Photocatalytic applications and modification methods of two-dimensional nanomaterials: a review

Abstract

Due to its unique electronic structure and special size effect, two-dimensional (2D) nanomaterials have shown great potential far beyond bulk materials in the field of photocatalysis. How to deeply explore the photocatalytic mechanism of 2D nanomaterials and design more efficient 2D semiconductor photocatalysts are research hotspots. This review provides a comprehensive introduction to typical 2D nanomaterials and discusses their current application status in the field of photocatalysis. The effects of material properties such as band structure, morphology, crystal face structure, crystal structure and surface defects on the photocatalytic process are discussed. The main modification methods are highlighted, including doping, noble metal deposition, heterojunction, thickness adjustment, defect engineering, and dye sensitization in 2D material systems. Finally, the future development of 2D nanomaterials is prospected. It is hoped that this paper can provide systematic and useful information for researchers engaged in the field of photocatalysis.

1 Introduction

With the development of science and technology, environmental pollution and energy shortage have become two major concerns of human beings. As a mild reaction, non-toxic, and reusable technology, photocatalysis can efficiently convert solar energy into chemical energy, and is considered to be one of the most forward-looking long-term solutions to global energy and environmental problems [1, 2].

As a promising technology, photocatalysis has been widely used in water treatment [3,4,5], hydrogen production by water splitting [6,7,8], CO₂ reduction [9, 10], nitrogen fixation [11, 12], and organic synthesis [13, 14] (Fig. 1). Fujishima and Honda published a paper on TiO₂ photoelectrochemical cells in 1972, showing that water can be decomposed into oxygen and hydrogen by visible light without any external voltage [15]. Since then, photocatalytic has come into public attention. In 1976, Carey et al. found that in the photocatalytic process of TiO₂, polychlorinated biphenyl, haloalkane, etc. can be effectively photocatalytically degraded [16]. This has led to the realization that photocatalysis can provide new approaches for environmental governance. Schrauzer and Guth determined that water splitting does not require external energy except for light energy in 1977, and demonstrated that TiO₂ can split water to generate H₂ under ultraviolet (UV) light irradiation [17]. Nearly, 50 years of researches have also proved that photocatalytic technology can be effectively applied in environmental governance and energy conversion and has great development potential. Even studies have shown that photocatalysis can degrade almost all organic pollutants in water [18].

Fig. 1

Application areas of photocatalytic technology

The principle of semiconductor photocatalyst is shown in Fig. 2: the energy band structure of semiconductor materials consists of a low-level valence band (VB) filled with electrons and a high-level conduction band (CB) that is not filled with electrons, and a gap called the forbidden band is located between the conduction band minimum (CBM) and valence band maximum (VBM). When the photocatalyst is irradiated by light with energy greater than or equal to the forbidden band width, the electrons in the VB absorb energy and transition to the CB, leaving holes in the VB, eventually forming “electron–hole pairs”, also known as “carriers”. Under the action of the space electric field, the electron–hole pairs partially migrate to the particle surface and participate in further catalytic reactions [19].

Fig. 2

Schematic diagram of photocatalytic schematic

As the first semiconductor material to be applied to photocatalysis, TiO₂ is a wide-bandgap semiconductor (3–3.2 eV), which can only absorb UV light with a wavelength less than 387 nm. UV light only accounts for 5% of sunlight, making TiO₂ less efficient for sunlight. This makes TiO₂ less than ideal in practice. At the same time, in the photocatalytic process, the high electron–hole pair recombination rate and low quantum efficiency also limit the photocatalytic activity of TiO₂ [20]. Hoffmann et al. studied TiO₂ in depth, and found that four processes are very important in the photocatalytic process: first, the spectral response range; second, whether the photogenerated electron holes can be effectively separated; the third is the speed at which electrons are transferred from the interior of an atom to the surface; the fourth is the strength of redox activity [21]. In order to better participate in the above four processes, the selection of semiconductor photocatalysts is the key to photocatalytic applications. High-efficiency photocatalysts must have a suitable bandgap to provide sufficient redox activity, excellent electron–hole separation ability, and visible-light response characteristics.

Aiming at these four key steps to improve the photocatalytic efficiency, two-dimensional (2D) nanomaterials have entered people’s field of vision as new photocatalysts. As shown in Fig. 3, carbon materials are used as an example to show different material structures from zero-dimensional (0D) to three-dimensional (3D) [22,23,24,25]. 2D nanomaterials refer to materials in which electrons can only move freely (plane movement) on the nano scale (1–100 nm) in two dimensions. Because of its unique thickness size and 2D structure characteristics, the intrinsic properties of the materials can be enhanced, resulting in many different properties from their corresponding 3D structures. For example, large specific surface area, porous surface with abundant reactive sites, good carrier mobility, good Young’s modulus, excellent electrical and thermal conductivity, and the advantages of easily forming new composite structures with other materials [26]. Meanwhile, this 2D structure can be assembled through spin processes or layer-by-layer, and layered artificial structures are created and designed on demand, with high flexibility in material design [27].

Fig. 3

Reproduced with permission from Ref. [22]. Copyright 2018, Royal Society of Chemistry. Reproduced with permission from Ref. [23]. Copyright 2020, Elsevier. Reproduced with permission from Ref. [24]. Copyright 2017, ACS Publications. Reproduced with permission from Ref. [25]. Copyright 2020, Elsevier

Schematic diagram of material dimensions.

However, the practical application of 2D nanomaterials is still limited due to the easy agglomeration of powders, the small scale of preparation, and the uncontrollability. At the same time, too narrow or too wide bandgap of a single 2D nanomaterials will lead to problems such as insufficient redox capacity, narrow photoresponse range, and fast recombination of electron–hole pairs, resulting in poor photocatalytic performance. On this basis, scientists have further researched, proposed a variety of modification methods to improve the photocatalytic efficiency of 2D nanomaterials, and revealed their mechanisms. Due to the important position of 2D materials in the field of photocatalysis, some researchers have reviewed in detail the basic properties of 2D materials, the synthesis strategies of 2D materials, and the surface modification of 2D materials in recent years, providing many interesting perspectives [28,29,30,31,32,33].

Compared with previous reviews [28,29,30,31,32,33], in this review, we introduce the progress of various typical 2D materials in the field of photocatalysis in a more comprehensive and detailed manner. Unlike other reviews, which often explore the advantages of 2D materials in the field of photocatalysis from the perspective of photocatalytic processes such as light absorption, charge separation and transfer, this review discusses the influence of intrinsic characteristics of 2D materials (bandgap, morphology, facet structure, crystal structure and defects) on the photocatalytic process, revealing the advantages of 2D systems in photocatalysis. Afterwards, various modification methods (element doping, noble metal deposition, heterojunction, thickness tuning, defect engineering, and dye sensitization) are further discussed on the basis of the above-mentioned influencing factors. Finally, this review summarizes and looks forward to the main difficulties and potential solutions in the current 2D materials research, looking forward to helping readers think about the ideas of photocatalyst design.

2 Typical 2D nanomaterials

Typical 2D nanomaterials are shown in Fig. 4, which can be mainly divided into two aspects. One is graphene and its derivatives, the other is graphene-like 2D materials, such as 2D metal oxides [34, 35], hexagonal boron nitride (h-BN) [36, 37], transition metal dichalcogenides (TMDs) [38], carbonitrides [39, 40], black phosphorus (BP) [41], layered double hydroxides (LDHs) [42], 2D metal organic frameworks (MOFs) [43] and other layered materials with unique physical and chemical properties. In this section, we introduce the structure, properties, synthesis process and research progress of these materials, in order to help readers better understand the 2D material system.

Fig. 4

Common 2D materials and their structures (COFs: covalent organic frameworks)

2.1 Graphene and its derivatives

Since graphene was prepared by micromechanical exfoliation method [44] in 2004, it has attracted wide attention in the field of photocatalysis due to its outstanding mechanical, thermal, optical properties and appearance characteristics [45]. With the deepening of research, some new methods for preparing graphene have been proposed, such as mechanical exfoliation method, liquid phase exfoliation method (LPE), redox method, chemical vapor deposition method (CVD), and electrochemical exfoliation method. Among them, the exfoliation preparation process represented by the mechanical exfoliation method can obtain high-quality graphene, but it also has the disadvantages of small area and low yield. Although the synthesis process represented by CVD method can prepare large-area graphene, the defects generated during the growth process reduce the performance of graphene. How to combine the advantages of CVD and mechanical exfoliation is the main idea for large-scale preparation of high-quality, large-area graphene [46].

The basic structure of graphene is a honeycomb-shaped planar film formed by sp² hybridization of carbon atoms, and its derivatives are graphene oxide (GO) and reduced graphene oxide (rGO). Due to its zero bandgap, graphene has excellent electrical conductivity, and has a wide spectral absorption range, including visible light and long-wave infrared, even to terahertz frequencies. At the same time, the special single-atom-layer planar structure enables it to have a high light transmittance (~ 97.7%), which can be combined with other catalysts without affecting the light absorption capacity of the photocatalyst. The theoretical specific surface area of graphene is as high as 2,630 m²·g⁻¹, which can provide more redox active sites [47]. However, the zero bandgap of graphene also makes its redox reaction ability poor, which is not conducive to the photocatalytic reaction. Therefore, researchers often combine it with other wide-bandgap photocatalytic materials to form composites with excellent photocatalytic properties, thereby exerting the excellent optoelectronic properties of graphene. Zhang et al. prepared graphene/ZnO composites (GR/ZnO) by ball milling and LPE [48]. The bandgap test results show that the bandgap of the GR/ZnO is narrowed to 3.10 eV compared with pure ZnO. Under simulated sunlight irradiation, GR/ZnO exhibited significantly enhanced photocatalytic activity, which was 6 times higher than that of pure ZnO (Fig. 5). This study shows that the addition of graphene can effectively narrow the bandgap and improve the photocatalytic performance. Huang et al. composited flake BiFeO₃ with graphene via a simple hydrothermal method. The composite material has excellent visible-light response, and the photocatalytic performance is greatly improved [49]. Zhang et al. prepared MoS₂/N-doped graphene composites, which constituted a Z-scheme photocatalytic system composed by two narrow-bandgap semiconductors [50]. Its transformed bandgap extends the photoresponse to 1,550 nm, enabling the composite to effectively degrade NH₃ under near-infrared irradiation. Although graphene can hardly be applied to photocatalysis independently due to the zero bandgap, Yang et al. found that coating a single layer of graphene on a metal co-catalyst can form a photocatalyst that can be applied to photocatalytic overall water splitting (POWS) [51]. Studies have shown that graphene plays an important role in suppressing the reverse action of H₂ and O₂. Importantly, among various types of graphene, γ-graphene (GY) is regarded as the graphene with the highest stability and semiconducting properties. It has a suitable optical bandgap (2.3 eV) and can be well applied in the field of photocatalysis [52]. Wu et al. prepared TiO₂/MoSe₂/GY ternary nanocomposites. GY acts as an electron collector and hole transfer channel, effectively separating photogenerated carriers [53].

Fig. 5

Reproduced with permission from Ref. [48].Copyright 2021, Elsevier

a Schematic diagram of the preparation of GR/ZnO. b Ultraviolet–visible absorption spectra (UV–Vis absorption spectra) of ZnO–Polyvinyl alcohol (PVA) and GR/ZnO. c Light energy relationship diagram of ZnO–PVA and GR/ZnO. d Rate diagram of degradation of methylene blue (MB) and rhodamine B (RhB) by GR/ZnO under simulated sunlight, and e pseudo-first order kinetic curve.

More studies have shown that functionalized graphene-based analogs, such as GO, may be potential materials for photocatalysts, because the band structure of GO is related to its oxidation degree, its bandgap can be regulated during the preparation process [54, 55]. Yeh et al. used the modified Hummers method to prepare GO with a bandgap of 2.4–4.3 eV, and the results showed that it can be used as a photocatalytic catalyst for water splitting [56]. Under ultraviolet or visible-light irradiation, 20% Methanol aqueous solution and pure water can stably generate H₂.

2.2 Graphene-like 2D materials

Graphene-like 2D materials are a general term for 2D single-layer or few-layer structural materials. There are many types of graphene-like materials, which exhibit superior surface properties compared with bulk materials due to their unique layered structure.

2.2.1 Metal oxides

In the past 4 decades, many metal oxides such as TiO₂ [57, 58], ZnO [59, 60], SnO₂ [61, 62], WO₃ [63, 64], and Fe₂O₃ [65, 66] have been extensively studied as photocatalysts. Due to the obvious advantages of 2D materials in the field of photocatalysis compared with bulk materials, the research of metal oxide nanosheets is also widely carried out. Taking TiO₂ as an example, the preparation methods of metal oxide nanomaterials mainly include CVD, sol–gel method, hydrothermal method, pulse laser deposition method (PLD) and sputtering method. Among them, the hydrothermal method is the most common method for preparing 2D TiO₂ nanosheets, and PLD and laser sputtering method are commonly used to prepare thinner TiO₂ films. Other methods are commonly used to prepare lower dimensional forms such as TiO₂ nanorods and nanospheres. It is notable that by controlling the reaction temperature during CVD, different states of TiO₂ from nanotubes to thin films can be obtained [67].

