Evidence of direct Z-scheme triazine-based g-C₃N₄/BiOI (001) heterostructures: a hybrid density functional investigation

Abstract

In this work, we systematically investigate the photocatalytic mechanism of g-C₃N₄/BiOI (001) through hybrid functional calculations based on first-principles theory. The staggered band structure is observed in the g-C₃N₄/BiOI (001); meanwhile, a built-in electric field exists from the g-C₃N₄ monolayer to the BiOI surface at the interface. BiOI has lower band edges, which bend downward at the interface; whereas g-C₃N₄ has higher band edges, which bend upward. With Coulomb interaction and the built-in electric field, photo-generated electrons in the conduction bands (CB) of BiOI recombine with photo-generated holes in the valence bands (VB) of g-C₃N₄. Meanwhile, the stronger reduction capacity for photo-excited electrons in the g-C₃N₄'s CB and the stronger oxidation capacity for photo-generated holes in the BiOI (001)'s VB are retained. Therefore, a direct Z-scheme heterostructure character is presented. As a result, the electrons and holes generated by the photons can be separated and migrate highly effectively at the interface. The separated electrons and holes can effectively participate in the redox reactions with water/pollutants to produce the photocatalytically reactive species superoxide ions (˙O₂⁻) and hydroxyl radicals (˙OH), respectively. This is consistent with the experimental results. It is also worth noting that the g-C₃N₄/BiOI (001) heterostructure shows a larger difference in the effective mass of carriers. Therefore, the direct Z-scheme charge transfer and separation mechanism and the larger effective mass difference of carriers lead to the superior photocatalytic activity of the g-C₃N₄/BiOI (001) in experiments. A few speculations and controversies that arose from the experiments are clarified.

---

1. Introduction

The rapid development of industry in recent decades has led to a gradual decrease in the earth's energy reserves, while pollution has become increasingly severe due to the rapid development of industry. Nowadays, how to solve the energy shortage and environmental pollution problems are significant and urgent challenges for scientists, scholars, and governments and have drawn a great deal of attention and research.^1–5 In this regard, semiconductor-based photocatalysis technology has become essential to solving these severe problems. This is because this technology exhibits excellent potential for splitting water to produce hydrogen and oxygen and transforming all kinds of liquid and gaseous pollutants into harmless matter, even valuable chemicals and fuels, using the inexhaustible solar energy source as a driving force.^6–10 Generally, adequate light absorption, effective photo-generated carrier separation, and high mobility of charge carriers are several critical factors for the applications of semiconductor photocatalysis.^11,12 However, satisfying all these conditions for a single and pure semiconductor photocatalyst is not easy. Therefore, a variety of strategies such as element doping,^13–15 heterostructure construction,^16–19 metal deposition,^20–22 and defect control^23,24 are applied to improve the separation efficiency of the electron–hole pairs generated by the photocatalysis and to increase the absorption of visible light so that more photocatalytic activity can take place. Among these methods, constructing a heterostructure^16–19 is considered the most promising method for enhancing photocatalytic performances.

The two-dimensional (2D) material g-C₃N₄ has a graphene-like structure but has an appropriate bandgap of about 2.7 eV for photocatalytic performance in experiments.^25–29 At the same time, g-C₃N₄ has high chemical and thermal stability and significant optical and photoelectrochemical properties as well as low cost.^30–32 It is a promising material for photocatalytic hydrogen production and decomposition of organic pollutants.¹⁶ More importantly, 2D g-C₃N₄ (001 or 002 crystal facet) is an excellent candidate for heterostructure construction with other 2D semiconductor nanosheets. For example, g-C₃N₄/SnS₂,³³ g-C₃N₄/ZnO,^34,35 and g-C₃N₄/TiO₂³⁶ present excellent photocatalytic activities in experiments, in contrast to their single counter parts. On the other hand, BiOI has also attracted great interest from researchers as a promising photocatalytic material because of its special merits, such as an appropriate band gap around 2 eV and a built-in electrical field between the cation (Bi₂O₂)²⁺ slab layer and anion I¹⁻ slap layer.^37–39 However, similar to the cases of many individual semiconductors, including g-C₃N₄, as mentioned above, BiOI material is also limited in its ability to be used as a photocatalyst due to some disadvantages. Among these disadvantages, the two major challenges are the slow charge migration and high recombination probability of the electrons and holes that are generated in response to photons. Building heterostructures (such as the direct Z-scheme) with other appropriate 2D semiconductors is the major way to enhance the activity of BiOI.⁴⁰ At the same time, because of the layered structure of BiOI materials and the stronger visible light absorption, BiOI is also an excellent candidate for building heterostructures with other appropriate 2D semiconductors to improve the activities, for example, BiOI/Bi₂O₃,^41,42 BiOI/SnS,⁴³ BiOI/SrTiO₃,⁴⁴ BiOI/MoS₂,⁴⁵ and so on.

A great deal of g-C₃N₄-based and BiOI-based heterostructures, the g-C₃N₄/BiOI, including the different crystal facet conjunctions, the different preparation methods, and the different ratios between g-C₃N₄ and BiOI, have been extensively investigated in experiments.^{25,26,28,29,46–55} For the different crystal facet conjunctions, the BiOI (001) and BiOI (110) crystal facets are mostly considered for building the heterostructures between BiOI and g-C₃N₄. Despite this, Tian et al.⁵⁰ reported that the photocatalytic performance of the g-C₃N₄/BiOI (001) is obviously superior to that of the g-C₃N₄/BiOI (110) in experiments. However, the photocatalytic mechanism for the significant g-C₃N₄/BiOI (001) heterojunction is unclear and controversial in experiments. Alam et al.²⁵ and Tian et al.⁵⁰ reported that the g-C₃N₄/BiOI (001) presents type II band alignment, while Zhang et al.⁴⁷ and He et al.⁴⁹ reported that the g-C₃N₄/BiOI (001) is the direct Z-scheme heterojunction. In view of the fact that the systematic and accurate first principles investigations and understandings of the mechanisms governing the photocatalytic reaction are lacking for g-C₃N₄/BiOI (001), it is desired to explore the reason for the high photocatalytic performance of the g-C₃N₄/BiOI (001) through first-principles calculations so that the experimental findings on the g-C₃N₄/BiOI(001) can be further clarified and our understanding can be further enhanced.

