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

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 (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-interfacebased 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.
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 þ to H 2 reduction potential (0 V vs NHE at pH ¼ 0), whereas the top of the VB must be located at a value more positive than the H 2 O to O 2 oxidation potential (1.23 V vs NHE at pH ¼ 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 TiO 2 , 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), semiconductormetal, semiconductor-semiconductor-metal, or semiconductormetal-semiconductor layers in combination.

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-semiconductormetal, 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 semiconductorgraphene, semiconductor(S)-graphitic carbon nitride (C 3 N 4 ), semiconductor-(RGO/metal)-graphitic carbon nitride (g-C 3 N 4 ); 3) recent fascinating investigations of CQDs incorporated into graphitic carbon nitride heterojunctions, and the advances achieved by each for enhancing the overall photocatalytic process.
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.
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 Cu 2 O/TiO 2 heterojunction photocatalyst in which p-type Cu 2 O and n-type TiO 2 contact directly. [29] Recently, C. Ding et al. synthesized a highly efficient Cu 2 O-TiO 2 heterojunction by a wet chemical method. The Cu 2 O-TiO 2 heterojunction prepared in this way achieves an improved methylene blue degradation rate of 93.67% in 45 min, more efficient than pristine Cu 2 O. At the p-n junction, free electrons diffuse from TiO 2 to Cu 2 O and holes from Cu 2 O to TiO 2 until equilibrium is reached. Consequently, they leave negative and positive regions at the Cu 2 O/TiO 2 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 Cu 2 O (E g ¼ 2À2.2 eV) [30][31][32] to the CB of TiO 2 (E g ¼ 3.2 eV) and photoexcited holes migrate from the VB of TiO 2 to the VB of Cu 2 O, a condition that favors reduction at TiO 2 and oxidation at Cu 2 O. 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 MoS 2 /TiO 2 (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 MoS 2 /TiO 2 heterojunction has 17.8 times higher activity than that of pristine TiO 2 . The photocatalytic degradation enhancement factor of the corresponding kinetic constant is about 5.2. Such a dramatic improvement originates: 1) from the incorporation of MoS 2 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.   [190] 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. Degradation of isopropyl alcohol hydrogen evolution [193] BiOCl-SrFe 12

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.
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] 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  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 Cu 2 O cubes at moderate Pd density produced hydrogen at a rate of 2.20 mmol g À1 . [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) BiVO 4 crystal facets. [38] Their results showed that metals or oxides can be deposited selectively on specific facets of BiVO 4 ; 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/TiO 2 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 TiO 2 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-TiO 2 are efficiently delivered to the lateral surfaces of the crystal through the TiO 2 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/TiO 2 NP systems. [42] Jiang et al. studied the key role of metals in the photocatalytic activity of Au-CeO 2 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 CeO 2 . 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 O 2 dissociation and activation. In general, this investigation shows that the design of efficient metalÀsemiconductor 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.
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 highenergy 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.