The 2D nanosheets of TiO₂ are exfoliated from layered titanates and exhibit similar semiconductor properties to rutile and anatase TiO₂. However, due to the reduced grain size, the bandgap of TiO₂ nanosheets is slightly larger (~ 3.8 eV), which is larger than that of anatase TiO₂ (~ 3.2 eV). This will lead to a decrease in the photoresponse range, which is not conducive to the photocatalytic reaction [68]. Studies have shown that photocatalytic performance of TiO₂ nanosheets can be further improved by modifying it by doping or carrying a narrow-bandgap photocatalyst. Such as Liu et al. prepared transition metal modified TiO₂ nanosheets by microwave-assisted hydrothermal method [69]. It is revealed that the addition of transition metals effectively narrows the bandgap of TiO₂ (Fig. 6a). Through UV–Vis absorption spectra, it can be seen that TiO₂ with transition metals has better visible-light absorption than pure TiO₂ (Fig. 6b). The results of photoluminescence spectra (PL spectra) (Fig. 6c) show that the addition of transition metals reduces the intensity of TiO₂, especially Cu/TiO₂. The decrease in intensity indicates faster carrier transfer, which corresponds to the optimal photocatalytic nitrogen fixation performance of Cu/TiO₂. Chen et al. prepared TiO₂ nanosheets/NiO nanorods composites by a hydrothermal method (Fig. 6d) [58]. TiO₂/NiO photocatalyst showed good MB degradation activity and hydrogen production efficiency under visible light, which were 12 times and 10 times that of TiO₂ nanosheets. This is due to the construction of its p–n heterojunctions, which enhances interfacial charge transfer and light absorption (Fig. 6e). At present, TiO₂-based nanocomposites have been studied in many fields such as sewage treatment, medical treatment, hydrogen energy, and CO₂ reduction. It is worth pointing out that in the field of anti-cancer, TiO₂ will have great potential due to its good biocompatibility and low toxicity.

Fig. 6

Reproduced with permission from Ref. [69]. Copyright 2020, Royal Society of Chemistry. d The schematic diagram of the synthesis processes of TiO₂/NiO. e The schematic diagram of the photocatalytic mechanism of TiO₂/NiO. Reproduced with permission from Ref. [58] Copyright 2020, Elsevier. Degradation curves of MB by MnO₂/Fe₃O₄ at different pH values f in the dark and g with UV−Vis light, h the speculated reaction principle of MnO₂/Fe₃O₄ and MB under different pH values. Reproduced with permission from Ref. [71]. Copyright 2014, ACS Publications

a UV–Vis absorption spectra of transition metal modified TiO₂ nanosheets. b The bandgap evaluation of transition metal modified TiO₂ nanosheets. c PL spectra of transition metal modified TiO₂ nanosheets.

In addition to TiO₂, the nanoscale research on other traditional metal oxide catalysts has also achieved success. Wang et al. prepared oxygen-rich vacancy-defective ZnO nanosheets with tunable Brunauer–Emmett–Teller (BET) surface area [70]. The analysis showed that the oxygen-rich vacancies enhanced the visible-light absorption of the ZnO nanosheets, and the increase in the surface oxygen vacancy concentration was positively correlated with the BET surface area. Compared with ZnO with fewer oxygen defects, the rate of RhB degradation by oxygen-enriched ZnO under visible light is increased by about 11 times. Zhang et al. prepared MnO₂/Fe₃O₄ nanocomposites [71]. The results show that the MnO₂/Fe₃O₄ nanocomposite exhibits excellent photocatalytic performance and high stability when degrading MB under UV–Vis light (Fig. 6f, g). In addition to the light factor, the photocatalytic performance of MnO₂/Fe₃O₄ also varied with pH, which was due to the different surface charge states at different pH (Fig. 6h). In addition, tests have shown that the MnO₂/Fe₃O₄ nanocomposite is ferromagnetic and can be effectively separated by an external magnetic field, enabling reuse and being more environmentally friendly.

2.2.2 h-BN

h-BN is a graphene-like hexagonal structure formed by covalently connecting N atoms and B atoms in equal proportions. It has a honeycomb lattice structure similar to graphene, so it is called “white graphite” and has excellent physical properties and surface atomic flatness [72, 73]. Since h-BN is an insulator with a bandgap of 6 eV and lacks dangling bonds on the surface, h-BN was studied as an ideal substrate and insulating layer for other 2D materials in the early stages of research [74]. As the research progressed, it was found that the bandgap of h-BN is very sensitive to size, thickness and defects. Meanwhile, h-BN has a high specific surface area and excellent conductivity [75]. Moreover, although h-BN is an insulator, its (002) crystal plane has abundant active sites, which can effectively participate in the photocatalytic process [76]. The above advantages have led researchers to explore the application of h-BN in photocatalysis. Currently, the mainstream methods for preparing h-BN include mechanical exfoliation, solvothermal, CVD, and molecular beam epitaxy (MBE). Due to the advantages of high quality and large scale, CVD has become the main preparation method of h-BN. At the same time, CVD also has excellent performance for the preparation of h-BN heterostructures [77].

In practical photocatalytic applications, due to the wide bandgap of h-BN, it is necessary to modify h-BN by defect engineering, doping or building a heterojunction structure to realize its application in the field of catalysis [78,79,80]. It is found that combining wide-bandgap h-BN with zero-bandgap graphene into an in-plane heterostructure is an effective approach to design efficient photocatalysts. Bevilacqua et al. used density functional theory (DFT) to study the stability and optical properties of monolayer h-BN when doped with graphene [81]. The results show that graphene (h-BNC) doped with wide-bandgap h-BN nanodomains exhibits a non-zero gap, and the width of its bandgap is related to the number of BN nanodomains. Therefore, by doping h-BN with graphene, the h-BNC that can respond to visible light can be obtained, which is a promising photocatalytic material. Besides graphene, h-BN can also be combined with other 2D materials to form photocatalysts. Li et al. constructed TiO₂/h-BN composite photocatalysts with TiO₂ nanosheets and h-BN nanosheets by a solvothermal method. The photocatalyst has greatly improved the degradation efficiency of rhodamine B, which is 6.9 times that of commercial TiO₂ [82]. Wang et al. prepared a composite photocatalyst of h-BN and g-C₃N₄ quantum dots. Due to the large specific surface area and good electron-harvesting properties of h-BN, h-BN/g-C₃N₄ has excellent photocatalytic performance [83].

2.2.3 Transitional metal dichalcogenides (TMDs)

TMDs are compounds composed of transition metal elements and chalcogen elements (S, Se or Te). The basic expression of TMDs is MX₂, where M is a transition metal and X is a chalcogen (Fig. 7a). The layers of TMDs are connected by weak van der Waals forces, which can be easily peeled from the bulk phase to a single sheet [38]. As Fig. 7b shows, depending on the coordination of elements and the stacking order of the lamellae, TMDs can form a variety of crystalline polytypes: 1T, 2H, and 1T’ [84], exhibiting different electronic properties from metals to semiconductors [85, 86]. TMDs exhibit excellent electronic, optical, and physicochemical properties, which mainly have the advantages of high electrical conductivity, narrow bandgap, abundant active sites, abundant earth reserves, and low cost [87,88,89]. At the same time, it is found that the theoretical catalytic performance of TMDs is second only to the noble metal Pt and some transition metals, which has great research value [90, 91].

Fig. 7

Reproduced with permission from Ref. [84]. Copyright 2023, John Wiley and Sons. c Evolution of the band structure of 2H‑MoS₂ calculated for samples of decreasing thickness. Reproduced with permission from Ref. [96]. Copyright 2017, Springer Nature. d Schematic diagram of direct band gap and indirect band gap. Reproduced with permission from Ref. [97]. Copyright 2020, ACS Publications. e Scanning electron microscope image of U-CN/MoS₂-3. f Histograms of hydrogen production for different ratios of U-CN/MoS₂. g Heterojunction mechanism of U-CN/MoS₂ photocatalyst. Reproduced with permission from Ref. [106]. Copyright 2019, Elsevier

a Composition of TMDs. b Schematic models of different phases of TMDs.

Most TMDs materials have typical bandgap tunable properties, such as MoS₂ [92], MoSe₂ [93], WSe₂ [94] and WS₂ [95]. According to DFT, as shown in Fig. 7c, the band structure of 2H-MoS₂ will evolve as it decreases from bulk layer to monolayer [96]. As the thickness decreases, the positions of VB and CB change, from an indirect bandgap semiconductor to a direct bandgap semiconductor. The difference between the two bandgaps is shown in Fig. 7d [97]. At present, the main methods for preparing 2D TMDs include mechanical exfoliation, LPE, CVD, and wet chemical synthesis. The above traditional methods have been widely used in the preparation of various TMDs materials. Yang et al. reviewed the preparation methods of TMDs materials in detail [84]. It is noteworthy that the electrochemical Li-ion intercalation method has been creatively proposed in recent years. This method can produce atomically thin TMDs nanosheets on a large scale and with high yield, and has been widely applied to the preparation of various TMDs materials [98].

As the most representative material among TMDs materials, MoS₂ is a hexagonal layered crystal structure, and each unit is a “sandwich” structure of S-Mo-S. The common crystal forms are 1T, 2H and 3R, of which the 2H crystal form is the most stable. Although some studies have revealed that the metal phase 1T-MoS₂ exhibits better photocatalytic performance, the research on 2H-MoS₂ catalysts is still the mainstream due to the requirement of stability in practical applications [99,100,101]. The bandgap of bulk MoS₂ is too small, only 1.2 eV, which cannot provide sufficient oxidation/reduction potential to activate the photocatalytic process. The single-layer MoS₂, due to the conversion of indirect bandgap to direct bandgap, has an increased forbidden band width to about 1.96 eV, which can be used as a photocatalyst [102, 103]. However, due to its still narrow bandgap, the oxidizing ability is weaker when used alone. The edge potentials of the CB and VB of MoS₂ are higher than those of most photosensitive semiconductors, and the potential difference between the CB of MoS₂ and other semiconductor materials can promote the transfer of electrons to MoS₂, thereby effectively separating the carriers. At the same time, MoS₂ can be used as a substitute for noble metals in the hydrogen evolution reaction and become a co-catalyst for the hydrogen evolution reaction [104, 105]. Therefore, MoS₂ is generally used as a co-catalyst or carrying other semiconductor photocatalytic materials to form a heterojunction to improve the photocatalytic performance. Li et al. prepared ultrathin C₃N₄/MoS₂ heterojunction photocatalysts (U-CN/MoS₂) [106]. The results show that the photocatalyst activity is significantly improved after MoS₂ is supported on C₃N₄ to form a heterojunction (Fig. 7e). The best photocatalytic hydrogen production efficiency can reach 385.04 μmol·h⁻¹·g⁻¹ (Fig. 7f). The photocatalytic mechanism of U-CN/MoS₂ is shown in Fig. 7g. When U-CN/MoS₂ is illuminated, the electrons on VB of both materials are excited to CB. Under the action of the heterojunction, the electrons on the CB of U-CN are transferred to the CB of MoS₂, and the holes on the VB of MoS₂ are transferred to the VB of U-CN at the same time. This process realizes the effective separation of electrons and holes, which promotes the improvement of photocatalytic activity.

In addition to MoS₂, other TMDs materials have also shown excellent performance in the field of photocatalysis. Kou et al. grew TiO₂ on MoSe₂ nanosheets [107]. Under visible light, the hydrogen production rate of MoSe₂/TiO₂ is much higher than that of pure MoSe₂ and TiO₂. The study shows that the increase in the hydrogen production rate is attributed to the MoSe₂ promoting the absorption of visible light and the high charge separation efficiency of the MoSe₂/TiO₂ interface. Zhang et al. synthesized WSe₂ nanosheets with a thickness of 0.7 nm by a colloidal method [108]. The results show that the synthesized WSe₂ nanosheets can efficiently catalyze the aerobic coupling reaction of amines and imines under visible-light irradiation, verifying the possibility of TMDs being applied in various catalytic processes.