We propose using hybrid functional calculations to explore the photocatalytic mechanism of the heterojunction between the BiOI(001) and the g-C₃N₄ monolayer by building a reasonable model. This paper will discuss the interfacial interactions, the effective masses of electrons and holes, charge transfer, the projected band structure, the work function, and the band alignment for the g-C₃N₄/BiOI(001).

2. Computational method

All calculations are carried out using the Vienna ab initio simulation package (VASP) with the projector augmented wave (PAW) approach, which is based on density functional theory.^56,57 The geometrical optimization is carried out via the generalized gradient approximation (GGA) of Perdew et al.⁵⁸ The DFT-D3(BJ) method⁵⁹ of Becke–Jonson is applied to describe the long-range effect of van der Waals (VdW) forces. The vacuum region of 20 Å is used in this study in order to avoid the interaction between the layers.⁶⁰ There is a set cut-off energy of 500 eV for the plane-wave basis. The k-point mesh of 4 × 2 × 1 is adopted for the heterojunction g-C₃N₄/BiOI (001) supercell model. The structural optimizations are finished when each atom's energy is less than 1.0 × 10⁻⁴ eV and all residual forces are less than 0.03 eV Å⁻¹. As we all know, the traditional DFT-GGA method usually underestimates the bandgap value of the semiconductor. Therefore, we use the Heyd–Scuseria–Ernzerhof (HSE) hybrid function⁶¹ to calculate the electronic structures accurately. As short-range Hartree–Fock exchange parameters, we applied 0.2 and 0.18 for bulk BiOI and monolayer g-C₃N₄. This led to the theoretical band gaps for bulk BiOI and monolayer g-C₃N₄ (001) being 2.04 and 2.79 eV, respectively, which are in good agreement with the corresponding experimental data 1.77–2.22^42,62 and 2.75–2.8 eV,^26,46,48 respectively. Thus, an average value α = 0.19 is used for the heterojunction. The thermostability of the heterojunction is evaluated by ab initio molecular dynamic simulation (AIMD)⁶³ with a canonical (NVT) ensemble and a simulation temperature of 300 K. The Anderson thermostat is employed, with a time step of 1 fs and the total AIMD time is 8 ps.

The optical absorption coefficient A(ω) of the 2D material is obtained from the formula A(ω) = ωLε²(ω)/c (where L is the thickness of vacuum and c is the speed of light in a vacuum).⁶⁴ The dielectric constant calculation is performed with this method in the module “Linear Optical Spectra for Two Dimensional Semiconductors” of the VASPKIT code.⁶⁵

3. Results and discussion

3.1 Geometric structural model of g-C₃N₄/BiOI(001) heterojunctions

In order to construct the appropriate g-C₃N₄/BiOI (001) structural model, the surface termination problem is first discussed and determined. This is due to the fact that the band alignment and type of heterostructure could potentially be affected by the different surface terminations.⁶⁶ For the termination surface of g-C₃N₄, there is only one termination circumstance including in triazine and heptazine structures, because of its 2D character. While the (001) crystal facet of BiOI is characterized by three different terminations, BiO, 2I and 1I, which are shown in Fig. 1(a)–(c), respectively. The relative stability of the different termination surfaces is judged via the calculations and comparisons of the surface energies:⁶⁷where E_slab is the total energy of the slab, E^formula_bulk is the energy per formula unit in bulk, n represents the number of formula units in the slab, n_i represents the excess atoms of species i, and A represents its surface area. Obviously, the chemical potential μ_i has to be decided before calculating the surface energy.

Fig. 1 The slab models employed for the surface energy calculations of (a) BiO-terminated, (b) 2I-terminated, and (c) 1I-terminated BiOI (001) surfaces.

Under thermodynamic constraints, the chemical potentials of μ_Bi, μ_O and μ_I are required to satisfy the following equation in order to avoid the formation of pure elemental phases in BiOI.

Δμ_Bi ≤ 0, Δμ_I ≤ 0, Δμ_O ≤ 0 (2)

where Δμ_Bi = μ_Bi − E_Bi, Δμ_I = μ_I − E_I and Δμ_O = μ_O − E_O are defined as the change in chemical potential with reference to the energy of the elemental bulk/gaseous state, depending on the growth environment; E_Bi is the energy of a Bi atom in the trigonal Bi-metal, E_O is half of the energy of an isolated O₂ molecule, E_I is the energy of a I atom in the space group of Cmca. We can express the optimized static energy E_BiOI per formula unit in BiOI as E_BiOI = E_Bi + E_O + E_I + ΔH_f(BiOI), where ΔH_f(BiOI) is the formation enthalpy for per formula BiOI. With the conditions of thermal equilibrium growth, the chemical potential of BiOI is equal to the optimized static energy per formula unit in BiOI, namely μ_BiOI = E_BiOI. On the other hand, the equation μ_Bi + μ_O + μ_I = μ_BiOI is required.⁶⁷ Then, we have the equation:

ΔH_f(BiOI) = (μ_Bi − E_Bi) + (μ_I − E_I) + (μ_O − E_O) = Δμ_Bi + Δμ_I + Δμ_O (3)

Additionally, Δμ_Bi, Δμ_I, and Δμ_O must satisfy further constraints to avoid the formation of secondary phases Bi₂O₃ and BiI₃:

2Δμ_Bi + 3Δμ_O < ΔH_f(Bi₂OI₃) (4)

Δμ_Bi + 3Δμ_I < ΔH_f(BiI₃) (5)

where ΔH_f(Bi₂OI₃) and ΔH_f(BiI₃) represent the formation enthalpy in every formula unit for Bi₂O₃ and BiI₃, respectively. The calculated formation enthalpies in every formula BiOI, Bi₂O₃ and BiI₃ are listed in Table 1. The acquired formation enthalpies through HSE approximation are consistent with the experimental data from the database (https://oqmd.org).

