Charge Steering in Heterojunction Photocatalysis: General Principles, Design, Construction, and Challenges

Abstract

Steering charge kinetics is a key to optimizing quantum efficiency. Advancing the design of photocatalysts (ranging from single semiconductor to multicomponent semiconductor junctions) that promise improved photocatalytic performance for converting solar to chemical energy, entails mastery of increasingly more complicated processes. Indeed, charge kinetics become more complex as both charge generation and charge consumption may occur simultaneously on different components, generally with charges being transferred from one component to another. Capturing detailed charge dynamics information in each heterojunction would provide numerous significant benefits for applications and has been needed for a long time. Here, the steering of charge kinetics by modulating charge energy states in the design of semiconductor–metal‐interface‐based heterogeneous photocatalysts is focused. These phenomena can be delineated by separating heterojunctions into classes exhibiting either Schottky/ohmic or plasmonic effects. General principles for the design and construction of heterojunction photocatalysts, including recent advances in the interfacing of semiconductors with graphene, carbon quantum dots, and graphitic carbon nitride are presented. Their limitations and possible future outlook are brought forward to further instruct the field in designing highly efficient photocatalysts.

Steering charge kinetics is a key to optimizing quantum efficiency. Advancing the design of photocatalysts that promise improved photocatalytic performance for converting solar to chemical energy, entails mastery of increasingly more complicated processes.

1. Introduction

Developing clean, low‐cost, and renewable fuel sources is a key challenge for meeting the energy demands of a fast‐growing population and increased industrialization, which at the same time presents a promising route for mitigating environmental issues caused by the combustion of fossil fuels.^[ 1, 2, 3 ^] However, obtaining sufficient efficiency in large‐scale applications for converting solar energy into fuels and/or electricity is still a challenge. Holding so much promise for contributing to a sustainable energy future, photocatalytic splitting of water for the production of hydrogen and oxygen has received a great deal of attention from all over the world^{[ 4 ]} since the first demonstration by Fujishima and Honda using TiO₂ and ultraviolet radiation (UV).^{[ 5 ]} As a result, several worthwhile single semiconductor photocatalysts utilizing Cu₂O,^[ 6, 7 ^] ZnO,^{[ 8 ]} WO₃,^[ 9, 10, 11 ^] etc. have been studied exhaustively aiming to optimize and achieve efficient semiconductor photocatalysis.

A semiconductor material is characterized by the highest occupied energy band (the valence band, VB), the lowest empty band (the conduction band, CB), and an energy bandgap (E _g) separating the two, i.e., the energy region where no electronic states exist due to quantization of energy. Semiconductor‐based photocatalysis generally depends upon the generation of photoexcited charge carriers. Specifically, electrons at the VB maximum of the semiconductor are driven to the CB minimum by incident light, leaving oxidative holes at the VB maximum. In general, the process of semiconductor‐based photocatalysis consists of four main steps: 1) absorption of photons with energies equal to or greater than the bandgap; 2) charge separation of photoexcited electron–hole pairs in the bulk catalyst; 3) migration of charge carriers from the bulk to active sites on the surface of the catalyst; and 4) occurrence of chemical redox reactions (such as the splitting of water into hydrogen and oxygen) driven by energetic electrons or holes at active sites.^[ 2, 4, 12, 13 ^]

However, in the case of single semiconductors, there exist severe performance bottlenecks preventing their use in practical photocatalytic applications. For example, there are mutual tradeoffs between the energy region of the solar spectrum to which the semiconductor responds and the energetics of the reduction/oxidation (redox) reactions it is capable of catalyzing. Photons cannot generate e–h pairs unless their energies exceed or equal the bandgap (E _g) of the semiconductor. It is also necessary to adjust band structures to enable a response to a wider region of the solar spectrum to achieve higher efficiency. The viability of the overall photocatalytic reaction requires the position of the CB (VB) edge to be higher (lower) than the potential of the reduction (oxidation) half‐reaction, respectively. For instance, in the splitting of water, the bottom of the CB should be at a more negative potential than the H⁺ to H₂ reduction potential (0 V vs NHE at pH = 0), whereas the top of the VB must be located at a value more positive than the H₂O to O₂ oxidation potential (1.23 V vs NHE at pH = 0), making many otherwise promising semiconductors unsuitable for water‐splitting applications.^[ 14, 15 ^]

In single semiconductors, the relationship between the energy of absorbed photons and the redox potential of charge carriers is irreconcilable and incompatible. For example, for TiO₂, only the UV region (≈5% of the solar spectrum) can be utilized. Moreover, the migration of photogenerated electrons and holes in single semiconductors often results in a high probability of unwanted charge recombination and low efficiency of charge migration to redox sites, along with an increased probability of encountering defects (trapping centers). In contrast to bulk single semiconductor photocatalysts, low‐dimensional crystals have charge carriers that can travel thousands of interatomic distances without scattering. The reduced dimensionality of these crystals maximizes not only their surface area but also the surface per quantity of electrons available for enhanced photocatalytic activity.^[ 16, 17 ^] In general, low‐dimensional materials have shown appreciable differences in minimizing the recombination rate of electron–hole pairs as well as other electronic, catalytic, optical, and mechanical properties compared to their bulk counterparts.^[ 18, 19, 20, 21, 22, 23, 24 ^]

Overall, the key issues for enhancing photocatalytic performance are the improvement of the light absorption characteristics of catalytic systems, the enhancement of the separation of effective charge carriers, and the enlargement of catalytic surface area— all of which are limited by the intrinsic nature of single semiconductors. Additional critical factors required by practical applications include high chemical stability, high precision and flexibility of combinations of crystal structures and defects, optimum photocatalyst band positions, and low‐cost. To overcome the shortcomings of single semiconductors, creation of heterojunction structures has been proposed, such as by interfacing more than one semiconductor (photocatalyst), semiconductor–metal, semiconductor–semiconductor–metal, or semiconductor–metal–semiconductor layers in combination.

2. Charge Steering in Heterojunction Photocatalysis

The fundamental principle of building a semiconductor‐based photocatalytic heterostructure is to make full use of the advantages of each component by rationally arranging component geometry to maximize overall photocatalytic performance. Heterojunctions of p–n semiconductor (p–n), semiconductor–metal (s–m), semiconductor–semiconductor–metal, and Z‐scheme semiconductor (Z‐scheme) hybrid systems are the basic structures of well‐documented designs for effective photocatalytic systems. The improved photocatalytic efficiency of these is mainly attributed to increased rates of charge separation and migration and utilization of a greater portion of the broad solar spectrum. Establishing relationships between the parameters characterizing these heterojunctions within each reaction step is important for gaining charge kinetics information of a given photocatalytic system.

To improve the design of heterojunction photocatalysts, tuning charge kinetics dynamics to improve charge separation and minimize loss of energy during charge carrier migration is critical for quantum yield optimization. Designs of multicomponent semiconductor junctions have long sought to steer charge flows and attain more efficient charge separation. Inherently, the charge kinetics becomes more complicated in these systems because both charge generation, as well as charge consumption, may simultaneously take place on different components and because charge is generally transferred from one component to another. However, capturing detailed charge dynamics information at each heterojunction benefits numerous important applications.

This review focuses on the steering of charge kinetics in different semiconductor heterojunction systems to improve charge separation as determined by the nature of the generated internal electric field and characteristics of band bending at the junction. Further, it covers recent work on heterojunctions including: 1) Schottky/ohmic junctions and plasmonic effects models; 2) materials incorporating semiconductor–graphene, semiconductor(S)–graphitic carbon nitride (C₃N₄), semiconductor–(RGO/metal)–graphitic carbon nitride (g–C₃N₄); 3) recent fascinating investigations of CQDs incorporated into graphitic carbon nitride heterojunctions, and the advances achieved by each for enhancing the overall photocatalytic process.

3. p‐Type–n‐Type (p–n) Junction photocatalysis

In the design of multicomponent semiconductor junctions, the steering of charge flow has long been recognized as the key for creating efficient charge separation. Yongsheng et al. first showed that the p–n heterojunction enhances photocatalytic activity more efficiently than conventional heterojunctions (i.e., heterojunctions composed of different materials or the same materials (see Figure  1 ) n‐type–n‐type or p‐type–p‐type heterojunctions);^{[ 25 ]} subsequently, this finding has been confirmed by many others.^[ 2, 15, 26, 27, 28 ^]

Figure 1.

a,b) Schematic representations of a p–n heterojunction between two semiconductors without an externally applied voltage (a), and electron–hole separation in a p–n heterojunction photocatalyst under the influence of the internal electric field upon light illumination (b).

The basic principle of a p–n system consists of two different components—a p‐type and an n‐type semiconductor—that are in direct contact, as depicted in Figure 1a. Each electron from the n‐type semiconductor that diffuses into the p‐type semiconductor leaves a positive charge behind; similarly, a hole migrating from the p‐type to the n‐type semiconductor leaves a negative charge. Electron–hole diffusion continues until the system achieves Fermi‐level equilibrium. As a result, a charged region forms close to the p–n interface, the so‐called internal electric field. With the formation of the internal electric field, after light excitation, photoexcited electrons transfer from high CB to low CB and holes from low VB to high VB, and these e–h pairs remain well separated, as shown in Figure 1b.

Conventionally, semiconductor 1–semiconductor 2 hybrid junctions are classified as Type I, Type II, or Type III, as shown in Figure  2 . When a Type‐I p–n heterojunction is irradiated with light with energy greater than or equal to the bandgap, the photogenerated electrons migrate from the higher CB to the lower CB and the holes from lower VB to higher VB, as depicted in Figure 2a. Similarly, migration of photoexcited electrons and holes in Type‐II and Type‐III p–n junctions occurs from high CB to low CB and low VB to high VB, and effective charge separation can be obtained in either, as shown in Figure 2b,c. In the latter two types, separation results from the migration of high‐energy electrons and the opposite movement of high energy holes, whereas in the straddling bandgap (Type I), both high‐energy electrons and holes move to the same semiconductor, which disfavors improvement of photocatalytic activity.

Figure 2.

a–c) Schematic of the three different types of p–n junction photocatalysts: (a) Type I, (b) Type II, and (c) Type III upon light exposure.

Notably, Type‐II p–n heterojunction photocatalysts offer favorable band alignments for efficient charge carrier separation. Moreover, rapid charge transfer assisted by the internal electric field formed at the interface junction is beneficial for producing rates of charge separation that lead to enhanced photocatalytic reactions. One of the most investigated Type‐II p–n multicomponent junctions is the Cu₂O/TiO₂ heterojunction photocatalyst in which p‐type Cu₂O and n‐type TiO₂ contact directly.^{[ 29 ]} Recently, C. Ding et al. synthesized a highly efficient Cu₂O–TiO₂ heterojunction by a wet chemical method. The Cu₂O–TiO₂ heterojunction prepared in this way achieves an improved methylene blue degradation rate of 93.67% in 45 min, more efficient than pristine Cu₂O. At the p–n junction, free electrons diffuse from TiO₂ to Cu₂O and holes from Cu₂O to TiO₂ until equilibrium is reached. Consequently, they leave negative and positive regions at the Cu₂O/TiO₂ interface, which forms a space charge region that results in an internal electric field. Upon irradiation of light, photoexcited electrons transfer from the CB of Cu₂O (E _g = 2−2.2 eV)^[ 30, 31, 32 ^] to the CB of TiO₂ (E _g = 3.2 eV) and photoexcited holes migrate from the VB of TiO₂ to the VB of Cu₂O, a condition that favors reduction at TiO₂ and oxidation at Cu₂O. The formation of the internal electric field minimizes the recombination probability of photoexcited electron–hole pairs and results in improved photocatalytic activity, as illustrated in Figure 1b.