2.2.4 Graphitic carbon nitride (g-C₃N₄)

As a semiconductor material, the catalytic hydrogen evolution performance of g-C₃N₄ was not discovered until 2008 by Wang et al. and it is the first non-metal organic semiconductor photocatalytic material with visible-light response in the field of photocatalysis with a suitable bandgap (2.7 eV) [109]. Since bulk g-C₃N₄ is formed by stacking 2D structures, g-C₃N₄ nanosheets can be obtained by exfoliating bulk g-C₃N₄. Studies have shown that the photocatalytic performance of g-C₃N₄ nanosheets is significantly improved compared with that of bulk g-C₃N₄, and are only composed of C and N elements, which will not cause metal pollution and are environmentally friendly [110]. The main methods for preparing bulk g-C₃N₄ include thermal polycondensation, hydrothermal method, template method and ball milling method. Among them, the preparation of g-C₃N₄ by thermal polycondensation of nitrogen-rich precursors is the most commonly used [111]. However, g-C₃N₄ prepared by this method is rich in defects due to incomplete deamination during calcination. Relatively speaking, the template method can effectively regulate the structural characteristics such as porosity and specific surface area of g-C₃N₄ through appropriate templates. However, the residual template will have a great impact on the photocatalytic performance of g-C₃N₄, and it also has a more cumbersome process. In response to the above problems, the researchers considered improving the structural defects by pretreatment of the nitrogen-rich precursor before calcination. If the precursor is treated with acid, the van der Waals force between the layers is weakened, which is beneficial to obtain thinner g-C₃N₄ sheets [112].

The photocatalytic activity of pure-phase g-C₃N₄ still needs to be improved due to the limitations of low quantum rate, fast recombination of photogenerated carriers, and insufficient response to visible light above 460 nm [113]. Generally, the photocatalytic activity of g-C₃N₄ can be enhanced by modification methods such as doping, noble metal deposition, and construction of heterojunctions. Peng et al. synthesized a ternary composite photocatalyst CdS/Au/g-C₃N₄ [114]. Due to the deposition of CdS and Au on the surface of g-C₃N₄, two different structures, CdS–g-C₃N₄ and CdS@Au-g-C₃N₄, were formed (Fig. 8a, b). Meanwhile, there is a strong interaction between the three components, so compared with the binary composites CdS/g-C₃N₄, Au/g-C₃N₄ and pure g-C₃N₄, the photocatalytic activity of CdS/Au/g-C₃N₄ for RhB degradation was significantly enhanced under visible-light irradiation. Zhang et al. prepared Mn-adsorbed g-C₃N₄ [115]. The change in electronic structure caused by the stable bond between Mn and N atoms makes the photocatalytic performance of Mn-adsorbed g-C₃N₄ three times that of pristine g-C₃N₄. Cao et al. introduced nitrogen vacancies (NVs) into n-type g-C₃N₄ to construct a p-n heterojunction, which significantly improved the photocatalytic degradation activity of g-C₃N₄ [116]. Introducing appropriate functional groups into the g-C₃N₄ framework is also an effective strategy to enhance the photocatalytic activity. Wang et al. designed a simple KOH-assisted sealing heating process to tune the electronic structure of g-C₃N₄ [117]. The introduction of cyano groups increases the charge carrier density of g-C₃N₄, reduces the surface transfer resistance, promotes charge separation and transfer, and improves the photocatalytic hydrogen production performance.

Fig. 8

Reproduced with permission from Ref. [114]. Copyright 2016, Royal Society of Chemistry

a CdS–g-C₃N₄ heterojunction. b CdS@Au–g-C₃N₄ heterojunction.

2.2.5 2D elemental material

2D elemental materials include 2D main group elements and 2D transition metals. In the field of catalysis, BP is one of the most concerned 2D elemental materials. BP is an allotrope of phosphorus with a monolayer pleated honeycomb structure. It was first extracted from white phosphorus in 1,914 under high pressure and moderate temperature (200 ℃), showing high density and good conductivity [118]. However, the high-pressure conditions of this method limit the development of this method. In 2007, a simple method for preparing BP by catalyzing red phosphorus was proposed, which further promoted the practical application of BP [119]. The current mainstream methods for preparing BP mainly include “top-down” methods such as mechanical exfoliation, LPE, and electrochemical exfoliation. BP is highly sensitive to substrates, so the “bottom-up” preparation of BP remains a challenge [120]. BP nanosheets have a flexible and tunable bandgap (0.3–2.0 eV), which increases with the decrease of the number of layers, 2D ultrathin structure and large specific surface area. These features can enhance the adsorption of photocatalysts, increase the number of active sites, and effectively improve the catalytic performance [121, 122]. At present, the latest research published in Nature has revealed a significant change in the shape of the VB of BP when it receives light radiation, which will provide a solid theoretical basis for its further application in many fields such as catalysis [123]. The narrow bandgap of BP enables it to absorb a large amount of visible light and even near-infrared light, but at the same time, it does not have enough driving force to participate in the photoreaction. Another limitation of BP is that it is unstable in the atmospheric environment, and usually requires the deposition of a coating on the surface to improve its stability [124]. Although monolayer BP has excellent theoretical photocatalytic activity, few experiments have proved the photocatalytic performance of pure BP, and researchers often combine it with other materials. It was not until 2017 that Zhu et al. prepared BP nanosheets with a size of 300–500 nm by ball milling [125]. The hydrogen evolution rate was 512 μmol·h⁻¹·g⁻¹, which was superior to g-C₃N₄.

In addition to BP, bismuth (Bi), a group V element, is also a popular element 2D material, which has the characteristics of metal and semiconductor. Bi-based semiconductor photocatalysts have been widely used in the field of catalysis due to their unique electronic, optical, magnetic, and structural properties. Unlike metal oxides such as TiO₂, the VB of Bi is formed by the hybridization of Bi 6s and O 2p orbitals, which effectively shortens the semiconductor band gap and is more conducive to the absorption of visible light [28]. At present, researchers have synthesized many Bi-based photocatalysts, such as bismuth oxide (Bi₂O₃), bismuth oxyhalide (Bi_aO_bX_c, where X = Cl, Br or I), and Bi-based polymetallic salts (BiVO₄, Bi₂WO₆, and Bi₂MoO₆, etc.). By adjusting the parameters to prepare Bi-based photocatalysts with ultrathin structures, materials with more excellent photocatalytic properties will be obtained. Its main preparation methods include exfoliation, solvothermal, colloidal synthesis, and CVD growth [126].

In addition to 2D Bi-based materials, the bismuthene exfoliated from bulk bismuth also has excellent 2D ultrathin structures. However, due to the diamond structure of bulk Bi, the layers may not be completely connected by weak van der Waals force, so it is difficult to prepare bismuthene by traditional mechanical exfoliation and ultrasonic exfoliation. Generally, it is prepared by other methods such as high-temperature hydrothermal and LPE [127]. Yu et al. successfully prepared Bi-nanobelts (~ 20 nm) through high-temperature hydrothermal synthesis [128]. Yang et al. designed a method for preparing bismuthene by sulfuric acid intercalation exfoliation, and finally successfully exfoliated bulk Bi into thin bismuthene materials [129]. Hussain et al. used a simple mechanical route to successfully prepare Bi-nanosheets (~ 2 nm) on highly polished silicon substrates by hot pressing [130]. Sun et al. prepared bismuthene nanosheets by LPE. They ground crystalline Bi for 5 h, then dispersed and centrifuged, and finally successfully prepared bismuthene nanosheets with a thickness of about 4 nm [131].

2.2.6 MOFs and covalent organic frameworks (COFs)

MOFs are a class of 2D materials composed of metal ions or metal clusters and organic ligand molecules. Due to their tunable metal centers and functionalizable ligands, MOFs can have diverse basic building blocks, monodisperse metal sites and organic ligands, and thus can form crystal structures with different space groups. At the same time, MOFs also have a periodic arrangement structure, a precisely controlled pore structure, and a high specific surface area. Compared with bulk materials, 2D MOFs nanosheets have more unsaturated coordination bonds and active centers, and have better catalytic performance [132, 133]. Due to the variability of the crystal structure of 2D MOFs, the electronic structure of MOFs can be adjusted by the construction of vacancies, which can cause atomic arrangement distortion, charge density redistribution, and generate a large number of active sites. Zou et al. constructed MOFs with abundant vacancies by pyrolyzing specific functional groups [134]. The results showed that the MOFs maintained the framework structure and more active sites were exposed. The electronic structure is optimized to facilitate the uniform formation of site angles along the defect interface. After fusion of metal phosphides with defective MOFs, they exhibit low overpotentials and long-term oxygen evolution reaction (OER) stability. The excellent electrochemical performance is mainly attributed to vacancies, which enlarge the electrochemically active surface area and accelerate the charge-transfer rate. On the other hand, due to their functionalizable ligands, MOFs can also improve light absorption properties by introducing large conjugated ligands, dual ligands and other ligand engineering methods. Degaga et al. investigated the structure–property relationship of pristine and functionalized zinc benzene-1,3,5-tricarboxylate (Zn-BTC) MOFs using DFT [135]. Figure 9a shows different Zn-BTC MOFs unit-cells constituted from the coordinatively saturated and unsaturated secondary building unit (SBU) and the organic linker BTC for pristine, amino-functionalized (–NH₂), and cyano-functionalized (–CN) MOFs. Density of state (DOS) calculations revealed the variation of VBM and CBM of the Zn-BTC constituted from the coordinatively unsaturated SBU（Zn-BTCU）and their relative relationship with the Fermi level (Fig. 9b). Figure 9c shows the relative positions of VBM and CBM of pristine and functionalized Zn-BTCU MOFs. It is not difficult to see, the –NH₂ can lead to an upward shift of the band edge, and the –CN leads to a shift of the band edge in the opposite direction, which narrows the bandgap and tunes the light absorption properties. This variation is due to the inhomogeneous distribution of N 2p and C 2p states at the band edge introduced by functionalizations.

Fig. 9

Reproduced with permission from Ref. [135]. Copyright 2019, Royal Society of Chemistry

a Multiple structures of Zn-BTC (Zn₃(BTC)₂) MOFs combined with different ligands. b DOS calculation of Zn-BTCU MOFs with unsaturated SBU. c Relative positions of the VBM and CBM of pristine and functionalized Zn-BTCU MOFs. (PDOS: partial density of states)

Notably, not all MOFs are suitable for photocatalytic water splitting. MOFs suitable for photocatalytic decomposition should satisfy three requirements: (1) low bandgap energy and suitable band edges, (2) water-stability, and (3) high conjugation backbone [136]. Meanwhile, the stability of MOFs in practical applications is related to the bonding strength between metal clusters and organic ligands. At present, the mainstream MOFs materials used in the field of photocatalysis are: Zeolitic imidazolate framework -67 (ZIF-67) [137, 138], Materials of Institut Lavoisier -125 (MIL-125) [139, 140], University of Oslo -66 (UIO-66) [141, 142], etc.

Since MOFs have strict requirements on the reaction environment, COFs have been proposed as their alternative materials. COFs are porous crystalline materials composed of organic units composed of light elements (C, N, B, etc.) covalently linked. They have the same low density and high specific surface area and are easy to functionalize as MOFs [143]. The design principle of COFs is to use covalent or non-covalent bonds to integrate structural units to form an extended structure in the internal space of the material. Since covalent bonds can be designed in advance, COFs have the conditions for directional design of pore structures, and various functional groups such as alkyl, phenyl, fullerene, acid, amine, organic groups can be integrated into the pore structure of COFs, thereby realizing rich functional applications [144]. Fuerte-Diez et al. prepared novel covalent triazine frameworks (CTFs) containing phenyl-extended naphthalene units [145]. This metal-free heterogeneous photocatalyst exhibits high chemical stability for selective aerobic oxidation and dehydrogenation cross-coupling reactions of sulfides under visible-light irradiation.

2.2.7 LDHs

LDHs, also known as hydrotalcite-like compounds, are a kind of 2D nanomaterials with a chemical composition of ${{[\mathrm{M}}_{1-x}^{2}{\mathrm{M}}_{x}^{3+}{\left(\mathrm{OH}\right)}_{2}]}^{x+}{({\mathrm{A}}^{n-})}_{x/n}\cdot {\mathrm{mH}}_{2}\mathrm{O}$, where M²⁺ and M³⁺ represent divalent and trivalent metal cations, respectively (such as Mg²⁺ and Al³⁺ and transition metal elements, Fe²⁺, Co²⁺, Ni²⁺, Zn²⁺, Cr³⁺, Fe³⁺, etc.), and Aⁿ⁻ represent interlayer anions(such as NO₃⁻, CO₃²⁻ and SO₄²⁻) [146, 147]. The structures of LDHs are shown in Fig. 10. Due to the diversity of cations and interlayer anions, LDHs have large interlayer spacing, so they are easy to exfoliate, and easy to combine with other materials to achieve specific functional designs. LDHs can be produced naturally or synthesized in the laboratory. At present, the main methods for preparing LDHs include co-precipitation, urea hydrolysis, ion exchange, hydrothermal and electrodeposition. The co-precipitation method is the most important preparation method of LDHs. By adjusting the pH values, temperature, solution concentration and precipitation time during the precipitation process, the control of the nucleation process of LDH materials can be realized [148]. The hydrothermal method is also often used in the preparation of LDHs materials. Compared with the co-precipitation method, the synthesized LDHs have the advantages of large area, good crystallinity, uniform shape and fine particles [149]. At the same time, the LDHs prepared by the hydrothermal method have a narrower bandgap (~ 2.6 eV) and structural advantages, which can be better used in photocatalysis [150].