Table 1 Compared the ΔH_f (eV) in every formula unit from calculation results through HSE approximations and experimental data

	ΔH^HSEf	ΔH^EXPf
BiOI	−2.89	−3.49
Bi2O3	−5.74	−5.91
BiI3	−2.12	−1.56

---

Fig. 2(a) shows the allowed chemical domain (shaded region) for BiOI in the Δμ_I and Δμ_O plane. The calculated values at four representative chemical potential points are labelled as A (O-rich, Bi/I-poor), B (Bi-rich, O/I-poor), C (Bi-rich, O/I-poor), and D (I-rich, Bi/O-poor) in Fig. 2(a). The Δμ_Bi, Δμ_I and Δμ_O are −2.87, −0.02 and 0 eV for A conditions, respectively, while 0, −0.98 and −1.91 eV for B conditions, 0, −0.71 and −2.18 eV for C conditions, and −2.12, 0 and −0.77 eV for D conditions. Furthermore, the corresponding chemical potentials for μ_Bi, μ_I and μ_O can be acquired. Then the surface energies for each of the atomic termination surfaces on the BiOI (001) are calculated in order to assess their feasibility. In Fig. 2(b), we show the results of our calculations. Under all four growth conditions (points A, B, C, and D), the 1I-terminated surface has been observed to have the lowest surface energy among all three surface types. Based on this, the BiOI (001) with 1I termination has the highest stability and is the easiest to expose and form among these termination surfaces. This conclusion is consistent with the results of BiOCl (001) via similar calculations.⁶⁷ At the same time, it also agrees with Kong et al.'s calculations⁶⁸ about the surface energy of the BiOI (001) facet using other judgment methods. Thus, only the 1I-terminated BiOI (001) surface is considered for constructing the heterostructure model between the BiOI and g-C₃N₄(001). The corresponding model is denoted as g-C₃N₄/BiOI (001).

Fig. 2 (a) An accessible range of chemical potentials (red/shaded region) for the equilibrium growth conditions of BiOI using the HSE functional. Specific points A, B, C and D are chosen as the representative chemical potentials (eV) for the following surface energy calculations. The A condition is O-rich, Bi/I-poor; the B and C conditions are Bi-rich, O/I-poor; the D condition is I-rich, O/Bi-poor. (b) Under four representative chemical potential growth conditions (A, B, C and D), the calculated surface energies (J m⁻²) for various atomic termination surfaces of BiOI (001).

In experiments, g-C₃N₄/BiOI nanocomposites were prepared with the use of triazine-based g-C₃N₄.^{25,26,28,29,46–55} A triazine structure is found in the hexagonal system of g-C₃N₄, and its space group can be described as P6̄m2. The relaxed lattice parameters for bulk g-C₃N₄ are a = b = 4.78 Å, c = 7.08 Å and the angles are α = β = 90°, γ = 120°. Those of tetragonal BiOI with space group P4/nmms are a = b = 3.98 Å, c = 9.13 Å and α = β = γ = 90°. To construct a reasonable heterojunction model, we cleave the (001) crystal facet of bulk g-C₃N₄ and redefine the lattices of g-C₃N₄ (001) to a = 8.29 Å, b = 4.78 Å, and α = β = γ = 90°. As a result, for the g-C₃N₄/BiOI (001) interface, a monolayer 1 × 5 g-C₃N₄ (001) surface (containing 70 atoms C₃₀N₄₀) is used to match a three-layer 2 × 6 BiOI (001) surface with 1I termination (containing 72 atoms Bi₂₄O₂₄I₂₄). The lattice mismatches of the g-C₃N₄/BiOI (001) interface along a and b directions are only 3.92% and 0.07%, respectively. The structural models for the g-C₃N₄/BiOI (001) before relaxation and after relaxation are presented in Fig. 3. As a measure of the stability of the heterojunction, the binding energy (E_b) of g-C₃N₄/BiOI(001) is calculated according to the following equation:⁶⁹

E_b = (E_g-C3N4/BiOI − E_g-C3N4 − E_BiOI)/S_heter (6)

where E_g-C3N4/BiOI is the total energy of the g-C₃N₄/BiOI heterostructure, S_heter is the cross-sectional area of the heterostructure, and E_g-C3N4 and E_BiOI are the total energies of the isolated g-C₃N₄ and BiOI, respectively. It has been calculated that the binding energies of the heterojunction g-C₃N₄/BiOI (001) is −32.05 meV Å⁻². Due to the negative binding energy, this heterostructure is thermodynamically stable, which means it can be prepared or grown in a practical way.^70,71 Indeed, the thermostability of the heterojunction g-C₃N₄/BiOI (001) is verified via an AIMD simulation at 300 K, shown in Fig. 4(a) and (b). The temperature and total energy of the heterostructure system fluctuate stably with small ups and downs in Fig. 4(a).⁶³ At the same time, it is seen clearly in Fig. 4(b) that the atoms are only displaced slightly near their equilibrium positions. No structural corruption is observed via the structure comparison before and after heating.

Fig. 3 Side view of the g-C₃N₄/BiOI (001) heterostructure models before relaxation and after relaxation. The gray, brown, red, white and blue balls represent bismuth, iodine, oxygen, nitrogen and carbon atoms, respectively.

Fig. 4 (a) Variations of temperature and energy plotted as a function of time for AIMD simulations of g-C₃N₄/BiOI(001), under 300 K for 8 ps with a time step of 1 fs. (b) The bottom parts show the g-C₃N₄/BiOI(001) structures at 0 ps and 8 ps.

