1 Introduction

Solar energy, as an abundant and environmentally friendly resource, presents a promising solution to meet the high consumption of finite fossil fuels and the growing energy demands of human society [1,2,3]. Photocatalytic process has been proven to be an efficient approach to harness and convert intermittent sunlight into transportable chemical fuels, reducing greenhouse gas emissions in the process [4, 5]. As an illustration, the photoelectrochemical (PEC) water-splitting reaction, which mirrors natural photosynthesis, emerges as a viable artificial strategy for generating hydrogen gas, representing a significant move towards cleaner energy sources, thereby mitigating global energy shortages and environmental issues. In 1972, Fujishima and Honda achieved a pioneering breakthrough by successfully employing PEC cells to generate hydrogen and oxygen gases separately when exposing the photoelectrode to light irradiation [6], thus preventing the dangerous mixture of oxygen and hydrogen into explosion gas. Consequently, PEC devices have been recognized as a straightforward method for attaining the solar-to-chemical energy objective, relying solely on semiconductors to integrate electrocatalytic and light-absorbing functionalities. In the realm of water-splitting PEC cells, semiconductor materials, such as TiO2 [7,8,9,10,11], BiVO4 [12,13,14], ZnO [15, 16], among others [17], are crucial, serving as primary components for capturing and converting solar energy into chemical fuels. Specifically, graphitic carbon nitride (g-CN or g-C3N4), due to its affordability, non-toxicity nature, and high stability, has garnered considerable interest in this field [18,19,20,21,22,23,24,25,26]. Additionally, g-CN can be easily synthesized from cost-effective nitrogen-rich and oxygen-free compounds containing C–N core structures, such as thiourea, dicyandiamide, and melamine [4]. Therefore, since its initial use in 2009 [26], significant dedication of effort has been focused on exploring the potential of g-CN in PEC devices.

In the PEC system, the absorption of sunlight by the g-CN materials possessing an appropriate bandgap (2.7 eV) is adequately substantial to overcome the endothermic energy of the water-splitting reaction [4]. The electrons are excited from the valence band (VB) or the highest occupied molecular orbital (HOMO) to the conduction band (CB) or the lowest unoccupied molecular orbital (LUMO) of the photoelectrode to reduce the H+ producing H2 gas (2H+ 2e → H2) in the water-splitting process (Fig. 1a) [1, 27]. Meanwhile, the remaining hole in the VB or HOMO facilitates the oxidation reaction, releasing O2 gas (2H2O → 4H+ + O2 + 4e). Therefore, the water-splitting process can be delineated into three fundamental steps: (1) light absorption, (2) transfer and separation of photoinduced electron and hole charge carriers, and (3) redox reactions. However, a noteworthy concern is the large bandgap of g-CN, limiting its utilization of visible light. Besides, the high recombination rate of partially separated photoinduced charge carriers also reduces the photocatalytic activity of g-CN for water splitting (Fig. 1b) [28]. To address these issues, surface junctions, which can separate photoinduced electrons and holes onto distinct surfaces, have been employed. For example, gold (Au) clusters were combined with g-CN (Au/g-CN) by electrophoretic deposition and ultrasonication methods [29]. The plasmonic effects induced by Au extend the efficiency of visible-light absorption and enhance photocurrent density. Semiconductor oxide nanoparticles, such as Cu2O [30], TiO2 [31], BiVO4 [32], and Fe2O3 [33], have been utilized to establish surface junction structures for high photocatalytic activity. Moreover, Xiong et al. used the multiple-step thermal vapor condensation (MSTVC) approach with thiourea (CH4N2S) as precursors, reacting with the fluorine-doped tin dioxide (FTO) layer to synthesize the SnS2/g-CN composite [34]. As S atoms possess a greater number of valence electrons compared to N, this method enables the excitation of more electrons, ultimately enhancing photocurrent density of 844.6 µA cm−2 at 1.23 V vs. RHE under AM 1.5G illumination. Hence, surface junction has been demonstrated as an effective strategy in enhancing the photocatalytic performance of g-CN films.