Fig. 10

Reproduced with permission from Ref. [147]. Copyright 2019, John Wiley and Sons

Schematic diagram of the structure of LDH.

At present, LDHs have shown great potential in the fields of supercapacitors, secondary batteries and catalysis. By regulating the M²⁺ and M³⁺ cations, the energy bandgap of LDHs ranges from 2.0 to 3.4 eV, which can improve the absorption of visible light, and generally does not introduce additional carrier recombination centers. The chemical composition of the inorganic compound layer and the intercalation layer of LDHs is highly tunable, which can effectively control the position of the catalytic active site, the activation mode of the photocatalytic reaction and the product selectivity. With the decrease of the number of LDHs layers, the thickness decreases continuously, which shortens the migration paths of photogenerated electrons and holes, and can effectively inhibit the recombination of photogenerated electrons and holes [151]. Li et al. briefly introduced the influence of the structural features of LDHs on the photocatalytic process and clarified the flexible and tunable properties of LDHs [152]. Based on the structural tunability of LDHs, researchers have carried out many studies on their ion composition. Starukh et al. prepared Zn-Al LDHs [153]. The results show that the photocatalytic activity of Zn-Al-LDHs for MB degradation under UV light originates from the existence of ZnO phase, and the content of ZnO can be adjusted by changing the Zn/Al ratio and heating temperature. Thus, its photocatalytic activity can be regulated. In addition to adjusting the composition of metal ions, Su et al.’s study showed that ZnAl-LDHs with different anions showed different photocatalytic nitrogen reduction activities, which means that the performance of LDHs can also be adjusted by adjusting the interlayer anions [154].

2.2.8 MXenes

MXenes are a new type of ultrathin 2D nanomaterials with great application potential, which contain a large number of transition metal carbides and carbonitrides. MXenes were discovered and synthesized by Gogotsi’s group and Barsoum’s group in 2011 [155]. Unlike graphene and phosphorene, which have natural 3D precursors of graphite and BP, respectively, MXenes have no direct 3D precursors in nature. In contrast, MXenes multilayer flakes are usually produced by selective removal of the A layer in the MAX phase [156]. The specific process is to use selective etching to etch the A atomic layer in the MAX phase to weaken the bonding between the layers, and then peel off the single-layer nanosheets by intercalation peeling or ultrasonic-assisted liquid phase peeling, and stably dispersed in a specific solvent [157,158,159]. Since the precursor of this new ultrathin 2D material comes from the MAX phase of the ternary layered ceramic material and exhibits properties similar to graphene, after unifying the names of the MAX phase and graphene, the newly discovered materials are named MXenes [160]. The general formula of the MAX phase is M_n+1AX_n, where “M” refers to an early transition metal, “A” refers to a main group element, “X” means C or N, and n = 1–4. Compared with the single-atom structure of graphene and the metallic bond between C–C, the M and X atoms of MXenes, and the covalent bond between M and X, make MXenes more stable and more diverse than graphene [161]. The general formula of MXenes after removing the A layer is M_n+1X_nT_x, where T represents the surface functional groups (–O, –OH and –F), and x is the number of surface functional groups. Studies have shown that surface functional groups are related to the electronic state of MXenes. MXenes without functional groups exhibit metallic properties, while it with functional groups exhibit semiconducting properties, making it more widely used [162].

The most widely used preparation method for MXenes is wet chemical etching, and fluorine-containing compounds such as hydrofluoric acid were first used for etching. Later, with the research on fluorine-free etching methods, electrochemical etching methods, alkali etching methods, and molten salt etching methods were also proposed one after another. Generally speaking, the MXenes obtained by etching have a stacked accordion morphology (Fig. 11a), and further intercalation and exfoliation are required to obtain single-layer MXenes [163]. At present, the main peeling methods are ultrasonic exfoliation method, mechanical exfoliation method, intercalation exfoliation method and so on. Wei et al. reviewed the etching methods and layering methods of MXenes, and discussed their advantages and disadvantages [164].

Fig. 11

Reproduced with permission from Ref. [163]. Copyright 2022, ACS Publications. b Structural diagrams of the four kinds of MXenes. Reproduced with permission from Ref. [165]. Copyright 2022, Elsevier. c The charge-transfer process in the Ti₃C₂/TiO₂ system. Reproduced with permission from Ref. [166]. Copyright 2020, Elsevier. d Schematic diagram of the reaction mechanism of 2D-Bi₂MoO₆@2D-Mxene. (NSs: nanosheets) Reproduced with permission from Ref. [169]. Copyright 2020 Elsevier

a Field emission scanning electron microcopy images of bulk Ti₃AlC₂ and Ti₃AlC₂ after HF treatment.

Since its discovery in 2011, researchers have predicted more than 100 possible MXenes compositions through theoretical calculations, of which more than 40 MXenes have been successfully prepared experimentally. As shown in Fig. 11b, in addition to the most studied M₃X₂T_x system, three typical MXenes systems have also been synthesized: M₂XT_x, M₄X₃T_x, and M₅X₄T_x. Among them, the research based on Ti₃C₂ is the most extensive [165]. The 2D planar topology endows MXenes nanosheets with a large surface area, and its layered structure makes it easy to carry other materials to form nanocomposite photocatalysts, enhance light absorption, and obtain better photocatalytic performance. Since the excellent synergy between TiO₂ and MXenes can effectively improve the electrical conductivity, electron mobility, and carrier separation efficiency of the system, the TiO₂/MXenes system has been extensively studied. Chen et al. synthesized Ti₃C₂/TiO₂ nanosheets with HF as auxiliary agent [166]. The Schottky barrier at the interface of Ti₃C₂/TiO₂ effectively hinders the electron backflow of TiO₂ (Fig. 11c). At the same time, the (001) crystal face of TiO₂ with higher catalytic activity is exposed by using HF as an auxiliary agent, which improves the photocatalytic activity of Ti₃C₂/TiO₂ nanosheets. Ti₃C₂/TiO₂ composite photocatalysts can also be successfully prepared by in-situ oxidation or self-assembly [167, 168]. In addition to composites with TiO₂, MXenes materials have also been extensively studied in other composite systems. Zuo et al. prepared layered 2D-Bi₂MoO₆@2D-MXene nanocomposites [169]. The photocatalytic activity of the composites for water splitting is 7.9 times higher than that of pristine Bi₂MoO₆. The analysis showed that there are two reasons for the enhanced photocatalytic activity. On the one hand, the composite structure inhibits the agglomeration of Bi₂MoO₆ nanosheets and increases its specific surface area, and on the other hand, the formed Schottky heterojunction can promote charge transport and hinder carrier recombination (Fig. 11d). Liu et al. successfully prepared a 2D/2D Schottky junction of Ti₃C₂/g-C₃N₄ by ultrasonic irradiation, which accelerated the charge separation [170]. g-C₃N₄ loaded with Ti₃C₂ exhibited excellent photocatalytic activity in nitrogen fixation, CO₂ reduction and degradation.

3 Main factors affecting photocatalysis

In general, photocatalytic material properties and photocatalytic reaction conditions are the two main factors affecting the photocatalytic performance. This paper mainly discusses the influencing factors of the material itself, such as energy band structure, morphology, and crystal face, and further explains the unique influence of these influencing factors in the 2D nanomaterial system.

3.1 Band structure of photocatalysts

In the photocatalytic process, the core reaction process is: when light is irradiated on the photocatalyst, the catalyst absorbs light energy to generate electrons (e⁻) with reducing ability and holes (h⁺) with oxidative ability. e⁻ and h⁺ further react with the system to generate free radicals, which reduce or oxidize substances in the environment such as H₂O, pollutants, heavy metal ions, and CO₂. This process is closely related to the energy band structure of the photocatalyst. On the one hand, the positions of the semiconductor photocatalysts CB and VB determine their redox capacity, that is, the more negative CB, the stronger the reduction capacity; the more positive VB, the stronger the oxidation capacity. In other words, the wider the forbidden band of the semiconductor photocatalyst, the stronger is its redox capacity [171]. On the other hand, the photoresponse range of semiconductor photocatalysts is also related to its forbidden bandwidth, as shown in Eq. (1):

$$E_{{\text{g}}} = {124}0/\lambda \, \left( {{\text{eV}}} \right)$$

(1)

E_g is the forbidden bandwidth of the semiconductor, and λ is the wavelength. The narrower the forbidden bandwidth, the wider the light absorption range, and the easier it is to be excited by light. According to this equation, the visible-light response range of the semiconductor photocatalytic material can be roughly estimated. The wavelength of 400–700 nm is the visible range. According to Eq. (1), the semiconductor photocatalyst that can respond to visible light has a bandgap range of about 1.77–3.1 eV. Therefore, in order to better respond to the visible-light range and improve the practical application potential of the photocatalyst, the bandgap of the photocatalyst should be selected in 3.1 eV or less.

Hydrogen production from water splitting is one of the main applications of photocatalytic technology. The water splitting process is closely related to the energy band of the photocatalyst. Figure 12 is a schematic diagram of the energy of photocatalytic water splitting when pH = 0. Protons are reduced by e⁻ in CB to generate hydrogen, while h⁺ left in VB oxidizes water molecules to oxygen. In order to realize the whole water splitting reaction, two points must be satisfied: (i) the CBM is more negative than the reduction potential (0 eV) of H⁺ to H₂, (ii) the VBM is larger than the oxidation potential of H₂O to O₂ (+ 1.23 eV). To satisfy the above two conditions, the minimum theoretical bandgap for driving water splitting is 1.23 eV. In theory, the bandgap of the semiconductor photocatalyst should be designed between 1.23 and 3.1 eV to ensure visible-light activity while overcoming the energy consumption of the water splitting reaction. In fact, some additional kinetic overpotentials are likely to be required to drive the electron transfer process in practical applications, and hydrogen evolution reaction (HER) and OER at reasonable reaction rates, semiconductors with bandgaps less than 1.6 eV will be limited by it [97]. At present, common semiconductors used for photocatalytic hydrogen production include TiO₂, ZnO, transition metal compounds (such as ZrO₂, CdS, and WO₃) and composite oxides with a layered perovskite structure. Since the energy band of 2D materials is very sensitive to external effects, it is an important research direction for photocatalysts to peel off traditional bulk semiconductor materials to prepare 2D materials.

Fig. 12

Schematic diagram of the energy of photocatalytic water splitting

Starting from the basic principles of photocatalysis, it can be known that the regulation of the bandgap is an extremely important modification method in photocatalysis. Many influencing factors and modification methods in photocatalysis basically change the bandgap of semiconductor catalysts, thereby changing the photocatalytic performance. Based on the semiconductor properties, its photocatalytic performance can be altered by tuning the VB, CB, and simultaneous tuning of the two. The methods for adjusting the bandgap of semiconductor materials include: thickness adjustment, element doping, heterojunction, external electric field, and strain engineering. Among them, for 2D materials with a layered structure, thickness adjustment is a very important mean of regulation, and many studies have proved that the bandgap change is strongly correlated with the thickness. We will go into more details in Sect. 4.

3.2 Catalyst morphology

The effect of material morphology on its properties is significant. In addition to the inherent morphology of the material, the catalysts can also design hollow structures [172], core–shell structures [173] and interlayer layered structures [174], showing flower-like [175], spherical [176] and other morphologies. The acquisition of different morphologies depends on their preparation methods, reaction times, and temperatures. Lu et al. synthesized CeO₂ materials with different morphologies by hydrothermal method and template method and have good applications in biological sterilization [177]. Figure 13a shows the transmission electron microscope (TEM) images of CeO₂ nanorods (NRs), nanocubes (NCs), nanosheets (NSs), hollow spheres (HSs), and mesoporous CeO₂ (m-CeO₂). Based on the area of the X-ray photoelectron spectroscopy (XPS) pattern, it is calculated that the CeO₂ nanorods have a higher concentration of Ce³⁺ and oxygen vacancies (Fig. 13b–d). Benefiting from this feature, CeO₂ nanorods have the most excellent tetracycline (TC) degradation performance (Fig. 13e, f). Free radical trapping experiments reveal that h⁺ and superoxide radicals (·O₂⁻) play a major role in the degradation of TC (Fig. 13g). Liu et al. prepared TiO₂ with different morphologies (porous structure, flower-like nanospheres, and granular) by solvothermal method [178]. TiO₂ with a porous structure exhibits better catalytic activity due to the unsaturated coordinated Ti atoms and the highest concentration of oxygen vacancies, as well as the good adsorption properties brought about by the high specific surface area and low pore size. The above experiments show that it is necessary to study the different morphologies of the same material. The difference in morphology will create different surface states, showing different surface defects, adsorption properties, specific surface area, etc.