3.2 Electronic structure, charge transfer, band alignment and optical absorption spectrum of the g-C₃N₄/BiOI(001) heterojunction

First, the calculated band structures, the total density of states (TDOS) and the projected density of states (PDOS) for the individual g-C₃N₄ (001) and BiOI (001) surfaces are shown in Fig. 5(a) and (b), respectively. It is seen that both monolayer 1 × 5 g-C₃N₄ (001) and 2 × 6 three-layer BiOI (001) surface slabs present the direct band gap at the high symmetry point Γ. Their direct band gap values from the HSE functional calculations are 2.48 and 2.01 eV, respectively, in contrast to the corresponding experimental data 2.7–2.8 eV^26,46,48 and 1.77–2.22 eV.^42,62 The valence band top of g-C₃N₄ is mostly composed of N 2p orbitals, while the conduction band bottom majorly comes from the contribution of C 2p and N 2p orbitals. For the BiOI (001), the valence band top is mainly composed of I 5p orbitals, and the conduction band bottom mainly originates from the Bi 6p orbitals. These results are in agreement with other calculations.^18,72

Fig. 5 The band structures, TDOS and PDOS for (a) monolayer 1 × 5 g-C₃N₄, and (b) three-layer 2 × 6 BiOI (001) surface with 1I termination. In order to make comparisons easier, the Fermi energy level is placed at 0 eV for the sake of convenience.

It is more interesting to explore the projected band structures (Fig. 6(a)) and TDOS and PDOS (Fig. 6(b)–(d)) for the coupled heterostructure g-C₃N₄/BiOI (001). The projected band structures indicate that the theoretical band gaps for g-C₃N₄ and BiOI (001) in the heterostructure are 3.24 and 2.01 eV, respectively. The interaction between the g-C₃N₄ and BiOI (001) layer seems to be clearly changing the band structure of the g-C₃N₄ and enhancing the conduction band minimum (CBM) level. This is similar to the case of g-C₃N₄/ZnO.¹⁷ Nevertheless, the band gap of the g-C₃N₄ part should not be increased markedly, in contrast to that before building the heterojunction. In experiments, the band gap of every component in a heterojunction is usually considered the same as the corresponding one before building.^26,46,48 Therefore, the acquired band gap of the g-C₃N₄ from the projected band structure should be overestimated. In contrast, the heterostructure's band gap is almost the same as the BiOI part but much smaller than the g-C₃N₄ part. Furthermore, the visible light absorption intensity and range of the g-C₃N₄/BiOI (001) in Fig. 7 are almost the same as the BiOI. As a result, visible light absorption should not be the significant factor in enhancing the photocatalytic performance in the g-C₃N₄/BiOI (001) in experiments,^25,47,49,50 in contrast to the individual BiOI. The efficient separation and migration of the photo-generated electrons and holes should be primarily responsible for the improvement. However, how does the process occur?

Fig. 6 Projected band structures, TDOS, and PDOS of the g-C3N4/BiOI (001) heterostructure from HSE functional calculations. The green and blue circles represent the contribution from BiOI and g-C3N4, respectively. The position of the Fermi energy level is set at 0 eV.

Fig. 7 Optical absorption spectra of isolated g-C₃N₄, BiOI (001), and g-C₃N₄/BiOI (001) from HSE hybrid functional calculations.

Fig. 6(a)–(d) show clearly that the conduction band bottom of the g-C₃N₄/BiOI (001) is determined by the contribution of the Bi 6p orbitals and I 5p orbitals in BiOI. Meanwhile, the N 2p orbitals in g-C₃N₄ mainly determine the valence band top of the heterostructure. There is no doubt that the conduction band minimum level (CBM) of the BiOI is lower than the CBM level of the g-C₃N₄. Comparing the average values of the highest valence band energies at the different K points for the g-C₃N₄ and BiOI in the heterostructure, the valence band maximum (VBM) level of the g-C₃N₄ part is considered to be about 0.15 eV higher than that of the BiOI part. The circumstance is similar to the case of the g-C₃N₄/2BN-Gyne.⁷³ Therefore, the valence band top and conduction band bottom originate from the different parts in the g-C₃N₄/BiOI (001), and a staggered projected band structure is presented. This is very expected for the applications of photocatalysis. Depending on the charge transfer, the staggered projected band structure could be a direct Z-scheme or type II band alignment.⁷⁴

To investigate the charge transfer in the interface between the g-C₃N₄ and the BiOI (001), we calculated the work functions of the monolayer g-C₃N₄ and three-layer BiOI (001) utilizing the HSE hybrid functional. This value is the difference between the vacuum level and Fermi level and can be calculated by using the formula:^72,75,76

Φ = E_vac − E_F (7)

where Φ, E_vac and E_F represent the energy of work function, vacuum level, and Fermi level, respectively. The calculated work functions of the monolayer g-C₃N₄ and three-layer BiOI (001) are 4.88 and 6.58 eV, respectively, which are consistent with other HSE calculations,^72,75,76 but are overestimated in contrast to the experimental data 4.3 (g-C₃N₄) and 5.02 eV (BiOI (001).⁴⁹ The overestimation could partially originate from the HSE hybrid functional calculation itself; if the Perdew–Burke–Ernzerhof (PBE) functional is applied, the calculated work functions for the monolayer g-C₃N₄ and three-layer BiOI (001) are 3.86 eV and 5.99 eV, respectively, which are closer to the experimental data. It is seen that both the HSE results and PBE results are qualitatively in accord with the experimental data. Thus, the electron will translate from g-C₃N₄ to BiOI when they are contacted. As can be affirmed from the calculations of the charge density difference, which can be expressed as follows:^63,72,75,76

Δρ = ρ_{(g-C3N4/BiOI)/BiOI (001)} − ρ_g-C3N4 − ρ_BiOI (001) (8)