The study of p–n heterojunctions has been extended to 2D p–n junction nanosheets. The combined experimental and theoretical work of Nan et al. on MoS₂/TiO₂ (a Type‐II p–n junction), demonstrates that enhanced visible light absorption can result in enhanced photocatalytic activity.^{[ 33 ]} Their photocurrent density analysis shows that the MoS₂/TiO₂ heterojunction has 17.8 times higher activity than that of pristine TiO₂. The photocatalytic degradation enhancement factor of the corresponding kinetic constant is about 5.2. Such a dramatic improvement originates: 1) from the incorporation of MoS₂ nanosheets that improve light‐harvesting performance; and 2) from the favorable band alignment that results in fast and efficient charge separation of photogenerated charge carriers. Additional comprehensive studies of many outstanding p–n junction designs are outlined in Table  1 .

Table 1.

Summary of representative p–n heterojunctions reported so far

Sample/model	Heterojunction class	Oxidation site	Reduction site	Charge migration direction across the interface	Activity test/Application	References
NiO–ZnO	Type II	NiO	ZnO	NiOZnO, ZnONiO	Degradation of rhodamine B (RhB)	[88, 89, 90, 91, 92, 93]
NiO–TiO2	Type II	NiO	TiO2	NiOTiO2, TiO2 NiO	Photocatalytic reduction of Cr2O7 2− and photocatalytic oxidation of RhB,hydrogen generation	[94, 95, 96]
NiO–SnO2	Type II	NiO	SnO2	NiOSnO2, SnO2 NiO	Degradation of RhB	[97]
Cu2O–TiO2	Type II	Cu2O	TiO2	Cu2OTiO2, TiO2 Cu2O	Decomposition of p‐Nitrophenol, Orange II oxidation, MB degradation, 2,4,6‐trichloro‐phenol degradation, enhanced photocurrents,hydrogen generation	[29, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108]
CuO–In2O3	Type II	CuO	In2O3	CuO In2O3, In2O3 CuO	Degradation of RhB	[109]
ZnS–Cu2S	Type I	–	Cu2S	ZnSCu2S	Photocatalytic water splitting	[110]
Cu2O–CdS	Type II	Cu2O	CdS	Cu2OCdS, CdSCu2O	Hydrogen Evolution	[111, 112]
CuO–ZnO	Type II	CuO	ZnO	CuOZnO, ZnOCuO	Photodegradation of Phenol, azo dye, (MB), degradation of Congo red, and benzoic acid, photocurrent production	[93, 113, 114, 115, 116, 117, 118]
CuO–SnO2	Type II	CuO	SnO2	CuOSnO2, SnO2 CuO	Decomposition of MB	[119]
Cu2O–ZnO	Type II	Cu2O	ZnO	Cu2O ZnO, ZnOCu2O	Methyl orange (MO) photodegradation, reduction of methylviologen (MV2+)	[120, 121, 122, 123, 124]
CuS–ZnO	Type II	CuS	ZnO	CuS ZnO, ZnOCuS	Decomposition of MB	[125]
CuS–TiO2	Type II	CuS	TiO2	CuSTiO2, TiO2 CuS	H2 production	[126]
CuS–CdS	Type II	CdS	CuS	CdSCuS, CuSCdS	Degradation of MB and H2 production	[127, 128]
CuS–CdS	Type II	CuS	CdS	CuSCdS, CdSCuS	H2 production	[129]
Cu2S–CdS	Type II	Cu2S	CdS	Cu2SCdS, CdSCu2S	H2 production	[130]
Cu2S–ZnO	Type II	Cu2S	ZnO	Cu2SZnO, ZnOCu2S	Hydrogen generation	[131]
Cu3P–CdS	Type II	Cu3P	CdS	Cu3PCdS, CdSCu3P	H2 production	[132]
Cu3P–TiO2	Type II	Cu3P	TiO2	Cu3PTiO2, TiO2 Cu3P	H2 production	[133]
Ag2O–TiO2	Type II	Ag2O	TiO2	Ag2OTiO2, TiO2 Ag2O	Degradation of MO	[134]
Ag2O–ZnO	Type II	Ag2O	ZnO	Ag2O ZnO, ZnOAg2O	MB degradation	[135]
SnO–SnO2	Type II	SnO	SnO2	SnOSnO2, SnO2 SnO	MB degradation	[136]
PbS –SnO2	Type II	PbS	SnO2	PbSSnO2, SnO2 PbS	Degradation of RhB	[137]
SnO–ZnO	Type II	SnO	ZnO	SnOZnO, ZnOSnO	MB degradation	[138]
SnS–ZnO	Type II	SnS	ZnO	SnSZnO, ZnOSnS	Degradation of MO, RhB	[139]
Ag2O–Bi2O2CO3	Type II	Ag2O	Bi2O2CO3	Ag2OBi2O2CO3, Bi2O2CO3 Ag2O	Degradation of RhB, MB, and MO	[140]
Ag2O–Bi2WO6	Type II	Ag2O	Bi2WO6	Ag2OBi2WO6, Bi2WO6 Ag2O	Degradation RhB	[141]
Co3O4/BiVO4	Type I	Co3O4	Co3O4	BiVO4 Co3O4, BiVO4 Co3O4	Degradation of Phenol,RhB	[142, 143]
m‐BiVO4‐γ‐Bi2O3	Type II	γ‐Bi2O3	m‐BiVO4	γ‐Bi2O3 m‐BiVO4, m‐BiVO4 γ‐Bi2O3	Degradation RhB	[144, 145]
Co3O4/Fe2O3	Type II	Co3O4	Fe2O3	Co3O4 Fe2O3, Fe2O3 Co3O4	Water Splitting	[146]
Co3O4/WO3	Type II	Co3O4	WO3	Co3O4 WO3, WO3 Co3O4	2‐propanol decomposition	[147]
CoP3/Ni2P	Type II	CoP3	Ni2P	CoP3 Ni2P, Ni2PCoP3	H2 production	[148]
Bi2O3–TiO2	Type II	Bi2O3	TiO2	Bi2O3 TiO2, TiO2 Bi2O3	Degradation of p‐chlorophenol, MB, pentachlorophenol	[99, 149]
Bi2O3–SnO2	Type II	Bi2O3‐	SnO2	Bi2O3 SnO2, SnO2 Bi2O3	Degradation of RhB	[150]
Bi2O3–ZnO	Type II	Bi2O3	ZnO	Bi2O3 ZnO, ZnO Bi2O3	Degradation of alizarin red (AR) dye	[151]
V2O5–TiO2	Type I/II	V2O5	TiO2	V2O5 TiO2, TiO2 V2O5	RhB degradation, H2 production	[152, 153, 154]
CuO–TiO2	Type II	CuO	TiO2	CuO TiO2, TiO2 CuO	H2 production, Orange II oxidation	[155, 156, 157]
CeO2–TiO2	Type II	CeO2	TiO2	CeO2 TiO2, TiO2 CeO2	Oxidative degradation of CV dye (Crystal Violet)	[158]
CeO2–Ag3PO4	Type II	CeO2	Ag3PO4	CeO2 Ag3PO4, Ag3PO4 CeO2	Degradation of MB, RhB, and ciprofloxacin (CIP)	[159]
NiO–SrTiO3	Type II	NiO	SrTiO3	NiOSrTiO3, SrTiO3 NiO	Enhanced photocurrents	[160]
CuO–BaTiO3	Type I	CuO	CuO	BaTiO3 CuO, BaTiO3 CuO	MO degradation	[161]
Cu2O–SrTiO3	Type II	Cu2O	SrTiO3	Cu2O SrTiO3, SrTiO3 Cu2O	Photodegradation of tetracycline (TC)	[162]
CuO–PbTiO3	Type II	CuO	PbTiO3	CuOPbTiO3, PbTiO3 CuO	Degradation of malachite green	[163]
α‐Fe2O3/Cu2O	Type II	α‐Fe2O3	Cu2O	α‐Fe2O3 Cu2O, Cu2Oα‐Fe2O3	Photoreduction of CO2	[38]
α‐Fe2O3/ZnO	Type II	α‐Fe2O3	ZnO	α‐Fe2O3 ZnO, ZnOα‐Fe2O3	H2 production	[164]
α‐Fe2O3/CuPc	Type II	α‐Fe2O3	CuPc	α‐Fe2O3 CuPc, CuPcα‐Fe2O3	Photoreduction of CO2	[165]
Cu2O–Fe2O3	Type II	Cu2O	Fe2O3	Cu2O Fe2O3, Fe2O3 Cu2O	Degradation of MB and RhB	[166]
MoS2–CeO2	Type II	MoS2	CeO2	MoS2 CeO2, CeO2 MoS2	H2 production	[167]
MoS2–CdS	Type I	MoS2	MoS2	CdSMoS2, CdSMoS2	H2 production	[168, 169, 170]
MoS2–TiO2	Type II	MoS2	TiO2	MoS2 TiO2, TiO2 MoS2	Photodegradation of 4‐CP, RhB, hydrogen evolution	[33, 127, 171]
MoS2–TiO2	Type II	TiO2	MoS2	TiO2 MoS2, MoS2 TiO2	Hydrogen evolution	[172]
MoS2–ZrO2	Type I	MoS2	MoS2	ZrO2 MoS2, ZrO2 MoS2	Degradation of MO	[173]
MoS2–SnO2	Type II	MoS2	SnO2	MoS2 SnO2, SnO2 MoS2	Methylene blue (MB) degradation	[174]
MoS2–WO3	Type II	MoS2	WO3	MoS2 WO3, WO3 MoS2	Degradation of Congo red (CR)	[175]
MoS2–ZnS	Type I	MoS2	MoS2	ZnSMoS2, ZnSMoS2	Malachite green dye degradation	[176]
α‐MoO3–MoS2	Type II	MoS2	α‐MoO3	MoS2 α‐MoO3, α‐MoO3 MoS2	RhB degradation	[177]
MoS2–WSe2	Type II	MoS2	WSe2	MoS2 WSe2, WSe2 MoS2	Hydrogen evolution	[178, 179]
BiOI–TiO2	Type I	BiOI	BiOI	TiO2 BiOI, TiO2 BiOI	MO degradation	[180, 181]
BiOBr–TiO2	Type II	TiO2	BiOBr	TiO2 BiOBr, BiOBrTiO2	Degradation RhB and MO	[182, 183]
BiOI–SnS2	Type II	BiOI	SnS2	BiOISnS2, SnS2 BiOI	RhB degradation	[184]
BiOI–TiO2	Type II	BiOI	TiO2	BiOITiO2, TiO2 BiOI	MO degradation, RhB degradation	[185, 186]
BiOI–WO3	Type II	BiOI	WO3	BiOIWO3, WO3 BiOI	Degradation tetracycline (TC)	[187]
ZnMn2O4–TiO2	Type I	ZnMn2O4	ZnMn2O4	TiO2 ZnMn2O4, TiO2 ZnMn2O4	Orange II oxidation	[99]
Ag3PO4–TiO2	Type II	Ag3PO4	TiO2	Ag3PO4 TiO2 TiO2 Ag3PO4	Degradation RhB and MB	[188]
Ag3PO4–BiVO4	Type II	Ag3PO4	BiVO4	Ag3PO4 BiVO4, BiVO4 Ag 3PO4	Degradation RhB and MB	[189]
CuInS2–TiO2	Type II	CuInS2	TiO2	CuInS2 TiO2, TiO2 CuInS2	4‐nitrophenol degradation	[190]
Bi12TiO20–TiO2	Type II	Bi12TiO20	TiO2	Bi12TiO20 TiO2, TiO2 Bi12TiO20	RhB degradation	[191]
YFeO3–TiO2	Type II	YFeO3	TiO2	YFeO3 TiO2, TiO2 YFeO3	Benzene oxidation, Orange II oxidation	[192]
CaFe2O4/MgFe2O4	Type II	CaFe2O4	MgFe2O4	CaFe2O4 MgFe2O4, MgFe2O4 CaFe2O4	Degradation of isopropyl alcohol hydrogen evolution	[193]
BiOCl–SrFe12O19	Type I	SrFe12O19	SrFe12O19	BiOClSrFe12O19, BiOClSrFe12O19	MB degradation	[194]
BiOCl–TiO2	Type II	TiO2	BiOCl	TiO2 BiOCl, BiOClTiO2	MB degradation	[195]
BiOCl–SnO2	Type II	BiOCl	SnO2	BiOClSnO2, SnO2 BiOCl	RhB degradation	[196]
BiOCl–BiVO4	Type II	BiOCl	BiVO4	BiOClBiVO4, BiVO4 BiOCl	MO photodegradation	[197]
BiOI–CdWO4	Type II	BiOI	CdWO4	BiOICdWO4, CdWO4 BiOI	Degradation of TC and RhB	[198]
BiOBr–ZnO	Type II	BiOBr	ZnO	BiOBrZnO, ZnOBiOBr	MO decolorization	[199]
BiOI–ZnO	Type II	BiOI	ZnO	BiOIZnO, ZnOBiOI	Degradation of CR,MO	[200, 201]
BiOBr–CeO2	Type II	CeO2	BiOBr	CeO2 BiOBr, BiOBrCeO2	Degradation of RhB, MB, and bisphenol A (BPA)	[202]
BiOBr–WO3	Type II	BiOBr	WO3	BiOBrWO3, WO3 BiOBr	Degradation of RhB, MO, and para‐chlorophenol (4‐CP)	[203]
BiOBr–BiVO4	Type II	BiOBr	BiVO4	BiOBrBiVO4, BiVO4 BiOBr	Degradation of MB	[204, 205]
BiOBr–BiVO4	Type I	BiVO4	BiVO4	BiOBrBiVO4, BiOBrBiVO4	Degradation of MB	[206]
Ag3VO4–WO3	Type II	Ag3VO4	WO3	Ag3VO4 WO3, WO3 Ag3VO4	Degradation of TC	[159]
Bi2O3–WO3	Type II	Bi2O3	WO3	Bi2O3 WO3, WO3 Bi2O3	Decomposition of RhB and 4‐nitroaniline (4‐NA)	[207]
Bi2O3–BaTiO3	Type II	BaTiO3	Bi2O3	BaTiO3 Bi2O3, Bi2O3 BaTiO3	Degradations of MO and MB	[208]
AgBr–BiPO4	Type II	AgBr	BiPO4	AgBrBiPO4, BiPO4 AgBr	MB degradation	[209]
Ag3PO4–AgBr	Type II	AgBr	Ag3PO4	AgBr Ag3PO4, Ag3PO4 AgBr	Degradation of MB and RhB	[210, 211]
BiOI–La2Ti2O7	Type II	BiOI	La2Ti2O7	BiOILa2 Ti2O7, La2Ti2O7 BiOI	Degradation of RhB, dye X‐3B, MO	[212]
Cu2O–Ta2O5	Type II	Cu2O	Ta2O5	Cu2OTa2O5, Ta2O5 Cu2O	Generated hydrogen and oxygen	[213]
CuCrO2–SnO2	Type II	CuCrO2	SnO2	CuCrO2 SnO2, SnO2 CuCrO2	Degradation of MB	[214]
CuCo2O4–TiO2	Type II	CuCo2O4	TiO2	CuCo2O4 TiO2, TiO2 CuCo2O4	Hydrogen evolution, photocurrent density	[215]
CdS–LaFeO3	Type II	CdS	LaFeO3	CdSLaFeO3, LaFeO3 CdS	Degradation MB, RhB, and MO	[216]
CuO–BiFeO3	Type II	CuO	BiFeO3	CuO BiFeO3, BiFeO3 CuO	Degradation of MO	[217]
BiFeO3–ZnO	Type II	BiFeO3	ZnO	BiFeO3 ZnO, ZnOBiFeO3	Photodegradation of 2,4‐dichlorophenol and RhB	[218]
BiVO4–Mn3O4	Type II	Mn3O4	BiVO4	Mn3O4 BiVO4, BiVO4 Mn3O4	Degradation MB	[219]
BiOI–ZnTiO3	Type II	BiOI	ZnTiO3	BiOI ZnTiO3, ZnTiO3 BiOI	Degradation of Rhodamine 6 G	[220]
LaCrO3–PbTiO3	Type II	LaCrO3	PbTiO3	LaCrO3 PbTiO3, PbTiO3 LaCrO3	Degradation of phenol	[221]
CdS–Ni2P	Type II	CdS	Ni2P	CdSNi2P, Ni2P CdS	Hydrogen Evolution	[222]
BiVO4@Cu3SnS4	Type II	BiVO4	Cu3SnS4	BiVO4 Cu3SnS4, Cu3SnS4 BiVO4	Degradation MB	[223]