Fig. 13

Reproduced with permission from Ref. [177]. Copyright 2023, Elsevier

a The TEM images of CeO₂ with different morphology. XPS spectra of synthesized CeO₂ samples, b Ce 3d, c O1s. d The Ce³⁺/O_ads concentrations of CeO₂ with different morphologies. e The reaction kinetic curves of TC degraded by CeO₂ sample. f The kinetic rate constant TC degraded by CeO₂ sample. g The free radical trapping experiments of CeO₂ NRs. (BQ: benzoquinone. AO: ammonium oxalate. IPA: isopropanol)

Due to the flexible and adjustable structure of 2D materials, the design of their morphology is easier than that of bulk materials. The commonly used methods are hydrothermal method and template method. The hydrothermal method can effectively prepare photocatalysts with different morphologies by changing the reaction conditions and reagents. The template method can design a more complicated specific shape through template induction.

When discussing the difference in morphology of 2D materials, the discussion of specific surface area is necessary. Generally speaking, the larger the specific surface area, the better the adsorption performance of the photocatalyst and the more active sites, so the catalytic performance is also better. In particular, the inherent layered structure of 2D nanomaterials can naturally lead to a large specific surface area, so that most of the active centers are exposed on the surface [179]. Du et al. prepared nanomesh B/O co-doped graphitic carbon nitride (BO-C₃N₄ nanomesh) [180]. The hydrogen production rate of BO-C₃N₄ nanomesh under visible-light irradiation is nearly 28 times higher than that of bulk g-C₃N₄. This is due to the unique 2D porous nanomesh structure, which enables BO-C₃N₄ to obtain a surface area of up to 160.58 m²·g⁻¹. Liu et al. constructed a porous SnO₂ precursor with obvious photocatalytic activity [181]. Then, using In₂O₃ as a sensitizer to increase the specific surface area of the porous SnO₂ matrix, the specific surface area of the porous SnO₂ was increased by 3 times, and the photocatalytic performance was 1.94 times that of the original porous SnO₂.

However, some scholars hold different views. Some studies have pointed out that the photocatalytic activity of the material may not be related to the specific surface area. Li et al. synthesized graphitic carbon nitride with low specific surface area by silica template method, and its photocatalytic activity was improved compared with the high specific surface area graphitic carbon nitride synthesized by traditional method [182]. Studies have shown that this is because the synthesized low specific surface area graphitic carbon nitride has more free electrons and a lower recombination rate of carriers. In contrast, the size of the specific surface area is not the main factor affecting the photocatalytic performance.

The above studies show that it is a feasible method to improve the photocatalytic performance by changing the morphology of the catalyst and increasing its specific surface area. However, it cannot be simply assumed that the specific surface area of the material must be positively correlated with the improvement of the photocatalytic performance. The changes in the catalytic process at the microscopic level caused by the specific surface area should be further explored, which in turn guides the design of catalyst morphology and structure.

3.3 The structure of crystal face

Photocatalytic reactions generally occur on the surface of photocatalysts, so the reaction process is closely related to the surface atomic structure [183]. Since different crystal faces of photocatalysts naturally have different atomic arrangements, they can further affect key steps in photocatalytic reactions such as adsorption, desorption, and electron transfer processes. The research on TiO₂ single crystal surface regulation published by Yang et al. in 2008 is the most representative work on semiconductor crystal surface regulation [184]. Its research found that the concentration of F ions can precisely control the exposure ratio of (001) crystal planes, significantly reduce the surface activation energy of (001) crystal planes, and promote the growth of (001) crystal planes. This work demonstrates the feasibility of finely tuned crystal planes. Subsequent studies have shown that by adjusting the concentration of specific ions, changing the pH values, adding surfactants, etc., specific crystal faces can be effectively exposed to further improve the photocatalytic performance of the system. Nowadays, facet engineering has become a common means of photocatalyst regulation. At present, the active crystal planes of some typical 2D materials have been identified, such as Li et al. synthesized WO₃ thin films with different exposure ratio (002) planes by a simple hydrothermal method [185]. The proportion of each crystal plane of different WO₃ samples was calculated by X-ray diffraction (XRD) spectrum (Fig. 14a), and the layer spacing corresponding to (002) crystal plane was also observed in TEM spectrum (Fig. 14b). Through a series of characterizations such as current density–voltage curve (J–V curve) (Fig. 14c), electrochemical impedance spectroscopy (EIS) (Fig. 14d) and Mott–Schottky plots (M–S plots) (Fig. 14e), it was found that with the increase of the (002) plane exposure rate, the separation of photogenerated carriers was promoted and the photoelectrochemical (PEC) performance was improved. Hu et al. prepared MoS₂ nanosheets with higher activity (100) surface concentration at lower pH values, and the photocatalytic performance was effectively improved [186]. CdS with exposed (002) crystal faces also showed better performance in catalyzing hydrogen evolution [187]. The regulation of specific crystal planes can also control the thickness of 2D materials. Guan et al. used polyvinyl pyrrolidone (PVP) as an active agent to synthesize ultrathin BiOCl nanosheets with a thickness of 2.7 nm, while the thickness of BiOCl nanosheets without PVP was 30 nm [188]. The generation of ultrathin nanosheets is due to the deposition of PVP on the bottom and top (001) planes, preventing the growth of the longitudinal planes (Fig. 14f). TEM images and atomic force microscopy (AFM) revealed its crystal planes and thickness (Fig. 14g, h). Ultrathin BiOCl nanosheets show superior RhB degradation activity (Fig. 14i). Figure 14j, k reveals the distribution of defects in the samples, triple Bi³⁺ oxygen vacancy dominates in ultrathin BiOCl nanosheets, while isolated Bi vacancies dominates in BiOCl nanoplates.

Fig. 14

Reproduced with permission from Ref. [185]. Copyright 2023, Elsevier. f Schematic illustration of the crystal orientation of the nanosheet. g The high resolution transmission electron microscopy (HRTEM) images of the ultrathin BiOCl nanosheet. h The atomic force microscopic image of the ultrathin BiOCl nanosheet. i The RhB degradation curve. j The positron annihilation spectroscopy. k The positron lifetime spectrum. Reproduced with permission from Ref. [188]. Copyright 2013, ACS Publications

a XRD patterns and calculated crystal facet ratios of different WO₃ samples. b The TEM image of WO₃-S4. c The J–V curve of different WO₃ samples. d The EIS of different WO₃ samples. e The M–S plots of different WO₃ samples.

Compared with traditional materials, 2D materials require more complex crystal facet engineering due to their 2D anisotropic growth and thermodynamic factors. In general, there are two factors influencing the formation of specific crystal planes: thermodynamic and kinetic driving forces. From a thermodynamic point of view, the material facets exposed to the lowest surface energy will eventually be preserved. Kinetically, the slowest growing facets will eventually be preserved. Therefore, how to find a balance between the minimum energy exposure surface and the minimum total area is the challenge that the crystal plane engineering of 2D materials needs to face. In other words, researchers need to think about how to expose specific crystal faces while maintaining the ultrathin atomic thickness unique to 2D materials [189].

3.4 Crystal structure

The same material has different crystal structures due to different atomic arrangements. Different crystal structures have different forbidden bandwidths, so they will show different catalytic properties. The choice of crystal structure of photocatalytic materials is also an important factor to improve their catalytic properties. Taking TiO₂ as an example, TiO₂ has three crystal structures: anatase, rutile and brookite. Among them, the most common structures are anatase and rutile. The forbidden bandwidths of anatase-type TiO₂ and rutile-type TiO₂ are 3.2 eV and 3.0 eV, respectively, so anatase-type TiO₂ has stronger redox ability. At the same time, anatase TiO₂ has a larger specific surface area, more lattice defects, and higher separation efficiency of electron–hole pairs, thus possessing stronger photocatalytic performance [190]. MoS₂ is a typical material with different crystal structures leading to different photocatalytic properties. 1T-MoS₂ brings better photocatalytic performance than 2H-MoS₂ due to the significantly lower charge-transfer resistance. Meanwhile, the active sites of 1T-MoS₂ are distributed in a large number in the plane, rather than in the edge zone as in 2H-MoS₂. However, since the 1T phase is not stable enough in nature, it will spontaneously transform into the 2H phase. How to utilize the 1T phase with better catalytic performance is one of the research focuses of MoS₂ catalytic application. At present, intercalation of alkali metal ions, heteroatom doping and laser beam irradiation are considered to be effective methods for phase modulation [191]. Among them, the hydrazine hydrate-induced phase transition method is a simple and effective method. Li et al. regulated and synthesized a stable 1T-MoS₂ phase by hydrazine hydrate without obvious transformation within 3 months, and its conductivity was 700 times that of 2H-MoS₂ [192]. Jia et al. easily adjusted the ratio of 1T phase to 2H phase of MoS₂ by simply adjusting the amount of hydrazine hydrate, and the obtained MoS₂ had a small Tafel slope with 56.6 mV·dec⁻¹ [101]. However, the current research on the photocatalytic performance of 1T-MoS₂ has not yet obtained satisfactory results. 1T-MoS₂ is usually used as a co-catalyst in photocatalysis to enhance the photocatalytic performance, which benefits from its metallicity. Liu et al. prepared a 1T/2H-MoS₂ mixed phase loaded TiO₂ nanorod system by a hydrothermal method [193]. The system exhibited excellent performance in the photocatalytic hydrogen evolution process. Mixed phase MoS₂ exhibits better co-catalyst performance than Pt. Through Perdew–Burke–Ernzerhof (PBE) calculation (Fig. 15), it can be seen that due to the embedding of 1T-MoS₂ layer, the mixed phase MoS₂ exhibits conductor characteristics, which is helpful for the catalytic hydrogen evolution reaction on TiO₂.

Fig. 15

Reproduced with permission from Ref. [193]. Copyright 2019, Elsevier

PBE calculated band structures of a 2H-MoS₂, b 1T-MoS₂, and c mixed phase MoS₂.

For 2D nanomaterial photocatalysts, the crystal structure is also a significant influencing factor in their photocatalytic applications. Due to the strong correlation between the geometric properties and performance of 2D nanomaterials, by controlling the thickness of 2D nanomaterials, the crystal structure can be adjusted, so that different bandgap widths, electronic properties, etc. can be obtained on the same material. Huang et al. selectively destroyed the hydrogen bonds between g-C₃N₄ layers to adjust the crystal structure of g-C₃N₄, and obtained layer plane ordered porous carbon nitride (LOP-CN) [194]. The photocatalytic hydrogen production performance of LOP-CN is 7.4 times that of pristine g-C₃N₄. Studies have shown that the improvement of photocatalytic performance is due to the long-range atomic order layer plane brought by the new crystal structure, which improves the diffusion length of electrons in the plane, thereby enhancing the charge separation ability. Meanwhile, the broken interlayer hydrogen bonds lead to the formation of a large number of slit holes, providing more active sites.

3.5 Catalyst defect

Defects in materials can generally be divided into intrinsic defects, electron defects and hole defects, and the intrinsic defects are mainly caused by the deviation of the stoichiometric ratio of the material itself. Material defects can affect photocatalysts in several ways. On the one hand, it affects the band structure of the catalyst. Defects will introduce new energy levels in the energy band structure of semiconductor photocatalysts, thereby shortening the forbidden band width and enhancing the absorption ability of visible light. On the other hand, due to the existence of defects, the adsorption of H₂O and O₂ is enhanced, which can further generate a large number of active free radicals, thereby improving the photocatalytic performance. At the same time, the defects can serve as electron-trapping centers, which are beneficial to the separation of electron–hole pairs and improve the photocatalytic performance [195]. In conclusion, the purposeful introduction of defects can enhance the performance of photocatalysts. Due to the reduction of its thickness, the local atomic structure of 2D nanomaterials is significantly different from that of bulk materials, such as coordination number, bond length, bond angle, and surface atomic disorder, so it is easy to form more surface defects [196]. Therefore, for 2D nanomaterials, it is necessary to study the effect of surface defects on the photocatalytic performance. As an effective modification method for photocatalysts, defect engineering has been extensively studied, and the details will be further introduced in Sect. 4.5 of this paper.