Among them, ρ_g-C3N4/BiOI, ρ_g-C3N4 and ρ_BiOI (001) represent the charge densities of the g-C₃N₄/BiOI (001) heterojunction, monolayer g-C₃N₄ and three-layer BiOI (001), respectively. Fig. 8(a) displays the calculated results. The cyan area is used to represent electron depletion, and the yellow area is used to represent electron accumulation. The cyan colour dominates a major portion of g-C₃N₄'s surface area, and the yellow colour covers a minor proportion. The yellow area dominates the BiOI (001) surface, but a few parts of the area are covered by cyan. This indicates that electrons are consumed in the g-C₃N₄ and accumulated in the BiOI (001). For a more visual response to the electron transfer, the planar-averaged charge density difference along the Z-axis of the g-C₃N₄/BiOI (001) heterostructure is also calculated. It is shown in Fig. 8(b). The positive value means the accumulation of electrons, and the negative value means the decrease of electrons in Fig. 8(b). In contrast to Fig. 8(a), it is seen that the positive value dominates the region of BiOI, but the negative value dominates the region of g-C₃N₄. Taking the above work function analysis into account, these results are in agreement with the above analysis. Consequently, in the g-C₃N₄/BiOI (001) heterostructure, there is a built-in electric field from the g-C₃N₄ to the BiOI, leading to band edge bending. On the low-Fermi-level semiconductor BiOI, the band edges bend downward at the interface; while on the high-Fermi-level semiconductor g-C₃N₄, the band edges bend upward at the interface.

Fig. 8 (a) The charge density difference of the g-C₃N₄/BiOI (001) heterojunction, and (b) the planar-averaged differential charge density along the Z direction. The cyan region represents electron depletion, and the yellow region represents electron accumulation. The gray, brown, red, white, and blue balls represent bismuth, iodine, oxygen, nitrogen, and carbon atoms, respectively. The isovalues are 0.00175 e Å⁻³.

As a result, the staggered projected band structure in Fig. 6 corresponds to a direct Z-scheme band alignment.⁴⁰ In conjunction with the calculated work function and the projected band structures, the direct Z-scheme band alignment and the charge separation schematic diagram are presented in Fig. 9, in contrast to the vacuum level.^5,16 It is noted that under sunlight irradiation, the electrons are excited from the valence bands (VB) to the conduction bands (CB). At the same time, the holes are generated in the VB. For the photo-generated electrons in the CB of g-C₃N₄/BiOI (001), except for going back to the VB and recombining with the corresponding holes (intralayer recombination), there are three possible migration paths between g-C₃N₄ and BiOI (Fig. 9). In path 1, the photo-generated electrons in the CB of the BiOI transfer towards the VB of the g-C₃N₄ and recombine with the photo-generated holes in the VB of the g-C₃N₄ under the action of the built-in electric field and Coulomb attraction between holes and electrons (interlayer recombination^77–79). In path 2, the photo-generated electrons in the CB of the g-C₃N₄ migrate towards the CB of the BiOI because of the conduction band offset (CBO) potential; similarly, the photo-generated holes in the VB of the BiOI migrate towards the VB of the g-C₃N₄ because of the valence band offset (VBO) potential (type II transfer mode⁴⁰). In path 3, the photo-generated electrons in the CB of the g-C₃N₄ migrate towards the VB of the BiOI because of Coulomb attraction. Comparing three possible transfer paths, path 1 is the most favourable, given the reasons mentioned above. In contrast, path 2 is suppressed with the operation of the built-in electric field from the g-C₃N₄ to BiOI. In addition to this, the suppression is also caused by the Coulomb repulsion between the electrons in BiOI's CB (holes in BiOI's VB) and the electrons in g-C₃N₄'s CB (holes in g-C₃N₄'s VB), as well as the extra potential barrier that is induced by band bending.^77–79 Regarding the long distance between the VBM level of the BiOI and the CBM level of the g-C₃N₄, pathway 3 is the least favourable.⁸⁰ In addition, there is no built-in electric field in the BiOI or g-C₃N₄ for intralayer recombination and the energy gap between the respective VBM level and CBM level is also larger than that between the VBM level of the g-C₃N₄ and the CBM level of the BiOI. Therefore, the intralayer recombination probability of the photo-generated electrons and holes should be much smaller than the interlayer recombination probability.

Fig. 9 The schematic diagram of the band alignment and the charge separation at the interface of the g-C₃N₄/BiOI (001).

The photo-excited electrons with higher reduction capacity in the g-C₃N₄'s CB (the g-C₃N₄'s CBM level is obviously higher than the reduction potential of O₂/˙O₂⁻ (−4.11 eV)⁵³ in Fig. 9) as well as the photo-generated holes with the stronger oxidation capacity in the BiOI (001)'s VB (the VBM level is much lower than the oxidation potential of H₂O/˙OH (−6.43 eV)⁵³ in Fig. 9) are retained. Furthermore, the photo-excited electrons in the g-C₃N₄'s CB have enough reduction ability to reduce O₂ to ˙O₂⁻ and the photo-generated hole in the BiOI (001)'s VB has a strong oxidation ability to produce ˙OH. Then, the produced superoxide ions (˙O₂⁻) and hydroxyl radical (˙OH) as the photocatalytic reactive species can degrade organic pollutants. In contrast, the electrons with the weaker reduction capacity and the holes with the weaker oxidation capacity are eliminated in the heterostructure. This way, the photo-generated electrons and holes can be separated and migrate efficiently. In addition, the separated electrons and holes in the heterostructure g-C₃N₄/BiOI (001) have been shown to be capable of taking part in reduction and oxidation reactions with water or pollutants in an effective manner. This analysis agrees with the experimental results of Zhang et al.⁴⁷ and He et al.,⁴⁹ where superoxide ions (˙O₂⁻), hydroxyl radicals (˙OH) and holes (h⁺) were affirmed as the main photocatalytic reactive species for pollutant degradation. In contrast, if the g-C₃N₄/BiOI (001) band alignment follows that of the type II band alignment, then the electrons accumulating in the CB of BiOI (Fig. 9) do not have enough reduction capacity to reduce O₂ to ˙O₂⁻. This is not consistent with experimental findings.^47,49,50 Therefore, the hybrid density functional calculations evidence the direct Z-scheme experimental speculation of Zhang et al.⁴⁷ and He et al.,⁴⁹ but negate the type II photocatalytic mechanism explanations in the experiment of Alam et al.²⁵ and in the experiment from Tian et al.⁵⁰ where ˙O₂⁻ and ˙OH were also reported as the main active species for pollutant degradation.