Even though better charge separation and improved photocatalysis are attained, some disadvantages do limit the photocatalytic performance of p–n heterojunctions. The joining of p‐type and n‐type semiconductors to build p–n junctions is negatively impacted by crystal lattice mismatching, which results in poor coupling at the interface and quenching of charge carrier migration. At the same time, migration of the energetic carriers that guarantee effective charge separation is achieved at the cost of partial loss of the energy obtained from the photons absorbed. This energy consumption loss reduces the reduction and oxidation potential of charge carriers, thus lowering the photocatalytic reaction rate.

4. Semiconductor–Metal (s–m) Heterojunctions

Interfacing a semiconductor with metal is a well‐known design strategy aimed at improving charge separation. The integration of metals with semiconductors results in Schottky/ohmic or plasmon junctions depending on the relative work functions. The minimum energy required to remove an electron from the Fermi (E _f) level to the vacuum level is defined as the work function. When n‐type/p‐type semiconductors with higher/lower work functions contact a metal directly, free electrons flow from the higher Fermi level to the lower Fermi level until the two Fermi levels equilibrate. The two main mechanisms involved at the interface of the s–m junction photocatalytic system can produce either a Schottky/ohmic junction or one characterized by plasmonic effects. The former can be formed whenever the work function difference between semiconductor and metal at the interface is appropriate. When an n‐type semiconductor has a lower work function than the metal contacting it (ϕ _m > ϕ _sn), free electrons migrate from the lower work function (higher Fermi level) to the higher work function (lower Fermi level) until the Fermi levels equilibrate, resulting in an accumulation of negative charge—with positive charge remaining in the semiconductor layer due to electrostatic induction, as shown in Figure  3a.

Figure 3.

a) Schematic of n‐type‐metal Schottky junction formation, followed by light absorption and charge separation. b) Schematic of p‐type‐metal Schottky junction formation, followed by light absorption and charge separation.

The electric field formed at the metal–semiconductor interface cannot be screened effectively in the semiconductor due to the low concentration of free charge carriers there. This causes the free charge carrier concentration near the semiconductor surface to be depleted compared with the bulk; a space charge region (also called the depletion layer) is formed on one side of the semiconductor, with the electric field direction pointing from semiconductor to metal. This makes the energy band bend upward, going from semiconductor to metal, thus forming the Schottky barrier. It should be noted that electron flow hardly affects E _fm due to the high density pool of free electrons in the metal. Instead, only the band level of the semiconductor shifts up or down. In the case of integrating, a p‐type semiconductor having a higher work function than the metal (ϕ _m < ϕ _sp), as shown in Figure 3b, free electrons migrate from metal‐to‐semiconductor; positive charge accumulates in the metal and negative charge in the semiconductor, forming a space charge region that results in bending the band downward, and in an electric field that points from metal to semiconductor. Generally, a Schottky barrier may promote charge separation that results in the enhancement of photocatalytic performance by preventing the recombination of electron–hole pairs. Normally, most noble metals possess higher work functions than n‐type semiconductors and lower than p‐type semiconductors, thus favoring the formation of Schottky barriers.

In contrast, when metal is in contact with an n‐type semiconductor of higher work function or a p‐type semiconductor of lower work function (ϕ _m < ϕ _sn or ϕ _m > ϕ _sp) free electrons from the n‐type (holes from the p‐type) semiconductor accumulate in the space charge region. The bands bend opposite the band in a Schottky junction, where no barrier is created between metal and semiconductor. The metal–semiconductor interface in the absence of a barrier results in an ohmic junction, as shown in Figure  4 . Generally, band bending and the creation of an inner electric field at the interface results in the migration and separation of low energy electrons (holes) to the metal and confines high energy charges to the semiconductors.^[ 12, 34, 35 ^]

Figure 4.

a) Schematic of p‐type‐metal ohmic junction formation, light absorption, and charge separation. b) Schematic of n‐type‐metal ohmic junction formation, light absorption, and charge separation.

Upon light absorption, photoexcited electrons from the CB of the n‐type semiconductor are transferred to the metal; this process induces oxidation at the semiconductor and reduction at the metal, as illustrated in Figure 4b. Conversely, in the case of a p‐type semiconductor, the photoexcited holes from the VB of the p‐type semiconductor migrate and collect on the metal, thus favoring oxidation on the metal and reduction on the semiconductor, as shown in Figure 4a.

Previously, our group combined experiments with theoretical simulations to demonstrate a set of design rules and working principles in p‐type semiconductor–metal hybrid structures incorporating Cu₂O(100)–Pd and Cu₂O(111)–Pd heterojunctions; these devices form Schottky barriers and ohmic interfaces, respectively, that result in improved photocatalysis. The Cu₂O(100) work function is 7.2 eV and that of Cu₂O(111) is 4.8 eV. At the Cu₂O(100)–Pd junction, the Pd metal work function is 1.714 eV lower than that of Cu₂O(100), inducing migration of free electrons from Pd to Cu₂O(100) that results in a Schottky barrier. The formation of such a barrier at the Cu₂O(100)–Pd interface facilitates the migration of photoexcited holes to Pd metal and retains photoexcited electrons on Cu₂O(100), which results in efficient charge separation and improved optoelectronic conversion. At the Cu₂O(111)–Pd interface, the Pd metal work function is 2.523 eV higher than that of Cu₂O(111), which disfavors formation of a Schottky barrier and allows ohmic contact formation. Our investigation showed that the design of the semiconductor surface facet matters both for the establishment of a Schottky barrier/ohmic contact and for spatial charge separation. By taking the advantage of Schottky barrier formation, Pd‐decorated Cu₂O cubes at moderate Pd density produced hydrogen at a rate of 2.20 mmol g⁻¹.^{[ 36 ]}

Since the work function plays a critical role in determining the best type of heterojunction, tuning the work function using different approaches, such as facet selection, should provide a method of tuning charge separation. For example, Li et al.^{[ 37 ]} designed a facet‐dependent n‐type‐metal heterojunction for spatial separation of photogenerated electrons and holes using Pt, Au, and Ag metals with (010) and (110) BiVO₄ crystal facets.^{[ 38 ]} Their results showed that metals or oxides can be deposited selectively on specific facets of BiVO₄; different facets result in selective accumulation of electrons (holes) and adsorption of metal ions, leading to selective photodeposition and efficient charge separation.