4 Main modification methods and their mechanisms

4.1 Elemental doping

Most of the semiconductor photocatalysts are n-type semiconductors, and they all have a special energy band structure different from that of metals or insulating substances: there is a forbidden band between the VB and the CB. The relationship between the light absorption wavelength range of the semiconductor and the bandgap can be expressed by the above-mentioned Eq. (1)

The absorption wavelength range of commonly used wide-bandgap semiconductors are mostly in the ultraviolet region. For example, TiO₂ is a typical material that can only respond to ultraviolet light due to the wide bandgap. Since light absorption is an important process in photocatalytic reactions, the widening of the visible-light response range is of great significance for the practical application of photocatalytic technology. By shortening the bandgap, the photoresponse range can be effectively broadened. Studies have shown that by doping metal or non-metal elements, the bandgap of the photocatalyst can be effectively adjusted, its energy band structure can be changed, and the photoresponse range of the catalyst can be broadened [197]. At the same time, the introduction of impurity elements can also effectively enhance the concentration of carriers. As effective electron acceptors, metal ions can become photogenerated electron capture sites and hinder the recombination of electron holes [198]. In conclusion, doping can improve the photocatalytic performance of materials in multiple dimensions. Next, the mechanism of metal element doping and non-metal element doping will be introduced in detail to understand the role of doping in bandgap regulation.

The doping of metal elements has been studied for a long time and is more mature. The main principle is to change the light absorption range by the introduction of impurity energy levels. Figure 16a, b is schematic diagrams of metal element doping. The introduction of donor level and acceptor level enables electrons not only to be excited from the VB to the CB, but also to transition from donor level to the CB, or from the VB to acceptor level. The change in the forbidden bandgap allows semiconductors to absorb visible light. The type of element and the position of doping are the factors that affect the catalytic activity. Common metal atom dopings include Fe [199], Ni [200], Mn [201], Ag [202], Au [203], and Pt [204]. Tonda et al. prepared Fe-doped g-C₃N₄ nanosheets, which had a 7 times higher degradation rate of RhB than bulk g-C₃N₄ [205]. Shah et al. prepared Mn-doped BiFeO₃ nanofilms (Mn-BFO) [206]. The oxygen evolution activity of Mn-BFO under visible-light irradiation is significantly improved compared to pure BFO. Bandgap studies show that the bandgap of Mn-BFO can be tuned from 2.1 to 1.36 eV by changing the doping amount of Mn, thereby broadening its optical response range. Qin et al. prepared Ag-doped g-C₃N₄ [207]. Under visible-light irradiation, Ag-g-C₃N₄ exhibits good photocatalytic hydrogen evolution activity, and the highest hydrogen evolution rate is 3.62 times that of the pure phase.

Fig. 16

a Doping of metal elements to introduce the donor energy level. b Doping of metal elements to introduce the acceptor energy level. c Doping with non-metallic elements shortens the energy band

The doping of non-metallic elements can also change the energy band structure, but unlike metallic elements, it does not introduce new energy levels, but shortens the energy band. The specific process is shown in Fig. 16c. Common non-metallic elements used for doping are N, B, O, P, S, etc. The doping of non-metal elements compared to metallic elements can well avoid the effect of thermochemical instability on the photocatalytic activity. Thus, doping method with non-metal elements is more widely used. For example, the reason for the better photocatalytic performance of N-doped TiO₂ can be simply explained as follows: N-doping forms a state in the bandgap that can absorb visible light, and N-doping induces oxygen vacancies, thereby enhancing the photocatalytic performance [208]. Wang et al. prepared S-doped g-C₃N₄ (Fig. 17a) [209]. By doping element S, the bandgap of g-C₃N₄ is narrowed from 2.7 to 2.63 eV (Fig. 17b), the photoresponse range is broadened to 475 nm, and the photocatalytic reduction of CO₂ shows more outstanding performance. Kong et al. prepared N-doped ZnO/g-C₃N₄ composites [210]. Compared with pure g-C₃N₄, ZnO and composite ZnO/g-C₃N₄, N-ZnO/g-C₃N₄ degrades MB and phenol better under visible-light irradiation. The studies showed that one of the reasons for the improvement of photocatalytic performance is the change of the energy band structure and the improvement in stability brought by doping N.

Fig. 17

Reproduced with permission from Ref. [209]. Copyright 2015, Elsevier. c The Tauc plots of different samples. d Photo-induced temperature rise rates of different samples. e The electron paramagnetic resonance (EPR) spectra of 2,2,6,6-tetramethylpiperidinyl (TEMP-¹O₂) for Co₁Zn₁₉-ZIF. f The EPR spectra of 5,5-dimethyl-1-pyrroline N-oxide (DMPO-O₂^-) for Co₁Zn₁₉-ZIF. Reproduced with permission from Ref. [214]. Copyright 2023, Elsevier

a Unit cell of pure C₃N₄ and sulfur-doped C₃N₄. b Schematic diagram of the band structures of pure g-C₃N₄ (left) and S-doped g-C₃N₄ (right). (BCB: bottom of CB. TVB: top of VB)

In summary, it can be seen that element doping is an effective mean to narrow the bandgap of photocatalysts, and the narrowed bandgap can broaden the visible-light absorption range. In addition to metal doping and non-metal doping, co-doping is also widely used as a doping method, and the synergistic effect between two different elements also has a profound impact on the performance of semiconductor catalysts [203, 211]. Currently, there is no research on controllable bandgap narrowing, and the bandgap change caused by element doping is still an unquantifiable thing.

Usually, doping is achieved by introducing dopants containing corresponding elements during the preparation process or post-treatment process of the semiconductor material precursor. Dopants can be liquid, solid or gaseous. Taking N element doping as an example, the common doping method is hydrothermal method using ammonia solution (such as formamide and ammonia water), calcination of precursors containing nitrogen elements, or nitriding materials in NH₃ atmosphere. The element type, element position, doping concentration and doping solubility are the main influencing factors in the doping process. The doping solubility refers to the solubility of the dopant in the host material. In general, solubility is limited due to interactions between elements. Even the maximum solubility sometimes fails to achieve effective photocatalytic performance. In addition to selecting elements with higher solubility, appropriate post-treatments such as annealing, pressurization, and thermal cycling can also increase doping solubility [212, 213]. However, the current solubility enhancement methods have problems such as cumbersome process and unclear mechanism, which need to be further explored and discussed. It is worth looking forward to that, compared with bulk surface doping, 2D nanomaterials have the characteristic of thin thickness. The doping process of 2D materials requires only a small diffusion depth to form uniform doping, which is more conducive to adjusting the electronic structure, light absorption and improving photocatalytic activity, and can significantly change its physicochemical properties. In general, doping alone cannot effectively improve the visible-light response, and the defects caused by doping may also become the recombination sites of electrons and holes, which has a negative effect on the photocatalytic performance. The direction to solve this problem is to explore the mechanism of doping more deeply through the combination of various characterization methods and computational simulations.

The study of element types and doping concentration in the doping process is very necessary to explore the catalytic mechanism of the system. Li et al. studied the doping of ZIF-8 with different concentrations of different metal ions (Co²⁺, Ni²⁺, Cu²⁺) [214]. All doped ZIFs have narrowed bandgap and enhanced light-harvesting and photogenerated carrier separation efficiencies (Fig. 17c). As the concentration increases, the photothermal conversion efficiency gradually increases, which is conducive to its application in sterilization (Fig. 17d). The introduction of Co sites makes the photoactive oxygen change from a single oxygen state (¹O₂) to multiple oxygen states (¹O₂ and ·O₂⁻), which can better participate in the sterilization. Therefore, Co²⁺ with a doping concentration of 5% has the highest bactericidal efficiency on Escherichia coli (Fig. 17e, f).

In addition to the study of bandgap changes and surface active sites, the researchers also conducted an in-depth study of the carrier dynamics through transient absorption spectroscopy, revealing the role of doping in the carrier dynamics. For example, Zhao et al. prepared 0.6% Re-doped monolayer MoS₂ [215]. Using transient absorption spectroscopy, it is characterized that the average exciton lifetime of undoped MoS₂ is 22 ps, while the exciton lifetime of Re-MoS₂ is as low as 1 ps. This work reveals the effect of Re impurities on the catalytic performance of native MoS₂ in terms of carrier dynamics. It is worth noting that the excitons in the 2D material system are more stable than other materials, which is more conducive to the in-depth exploration of optoelectronic technology. The dark excitons that cannot be detected because they do not interact with light can be observed in 2D material systems [216].

4.2 Noble metal deposition

Studies have shown that by depositing noble metals on semiconductor photocatalysts, a “junction”, the Schottky junction, can be constructed between the semiconductor photocatalyst and the noble metal, thereby controlling the directional flow of electrons to improve the separation efficiency of carriers [217, 218]. In addition, the local surface plasmon resonance (LSPR) effect of noble metals can also effectively promote the absorption of light by semiconductor photocatalysts. Its mechanism is shown in Fig. 18a. Metal nanostructures with LSPR effect, when the oscillation frequency of electrons is consistent with the frequency of incident light waves, a strong resonant absorption peak will appear [219]. Hot electrons are generated during the decay of the oscillation, and the injection of hot electrons into the CB of adjacent semiconductors will contribute to the improvement of photocatalytic performance. Figure 18b reveals the injection mechanism of hot electrons in different cases [220]. In order to explore how the LSPR effect acts on the photocatalytic process and the formation of the Schottky junction, the researchers carried out many experimental investigations. Fang et al. prepared Ag nanoparticles (NPs)-doped BiVO₄ photonic crystal inverse opal (Ag/io-BiVO₄) [221]. The photocurrent density of Ag/io-BiVO₄ is 4.7 times higher than that of the disordered samples without Ag NPs. In the degradation of MB, Ag/io-BiVO₄ also showed the most excellent activity. With the in-depth exploration of the photocatalytic mechanism, compared with d-BiVO₄, the enhanced photocatalytic activity of Ag/io-BiVO₄ can be explained from the following three aspects. First, due to the slow light effect, the interaction between io-BiVO₄ and light-matter is enhanced, and the absorption intensity of light is higher than that of d-BiVO₄ (Fig. 18c). Second, the absorption range of the electronic bandgap of BiVO₄ coincides with the LSPR wavelength of Ag NPs (Fig. 18d), which facilitates the generation of carriers. The incident monochromatic photon-electron conversion efficiency test (IPCE test) also confirmed this (Fig. 18e). Finally, the Schottky junction constructed between Ag and io-BiVO₄ realizes the effective separation of carriers. Its formation mechanism is shown in Fig. 18f. When Ag NPs are in contact with BiVO₄, free electrons are redistributed on the interface until the Fermi level reaches an equilibrium state and cause upward bending of the semiconductor band edge, so a Schottky barrier is formed between Ag and BiVO₄. Ag acts as an electron trap, which helps the effective separation of electron–hole pairs. At present, Ag, Au, Pd, Pt, etc. are commonly used in noble metal deposition. The modification method of noble metal deposition is also widely applicable to a variety of material systems [222,223,224]. Same as doping, the photocatalytic activity of the material is related to the type of precious metal, the amount of deposition and the preparation method. Excessive deposited noble metals may also become the recombination center of carriers, hindering the separation of photogenerated carriers, thereby reducing the photocatalytic activity.

Fig. 18

Reproduced with permission from Ref. [219]. Copyright 2014, ACS Publications. b Schematic illustration of the hot-electron-injection effect: excitation of the electrons from thermal equilibrium to a high-energy state upon absorbing photons and injection of the electrons to the CB of the semiconductor (I). Redistribution of electron energy and formation of a Fermi–Dirac distribution at a high-temperature Fermi level after the injection of the electrons (II). Restoration of the standard electron distribution with electrons and holes flowing to different regions in the semiconductor (III). Reproduced with permission from Ref. [220]. Copyright 2015, John Wiley and Sons. c The absorption spectra of different BiVO₄ samples. d The absorption spectrum of the Ag NPs and the reflectance spectrum of the io-BiVO₄ sample. e The IPCE spectra of different BiVO₄ samples. f The schematic diagram of Schottky junction of Ag/BiVO₄. Reproduced with permission from Ref. [221]. Copyright 2016, AIP Publishing. (FTO: F-doped tin oxide) g The HRTEM images of different Ag/TiO₂ samples. h The schematic diagram of the formation mechanism of Ag NPs different sizes. Reproduced with permission from Ref. [226]. Copyright 2020, Royal Society of Chemistry

a Schematic diagram of LSPR.