On the other hand, the large difference in electron–hole mobility and directional separation is also vital for improving photo-conversion efficiency.⁸¹ The mobility of carriers can be judged by calculating their effective masses. In this case, the effective masses of the electrons and holes in monolayer g-C₃N₄, three-layer BiOI (001) surface, and g-C₃N₄/BiOI (001) heterojunction are calculated, and they are expressed as follows:¹⁶

where E is the band edge energy, k is the wave vector, and ħ is the reduced Planck constant. The calculated results are shown in Table 2. The ‘averaged’ means the arithmetic mean value of the effective masses along the different directions.⁸² From the calculated average value of the effective masses, it is seen that an obvious change after constructing a heterojunction is the significant increase in the difference between the electron and hole's effective masses. While the larger difference in the effective masses of electrons and holes means that they are easier to separate. This is beneficial in decreasing their recombination rate and improving the photocatalytic activity.^81,83 Therefore, the direct Z-scheme charge transfer and separation mechanism, as well as the larger effective mass difference of carriers, excellently explains the superior photocatalytic activity of the g-C₃N₄/BiOI (001) in experiments.^25,47,49,50

Table 2 Calculated effective masses of electrons and holes in monolayer g-C₃N₄ (001), monolayer BiOI (001), and g-C₃N₄ (001)/BiOI (001) heterojunction

System		Γ → X	Γ → Y	X → S	X → Γ	Averaged	or
g-C3N4(001)		0.99	0.42			0.71	3.64
		0.39	1.53			0.96
BiOI(001)		0.20	0.21			0.21	7.5
		1.50	0.28			0.89
g-C3N4/BiOI(001)		0.19	0.28			0.24	23.08
				8.69	2.38	5.54

---

4. Conclusions

Hence, we concluded that the structural model for the heterostructure g-C₃N₄/BiOI (001) could be reasonably determined. Employing first-principles hybrid functional calculations, the mechanism of the photocatalytic performance of g-C₃N₄/BiOI (001) has been systematically investigated. We find that the g-C₃N₄/BiOI (001) structure exhibits a staggered projected band structure with a built-in electric field from the g-C₃N₄ surface to the BiOI surface as a result of our calculations. High-Fermi-level semiconductor g-C₃N₄ has higher band edges, which bend upward, while low-Fermi-level semiconductor BiOI has lower band edges, which bend downward at the interface. The electrons with weaker reduction capacity in the valence bands (VB) of the g-C₃N₄ and the holes with weaker oxidation capacity in the conduction bands (CB) in the BiOI are eliminated in the heterostructure. In contrast, the photo-excited electrons with higher reduction capacity in the CB of the g-C₃N₄ and the photo-generated holes with greater oxidation capacity in the VB of the BiOI (001) are retained. Therefore, it can be said that the g-C₃N₄/BiOI (001) heterostructure has a direct Z-scheme character, and the photo-generated electrons and holes can be separated and migrate highly effectively. The separated electrons and holes can effectively participate in the reduction and oxidation reactions with water/pollutants to produce the photocatalytically reactive species of superoxide ions (˙O₂⁻) and hydroxyl radicals (˙OH), respectively. This is consistent with the experimental results. On the other hand, it is found that an obvious change after constructing a heterojunction is the significant increase in the difference between the electron and hole's effective masses. The direct Z-scheme charge transfer and separation mechanism as well as the larger effective mass difference of carriers excellently explain the superior photocatalytic activity of the g-C₃N₄/BiOI (001) in experiments. In contrast, the theory does not support the speculation from some experiments that the g-C₃N₄/BiOI (001) is attributed to the type II heterostructure.

Conflicts of interest

The authors declare that there are no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

This work is supported by the Doctor Research Fund of China West Normal University (Grants No. 412773), the Sichuan Youth Science and Technology Innovation Research (Grant No. 21XTD0038), the Sichuan Natural Science Foundation (Grant No. 2022NSFSC1841), and the National Natural Science Foundation of China (Grant No. 52262031). This work is carried out at the Shanxi Supercomputing Center of China, and the calculations are performed on TianHe-2. The authors also thank the Supercomputing Center of USTC for providing the computational time.