It is universally concluded that Schottky barriers created at the interface of a semiconductor and metal allow a one‐way flow of charges that ensure efficient charge separation of photogenerated charge carriers, thus enhancing photocatalytic performance. Although this principle is well accepted for bulk heterojunctions, it may not work for nanosized ones. For bulk heterojunction materials, the Fermi level—responsible for charge migration—greatly depends on carrier concentration. However, Fermi levels of nanosized heterojunction materials are greatly affected by quantum‐size effects, surface terminations/states, lattice distortions, and impurity doping. Mechanisms behind the improved photocatalytic activity of nanosized heterojunction systems have been investigated.^[ 39, 40 ^]

In parallel, Yan et al. have studied the formation of nanosized Schottky or ohmic junctions that greatly improve photocatalytic activity. Their work on the nanosized ohmic junction Ag/ZnO and Schottky junction Pt/ZnO shows that the Ag/ZnO ohmic junction exhibits higher photocatalytic efficiency than the Schottky junction Pt/ZnO model system. The separation of the photogenerated charge carriers that results in improved efficiency is greatly influenced by quantum size effects and the direction of the electric fields within the semiconductor–metal interface, which together strongly influence photocatalytic efficiency.^{[ 41 ]}

In contrast, the plasmonic effect occurs only under certain controlled conditions, when plasmonic metal bands are located in the visible or the near‐infrared (NIR) regions. Materials incorporating Au, Ag, or Cu—termed “plasmonic metals”—have strong plasmonic properties and have bands that are indeed located in the visible or NIR region. Importantly, the plasmonic properties of a metal are highly dependent on the size and shape of its layer, which can range from tens to hundreds of nanometers; thus, by increasing metal particle size and the dielectric properties of the surrounding medium, absorption can be shifted into the visible light region. Some other metals, such as Pt and Pd, possess very small excitation cross‐sections for surface plasmons and in small particle sizes their plasmonic bands are mainly located in the UV region.^{[ 12 ]}

Plasmonic Au/TiO₂ heterojunction photocatalysts that exhibit efficient charge separation have been synthesized by Bian et al. Due to surface plasmon resonance (SPR), Au NPs layered on the basal and lateral surfaces of TiO₂ impart a strong photoelectrochemical response in the visible light region (400–800 nm). Electrons injected from excited Au NPs (nanoparticles) on the basal surfaces of meso‐TiO₂ are efficiently delivered to the lateral surfaces of the crystal through the TiO₂ nanocrystal network. This feature allows for reduced loading of the metal on the semiconductor, which is especially advantageous for NIR‐active metallic nanostructures requiring larger sizes (e.g., Au nanorods). This anisotropic electron flow appreciably minimizes the recombination of electrons and holes in the Au NPs and enhances visible light photocatalytic activity by more than an order of magnitude, as compared to that of conventional Au/TiO₂ NP systems.^{[ 42 ]}

Jiang et al. studied the key role of metals in the photocatalytic activity of Au–CeO₂ junctions. They explored the effects on the equilibrium between plasmon resonance and surface catalysis by loading various amounts and particle sizes of Au NPs on CeO₂. Photoexcitation and surface catalysis vary inversely with Au NP size, but both NP loading and size determine the final photocatalytic performance of propylene oxidation under visible (>420 nm) light illumination. Increasing Au loading seems to promote photoabsorption, charge separation, and resonant energy transfer due to enhanced Au SPR. However, increased Au particle size leads to saturation and also decreases the number of exposed active sites that can adsorb reactant species. In addition, large Au NPs (>10 nm) demonstrate distinct passivity toward O₂ dissociation and activation. In general, this investigation shows that the design of efficient metal−semiconductor systems for ideal solar energy conversion requires medium‐sized particles (6–12 nm) for optimizing the plasmonic and catalytic properties of metallic nanostructures.^{[ 43 ]}

Despite that so many different junction designs have been investigated, the overall principles are more or less similar to those discussed earlier. Representative works on Schottky/ohmic junctions and plasmonic effects are summarized in Table  2 .

Table 2.

Summary of representative semiconductor–metal junctions

Model/sample	Junction type	Oxidation site	Reduction site	Charge migration direction across the junction	Absorbance	Application	References
Pt–PbS	Schottky	PbS	Pt	PbSPt	532 nm	Water splitting	[224]
Co–TiO2	Schottky	TiO2	Co	TiO2 Co	–	Hydrogen evolution	[225]
Au–Bi2S3	Schottky/Plasmon	Bi2S3	Au	Bi2S3 Au	560 nm	Degradation of MB	[226]
Pd–SiC	Schottky	SiC	Pd	SiCPd	530 nm	Hydrogenation of furan derivatives	[227]
Ag–ZnO	Plasmon/Schottky	ZnO	Ag	AgZnO	400 nm	Degradation of RhB	[228]
Ag–TiO2	Schottky	TiO2	Ag	TiO2 Ag	380–500 nm	Degradation of RhB	[229]
Au–TiO2	Plasmon/Schottky	TiO2	Au	Au TiO2	540–550 nm	Ethanol–water	[230]
Au–ZnO	Schottky	ZnO	Au	ZnOAu	350 nm	Degradation of RhB	[231]
Au–TiO2	Schottky/Plasmon	Au	TiO2	AuTiO2	λ > 410 nm	Degradation MO	[232]
W–WO3	Ohmic	WO3	W	WO3 W	350–400 nm	Degradation of gaseous acetaldehyde	[233]
Au–BiVO4	Plasmon/Schottky	Au	BiVO4	AuBiVO4	λ > 420 nm	Dye degradation and water oxidation	[234]
Ni–TiO2	Plasmon	Ni	TiO2	NiTiO2	495 nm < λ < 800 nm	Degradation of MB	[235]
Au–TiO2	Plasmon	Au	TiO2	AuTiO2	λ > 515 nm	Water splitting	[236]
Ag–ZnO	Plasmon	Ag	ZnO	AgZnO	438 nm	Degradation of MO	[237]
Ag–TiO2	Plasmon	Ag	TiO2	AgTiO2	553 nm	Degradation of RhB	[238]
Au–SnO2	Plasmon	SnO2	Au	AuSnO2	550 nm	Degradation of RhB	[239]
Au–TiO2	Plasmon	Au	TiO2	AuTiO2	550 nm	Decomposition of MB	[240]
M(Au, Ag, Cu) –TiO2	Plasmon	M	TiO2	MTiO2	Au (521 nm), Ag (540 nm) and Cu (800 nm)	Photo‐oxidation of benzaldehyde and nitrobenzaldehyde	[241]
Bi–TiO2	Plasmon	Bi	TiO2	BiTiO2	λ > 420 nm	Removal of ppb‐level NO in air	[242]
Au–WO3	Plasmon	Au	WO3	AuWO3	450–700 nm	RhB photodegradation	[243]
Au–TiO2	Plasmon	Au	TiO2	AuTiO2	λ > 450 nm	Oxidation of 1‐phenylethanol	[244]
Ag–TiO2	Plasmon	Ag	TiO2	AgTiO2	450–700 nm	Water splitting and methanol oxidation	[245]
Ag–TiO2	Plasmon	Ag	TiO2	AgTiO2	λ = 400–750 nm	MB photodegradation	[246]
Au–CeO2	Plasmon	Au	CeO2	AuCeO2	λ > 520 nm	Aerobic oxidations of propylene	[43]
Au–TiO2	Plasmon	Au	TiO2	AuTiO2	548 nm	MB, RhB degradation	[42]
Au–TiO2	Plasmon	Au	TiO2	AuTiO2	Visible light	Degradation MB	[247]
Au–CdSe	Plasmon	Au	CdSe	AuCdSe	λ > 700 nm	Hydrogen generation	[248]

Even though a great deal of effort has been exerted to understand the selective steering of charges at s–m junctions, it is poorly known how this weakens the redox capability of high‐energy electrons and holes at the reaction site. Apparently, charge separation and transfer works at the expense of the redox ability of charge carriers, and energy is lost during their transfer from semiconductor to metal due to the difference between the E _fm energy levels and the CB of an n‐type semiconductor (or the VB of a p‐type semiconductor). Still, a significant amount of leading‐edge research is being undertaken to overcome the challenges of attaining the efficient charge separation required for improved optoelectronic conversion. Semiconductor–semiconductor–metal and semiconductor–metal–semiconductor heterojunctions are other well‐studied heterojunctions for minimizing the drawbacks of p–n and semiconductor–metal heterojunctions.

5. All‐Solid‐State Ternary Heterojunctions (Z‐Schemes)

It has been a long‐standing challenge to realize the high degree of charge separation in semiconductor‐based heterojunction photocatalysts necessary for efficient optoelectronic conversion. In contrast to Type‐II and Type‐III semiconductor–semiconductor (s1–s2) junctions that steer charge flow favorably and ensure charge separation onto separate semiconductors, most s1–s2 heterojunctions fail due to their straddling bandgaps (Type I) (Figure 2a) in which both photogenerated electrons and holes are deposited in the same semiconductor with a small bandgap; this circumstance results in high recombination.

Inspired by the natural process of photosynthesis in green plants, the Z‐scheme has become prominent recently as an additional model system for water splitting. This scheme includes two different semiconductors (s1 and s2) and an appropriately reversible acceptor/donor pair (AD‐species) and carries out two half‐reactions on the corresponding surfaces of the semiconductors, as shown in Figure  5a.

Figure 5.

a) Schematic of band alignments and charge flows in the Z‐scheme. δE ₁ (δE ₂) denotes the energy barrier for diffusion of high energy holes (electrons) across the interface or is the relative potential between VB of S1 (CB of S2) and the oxidation (reduction) potential of acceptor (A) and donor (D)‐species. b) Schematic showing the band bending and charge separation mechanism in an all‐solid‐state S1–m–S2 Z‐scheme heterojunction.

In other words, the Z‐scheme guarantees that each of two half‐reactions can be realized on one of two spatially separated semiconductors possessing moderate E _g . Essentially, photons with moderate energy can drive this system, which means that the visible light portion of the solar spectrum can be utilized. Importantly, AD‐species (such as Fe³⁺/Fe²⁺ and IO₃ ⁻/I⁻) can be interposed to effect the separation of charge carriers formed by the consumption of energetic holes from s1 and electrons from s2. There are also some Z‐scheme systems without AD‐species in which two different semiconductors are in direct contact with each other, and in which energetic holes from the CB of s1 recombine with high‐energy electrons from the VB of s2 at the interface junction. Unfortunately, there remain drawbacks restricting further photocatalytic enhancement of Z‐scheme systems. For instance, there is no driving force to accelerate the process of reversible transition of redox mediators for realizing effective spatial separation of charges or for promoting the recombination of e–h, especially for high energy holes with low migration rates in s1. Hence, this architecture relies strongly on the band positions between the VB of s1 and the CB of s2 relative to the redox potential of the AD species (E ₁ and E ₂, respectively); this places restrictions on which combinations of semiconductors will work synergistically to create efficient photocatalytic heterostructures. A detailed tutorial and reviews of such conventional Z‐scheme systems exist.^[ 2, 44, 45, 46 ^] Another well‐known Z‐scheme heterojunction is an all‐solid‐state semiconductor–metal–semiconductor/semiconductor–semiconductor–metal ternary Z‐scheme that pays considerable attention to overcoming the aforementioned drawbacks and puts forth a solution to the two central issues of charge transport—the steering of charge flows across the interface and the assurance of charge separation—so as to improve photocatalytic efficiency (shown schematically in Figure 5b). This architecture alleviates the problem of band edge position mismatch between two semiconductors by aligning their Fermi levels and shifting the bands up or down in a Z‐scheme while also minimizing lattice mismatch and transforming the Type‐I and Type‐III p–n junctions into Type II—thus making efficient charge separation more favorable and hence improving photocatalysis. All‐solid‐state Z‐scheme heterojunction designs have been reviewed specifically.^[ 28, 46, 47, 48, 49, 50 ^] Here, we focus mainly on the recent advances and progress in these all‐solid‐state Z‐scheme designs.