In general, the deposition of noble metals can be achieved by wet chemical method, physical vapor deposition, photodeposition, and atomic layer deposition (ALD). Among them, the wet chemical method is easy to operate. Noble metals are deposited by deposition, wet impregnation, drying, heat treatment, etc. In order to avoid the occurrence of agglomeration during the deposition process, low concentrations of metal precursors should usually be selected. In alkaline reagents, due to the negative charge on the surface of metal oxides, noble metal cations can be attracted and deposited by electrostatic interaction, so this method is very suitable for the deposition of noble metals on the surface of metal oxides [225]. Photodeposition is also a common method of noble metal deposition. Its basic process is to use photogenerated electrons and holes to reduce the noble metal ions adsorbed on the semiconductor surface, so as to achieve noble metal deposition at specific active sites. It is a method that can accurately design noble metal deposition sites [226]. ALD technology is a new vapor deposition method. The basic process is to expose the base material to the vapor phase metal precursor and deposit it layer by layer. The advantage of this method is that the composition and thickness of the noble metal atomic film can be controlled to the atomic level, and it is a potential preparation method for exploring the deposition of single-atom noble metals and improving the utilization rate of noble metals. However, due to the high vacuum environment and expensive preparation price, ALD technology cannot be put into mass production [227]. During noble metal deposition, proper noble metal size and distribution also play a key role in the photocatalytic performance. Since noble metals usually have a more negative Fermi level than host materials, their growth process is fast, and they often have large sizes or uneven distribution during deposition process. Huang et al. changed the surface state of TiO₂ by adjusting the pH values, and studied its effect on the deposition of noble metals by Photodeposition [226]. When the pH values was low, the size of the deposited Ag Nps was larger than 10 nm, and when the pH values was 9, the small size Ag Nps of 0.5–2.0 nm appeared (Fig. 18g). As Fig. 18h shown, the formation of small-sized Ag NPs may be due to that some of photoelectrons are trapped by –OH under alkaline conditions during the electron transition to CB. Both photoelectrons located at CB and trapped by –OH can perform reductive deposition of Ag ions. Since the photoelectrons trapped by –OH have smaller energy and area than those on CB, smaller Ag NPs are generated. TiO₂ deposited by large-sized Ag NPs showed better photocatalytic hydrogen evolution activity. In summary, the size and distribution of noble metals are related to the nucleation and growth process regulated by the surface state and charge, and the size has a significant impact on the photocatalytic performance.

Strictly speaking, the Schottky junction formed by the recombination of noble metal and semiconductor is essentially a kind of heterojunction. Compared with the heterojunction constructed by semiconductors and semiconductors, it has obvious rectification properties, which can adjust the current shape, form a Schottky barrier at the interface, leave the negative charges on the metal, and separate the positive charges in the semiconductor, so as to effectively separate the carriers. At the same time, due to the metal-specific LSPR effect, when visible light irradiates the metal surface, the free electrons on the metal surface are resonated by the light wave, which will enhance the oscillating electric field around it, causing the electrons to be easily excited, thereby improving the catalysis efficiency. Due to the above characteristics, noble metal deposition has great research significance in regulating the photocatalytic performance of semiconductor catalysts.

4.3 Heterojunction

In addition to the noble metal deposition method mentioned above, which can effectively solve the problem of the fast electron–hole recombination speed of 2D nanomaterials, the construction of heterojunctions is also one of the most widely used strategies to solve this problem. As shown in Fig. 19a, heterojunctions can be basically divided into three categories. In type I, the CB and VB of semiconductor B are higher than A, holes in the VB of B can be transferred to the VB of A, and electrons in the CB can be transferred to the CB of A. But such heterojunctions did not promote separation of photogenerated carriers. There are two different mechanistic explanations for type II heterojunctions. One view is that holes in the VB of A can be transferred to the VB of B, and electrons in the CB of B can be transferred to the CB of A. Electrons are enriched in the CB of A, while holes are enriched in the VB of B, so that electron–hole pairs can be effectively separated. Another point of view is that the electrons in the CB of A may directly recombine with the holes in the VB of B, leaving the holes in the VB of A and the electrons in the CB of B to play a photocatalytic role. The most popular Z-type heterojunction theory in recent years. Type III can be regarded as another Z-type heterojunction, where electrons in the CB of A recombine with holes in the VB of B, thereby separating electron–hole pairs [228].

Fig. 19

Reproduced with permission from Ref. [231]. Copyright 2016, Elsevier. d SEM images of pure TiO₂ nanofibers and TiO₂/MoS₂ composite nanofibers. e The band structures of TiO₂/MoS₂ heterostructure. Reproduced with permission from Ref. [232]. Copyright 2020, Elsevier. f TEM images of WO₃ nanosheets, g-C₃N₄ nanosheets and WO₃/g-C₃N₄ samples. g The band structures of pure g-C₃N₄ and WO₃ before bonding (left), the internal electric field and band-edge bending of the WO₃/g-C₃N₄ interface after bonding (middle), and the schematic diagram of the S-heterojunction charge-transfer process between WO₃ and g-C₃N₄ under illumination (right). h the Schematic illustration of type II heterojunction and S-scheme heterojunction. Reproduced with permission from Ref. [233]. Copyright 2019, Elsevier. i Schematic illustration of NMS/SCN S-scheme heterojunction. (TEOA: Triethanolamine) Reproduced with permission from Ref. [234]. Copyright 2021, Elsevier

a Schematic diagram of three heterojunction structures. b TEM images of g-C₃N₄, CQDs, and CQDs/g-C₃N₄. c Schematic diagram of the photocatalytic mechanism of UCN/CQDs.

Heterojunctions composed of 2D materials and other semiconductor interfaces of different dimensions can integrate their respective advantages. For example, when 2D materials are composited with 0D materials, they can exist as good base materials, so that the 0D materials can be dispersed more uniformly. At the same time, the heterogeneous interfaces have various forms and a large number of active sites, which can effectively improve the catalytic performance. 2D materials are composited with 1D materials, and their heterointerfaces can capture photons and utilize sunlight more efficiently, thereby improving catalytic performance. The heterostructure of 2D materials and 2D materials has many special catalytic advantages due to its unique face-to-face contact structure. The close face-to-face contact creates a high specific surface area, strong coupling effect, and promotes current carrying. At the same time, due to the electron tunneling effect, the charge transport path is shortened and the transfer speed is accelerated [229, 230]. Therefore, rational design of heterojunctions between 2D materials and different materials is an effective means to synthesize excellent photocatalysts. Liu et al. prepared a 2D/0D photocatalyst composited with carbon nanodots (CQDs) and ultrathin g-C₃N₄(UCN) nanosheets [231] (Fig. 19b). CQDs/g-C₃N₄ exhibited excellent hydrogen evolution photocatalytic activity under visible-light irradiation. Figure 19c demonstrates the role of the heterojunction in the CQDs/g-C₃N₄ composite system. When illuminated by visible light, the electrons in g-C₃N₄ are transferred into the CQDs, which facilitates the charge separation. At the same time, CQDs have a narrow bandgap that can respond to visible light and provide electrons to the CB of g-C₃N₄, which broadens the visible-light absorption region of the system. Finally, CQDs absorb long-wavelength visible light and emit short-wavelength light, which can excite g-C₃N₄ to generate more carriers. Li et al. fabricated 2D/1D heterojunctions by assembling MoS₂ nanosheets onto ultralong TiO₂ nanofibers [232] (Fig. 19d). This heterostructure shows excellent photocatalytic hydrogen generation activity, which is approximately 24 times higher than that of pure TiO₂. As shown in Fig. 19e, band bending occurs between the interface of TiO₂ nanofibers and MoS₂ nanosheets, which facilitates photogenerated electron transfer to MoS₂ and enhances the photocatalytic activity. Fu et al. prepared ultrathin 2D/2D WO₃/g-C₃N₄ composite heterojunction photocatalysts [233] (Fig. 19f). The ultrathin layered heterojunction (Fig. 19g) makes it easier for photogenerated electrons and holes to transfer to the surface of the photocatalyst, which improves the surface photocatalytic rate and exhibits better hydrogen production activity than pure g-C₃N₄ and WO₃. Meanwhile, the heterojunction constructed in this study is a new stepped (S-scheme) heterojunction. Its structure is similar to the type II heterojunction, but the electron transport mode is different, which improves the disadvantage of the weakened carrier oxidation ability in the type II heterojunction. As shown in Fig. 19h, since electrons in CB and holes in VB of the two semiconductors are difficult to transfer due to Coulomb repulsion, and the Coulomb attraction between CB and VB of the same semiconductor hinders the transfer process, the transfer mechanism of the traditional type II heterojunction is thermodynamically unfavorable for the occurrence of photocatalytic oxidation and reduction reactions, which has obvious limitations. The S-type heterojunction can realize the elimination of electrons and holes between CB and VB through the internal electric field between the two semiconductors. Finally, the holes in the VB of semiconductor I and the electrons in the CB of semiconductor II are left to participate in the photocatalytic reaction. Obviously, this system will be more conducive to the separation of photogenerated carriers. Chen et al. also constructed S-type heterojunctions of S-doped g-C₃N₄ (SCN) and N-doped MoS₂ (NMS), and the obtained S-g-C₃N₄/N-MoS₂ photocatalyst has excellent photocatalytic hydrogen evolution performance [234]. The mechanism of charge transfer and separation in S-type heterojunction photocatalysts is shown in Fig. 19i. Under the action of the internal electric field directed from the SCN to the NMS, the holes on the VB of the SCN and the electrons on the CB of the NMS are eliminated, and finally the electrons on the CB of the SCN and the holes on the VB of the NMS are left to participate in subsequent reactions.

Whether it is a Schottky junction constructed by noble metal deposition or a heterojunction constructed between semiconductors, the essential regulation mechanism is to regulate the bandgap structure of the photocatalyst to better promote the transfer of photogenerated carriers. It is not difficult to see that there are two key aspects in the design of photocatalyst heterojunctions, one is to build a suitable two-semiconductor interlaced energy band structure, and the other is to provide an ideal interface for charge separation.

4.4 Thickness adjustment

As mentioned in Sect. 3 of this paper, the morphology of the photocatalyst is different, and its surface area, the number of exposed active sites, the contact area between the catalyst and the reactant, and the adsorption performance of the reactant will be different. Compared with bulk materials, the ultrathin thickness of 2D nanomaterials can provide more active sites and make full use of sunlight. At the same time, the charge transport path can be shortened and the photocatalytic activity can be significantly improved. Many experiments have confirmed that by changing the thickness of the material, it is possible to obtain much higher catalytic performance than the bulk material. That is, adjustment of thickness is an effective means of modification. Ping et al. exfoliated the bulk g-C₃N₄ into g-C₃N₄ nanosheets with a thickness of about 2 nm by thermal oxidation etching, and its photocatalytic activity was significantly improved [235]. This is due to the increased specific surface area after thickness reduction and the quantum confinement effect of ultrathin nanosheets, which enhances the electron transport rate and prolongs the carrier lifetime. In addition to the significant increase in specific surface area, the CB of g-C₃N₄ shifts to a more negative position during the thickness reduction process, making it have a better reducing ability. This will benefit its application in reducing CO₂. Song et al. prepared g-C₃N₄ with different thicknesses by high-temperature exfoliation [236]. As the calcination time increased, the thickness of the obtained g-C₃N₄ became thinner. UV–Vis diffuse reflectance spectroscopy (DRS) spectra indicated that the bandgap of g-C₃N₄ increases with decreasing thickness (Fig. 20a). As shown in Fig. 20b, a more negative CB position would be beneficial for the application of g-C₃N₄ in CO₂ reduction. The experimental characterization results also confirmed this point (Fig. 20c). During experiments, the thickness of 2D materials is usually characterized using AFM. As shown in Fig. 20d, CdS nanosheets with different thicknesses from 1.5 to 5.0 nm were characterized by AFM. Benefiting from the shorter charge transport distance and more negative CB edge (Fig. 20e) brought about by the smaller thickness, the thinnest CdS nanosheet at 1.5 nm has the most excellent catalytic hydrogen evolution performance [237] (Fig. 20f). In addition to thickness, AFM can characterize roughness and surface morphology. Due to the influence of possible attachments on the material and the influence of the substrate itself, the thickness of the material measured by AFM will generally be thicker. Due to equipment limitations, AFM can only characterize small pieces of 2D materials, and the characterization takes a long time, making it impossible to quickly characterize large-area 2D materials. In addition to AFM, the thickness of 2D materials can also be characterized by optical reflection contrast, Raman spectroscopy, photoluminescence, optical interference effects and other methods, which were reviewed in detail by Jin and Yu [238].

Fig. 20

Reproduced with permission from Ref. [236]. Copyright 2022, Elsevier. d AFM images and height cutaway views of different CdS samples. e Kubelka–Munk function vs. photon energy curves of different CdS samples. f Rates of photocatalytic produced hydrogen of different CdS samples. Reproduced with permission from Ref. [237]. Copyright 2018, Elsevier

a Tauc plots of different CN samples. b Bandgap structure of different CN samples and the reduction potential of some CO₂-derived products. c Photocatalytic performance of different CN samples.

