a great potential as co-catalysts for photocatalytic CO₂ reduction and that intensive research along this line is needed.

Table 7

\[ \]

Review of some studies on CO2 reduction in the presence of photocatalysts based on MXene/g-C3N4. Refs. [93, 108, 126, 200, 201, 208, 214-218]

(7)

As mentioned above, partially reduced graphene oxide (rGO) is a promising material for photocatalysis due to its excellent conductivity and mechanical and optical properties combined with ready availability and the ease of synthesis from graphene oxide (GO)[138]. Moreover, a change in the reduction conditions makes it possible to obtain rGO samples that markedly differ in properties[219]. A traditional method for rGO production is the modified Hummers method, in which graphite is chemically oxidized and then exfoliated to produce graphene oxide[220]. In this stage, oxygenated functional groups appear in the graphene layers (Figure 18)[221]. During the subsequent reduction, some of these groups are removed, and the degree of reduction varies depending on the conditions[222]. It is noteworthy that the production of graphene oxide is accompanied by not only the formation of numerous oxygenated groups on the surface, but also by disruption of the conjugated structure of graphene, which provides for high mobility of charge carriers. The reduction of graphene oxide leads to the partial removal of oxygenated groups and restores the conjugated structure[223].

Figure 18

Schematic image of graphite oxidation and subsequent reduction to rGO[221].

Li et al.[224] performed a two-step process consisting of calcination and hydrothermal treatment and thus synthesized a composite photocatalyst by combining g-C₃N₄ and rGO with pre-treated multi-walled carbon nanotubes (P-MWNT)[224]. The photocatalytic CO₂ reduction was carried out under xenon lamp irradiation with a 420 nm cut-off filter in a TEOA and acetonitrile solution. The highest rates of CO and CH₄ formation were 180 and 120 mmol g^–1 h^–1, respectively. The authors proposed a possible structure of the photocatalyst in which carbon nanotubes act as a mediator between g-C₃N₄ and rGO and increase the rate of electron transfer. Presumably, photogenerated electrons migrate from g-C₃N₄ to rGO (either directly or via P-MWNT) as a result of generation of a built-in electric field, which considerably increases the efficiency of separation of photogenerated charge carriers[225, 226].

Bafaqeer et al.[227] used the three-component rGO-bridged ZnV₂O₆/g-C₃N₄ photocatalyst for the photocatalytic conversion of CO₂ to CH₃OH in water. For experiments, the authors designed an externally reflected photoreactor to increase the efficiency of photon energy utilization. As a result, the highest CH₃OH formation rate on exposure to a xenon lamp was 630 mmol g^–1 h^–1. Meanwhile, the CH₃OH formation rate in a reactor without external reflection was about 520 mmol g^–1 h^–1. Thus, even without the use of reflected light energy, the obtained photocatalyst activity is quite high, which is attributable to implementation of the S-scheme between two narrow-band-gap semiconductors and to participation of rGO as a mediator for efficient photogenerated charge carrier separation.

Data on the photocatalysts based on rGO/g-C3N4 reported in recent years are summarized in Table 8.

Table 8

\[ \]

Review of some studies on CO2 reduction in the presence of rGO/g-C3N4-based photocatalysts. Refs. [137, 224, 227-232]

(8)

Only ternary systems, mainly containing some type of vanadate or layered double hydroxide (LDH) as the third component, are considered here, because in the case of binary rGO/g-C₃N₄ photocatalysts, high СО₂ reduction rates are not attained. Layered double hydroxides are a group of layered solids formed by different-valence metal ions and hydroxide ions, which possess a lot of unique properties such as possibility of surface functionalization, intercalation of anions and high chemical stability; this makes LDHs promising photocatalytic materials[233-235]. Most experiments on СО₂ reduction catalyzed by rGO/g-C₃N₄-based photocatalysts were carried out in the liquid phase and the highest W_e values were found for ZnV₂O₆/rGO/g-C₃N₄ systems with Z-scheme charge transfer.