In 2015, Li et al. from our group used a combination of theory and experiment to achieve a perfect Z‐scheme by properly aligning the bandgap of Ag₂S in a TiO₂–Ag–Ag₂S heterojunction. The work function difference between Ag and Ag₂S caused free electrons to flow from the low work function (high Fermi level) of Ag to the high work function (low Fermi level) of Ag₂S until their Fermi levels equilibrated, resulting in an upshift and downward band bending. As a consequence, the TiO₂–Ag–Ag₂S heterostructure aligned to form a perfect Z‐scheme. Upon exposure of the heterojunction to light, high energy photoexcited holes and electrons residing on TiO₂ and Ag₂S favor oxidation and reduction, respectively, on the separate semiconductors, whereas low‐energy electrons and holes are confined to the intermediate Ag metal. The rate of hydrogen generation by this system under full solar insolation is 6.3 μmol h⁻¹—higher than the sum of λ < 400 nm (1.3 μ molh⁻¹) and λ > 400 nm (0.19 μmol h⁻¹), as shown in Figure  6 .^{[ 51 ]}

Figure 6.

a) Average rates of photocatalytic hydrogen production and b) photocurrents versus time (I–t) curves of Ag₂S–(Ag)–TiO₂ hybrid structures under various light illumination conditions (full spectrum, λ < 400 nm and λ > 400 nm). a,b) Reproduced with permission.^{[ 51 ]} Copyright 2015, Tsinghua University Press and Springer‐Verlag Berlin Heidelberg.

The performance enhancement suggests that the TiO₂–Ag–Ag₂S Z‐scheme substantially improved the charge generation and charge separation processes under full spectrum illumination. Average lifetime analysis of the photogenerated carriers from open‐circuit voltage decay measurements under full light spectrum illumination confirms the improvement in charge separation. The carrier lifetime of TiO₂–Ag–Ag₂S is prolonged, demonstrating the key role trace Ag plays in the Z‐scheme architecture.

Zhuang et al. from our group, successfully synthesized ZnS–CdS binary and ZnS–(CdS–Au, Pt, and Pd) ternary heterojunctions; these nanosystems justify and confirm that a transition from Type I to Type II is possible and established a promising approach for improving the efficiency of such systems. The work function difference (W _CdS > W _m) of the ZnS–(CdS–Au, Pt, and Pd) heterojunction allows free electrons to flow from metal to CdS until the Fermi levels align and results in the upshifting and downward band bending of CdS. When the heterostructure is exposed to light, high energy photogenerated electrons migrate to ZnS while photogenerated holes migrate to the CdS/metal interface, ensuring efficient charge separation. Activities of these nanosystems, measured as the rate of hydrogen evolution, are 13.5, 36.5, 83.5, and 101 μmol h⁻¹ for ZnS–CdS, ZnS–(CdS–Au), ZnS–(CdS–Pd), and ZnS–(CdS–Pt), respectively.^{[ 52 ]}

Li et al. compared work functions of the CdS–Au–WO₃ and CdS–Pt–WO₃ heterostructures, determining that the sequential work functions of these two heterostructures are 4.9 < 5.0 < 5.05 and 4.9 < 5.2 > 5.05 eV, respectively. These results imply that the Fermi level of CdS is higher than that of either Au or Pt and that the Fermi level of WO₃ is between the levels of these two. In this system, free electrons transfer from CdS to Au (or Pt), from Au to WO₃, and also from WO₃ to Pt; this results in the formation of depletion and accumulation layers at the CdS/Au and Au/WO₃ interfaces, respectively, and depletion layers on both sides of the CdS/Pt and Pt/WO₃ interfaces.

When the CdS–Au–WO₃ heterojunction is exposed to light, photoinduced electrons from WO₃ transfer to the intermediate Au layer, and holes are created at the WO₃ semiconductor layer under the influence of an interfacial electric field, favoring oxidation. At the CdS–Au interface, photoinduced holes migrate to the intermediate Au layer and electrons collect at CdS, where reduction can then take place. The CdS–Pt–WO₃ heterojunction functions similarly.

The finding shows explicitly that work function differences result in a lower Schottky barrier height (W _CdS−W _Au < W _CdS–W _Pt) at the Au/CdS interface than at Pt/CdS; thus charge carrier transfer in the WO₃/Au/CdS heterostructure is smoother, and H₂ evolution efficiency better than in WO₃/Pt/CdS. This work demonstrates that modulation of the intermediate metal work function plays a crucial role in Z‐scheme photocatalysis.^{[ 53 ]}

As promising types of photocatalysts, generally, all Z‐scheme (i.e., including all‐solid‐state Z‐scheme and p–n junctions) heterojunction designs have been proposed to harvest energy over a broad solar spectral range by interfacing two semiconductors with a staggered bandgap. Their utilization is, however, often limited by the difficulty of achieving selective collection of low energy photogenerated charges at the interface while excluding high energy ones. Building on semiconductor–metal, p–n, and Z‐scheme designs, we and other groups have performed experimental and theoretical first‐principles investigations on energy‐dependent Z‐scheme designs of p‐type–metal–n‐type and n‐type–metal–p‐type heterostructures, which hold the promise of being able to steer charges selectively to the intermediate metal for attaining more effective charge separation. Representative summary of semiconductor–semiconductor–metal and semiconductor–metal–semiconductor photocatalyst junction studies are presented in Table  3 .

Table 3.

Summary of representative semiconductor–semiconductor–metal and semiconductor–metal–semiconductor photocatalyst junction studies

Sample/model	Heterojunction type	Oxidation site	Reduction site	Charge‐transfer direction across the junction	Activity test/application	References
Ag2O‐Fe‐TiO2	p‐S1‐m‐n‐S2	TiO2	Ag2O	TiO2 Fe, Ag2OFe	CO2 Conversion	[249]
BiOBr–Ag–AgBr	p‐S1‐m‐n‐S2	BiOBr	AgBr	BiOBrAg, AgBrAg	MO degradation	[250]
Ag3PO4–Ag–Bi2MoO6	p‐S1‐m‐n‐S2	Ag3PO4	Bi2MoO6	Ag3PO4 Ag, Bi2MoO6 Ag	RhB degradation	[251]
TiO2–Pt–CdS	n‐S1‐m‐n‐S2	TiO2	CdS	TiO2 Pt, CdSPt	CO2 Photoreduction	[252]
AgCl–Ag–BiOCl	n‐S1‐m‐p‐S2	AgCl	BiOCl	AgBiOCl, AgAgCl	MO Degradation	[253]
CdS–Au–ZnO	n‐S1‐m‐n‐S2	ZnO	CdS	ZnOAu, CdSAu	H2 generation	[254]
CdS–Au–TiO2	n‐S1‐m‐n‐S2	TiO2	CdS	TiO2 Au, AuTiO2	MV2+reduction	[255]
CdS–Au–BiOCl	n‐S1‐m‐p‐S2	BiOCl	CdS	BiOClAu, CdSAu	MO, RhB, and Phenol Degradation	[256]
BiVO4–Au–Cu2O	n‐S1‐m‐p‐S2	BiVO4	Cu2O	BiVO4 Au, Cu2OAu	CO2 Photoreduction	[257]
ZnO–Pt–CdS	n‐S1‐m‐n‐S2	CdS	Pt	CdSZnO, ZnOPt	H2 generation	[258]
CdS–Ag–Bi2MoO6	n‐S1‐m‐n‐S2	Bi2MoO6	CdS	Bi2MoO6 Ag, CdSAg	RhB degradation	[259]
BiVO4–Ag–Cu2O	n‐S1‐m‐p‐S2	BiVO4	Cu2O	BiVO4 Ag, AgCu2O	TC degradation	[260]
WO3–Ag–AgCl	n‐S1‐m‐n‐S2	WO3	AgCl	AgAgCl, AgWO3	RhB degradation	[261]
In2S3–Au–BiVO4	n‐S1‐m‐n‐S2	BiVO4	In2S3	BiVO4 Au, In2S3 Au	RhB and phenol Degradation	[262]
TiO2–Pt–SnO2	n‐S1‐m‐n‐S2	SnO2	TiO2	SnO2 Pt, TiO2 Pt	MB degradation	[263]
CdS–Pt–Mo2C	n‐S1‐m‐p‐S2	CdS	Pt	CdSMo2C, Mo2CPt	H2 generation	[264]
α‐Fe2O3–Ag–AgCl	n‐S1‐m‐p‐S2	a‐Fe2O3	AgCl	AgClAg, AgClα‐Fe2O3	RhB degradation	[265]
Ag3PO4–Ag–SiC	n‐S1‐m‐n‐S2	Ag3PO4	SiC	Ag3PO4 Ag, SiCAg	MO and Phenol degradation	[266]
α/β‐Bi2O3–Ag–AgCl	n‐S1‐m‐n‐S2	Bi2O3	AgCl	Bi2O3 Ag, AgAgCl	RhB and acid orange dyes degradation	[267]
CdS–Au–WO3	n‐S1‐m‐n‐S2	WO3	CdS	WO3 Au, AuCdS	Hydrogen and oxygen evolution	[268]
Pt–CeO2–ZnO	m‐n‐S1‐n‐S2	Pt	CeO2	CeO2 ZnO, ZnOPt, ZnOCeO2	Phenol degradation	[269]
Au–CuO–Co3O4	m‐p‐S1‐p‐S2	CuO,Co3O4	–	CuOAu, Co3O4 Au	O2 evolution	[270]
Ag–AgCl–BiPO4	m‐n‐S1‐n‐S2	Ag	BiPO4	AgAgCl, AgClBiPO4, BiPO4 AgCl	RhB degradation	[271]
Ag–AgVO3–BiOCl	m‐n‐S1‐n‐S2	AgVO3	BiOCl	AgVO3 Ag, AgVO3 BiOCl	MB degradation	[272]
Au–ZnO–TiO2	m‐n‐S1‐n‐S2	ZnO	Au	AuTiO2, ZnOTiO2, TiO2 ZnO	Photo‐oxidation of phenol and 4‐chlorophenol	[273]
Ag–AgBr–InVO4	m‐n‐S1‐n‐S2	InVO4	AgBr	InVO4 AgBr, AgAgBr, AgBrInVO4	RhB degradation	[274]
Ag–TiO2–ZnO	m‐n‐S1‐n‐S2	ZnO	Ag	ZnOTiO2, TiO2 Ag, TiO2 ZnO	Phenol degradation	[275]
Ag–SnS–TiO2	m‐n‐S1‐n‐S2	SnS	Ag	SnSAg, TiO2 Ag	RhB and MB degradation	[276]
Ag–AgBr–Ag3VO4	m‐n‐S1‐n‐S2	AgBr‐	Ag3VO4	AgAgBr, AgBrAg3VO4, Ag3VO4 AgBr	MO degradation	[277]
Ag–Ag2O–BiOCl	m‐p‐S1‐p‐S2	Ag2O	BiOCl	Ag2OAg, BiOClAg	RhB degradation	[278]
Ag–Ag2S–CuS	m‐n‐S1‐p‐S2	Ag2S	Ag	CuSAg, Ag2SAg, CuSAg2S	2,4‐Dichlorophenol degradation	[279]
Ag–AgCl–ZnO	m‐n‐S1‐n‐S2	AgCl	ZnO	AgClZnO, AgZnO	RhB and MO degradation	[280]
Cu–Cu2O–ZnO	m‐p‐S1‐n‐S2	Cu2O	ZnO	CuCu2O, Cu2OZnO, ZnOCu2O	H2 generation	[281]
Pt–In2O3–TiO2	m‐n‐S1‐n‐S2	In2O3	TiO2	PtIn2O3, In2O3 TiO2	CR degradation and H2 generation	[282]
Pt–In2O3–TiO2	m‐n‐S1‐n‐S2	In2O3	Pt	In2O3 TiO2, TiO2 Pt, TiO2 In2O3	RhB degradation	[283]
Ag‐AgBr‐Bi2MoO6	m‐n‐S1‐n‐S2	AgBr	Bi2MoO6	AgAgBr, AgBi2MoO6	MB degradation	[284]
Au–TiO2–SnO2	m‐n‐S1‐n‐S2	TiO2	SnO2	AuTiO2, AuSnO2, TiO2 SnO2	RhB degradation	[285]
TiO2–Fe2O3–Cu	n‐S1‐n‐S2‐m	TiO2	Fe2O3	TiO2 Fe2O3, Fe2O3 Cu, Fe2O3 TiO2	MO degradation	[286]
Cu–Cu2O–ZnO	m‐p‐S1‐n‐S2	Cu2O	ZnO	CuCu2O, CuZnO, Cu2OZnO	MB degradation	[287]
Pt–CdS–h–BiOBr	m‐n‐S1‐p‐S2	BiOBr	CdS	Pth‐BiOBr, h‐BiOBrCdS, CdSh‐BiOBr	MB degradation	[288]
Ag–AgBr–TiO2	m‐n‐S1‐n‐S2	AgBr	TiO2	AgAgBr, AgBrTiO2, TiO2 AgBr	MB degradation	[289]
Ag–SrTiO3–TiO2	m‐n‐S1‐n‐S2	SrTiO3	TiO2,Ag	SrTiO3 Ag, SrTiO3 TiO2, TiO2 SrTiO3	RhB degradation	[290]
Cu–SrTiO3–TiO2–Cu	m‐n‐S1‐n‐S2‐m	Cu	Cu	SrTiO3 TiO2, TiO2 Cu, TiO2 SrTiO3, SrTiO3 Cu	H2 generation	[291]
Fe3O4–TiO2–Ag	n‐S1‐n‐S2‐m	Ag	Fe3O4	AgTiO2, TiO2 Fe3O4	AMP degradation	[292]
CuS–TiO2–Pt	p‐S1‐n‐S2‐m	CuS	Pt	TiO2 Pt, TiO2 CuS	H2 production	[293]
In2O3–ZnO–Ag	n‐S1‐n‐S2‐m	In2O3	ZnO	In2O3 ZnO, AgZnO, ZnOAg nanowire	MO and 4‐nitrophenol degradation	[294]
Ag3VO4–Ag3PO4–Ag	p‐S1‐p‐S2‐m	Ag3VO4	Ag	Ag3VO4 Ag3PO4, Ag3PO4 Ag, Ag3PO4 Ag3VO4	Acid blue 92 (AB92) degradation	[295]
NiO–ZnO–Au	p‐S1‐n‐S2‐m	ZnO	NiO, Au	ZnOAu, ZnONiO, NiOZnO	RhB degradation	[296]
NiO–ZnO–Pt	p‐S1‐n‐S2‐m	ZnO	NiO, Pt	ZnOPt, ZnONiO, NiOZnO	MO degradation	[297]
MoS2–TiO2–Au	p‐S1‐n‐S2‐m	MoS2	TiO2	AuTiO2, AuMoS2, MoS2 TiO2, TiO2, MoS2	Current density and H2 generation	[298]