Fig. 1
figure 1

a Schematic illustration of water splitting over semiconductor photocatalyst. Reprinted with permission from Ref. [1]. Copyright 2011 Elsevier. b Frontier orbitals display of the g-CN in the ground state of various dimensions. Dark brown (dark green) isosurface color indicates HOMO (LUMO) wave functions. Reprinted with permission from Ref. [27]. Copyright 2018 Wiley


Doping is another widely used approach for modifying the bandgap and augmenting the charge transfer process in g-CN materials. Typically, metal or nonmetal dopants replace nitrogen or carbon atoms within the g-CN structure, broadening the light absorption range and effectively curbing charge recombination. Ag, as the first metal dopant, doped g-CN was successfully synthesized by Liu et al., with current density increased to 6.4 μA/cm2 [35]. A higher current density of 69.8 μA/cm2 was achieved using a liquid-based growth method where Ni atoms were strategically positioned within the vacancies of g-CN [36]. For metal-free doping, B was incorporated into g-CN by replacing carbon atoms through a thermal vapor condensation (TVC) process, employing a mixture of melamine and boron as precursors [21, 37]. This approach resulted in an almost fourfold increase in photocurrent density compared to pristine g-CN films. Similarly, O-doped g-CN also exhibited improved photocatalytic performance [25], with the introduction of O heteroatoms creating new bandgap states that enable the absorption of visible light across the spectrum. Apart from atom substitution, nonmetal atoms can also bind to the g-CN film through van der Waals (vdW) forces, electrostatic interactions, hydrogen bonding, or π–π stacking, resulting in a charge transfer complex in the stacking structure. This method was known as the charge transfer doping method [38]. For example, I3 and I5 clusters, as well as Ag atoms, were implanted into the g-CN film by noncovalent attachment [39]. The p orbitals of polyiodides facilitated electron tunneling, thus enhancing charge carrier transfer and separation. Additionally, Tian et al. conducted an in-depth investigation of the noncovalently iodinated g-CN film [40]. The I species introduced additional bandgap states and subsequently reduced the bandgap, leading to enhanced light absorption efficiency and improved charge carrier separation. Therefore, both conventional doping and charge transfer doping approaches have the potential to enhance the photocatalytic performance of g-CN films.

Due to its good electrical conductivity owing to C–C π bonds, cost-effectiveness, widespread availability, and relatively high chemical stability, carbon (C) is regarded as one of the premier materials employed for the modification of carbon-based catalysts [41]. Additionally, nitrogen vacancies (NVs) can lead to the C-rich g-CN, and it has been demonstrated that C–C π bonds have the capacity to reduce the bandgap of g-CN, thereby extending its light absorption capabilities into the visible spectrum [23, 24]. These bonds also play a vital role in enhancing conductivity, resulting in improved transfer of photoinduced charge carriers. Therefore, to construct C-modified g-CN (C/g-CN) structures, besides the C–C π bonds introduced by NVs, C has also been extensively applied in both surface junction and doping methods, leading to the introduction of various C structures, including C surface [42,43,44,45,46,47,48,49,50], C crystalline [51,52,53,54,55], and C atoms [56,57,58,59,60], into g-CN films. These kinds of C/g-CN photocatalysts are summarized in Table 1. In this review, we provide an overview of the recent developments in C/g-CN films.

able 1 Summary of different kinds of C-modified g-CN photocatalysts

Photocatalyst Synthesis method Structure Photocurrent (μA cm−2)
NV/g-CN Annealing [24], temperature-controlling route method [61] NVs in the plane About 75 (0.1-m Na2SO4, 0.1-m Na2SO3, and 0.01-m Na2S under a NEWBET AM 1.5-G solar simulator at 1.23 V vs. RHE) [24]
Graphene/g-CN Impregnation-chemical reduction and calcine [42] Surface junction About 25 (1-M Na2 SO4 under visible-light irradiation at 0.5 V vs Ag/AgCl) [42]
CNT/g-CN Combined solvothermal and ultrasonic methods [44], steaming treatment [45] Combined through hydrogen bonds [44] or covalent bonds [45] About 1 (under a 300-W Xe lamp) [44]
CQD/g-CN Thermal polymerization and ultrasonic treatment [46] Surface junction About 0.4 (0.50-m Na2SO4 under Xe lamp illumination at 0.8 V vs Ag/AgCl) [46]
Cring/g-CN Multiple-step thermal treatment [52], modified gas-shocking strategy [53] π-conjugated bonds in the plane Over 1 (0.3-m Na2SO4 under 300-W Xe lamp) [53]
Catom/g-CN Thermal vapor condensation method [23], one-step thermal condensation [58], impregnation-assisted calcination [60] C dopants in the plan About 30 (0.5-m Na2SO4 under a Xenon lamp at 1.0 V versus RHE) [60]