Development of g-C₃N₄-based photocatalysts for СО₂ reduction has been a significant research subject in recent years, which is confirmed by the steadily increasing number of relevant publications. The high interest of researchers in the g-C3N₄-based photocatalysts is caused by unique properties of g-C₃N₄, mostly, narrow band gap, which allows activation by visible light. However, narrow-band-gap semiconductors are characterized by high recombination rate of photogenerated charge carriers. Other drawbacks of unmodified g-C₃N₄ are low adsorption capacity towards CO₂ and, as a rule, low specific surface area, which result in moderate rate of CO₂ reduction. Modification of g-C₃N₄ mitigates the effect of these drawbacks on the photocatalytic process; the fabrication of heterostructures with other semiconductors represents the most popular modification method, as it allows for the control of the properties of photocatalysts and the subsequent increase in the CO₂ reduction rate.

Currently g-C₃N₄ heterostructures with metal oxides and sulfides, MXenes, and rGO are the most studied composite photocatalysts based on g-C₃N₄. Among these materials, TiO₂ is mostly studied, because it is a traditional material for many photocatalytic applications owing to its ready availability, lack of toxicity and stability. One more promising oxide semiconductor is CeO₂ owing to its unique feature, that is, change in the cerium oxidation state, giving rise to electron traps and additional adsorption sites. MXenes are a new class of 2D compounds with promising properties for the formation of heterostructures and photocatalytic applications. Various approaches to MXene synthesis make it possible to vary their surface properties, which opens up the way for development of a broad range of photocatalysts. Partially reduced graphene oxide rGO has also gained popularity in recent years, and a great potential of rGO for the use in photocatalytic CO₂ reduction systems has already been shown.

It is worth mentioning that comparison of the current studies on the photocatalytic CO₂ reduction is difficult because no unified experimental methodology has yet been developed. Studies of similar photocatalysts under similar reaction conditions by different research groups often result in significant differences not only in the rates of product formation, but also in the set of products, which may probably be caused by incomplete removal of organic impurities from the reaction medium or by low level of experimentation. Nevertheless, even though establishing of a strict correlation is problematic, it is still possible to determine the average productivity of particular systems to identify the most promising photocatalysts. The results considered in the review provide the conclusion that TiO₂ is most promising among the traditional materials used in photocatalysis for the formation of heterostructures with g-C₃N₄; this provides W_e values at the level of 200 – 600 mmol g^–1 h^–1, and in some cases, up to 6700 mmol g^–1 h^–1. In addition, there is a persistent trend towards the use of 2D materials such as MXenes and rGO for the formation of heterostructures. Indeed, in some studies, fairly high CO₂ reduction rates have already been attained by using Ti₃C₂ or rGO to modify g-C₃N₄. The major products formed upon СО₂ reduction catalyzed by heterostructures based on traditional semiconductors and those based on new 2D materials are CO and CH₄. However, in some cases, outstanding rates of CH₃OH formation have been achieved using rGO.

Thus, data on the activity of photocatalysts reported in the literature indicate that practical implementation of the photocatalytic CO₂ reduction requires much more active systems and, probably, the main trend of research in the photocatalytic CO₂ reduction for the next decade will be the search for and development of composite photocatalysts based on g-C₃N₄ combined with various 2D semiconductor materials and with materials (e.g., bentonite) that may significantly improve the textural properties of composites.

AQE — apparent quantum efficiency,
bpy — bipyridine,
CB — conduction band,
DETA — diethylenetriamine,
DMF — dimethylformamide,
g-C₃N₄ — graphitic carbon nitride,
GO — graphene oxide,
GP — gas phase,
LDH — layered double hydroxide,
LP — liquid phase,
m — mesoporous,
MXene — a class of two-dimensional metal carbides, nitrides or carbonitrides,
NHE — normal hydrogen electrode,
P-MWNT — pre-treated multi-walled carbon nanotubes,
PNS — porous nanosheets,
rGO — partially reduced graphene oxide,
SC — semiconductor,
TEOA — triethanolamine,
VB — valence band,
W_e — total number of electrons consumed in a photocatalytic process per unit reaction time and per unit photocatalyst mass.