Recently, our research group has proposed a promising ternary energy‐dependent Z‐scheme incorporating both an n‐type semiconductor–metal–p‐type–semiconductor and a p‐type semiconductor–metal–n‐type semiconductor heterojunction photocatalyst using first‐principles calculations. This design features a work function cascade with W _n < W _m < W _p and W _p < W _m < W _n for the two heterojunctions, respectively (Figure  7 ). These systems are designed to steer charge flow and enhance effective charge separation by combining the merits of traditional Z‐scheme, p–n, and s–m systems so that they work synergistically to realize high photocatalytic performance. The intermediate metal not only lowers the lattice matching and band alignment requirements between the two semiconductors but also induces band bending at the p–m and m–n interfaces via charge migration that lines up the Fermi levels. Such band bending enables selective steering of low energy charges from the two semiconductors to the intermediate metal while confining high‐energy electrons and holes to the individual p‐type and n‐type semiconductors, respectively, so as to enable oxidation and reduction reactions to take place.^[ 54, 55 ^]

Figure 7.

Energy‐dependent Z‐scheme designs of n‐type–metal–p‐type Schottky junction and p‐type–metal–n‐type ohmic junction models in which: a) An energy‐dependent Z‐scheme n–m–p design allows free electron migration from n‐type to metal and then to p‐type until Fermi levels are equilibrated. b) Upon illumination low‐energy photoexcited electrons from the n‐type layer and holes from p‐type are transferred to the intermediate metal layer while high‐energy electrons are confined to their semiconductors. a,b) Reproduced with permission.^{[ 54 ]} Copyright 2020, American Chemical Society. c) An energy‐dependent Z‐scheme p–m–n design allows free electrons to migrate from p‐type to metal to n‐type until the Fermi levels are equilibrated. d) Upon illumination, low‐energy photoexcited holes from the p‐type layer and electrons from n‐type are transferred to the intermediate metal and high‐energy electrons are confined to the individual semiconductors. c,d) Reproduced with permission.^{[ 55 ]} Copyright 2019, Royal Society of Chemistry.

A combined experimental and theoretical study of an n‐type–metal–p‐type TiO₂–Pd–Cu₂O Z‐scheme has been performed recently by Ye et al. Their design achieved improved photocatalysis by interfacing n‐type TiO₂(001) and p‐type Cu₂O(100) facets with a layer of Pd metal placed between them; this design produced the desired bandgap alignment for inducing migration of photoexcited electrons from TiO₂(001) to Pd and holes from Cu₂O(100) to Pd. The TiO₂(001) < Pd < Cu₂O(100) work functions ordering allows free electrons to flow from TiO₂(001) to Pd and from Pd to Cu₂O(100) and equilibrates the Fermi levels. Upon light illumination, this facet hybrid junction facilitates low energy electron transfer from the TiO₂(001) facet to the intermediate Pd layer and hole migration from Cu₂O(100) to Pd; good charge separation is produced by confining high energy holes at TiO₂(001)—enabling oxidation—and high‐energy electrons at Cu₂O(100)—enabling reduction. This hybrid TiO₂(001) < Pd < Cu₂O(100) design has 1.37–3.12 times higher photocurrent density and 1.22–2.06‐fold higher phenol degradation efficiency than another three‐hybrid design that also incorporates a TiO₂(101) facet or Cu₂O(111) facet in contact with Pd at an interface.^{[ 56 ]}

Recently, an interesting n–metal–p Janus plasmonic heteronanocrystals of Au/(PbS‐CdS) have been synthesized by Wan et al. The construction of these heterojunctions investigates that hot plasmonic electrons and holes collected simultaneously on individual semiconductors. The minimal Schottky barrier and ohmic contact created at the Au–CdS and Au–PbS interfaces enhance efficient separation of plasmonic electrons and smooth transfer of hot plasmonic holes, respectively. The result shows an extended lifetime of the charge separated state, and superior in photocatalytic CO₂ performance reduction.^{[ 57 ]}

The extension of ternary heterojunction designs to include an energy‐dependent Z‐scheme p–m–n and n–m–p heterojunction that can steer charges selectively across the interface presents a better opportunity to achieve good charge separation and thereby enhance photocatalytic activity. The inner electric field established at the junction assisted by band bending drives low energy holes and electrons to the intermediary metal and thereby lowers the probability of undesirable electron–hole pair recombination and keeps high‐energy electrons and holes apart by confining them to the individual semiconductors, finally resulting in good charge separation. The intermediary metal not only serves as a center of recombination but also equilibrates the Fermi levels across the interface and minimizes lattice mismatch between the two semiconductors. In general, this p–m–n and n–m–p strategy may, thus, pave the way for making more efficient Z‐scheme photocatalysts for practical applications. Our extension of this design that utilizes thin layers in an energy‐dependent Z‐scheme, the Cu₂S–Pt–WO₃ (p‐type–metal–n‐type) heterojunction, permits the exploration of an alternative way of realizing efficient charge separation. This system demonstrates increased charge flow across the junction compared to its bulk counterpart and exhibits a higher electron density on each surface that should produce enhanced optoelectronic conversion.^{[ 58 ]}

6. Semiconductor–Graphene Heterojunctions

Until now, achieving good charge separation and absorption of broad‐spectrum visible light either in a single semiconductor or by the interfacing of two semiconductors either alone or with metal is a challenge that has prevented their practical application. Shifting optical absorption from UV to the visible light region to maximize the quantum efficiency of single semiconductors in combination with graphene has emerged as a new prospect. Graphene is a well‐known material that possesses high surface area, high conductivity, and good adsorptive properties; these lead to improved accumulation of charge and effective charge separation.^[ 20, 59, 60, 61, 62, 63, 64 ^]

Williams et al. synthesized a photoactive graphene–TiO₂ nanocomposite with graphene oxide (GO) suspended in ethanol. Upon UV light irradiation, photoexcited electrons transferred from TiO₂ to GO; this was confirmed by the reduction accompanying changes in the absorption of the GO, as evidenced by the shift in color of the suspension from brown to black. No significant change was observed upon UV light illumination when TiO₂ was excluded from the solution, confirming that surface electrons migrate from TiO₂ and reduce GO.^{[ 65 ]} Similar studies on graphene–TiO₂ photocatalysis using different experimental methods show improved pollutant degradation and hydrogen production.^[ 66, 67, 68, 69, 70, 71, 72 ^]

Yu et al. successfully synthesized RGO–CdS nanorod composite by a one‐step microwave‐assisted hydrothermal process for carrying out the photocatalytic reduction of CO₂ to CH₄. Their analysis revealed that the RGO–CdS nanorod junction at 0.5 wt% of RGO achieved a high production rate (2.51 μmol h⁻¹ g⁻¹), 10 times greater than that of pristine CdS nanorods (Figure  8 ).

Figure 8.

a) Comparison of photocatalytic CH₄ production rates of G0, G0.1, G0.25, G0.5, G1.0, G2.0, N0.5, P0.5, and RGO samples under visible light irradiation. b) Transient photocurrent responses of the G0 and G0.5 samples in 0.4 m Na₂SO₄ aqueous solution under visible light (λ = 420 nm) irradiation. a,b) Reproduced with permission.^{[ 73 ]} Copyright 2014, Royal Society of Chemistry.

The high production rate is due to the presence of RGO that acts as an acceptor of photogenerated electrons from CdS upon irradiation.^{[ 73 ]} Similar metal‐free RGO‐Ss, such as Ag₂CrO₄–GO,^{[ 74 ]} TiO₂–S/rGO,^{[ 75 ]} and GO/CuFe₂O₄,^{[ 76 ]} have been investigated as materials that can replace expensive noble metals with carbon, and ultimately enhance the photocatalytic activity of single semiconductors.