Notes and references

1. A. Kubacka, M. Fernandez-Garcia and G. Colon, Chem. Rev., 2012, 112, 1555–1614.
2. J. Fu, J. Yu, C. Jiang and B. Cheng, Adv. Energy Mater., 2018, 8, 1701503.
3. H. Khojasteh, M. Salavati-Niasari, H. Safajou and H. Safardoust-Hojaghan, Diamond Relat. Mater., 2017, 79, 133–144.
4. H. Haghjoo, F. S. Sangsefidi and M. Salavati-Niasari, J. Mol. Liq., 2017, 247, 345–353.
5. C. Li, Y. Xu, W. Sheng, W. J. Yin, G. Z. Nie and Z. Ao, Phys. Chem. Chem. Phys., 2020, 22, 615–623.
6. J. Fu, B. Zhu, C. Jiang, B. Cheng, W. You and J. Yu, Small, 2017, 13, 1603938.
7. J. Low, B. Cheng, J. Yu and M. Jaroniec, Energy Storage Mater., 2016, 3, 24–35.
8. J. Jin, J. Yu, D. Guo, C. Cui and W. Ho, Small, 2015, 11, 5262–5271.
9. S. Sorcar, Y. Hwang, C. A. Grimes and S.-I. In, Mater. Today, 2017, 20, 507–515.
10. X. Li, J. Wen, J. Low, Y. Fang and J. Yu, Sci. China Mater., 2014, 57, 70–100.
11. R. M. Navarro Yerga, M. C. Alvarez Galvan, F. del Valle, J. A. Villoria de la Mano and J. L. Fierro, ChemSusChem, 2009, 2, 471–485.
12. K. Maeda and K. Domen, J. Phys. Chem. C, 2007, 111, 7851–7861.
13. G. Dong, K. Zhao and L. Zhang, Chem. Commun., 2012, 48, 6178–6180.
14. F. Hu, W. Luo, Y. Hu, H. Dai and X. Peng, J. Alloys Compd., 2019, 794, 594–605.
15. F. Raziq, M. Humayun, A. Ali, T. Wang, A. Khan, Q. Fu, W. Luo, H. Zeng, Z. Zheng, B. Khan, H. Shen, X. Zu, S. Li and L. Qiao, Appl. Catal., B, 2018, 237, 1082–1090.
16. J. Liu and E. Hua, J. Phys. Chem. C, 2017, 121, 25827–25835.
17. J. Cui, S. Liang, S. Sun, X. Zheng and J. Zhang, J. Phys.: Condens. Matter, 2018, 30, 175001.
18. F. Ling, W. Li and L. Ye, Appl. Surf. Sci., 2019, 473, 386–392.
19. X. Yu, J. Xie, Q. Liu, H. Dong and Y. Li, Colloids Interface Sci., 2021, 593, 133–141.
20. Y. Fu, T. Huang, L. Zhang, J. Zhu and X. Wang, Nanoscale, 2015, 7, 13723–13733.
21. K. Li, Z. Zeng, L. Yan, S. Luo, X. Luo, M. Huo and Y. Guo, Appl. Catal., B, 2015, 165, 428–437.
22. Z. Jin and L. Zhang, J. Mater. Sci. Technol., 2020, 49, 144–156.
23. J. Liao, W. Cui, J. Li, J. Sheng, H. Wang, X. a Dong, P. Chen, G. Jiang, Z. Wang and F. Dong, Chem. Eng. J., 2020, 379, 122282.
24. Y. Li, M. Gu, T. Shi, W. Cui, X. Zhang, F. Dong, J. Cheng, J. Fan and K. Lv, Appl. Catal., B, 2020, 262, 118281.
25. K. M. Alam, P. Kumar, P. Kar, U. K. Thakur, S. Zeng, K. Cui and K. Shankar, Nanoscale Adv., 2019, 1, 1460–1471.
26. X. Zhou, C. Shao, S. Yang, X. Li, X. Guo, X. Wang, X. Li and Y. Liu, ACS Sustainable Chem. Eng., 2018, 6, 2316–2323.
27. K. Dai, L. Lu, C. Liang, G. Zhu, Q. Liu, L. Geng and J. He, Dalton Trans., 2015, 44, 7903–7910.
28. J. Jiang, Z. Mu, P. Zhao, H. Wang and Y. Lin, Mater. Chem. Phys., 2020, 252, 123194.
29. S. Vinoth, P. M. Rajaitha and A. Pandikumar, New J. Chem., 2021, 45, 2010–2018.
30. W. Zhang, Z. Zhao, F. Dong and Y. Zhang, Chin. J. Catal., 2017, 38, 372–378.
31. X. Bai, L. Wang, R. Zong and Y. Zhu, J. Phys. Chem. C, 2013, 117, 9952–9961.
32. B. Lin, S. Chen, F. Dong and G. Yang, Nanoscale, 2017, 9, 5273–5279.
33. T. Di, B. Zhu, B. Cheng, J. Yu and J. Xu, J. Catal., 2017, 352, 532–541.
34. X. Tan, X. Wang, H. Hang, D. Zhang, N. Zhang, Z. Xiao and H. Tao, Opt. Mater., 2019, 96, 109266.
35. Y. Zhang, S.-M. Zhao, Q.-W. Su and J.-L. Xu, Rare Met., 2019, 40, 96–104.
36. R. Hao, G. Wang, H. Tang, L. Sun, C. Xu and D. Han, Appl. Catal., 2016, 187, 47–58.
37. X. Xiao and W.-D. Zhang, RSC Adv., 2011, 1, 1099–1105.
38. X. Xiao, R. Hao, M. Liang, X. Zuo, J. Nan, L. Li and W. Zhang, J. Hazard. Mater., 2012, 233–234, 122–130.
39. X. Yu, J. Yang, K. Ye, X. Fu, Y. Zhu and Y. Zhang, Inorg. Chem. Commun., 2016, 71, 45–49.
40. Q. Xu, L. Zhang, J. Yu, S. Wageh, A. A. Al-Ghamdi and M. Jaroniec, Mater. Today, 2018, 21, 1042–1063.
41. J. Bai, Y. Li, P. Wei, J. Liu, W. Chen and L. Liu, Small, 2019, 15, e1900020.
42. S. Han, J. Li, K. Yang and J. Lin, Chin. J. Catal., 2015, 36, 2119–2126.
43. S. Juntrapirom, D. Tantraviwat, S. Suntalelat, O. Thongsook, S. Phanichphant and B. Inceesungvorn, J. Colloid Interface Sci., 2017, 504, 711–720.