4.5 Defect engineering

During the exfoliation of the material from the bulk to the nanosheet, the internal atoms are exposed on the surface of the material, and various defects are inevitably formed. Fortunately, for catalytic processes, the creation of defects is not a disadvantage. Appropriate defects can help build new energy bands and increase active sites, thereby regulating electron capture. Defect engineering is an effective way to improve the light absorption performance of catalysts. By constructing defects, the efficiency of electron–hole separation is enhanced, and the light absorption region is also enlarged. In some experiments, even its photoresponse range was extended to the entire visible-light region. In addition to single defects that can improve catalytic performance, there are also interactions between defects. If this synergy can be exploited, the overall photocatalytic efficiency can also be improved. Due to the layered structure of 2D materials, most of the defects generated in the material are surface defects rather than bulk defects in bulk materials. This feature avoids the disadvantage of bulk defects becoming carrier recombination centers, and more shows the side that is helpful for photocatalysis. Wang et al. prepared an oxygen vacancy (OV)-rich ultrathin sulfur-doped BiOBr nanosheet with excellent 4-chlorophenol degradation performance [239]. As shown in Fig. 21a, there exists a sub-band at 0.38 eV below the CBM. The partial density of states (PDOS) further reveals that the sub-band is composed of Bi 6p state and S 2p state, and the Bi 6p state mainly originate from the Bi atoms nearest to the OVs (Fig. 21b). That is, this sub-band is generated by the synergy between oxygen vacancies and sulfur doping. Figure 21c shows how the sub-bands participate in the degradation process. The generation of sub-bands effectively narrows the bandgap and improves the visible-light absorption capacity. Niu et al. obtained porous g-C₃N₄ (CN) with a large number of defects (CNQ) by a simple heat treatment in air (Fig. 21d) [240]. As shown in Fig.21e, compared with pristine g-C₃N₄, the bandgaps of CNQ samples are all reduced, implying the generation of new sub-bands (Fig. 21f). The new peak at 2,175 cm⁻¹ in the FTIR spectrum corresponds to the appearance of the cyano group (–CN) that can disrupt the stacking structure and cause layer fluctuations (Fig. 21g). Figure 21h, i reveals the appearance of NVs after rapid thermal treatment. The above characterizations confirmed that the appearance of –CN and NVs caused by heat treatment was the cause of the structural changes.

Fig. 21

Reproduced with permission from Ref. [239]. Copyright 2019, Elsevier. d The typical TEM images of CN and CNQ680 and the schematic diagram of the morphology change from CN to CNQ. e The converted Kubelka–Munk vs. light energy plots of CN and CNQs. f Schematic diagram of band structure change from CN to CNQ680. g The FTIR spectra of CN and CNQs. h the (N_2c)/(N_3c) ratio of CN and CNQs. i The C/N ratio of CN and CNQs. Reproduced with permission from Ref. [240]. Copyright 2017, Elsevier. j The total and atomic projected DOS analyses of different h-BN samples. Reproduced with permission from Ref. [241]. Copyright 2021, ACS Publications

a Band structure of OV-rich ultrathin S-doped BiOBr nanosheet. b PDOS of OV-rich ultrathin S-doped BiOBr nanosheet. c Schematic diagram of the photocatalytic degradation 4-CP over OV-rich ultrathin S-doped BiOBr nanosheet.

In the investigation of the electronic structure changes caused by defects, the calculation of the DOS by DFT is an effective method to explore the influence of defects. Figure 21j is the DOS calculation for pristine h-BN, h-BN with boron vacancies, h-BN with NVs, and h-BN with oxygen-filled nitrogen vacancy [241]. The analysis shows that pristine h-BN has a wide bandgap of 5.42 eV, which is not conducive to the photoresponse. The band structure of h-BN with boron vacancies has changed greatly, and the mid-gap states has expanded beyond the Fermi level, making it an excellent electron acceptor similar to p-type semiconductors. Meanwhile, h-BN with NVs and h-BN with oxygen-filled nitrogen vacancy is a good electron donor similar to n-type semiconductors . Research has proved that the ubiquitous defects in materials affect the electrical, thermal, optical, magnetic, acoustic and mechanical properties of materials. The catalytic process is very sensitive to the change of electronic structure, so the introduction of defects is an effective means of catalytic modification. First, defects can form additional energy levels to expand the photoresponse range, and second, defects can affect and change the distribution of catalytic active sites and charge density, thereby changing the catalytic activity of the catalyst. However, due to the complexity and diversity of defects, the concentration, distribution, types of defects, and the interaction between different defects have not been systematically studied. But it is worth looking forward to, with the rapid development of computer technology and material characterization methods, the mechanism of defects in the catalytic process will be further revealed.

4.6 Dye sensitization

Dye sensitization is a method to expand the light absorption range by adsorbing dyes that can absorb various visible light and even near-infrared light on the semiconductor surface. When the dye is irradiated with visible light, electrons on the dye will transition from the ground state to the excited state. When the free electron potential of the excited state dye is higher than the potential of the CB of the semiconductor photocatalyst, electrons will be transferred from the dye to the semiconductor and react with O₂ to generate active radicals. At the same time, the dye itself becomes a positive ion radical. The schematic diagram of the reaction process of dye sensitization is shown in Fig. 22a [242]. The unique layered structure of 2D nanomaterials provides convenience for the adsorption of dye sensitizers. At the same time, the changeable structure of 2D nanomaterials has higher flexibility and can better combine with different dyes to expand the range of light absorption and enhanced charge-transfer rate. Li et al. loaded three different dye sensitizers: BY24, BR14, and MB on porous Cu-MOFs, resulting in composites basic yellow 24 (BY24@1), basic red 14 (BR14@1), and MB@1 [243]. It can be seen that the energy bandgaps of the three composites are respectively reduced to 2.19 eV, 2.10 eV and 1.76 eV (Fig. 22b). Due to the significantly enhanced visible-light response range, MB @1 with the narrowest bandgap shows the best catalytic performance. Li et al. found that dye sensitization can effectively inject electrons into Fe-MOFs (MIL-53), and increase its quasi-Fermi level under light, so that Fe-MOFs can participate in photocatalytic hydrogen evolution [244]. Figure 22c shows that the CBM of MIL-53 is about − 0.04 V, which is in the range of hydrogen evolution potential from the thermodynamic point of view. However, linear sweep voltammetry (LSV) characterization revealed that MIL-53 had an onset overpotential of 0.27 V, implying a kinetic barrier for HER (Fig. 22d). Therefore, as shown in Fig. 22e, the CBM at − 0.04 V cannot provide enough energy for the HER reaction. After the introduction of eosin Y (EY) by dye sensitization, the CBM position was significantly more negative enough to overcome the onset potential of 0.27 V (Fig. 22f). Therefore, EY-MIL-53 has shown significantly better performance than MIL-53 in catalytic hydrogen evolution (Fig. 22g). Wei et al. revealed two contributions to the photocatalytic process by the electrons generated after the dye was excited in the Bi₂MoO₆@RhB system [245]. One is that electrons are injected into the CB of Bi₂MoO₆ to participate in the reduction reaction, and the other is to suppress the recombination of carriers in the VB of Bi₂MoO₆ through the “4–5–6 cycle” shown in Fig. 22h. The sacrificial agent detection experiment found that the degradation activity decreased significantly when AgNO₃ which eliminates electrons or triethanolamine (TEOA) which eliminates holes was added. This shows that electrons and holes are directly involved in the process of degrading levofloxacin (Fig. 22i).

Fig. 22

Reproduced with permission from Ref. [242]. Copyright 2018, Elsevier. b energy bandgaps of different samples. Reproduced with permission from Ref. [243]. Copyright 2019, Academic Press Inc. c M − S plots of MIL-53 (Fe). d LSV curve of MIL-53 (Fe) for HER. e Illustration of band structure of MIL-53 (Fe). f M − S plots of EY-sensitized MIL-53 (Fe). g Photocatalytic hydrogen evolution activity of different samples. Reproduced with permission from Ref. [244]. Copyright 2022, Elsevier. h The photoexcited charge-transfer processes that may occur in the Bi₂MoO₆@RhB samples: 1. direct bandgap transition; 2. intrinsic charge recombination; 3. electron trapping/detrapping; 4. intramolecular photoexcitation; 5. interfacial electron injection; 6. interfacial electron recombination; 7. radiationless decay. i sacrificial agent experiments over Bi₂MoO₆@RhB under LED light irradiation. (TBA: tert-Butanol) Reproduced with permission from Ref. [245]. Copyright 2023, Elsevier

a Schematic diagram of the dye-sensitized electron transfer process.

In summary, dye sensitization is an effective modification method that can change the bandgap structure, broaden the photoresponse, and help photocatalysts overcome the kinetic barrier by injecting electrons. In the actual experiment process, how to effectively adsorb the dye on the surface of the material is an important process. The rich surface defects of 2D materials allow dyes to be adsorbed on the surface through a simple electrostatic self-assembly process. In order to further obtain better adsorption, the adsorption can also be promoted by means of light or an external electric field.

5 Conclusions and prospects

With the development of industry, environmental and energy challenges emerge one after another, and it is a global issue to explore how to solve these two problems. Photocatalysis has become the most potential technology to solve environmental and energy problems due to its green and low-cost characteristics. As the core medium of photocatalytic technology, semiconductor photocatalyst, how to modify it and improve its photocatalytic performance has become the focus of research.

Traditional semiconductor materials have a large bandgap, narrow light absorption range, and low separation and migration efficiency of photogenerated carriers, which are limited in practical applications and have insufficient reaction sites. 2D nanomaterials play an important role in the field of photocatalysis due to the unique effects brought about by their nanometer size and layered structure, bringing unprecedented opportunities for improving photocatalytic activity. The main modification methods of semiconductor photocatalysts include element doping, noble metal deposition, heterostructure construction, thickness adjustment, defect engineering, and dye sensitization. Due to the sensitivity of 2D nanomaterials to thickness changes, the modification process is more convenient than bulk materials, and the above modification methods can also play a more significant and effective role in 2D nanomaterial systems. The wide variety of 2D nanomaterial families with rich atomic structure and high flexibility also provides more new ideas for designing photocatalysts. However, the practical use of 2D nanomaterials is limited for the following reasons:

• (1) 2D nanomaterials have high surface energy, abundant active sites, easy agglomeration and chemical changes, and lack of long-term stability and durability, which limits their practical application;

• (2) At present, the production and preparation of 2D photocatalytic nanomaterials mainly centralized in the laboratory, low-cost, environmentally friendly mass production remains challenging;

• (3) The relationship between the surface structure of 2D nanomaterials and its photocatalytic activity is close, and the mechanism of the reaction process is complex, which needs to be explored with the help of a large number of microscopic characterization methods.

In the face of nanoscale and low-latitude photocatalysts, which are difficult to observe the microscopic reaction process, researchers can look forward to two directions:

• (1) Development of in-situ characterization technology. Compared with ex-situ characterization technology, in-situ characterization technology can intuitively obtain the intermediate process of material changes in the reaction, and obtain information related to transient reactions, such as atomic structure, element arrangement, surface defects, and molecular adsorption. In-situ characterization technology is an indispensable means to explore the catalytic mechanism of catalysts. At present, many in-situ characterization methods such as in-situ infrared, in-situ Raman, in-situ XPS, and in-situ synchrotron radiation techniques have been used in the study of photocatalytic reaction mechanisms, making important contributions to catalyst design. It is foreseeable that in the future, with the proficient analysis of existing in-situ characterization techniques and the further development of in-situ characterization techniques, the structure of 2D nanophotocatalysts will be accurately understood, and researchers will design and prepare more applicable for practical catalyst systems;

• (2) The development of computer simulation technology, such as first-principles calculation and so on. Computer simulation can start from the electronic structure of the material itself, and quantitatively analyze the fundamental relationship between photocatalyst modification methods and performance improvement. With further simulation studies on the four main steps of the photocatalytic reaction: visible-light absorption, carrier separation, carrier transport and redox reaction, researchers can more accurately simulate the theoretically optimal material system, morphology and structure, thus providing guidance for the stable and efficient production of industrial 2D nanophotocatalysts.

In the future, while focusing on catalytic performance and stability, it is necessary to further pay attention to the research on the environmental protection, cost and life of photocatalysts, so that photocatalytic technology can become a powerful means to solve environmental and energy problems. In summary, although there are still some challenges in the application of 2D nanomaterials in the field of photocatalysis, it is undeniable that with the deepening of research, more new 2D photocatalytic nanomaterials will be designed and prepared, and 2D nanophotocatalysts will become the mainstream catalyst in the field of photocatalysis.