7. Semiconductor (S)–Graphitic Carbon Nitride (C₃N₄) Heterojunctions

Recently, other heterojunctions composed of S–C₃N₄ have been widely investigated by many researchers for optimizing single semiconductor photoelectronic conversion. C₃N₄ is inexpensive, harmless, and has a narrow bandgap of 2.7 eV, qualifying it for consideration as a photocatalytic material for H₂ production, O₂ evolution, and CO₂ reduction upon visible light irradiation.^[ 77, 78 ^]

In a combined experimental and theoretical study, Yu et al. have demonstrated enhanced photocatalytic activity for selective reduction of CO₂ to CH₃OH on g‐C₃N₄–ZnO. The lower potential photogenerated electrons from the CB of ZnO recombine at a minimal rate with the higher potential VB‐excited holes of g‐C₃N₄ resulting in enhanced photocurrent production as confirmed experimentally. Indeed, this g‐C₃N₄–ZnO photocatalyst attains 2.3 times higher photocatalytic activity than does pure g‐C₃N₄.^{[ 79 ]}

Li et al. synthesized a direct Z‐scheme g‐C₃N₄–TiO₂ heterojunction that shows improved photocatalytic activity for degradation of the organic pollutant propylene under visible light irradiation. Upon light irradiation, photoexcited electrons migrate from the TiO₂ layer and recombine with holes from g‐C₃N₄; this favors reduction at g‐C₃N₄ and oxidation at the TiO₂ surface. This research attempts to demonstrate that a larger specific surface area and stronger UV–vis light absorption can result in improved photocatalytic activity due to increased numbers of active sites and photogenerated carriers. By considering two samples—20%g‐C₃N₄–TiO₂–400 and 30%g‐C₃N₄–TiO₂–600—they showed that increased light absorption does not always lead to improved photocatalytic activity. Rather, 20%g‐C₃N₄–TiO₂–400, which possesses a larger specific surface area and better visible light absorption, results in lower photocatalytic activity than 30%g‐C₃N₄–TiO₂–600. Thus, in addition to large surface area and broad light spectrum absorption, better charge transfer and separation are also crucial factors for enhanced photocatalysis.^{[ 80 ]}

CQDs incorporating carbon nitride show interesting effects and attain improved quantum efficiency. For instance, Liu et al. synthesized a metal‐free carbon nanodot–carbon nitride (C₃N₄) nanocomposite for photocatalytic water splitting. Such a CQDs–C₃N₄ hybrid is made of low‐cost and environmentally friendly materials. This device attained quantum efficiencies of 16% for wavelength λ = 420 ± 20 nm, 6.29% for λ = 580 ± 15 nm, and 4.42% for λ = 600 ± 10 nm, and an overall solar energy conversion efficiency of 2.0%.^{[ 81 ]}

Another facile experimental method has been developed by Guo et al. to synthesize an infrared‐responsive photocatalyst—a CQD/carbon nitride nanocomposite—at 450 °C, designed to enhance photocatalytic activity. In this hybrid photocatalyst, the CQD converts infrared to visible light, whereas the carbon nitride utilizes the visible light emitted by the CQD to degrade pollutants. This composite photocatalyst degrades MO efficiently under infrared light irradiation (λ > 800 nm).^{[ 82 ]}

Recently, our group has performed a detailed theoretical study of metal‐free CQD/carbon‐nitride hybrid systems to investigate the mechanism across the interface for isolating hydrogen from oxygen in photocatalytic water splitting.

The work functions of C₃N and the CQDs were calculated to be 3.25 and 4.65 eV, respectively, so that electron flow is driven from C₃N to the CQDs. The computed CQD/C₃N differential charges indicate that holes collect at C₃N and electrons at the CQDs. The computed absorption coefficients, as shown in Figure  9b, reveal that while bare C₃N absorbs mainly short‐wavelength visible light (<600 nm) and CQD absorption extends to ≈700 nm, absorption of the CQDs/C₃N composite extends to ≈900 nm.

Figure 9.

a) Differential charge distribution of the carbon quantum dot (CQD)/C₃N structure. The yellow and blue regions represent electron and hole charge distributions, respectively, with an isosurface value of 0.0006 e Å⁻³. b) Photoabsorption spectrum of the bare C₃N (red), CQD (blue), and CQD/C₃N composite structures (black). c) PDOS of the CQDs/C₃N hybrid. a–c) Reproduced with permission.^{[ 83 ]} Copyright 2018, Royal Society of Chemistry.

Extended time‐dependent density functional theory (TDDFT) calculations reveal that photoexcited electrons transfer from C₃N to CQD, which is in good agreement with ultrafast charge evolution results, which favor oxidation at C₃N and reduction at CQD sites. This CQDs/C₃N hybrid photocatalyst design enables the full use of visible and IR light to generate and distribute high energy holes and electrons on the C₃N and CQDs layers. Representative works on S–g‐C₃N₄ photocatalyst junctions are summarized in Table  4 .^{[ 83 ]}

Table 4.

Summary of representative S‐g‐C₃N₄ photocatalyst junctions reported to date

Model/sample	Oxidation site	Reduction site	Charge‐transfer direction across the junction	Activity test/application	References
g‐C3N4/TiO2	TiO2	g‐C3N4	TiO2 Interface, g‐C3N4 Interface	Degradation of propylene and hydrogen generation, TC, degradation of diclofenac	[299, 300]
C3N4/BiOIO3	BiOIO3	C3N4	BiOIO3 Interface, C3N4 Interface	Photodegradation of MO, RhB	[301]
g‐C3N4/Bi4O7	Bi4O7	g‐C3N4	Bi4O7 Interface, g‐C3N4 Interface	Degradation of MB, phenol, RhB, and BPA	[302]
g‐C3N4/Bi2MoO6	Bi2MoO6	g‐C3N4	Bi2MoO6 Interface, g‐C3N4 Interface	Photodegradation of MB	[303]
g‐C3N4/AgFeO2	AgFeO2	g‐C3N4	AgFeO2 Interface, g‐C3N4 Interface	Acid red G (ARG)	[304]
g‐C3N4/Ag2WO4	Ag2WO4	g‐C3N4	Ag2WO4 Interface, g‐C3N4 Interface	Degradation of MO	[305]
g‐C3N4/Ag2CrO4	Ag2CrO4	g‐C3N4	Ag2CrO4 Interface, g‐C3N4 Interface	Degradation of MO	[306]
g‐C3N4/Bi2O3	Bi2O3	g‐C3N4	Bi2O3 Interface, g‐C3N4 Interface	Photodegradation of MB, RhB	[307]
g‐C3N4/BiVO4	BiVO4	g‐C3N4	BiVO4 Interface, g‐C3N4 Interface	Degradation of toluene	[308]
g‐C3N4/WO3	WO3	g‐C3N4	WO3 Interface, g‐C3N4 Interface	Decomposition of acetaldehyde, MB	[309, 310, 311, 312]
g‐C3N4/ZnO	ZnO	g‐C3N4	ZnOInterface, g‐C3N4 Interface	Photocatalytic CO2 conversion to fuel, CO2 reduction to CH3OH	[229, 313]
g‐C3N4/V2O5	V2O5	g‐C3N4	V2O5 Interface, g‐C3N4 Interface	Degradation of RhB and TC	[314]
g‐C3N4/Ag2CrO4	Ag2CrO4	g‐C3N4	Ag2CrO4 Interface, g‐C3N4 Interface	Photodegradation of MB, RhB, MO	[315, 316]
g‐C3N4/Bi12GeO20	Bi12GeO20	g‐C3N4	Bi12GeO20 Interface, g‐C3N4 Interface	Degradation of microcystin‐LR and RhB, and for reduction of aqueous Cr(VI)	[317]
g‐C3N4/Bi2Sn2O7	Bi2Sn2O7	g‐C3N4	Bi2Sn2O7 Interface, g‐C3N4 Interface	Degradation of MB and acid red 18 (AR 18)	[318]
g‐C3N4/Ag2CO3	Ag2CO3	g‐C3N4	Ag2CO3 Interface, g‐C3N4 Interface	Degradation of MO	[319]
g‐C3N4/BiOI	BiOI	g‐C3N4	BiOIInterface, g‐C3N4 Interface	Reduction of CO2	[320]
g‐C3N4/Ag3PO4	Ag3PO4	g‐C3N4	Ag3PO4 Interface, g‐C3N4 Interface	CO2 conversion to fuel	[321, 322]
g‐C3N4/MoO3	MoO3	g‐C3N4	MoO3 Interface, g‐C3N4 Interface	Degradation of MO	[323]
g‐C3N4/DyVO4	g‐C3N4	DyVO4	g‐C3N4 DyVO4, DyVO4 g‐C3N4	Degradation of RhB	[324]
g‐C3N4/Bi2WO6	Bi2WO6	g‐C3N4	Bi2WO6 Interface, g‐C3N4 Interface	Photoreduction of CO2 to CO	[325]
g‐C3N4/ZrO2	g‐C3N4	ZrO2	g‐C3N4 ZrO2, ZrO2 g‐C3N4	Degradation of RhB	[326]
AgCl/g‐C3N4	AgCl	g‐C3N4	AgClInterface, g‐C3N4 Interface	H2 production	[327]
BiOIO3/g‐C3N4	BiOIO3	g‐C3N4	BiOIO3 Interface, g‐C3N4 Interface	Degradation of NO	[328]
Bi2S3/g‐C3N4	Bi2S3	g‐C3N4	Bi2S3 Interface, g‐C3N4 Interface	Photoreduction of CO2 to CO	[329]
MnCo2O4/g‐C3N4	g‐C3N4	MnCo2O4	g‐C3N4 MnCo2O4, MnCo2O4 g‐C3N4	H2 production	[330]
WS2/g‐C3N4	g‐C3N4	WS2	g‐C3N4 WS2	H2 production	[331]
g‐C3N4/SnS	g‐C3N4	SnS	g‐C3N4 SnS, SnSg‐C3N4	Reduction of aqueous Cr(VI)	[332]
g‐C3N4/BiOBr	g‐C3N4	BiOBr	g‐C3N4 BiOBr, BiOBrg‐C3N4	Oxidation of NO and reduction of CO2	[333]

8. Semiconductor–(RGO/Metal)–Graphitic Carbon Nitride (g‐C₃N₄) Heterojunctions

Recently, much effort has been devoted to the construction of all‐solid‐state Z‐scheme semiconductor–metal–g‐C₃N₄ and semiconductor–RGO–g‐C₃N₄ heterojunctions. These designs aim to minimize recombination and lattice mismatch by including metal and RGO mediators for efficient optoelectronic conversion.