44. X.-J. Wen, C.-G. Niu, L. Zhang, C. Liang and G.-M. Zeng, J. Catal., 2017, 356, 283–299.
45. S. Guo, H. Luo, Y. Li, J. Chen, B. Mou, X. Shi and G. Sun, J. Alloys Compd., 2021, 852, 157026.
46. H. An, B. Lin, C. Xue, X. Yan, Y. Dai, J. Wei and G. Yang, Chin. J. Catal., 2018, 39, 654–663.
47. J. Zhang, J. Fu, Z. Wang, B. Cheng, K. Dai and W. Ho, J. Alloys Compd., 2018, 766, 841–850.
48. J. Di, J. Xia, S. Yin, H. Xu, L. Xu, Y. Xu, M. He and H. Li, J. Mater. Chem. A, 2014, 2, 5340–5351.
49. R. He, K. Cheng, Z. Wei, S. Zhang and D. Xu, Appl. Surf. Sci., 2019, 465, 964–972.
50. N. Tian, H. Huang, S. Wang, T. Zhang, X. Du and Y. Zhang, Appl. Catal., B, 2020, 267, 118697.
51. H. Zhang, J. Yang, L. Guo, R. Wang, S. Peng, J. Wang, J. Wan and J. Xu, Chem. Phys. Lett., 2021, 762, 138143.
52. J. Hou, T. Jiang, X. Wang, G. Zhang, J.-J. Zou and C. Cao, J. Cleaner Prod., 2021, 287, 125072.
53. C. R. Chen, X. D. Wang, S. L. Wang, J. L. Zhang and B. Z. Tian, ChemistrySelect, 2020, 5, 15084–15090.
54. R. Wang, J. Wu, X. Mao, J. Wang, Q. Liu, Y. Qi, P. He, X. Qi, G. Liu and Y. Guan, Appl. Surf. Sci., 2021, 556, 149804.
55. J. C. Wang, H. C. Yao, Z. Y. Fan, L. Zhang, J. S. Wang, S. Q. Zang and Z. J. Li, ACS Appl. Mater. Interfaces, 2016, 8, 3765–3775.
56. G. Kresse and J. Furthmuller, Phys. Rev. B: Condens. Matter Mater. Phys., 1996, 54, 11169–11186.
57. G. Kresse and J. Furthmüller, Comput. Mater. Sci., 1996, 6, 15–50.
58. J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1996, 77, 3865–3868.
59. S. Grimme, S. Ehrlich and L. Goerigk, J. Comput. Chem., 2011, 32, 1456–1465.
60. Z. Zhang, Q. Qian, B. Li and K. J. Chen, ACS Appl. Mater. Interfaces, 2018, 10, 17419–17426.
61. J. Heyd, G. E. Scuseria and M. Ernzerhof, J. Chem. Phys., 2003, 118, 8207–8215.
62. C. Chang, H.-C. Yang, N. Gao and S.-Y. Lu, J. Alloys Compd., 2018, 738, 138–144.
63. H. Guo, Z. Zhang, B. Huang, X. Wang, H. Niu, Y. Guo, B. Li, R. Zheng and H. Wu, Nanoscale, 2020, 12, 20025–20032.
64. L. Matthes, P. Gori, O. Pulci and F. Bechstedt, Phys. Rev. B: Condens. Matter Mater. Phys., 2013, 87, 035438.
65. V. Wang, N. Xu, J.-C. Liu, G. Tang and W.-T. Geng, Comput. Phys. Commun., 2021, 267, 108033.
66. G. Di Liberto, L. A. Cipriano, S. Tosoni and G. Pacchioni, Chemistry, 2021, 27, 13306–13317.
67. T. Jing, Y. Dai, X. Ma, W. Wei and B. Huang, Phys. Chem. Chem. Phys., 2016, 18, 7261–7268.
68. T. Kong, X. Wei, G. Zhu and Y. Huang, J. Mater. Sci., 2017, 52, 5686–5695.
69. M.-Y. Yan, Z.-Y. Jiang, J.-M. Zheng, Y.-M. Lin and Z.-Y. Zhang, Appl. Surf. Sci., 2020, 531, 147318.
70. B. Kong, Z. Liu, J. Wang, T. Chen, M. Zhang, T. Zeng and W. Wang, J. Phys. Chem. C, 2022, 126, 11788–11799.
71. H. Chen, J. Zhao, X. Wang, X. Chen, Z. Zhang and M. Hua, Nanoscale, 2022, 14, 5551–5560.
72. J. Wang, Y. Huang, J. Guo, J. Zhang, X. Wei and F. Ma, J. Solid State Chem., 2020, 284, 121181.
73. J. Yun, Y. Zhang, Y. Ren, P. Kang, J. Yan, W. Zhao, Z. Zhang and H. Guo, Sol. Energy Mater. Sol. Cells, 2020, 210, 110516.
74. Z. Zhang, B. Huang, Q. Qian, Z. Gao, X. Tang and B. Li, APL Mater., 2020, 8, 041114.
75. J. Liu, B. Cheng and J. Yu, Phys. Chem. Chem. Phys., 2016, 18, 31175–31183.
76. W. Yao, J. Zhang, Y. Wang and F. Ren, Appl. Surf. Sci., 2018, 435, 1351–1360.
77. J. Abdul Nasir, A. Munir, N. Ahmad, T. U. Haq, Z. Khan and Z. Rehman, Adv. Mater., 2021, 33, e2105195.
78. L. Zhang, J. Zhang, H. Yu and J. Yu, Adv. Mater., 2022, 34, e2107668.
79. X. Li, C. Garlisi, Q. Guan, S. Anwer, K. Al-Ali, G. Palmisano and L. Zheng, Mater. Today, 2021, 47, 75–107.
80. X. Jiang, Q. Gao, X. Xu, G. Xu, D. Li, B. Cui, D. Liu and F. Qu, Phys. Chem. Chem. Phys., 2021, 23, 21641–21651.
81. Y. Liu, P. Lv, W. Zhou and J. Hong, J. Phys. Chem. C, 2020, 124, 9696–9702.
82. S. Tian, Y. F. Ding, M. Q. Cai, L. Chen, C. T. Au and S. F. Yin, Phys. Chem. Chem. Phys., 2021, 23, 17894–17903.
83. M. Dong, J. Zhang and J. Yu, APL Mater., 2015, 3, 104404.

---