Peng et al. synthesized a CdS/Au/g‐C₃N₄ ternary heterojunction by a facile two‐step photoreduction method; this device exhibited enhanced visible light photocatalytic performance. When the CdS/Au/g‐C₃N₄ heterojunction was irradiated, photogenerated electrons from CdS and holes from g‐C₃N₄ transferred to the intermediate metal; this process favored oxidation and reduction separately at CdS and g‐C₃N₄, respectively. Experimentally, the ternary CdS/Au/g‐C₃N₄ heterojunction exhibited better photocatalytic activity than did CdS/g‐C₃N₄. This may be due to the effect of the smaller sized CdS nanoparticles in the ternary than in the binary heterojunction, resulting in enhanced charge separation of photogenerated electron–hole pairs. Another possibility is that Z‐scheme formation in CdS/Au/g‐C₃N₄ imparts greater redox ability to excited charges than in the binary Au/g‐C₃N₄.^{[ 84 ]}

Lately, RGO has been investigated as a mediator in semiconductor–g‐C₃N₄ heterojunctions. Wu et al. demonstrated an all‐solid‐state g‐C₃N₄–RGO–TiO₂ Z‐scheme device that performs enhanced photocatalytic degradation of methylene blue. Upon light illumination photogenerated electrons from TiO₂ and holes from g‐C₃N₄ collect in RGO, enabling oxidation at TiO₂ and reduction at g‐C₃N₄. This indirect g‐C₃N₄–RGO–TiO₂ Z‐scheme device exhibited a maximal degradation rate of 0.0137 min⁻¹—about 4.7 and 3.2 times greater than in either pure g‐C₃N₄ (0.0029 min⁻¹) or direct Z‐scheme g‐C₃N₄–TiO₂ (0.0043 min⁻¹), respectively. The improved charge transfer and separation due to incorporating RGO in this nanoheterojunction results in an enhanced reduction of oxygen at g‐C₃N₄ and oxidation of hydroxyl radicals at TiO₂.^{[ 85 ]}

Marchal et al. produced Au/TiO₂‐g‐C₃N₄ nanocomposites with very low amounts of sacrificial agents present to split water for hydrogen production. When this heterojunction was exposed to light, photogenerated electrons from TiO₂ and g‐C₃N₄ migrated to and deposited in Au nanoparticles, favoring the production of H₂. High energy holes in the TiO₂ and g‐C₃N₄ layers oxidize H₂O and produce oxygen. This composite photocatalyst yields high H₂ production (350 μmol h⁻¹ g⁻¹ catalyst) using minimal amounts of sacrificial agent (≤1 vol%); its performance exceeds that of either binary Au/TiO₂ or Au/g‐C₃N₄ under solar and visible light irradiation. This enhanced performance is due to the homogeneous deposition of Au NPs onto both semiconductors, the SPR of the Au NPs, and the perfect VB and CB alignment forming a Type‐II heterojunction; these factors facilitate the charge transfer and charge separation that induces photogenerated electrons to transfer from the CB of g‐C₃N₄ to the CB of TiO₂.^{[ 86 ]} Recently, an investigation on ternary rGO@g‐C₃N₄/ZnO by Saeed et al.^{[ 87 ]} revealed increased photocatalytic MB degradation of 91.5% under visible light. The experimental finding showed an improved ability to harvest visible light absorption due to the addition of rGO and g‐C₃N₄ to ZnO by reducing the charge recombination. Representative works on S–(RGO/metal)–g‐(C₃N₄) are summarized in Table  5 .

Table 5.

Summary of representative S–(RGO/metal)–g‐(C₃N₄) photocatalyst junction publications

Model/sample	Oxidation site	Reduction site	Mediator	Charge‐transfer direction across the junction	Activity test/application	References
g‐C3N4/CdS@rGO	CdS	g‐C3N4	RGO	CdSRGO, g‐C3N4 RGO	CO2 reduction	[334]
g‐C3N4/GO/AgBr	AgBr	g‐C3N4	GO	AgBrGO, g‐C3N4 GO	Degradation of RhB	[335]
g‐C3N4/Au/CdS	CdS	g‐C3N4	Au	CdSAu, g‐C3N4 Au	Degradation of RhB, H2 production	[84, 336]
g‐C3N4/GO/TiO2	TiO2	g‐C3N4	GO	TiO2 GO, g‐C3N4 GO	Degradation of MB	[85]
g‐C3N4/Ag/BiVO4	BiVO4	g‐C3N4	Ag	BiVO4 Ag, g‐C3N4 Ag	Decomposition of TC	[337]
g‐C3N4/RGO/Bi2WO6	Bi2WO6	g‐C3N4	RGO	Bi2WO6 RGO, g‐C3N4 RGO	Degradation of 2,4,6‐trichlorophenol (TCP)	[338]
g‐C3N4/RGO /Bi2MoO6	Bi2MoO6	g‐C3N4	RGO	Bi2MoO6 RGO, g‐C3N4 RGO	Degradation of RhB	[339]
g‐C3N4/RGO/CdS	CdS	g‐C3N4	RGO	CdSRGO, g‐C3N4 RGO	H2 generation and degradation of atrazine	[340]
g‐C3N4/BiOI/RGO	g‐C3N4	BiOI	RGO	g‐C3N4 RGO, BiOIRGO	CO2 reduction	[341]
Co3O4–rGO–gC3N4	Co3O4	g‐C3N4	RGO	Co3O4 RGO, g‐C3N4 RGO	H2 generation	[342]
g‐C3N4/RGO/NiFe2O4	NiFe2O4	g‐C3N4	RGO	NiFe2O4 RGO, g‐C3N4 RGO	Degradation of MO	[343]
g‐C3N4/Cu/Cu2O	g‐C3N4	Cu2O	Cu	g‐C3N4 Cu, Cu2OCu	Degradation of MO and phenol	[344]
g‐C3N4/Ag/MoS2	MoS2	g‐C3N4	Ag	g‐C3N4 Ag, Cu2OAg, AgMoS2	Degradation of RhB	[345]
g‐C3N4/Au/CdZnS	CdZnS	g‐C3N4	Au	CdZnSAu, g‐C3N4 Au	H2 generation	[346]
g‐C3N4/Ag /Ag3VO4	Ag3VO4	g‐C3N4	Ag	Ag3VO4 Au, g‐C3N4 Au	Degradation of RhB	[347]
g‐C3N4–Ag–AgVO3	AgVO3	g‐C3N4	Ag	AgVO3 Ag, g‐C3N4 Ag	Degradation of RhB and bacterial inactivation	[348]
g‐C3N4/Ag/Ag3PO4	Ag3PO4	g‐C3N4	Ag	Ag3PO4 Ag, g‐C3N4 Ag	Degradation of RhB	[349]
g‐C3N4–Bi–BiOCl	BiOCl	g‐C3N4	Bi	BiOClBi, g‐C3N4 Bi	Degradation of RhB and Cr(VI) reduction	[350]
g‐C3N4/Ag/Ag2CO3	Ag2CO3	g‐C3N4	Ag	Ag2CO3 Ag, g‐C3N4 Ag	Degradation of RhB	[351]
g‐C3N4/Au/ZnIn2S4	ZnIn2S4	g‐C3N4	Au	ZnIn2S4 Au, g‐C3N4 Au	Nitric oxide removal and carbon dioxide conversion	[352]
g‐C3N4/MoS2(Ni, Co)	MoS2	MoS2	–	g‐C3N4 MoS2, g‐C3N4 MoS2	H2 generation	[353]

9. Conclusions and Outlook

In this critical review, we have first presented an overview and introduction of photocatalysis using typical bare semiconductor and semiconductor‐based heterojunction photocatalysts, and then thoroughly reviewed current state‐of‐the‐art heterojunction photocatalytic devices including semiconductors with graphene, CQDs, and graphitic carbon nitride. The review's primary focus—on the steering of charge kinetics, construction principles, and mechanisms of heterojunction photocatalysis—is presented as an elaborative tutorial on the advantages and disadvantages of Schottky/ohmic and plasmonic junctions. Even though our review primarily focuses on theoretical modeling principles and recent progress in charge transfer and separation proposals for enhanced heterojunction photocatalysis, it also combines experimental observations in parallel for better understanding and inspiration.

Although photocatalysts hold promise for enabling a sustainable future world, there are several roadblocks preventing their use in practical applications. Notable factors that can hinder the achievement of highly efficient photocatalysis include: 1) failure to size the photocatalyst bandgap small enough so that the energies needed to convert photons into e–h pairs match solar spectrum irradiance; 2) an excessively high recombination rate; 3) insufficiently large surface area; 4) insufficiently high chemical stability; 5) suboptimum band positions of the photocatalyst; 6) insufficiently low cost.

To alleviate these shortcomings, a number of well‐known designs have been proposed as holding promise; apart from ones utilizing a single semiconductor—these designs either incorporate the interfacing of a semiconductor with metal, semiconductor–semiconductor, semiconductor–graphene, semiconductor–graphitic carbon nitride, or use metal and RGO as a mediator in the heterojunction structure. Controlling the formation and making use of the advantages of semiconductor–metal designs—with emphasis on their Schottky barrier/ohmic contacts and plasmonic effects—have been the main areas of focus in the past few years. Despite great effort, inadequate charge separation and loss of energy during migration of charge carriers from semiconductor to metal greatly quench the redox ability of these photocatalysts. Recently, a conventional Z‐scheme that uses acceptor and donor mediators and an all‐solid‐state Z‐scheme that uses metals as a mediator have been investigated, with no emphasis to date on the selective steering of charges. In general, Z‐scheme designs mainly aim to minimize both the recombination rate and lattice mismatch at the p–n junction and improve charge transfer dynamics. Recently, a few promising experimental and theoretical simulation studies have been undertaken to enhance the selective steering of charges in some all‐solid‐state Z‐scheme designs to increase the efficiency of charge separation for improved optoelectronic conversion.

Also recently, heterojunction photocatalysis studies have been extended to consider semiconductor–RGO–graphene, graphene–graphitic carbon nitride, and graphene–graphitic carbon nitride designs for better optimization of metal‐free, inexpensive, and environmentally harmless heterojunctions. Such heterojunctions mainly aim to maximize quantum efficiency by increasing the surface area of the photocatalyst, broadening the range of light absorption, and increasing the rate of charge transfer across the interface. Although few studies on the incorporation of CQDs with carbon nitride have been performed, the results reveal interesting effects and attain enhanced quantum efficiency, identifying this design as an emerging promising candidate for heterojunction photocatalysis.

Despite some success in heterojunction photocatalysis design and tuning charge kinetics dynamics, this area of development still faces many challenges. First, the interplay of dynamics at the interface is not sufficiently considered; this process can quench photocatalyst performance. Primarily, the connecting of the energy‐dependent selective steering of charges that facilitates charge migration to the ultimate goal of efficient charge separation has not received enough attention. Another challenge is the failure to apply experimental and theoretical principles at the interface to take proper advantage of the properties of Schottky/ohmic and plasmonic junctions. Especially, insufficient effort has been made at applying the merits of a metal's SPR effect so as to make use of its dual role. Even though experimental and theoretical study methods show advances in studying dynamical properties at surfaces and interfaces, they are limited. Most studies do not present refined charge kinetics dynamics at the interface, especially those related to charge transfer and high energy e–h pair separation; these areas need much more work in the future. In general, to further insight into the area of semiconductor based heterojunctions, the following points can be considered. 1) Less lattice mismatch at the interface with favorable band alignment between different combining layers is crucial. In addition, the area should also focus on an energy‐dependent Z‐scheme p–m–n and n–m–p heterojunction to steer charges selectively across the interface to achieve good charge separation; and 2) In synthesizing metal‐free photocatalysts, combining g‐C₃N₄ with large bandgap semiconductors is a promising strategy to extend its light absorption region and increase its surface area. In addition, as experts and reviewers of this area, we grasp that the proportional composition of materials at the interface, their morphology, and crystal structures have not been properly considered, despite the fact that these properties greatly affect the photocatalytic activity of the photocatalyst.

Conflict of Interest

The authors declare no conflict of interest.

Biographies

Mesfin Eshete is an assistant professor of physical chemistry at Addis Ababa Science and Technology University in the School of Applied Sciences, Department of Industrial Chemistry. He honored his Doctor of Philosophy in physical chemistry (Ph.D.) from the University of Science and Technology of China (China) in 2020. His main research interests are structural and electronic properties of novel materials, electron dynamics at the surface and interface in photocatalytic systems (composition structures with semiconductor, metal, RGO, graphene), and designing of novel heterojunctions in energy conversion for application of photocatalysis, catalysis, pervoskites.

Jun Jiang is a professor at the School of Chemistry and Materials Science, University of Science and Technology of China (USTC). He received a B.S. degree in theoretical physics in 2000 at Wuhan University, China, a Ph.D. degree in theoretical chemistry under the tutelage of Prof. Yi Luo in 2007 at the Royal Institute of Technology, Sweden, a Ph.D. degree in solid state physics under the tutelage of Prof. Wei Lu in 2008 at Shanghai Institute of Technical Physics, Chinese Academy of Science. From 2008 to 2011, he worked as a post doc at the Royal Institute of Technology, Sweden, and the University of California Irvine under the tutelage of Prof. Shaul Mukamel. He joined the University of Science and Technology of China in December 2011 as a professor of physical chemistry.