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.
In other words, the Z-scheme guarantees that each of two halfreactions 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 Fe 3þ /Fe 2þ and IO 3 À /I À ) 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 ADspecies 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. Au-CdSe Plasmon Au CdSe Au CdSe λ > 700 nm Hydrogen generation [248] www.advancedsciencenews.com www.small-science-journal.com 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 Ag 2 S in a TiO 2 -Ag-Ag 2 S heterojunction. The work function difference between Ag and Ag 2 S caused free electrons to flow from the low work function (high Fermi level) of Ag to the high work function (low Fermi level) of Ag 2 S until their Fermi levels equilibrated, resulting in an upshift and downward band bending. As a consequence, the TiO 2 -Ag-Ag 2 S heterostructure aligned to form a perfect Z-scheme. Upon exposure of the heterojunction to light, high energy photoexcited holes and electrons residing on TiO 2 and Ag 2 S 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 À1 -higher than the sum of λ < 400 nm (1.3 μ molh À1 ) and λ > 400 nm (0.19 μmol h À1 ), as shown in Figure 6. [51] The performance enhancement suggests that the TiO 2 -Ag-Ag 2 S 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 TiO 2 -Ag-Ag 2 S 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 À1 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-WO 3 and CdS-Pt-WO 3 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 Figure 6. a) Average rates of photocatalytic hydrogen production and b) photocurrents versus time (I-t) curves of Ag 2 S-(Ag)-TiO 2 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. 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.
Fermi level of CdS is higher than that of either Au or Pt and that the Fermi level of WO 3 is between the levels of these two. In this system, free electrons transfer from CdS to Au (or Pt), from Au to WO 3 , and also from WO 3 to Pt; this results in the formation of depletion and accumulation layers at the CdS/Au and Au/WO 3 interfaces, respectively, and depletion layers on both sides of the CdS/Pt and Pt/WO 3 interfaces.
When the CdS-Au-WO 3 heterojunction is exposed to light, photoinduced electrons from WO 3  The finding shows explicitly that work function differences result in a lower Schottky barrier height (W CdS ÀW Au < W CdS -W Pt ) at the Au/CdS interface than at Pt/CdS; thus charge carrier transfer in the WO 3 /Au/CdS heterostructure is smoother, and H 2 evolution efficiency better than in WO 3 /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-semiconductormetal and semiconductor-metal-semiconductor photocatalyst junction studies are presented in Table 3. AgCl-Ag-BiOCl n-S1-m-p-S2 AgCl BiOCl Ag BiOCl, Ag AgCl MO Degradation [253] CdS-Au-ZnO n-S1-m-n-S2 ZnO CdS ZnO Au, CdS Au H 2 generation [254] CdS-Au-TiO 2 n-S1-m-n-S2 TiO 2 CdS TiO 2 Au, Au TiO 2 MV 2þ reduction [255] CdS-Au-BiOCl n-S1-m-p-S2 CdS-Au-WO 3 n-S1-m-n-S2 WO 3 CdS WO 3 Au, Au CdS Hydrogen and oxygen evolution [268] Pt-CeO 2 -ZnO m-n-S1-n-S2 Pt Ag-AgCl-BiPO 4 m-n-S1-n-S2 Ag BiPO 4 Ag AgCl, AgCl BiPO 4, BiPO 4 AgCl RhB degradation [271] Ag-AgVO 3 -BiOCl m-n-S1-n-S2 AgVO 3 BiOCl AgVO 3 Ag, AgVO 3 BiOCl MB degradation [272] www.advancedsciencenews.com www.small-science-journal.com 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 highenergy 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] A combined experimental and theoretical study of an n-typemetal-p-type TiO 2 -Pd-Cu 2 O Z-scheme has been performed recently by Ye et al. Their design achieved improved photocatalysis by interfacing n-type TiO 2 (001) and p-type Cu 2 O(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 TiO 2 (001) to Pd and holes from Cu 2 O(100) to Pd. The TiO 2 (001) < Pd < Cu 2 O(100) work Ag-AgBr-InVO 4 m-n-S1-n-S2 InVO 4 AgBr InVO 4 AgBr, Ag AgBr, AgBr InVO 4 RhB degradation [274] Ag-TiO 2 -ZnO m-n-S1-n-S2 ZnO Ag ZnO TiO 2 , TiO 2 Ag, TiO 2 ZnO Phenol degradation [275] Ag-SnS-TiO 2 m-n-S1-n-S2 SnS Ag SnS Ag, TiO 2 Ag RhB and MB degradation [276] Ag-AgBr-Ag 3 VO 4 m-n-S1-n-S2 AgBr-Ag 3 VO 4 Ag AgBr, Ag BrAg 3 VO 4 , Ag 3 VO 4 AgBr MO degradation [277] Ag-Ag 2 O-BiOCl m-p-S1-p-S2 Ag 2 O BiOCl Ag 2 O Ag, BiOCl Ag RhB degradation [278] Ag-Ag 2 S-CuS m-n-S1-p-S2 Ag 2 S A g CuS Ag, Ag 2 S Ag, CuS Ag 2 S 2,4-Dichlorophenol degradation [279] Ag-AgCl-ZnO m-n-S1-n-S2 AgCl ZnO AgCl ZnO, Ag ZnO RhB and MO degradation [280] Cu-Cu 2 O-ZnO m-p-S1-n-S2 Ag-AgBr-TiO 2 m-n-S1-n-S2 AgBr TiO 2 Ag AgBr, Ag BrTiO 2 , TiO 2 AgBr MB degradation [289] Ag-SrTiO 3 -TiO 2 m-n-S1-n-S2 Acid blue 92 (AB92) degradation [295] NiO-ZnO-Au p-S1-n-S2-m ZnO NiO, Au ZnO Au, ZnO NiO, NiO ZnO RhB degradation [296] NiO-ZnO-Pt p-S1-n-S2- functions ordering allows free electrons to flow from TiO 2 (001) to Pd and from Pd to Cu 2 O(100) and equilibrates the Fermi levels. Upon light illumination, this facet hybrid junction facilitates low energy electron transfer from the TiO 2 (001) facet to the intermediate Pd layer and hole migration from Cu 2 O(100) to Pd; good charge separation is produced by confining high energy holes at TiO 2 (001) -enabling oxidation-and high-energy electrons at Cu 2 O(100)enabling reduction. This hybrid TiO 2 (001) < Pd < Cu 2 O(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 TiO 2 (101) facet or Cu 2 O(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 CO 2 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 Zscheme photocatalysts for practical applications. Our extension 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.
www.advancedsciencenews.com www.small-science-journal.com of this design that utilizes thin layers in an energy-dependent Zscheme, the Cu 2 S-Pt-WO 3 (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-TiO 2 nanocomposite with graphene oxide (GO) suspended in ethanol. Upon UV light irradiation, photoexcited electrons transferred from TiO 2 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 TiO 2 was excluded from the solution, confirming that surface electrons migrate from TiO 2 and reduce GO. [65] Similar studies on graphene-TiO 2 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 CO 2 to CH 4 . Their analysis revealed that the RGO-CdS nanorod junction at 0.5 wt% of RGO achieved a high production rate (2.51 μmol h À1 g À1 ), 10 times greater than that of pristine CdS nanorods ( Figure 8).
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 Ag 2 CrO 4 -GO, [74] TiO 2 -S/rGO, [75] and GO/CuFe 2 O 4 , [76] have been investigated as materials that can replace expensive noble metals with carbon, and ultimately enhance the photocatalytic activity of single semiconductors.

Semiconductor (S)-Graphitic Carbon Nitride (C 3 N 4 ) Heterojunctions
Recently, other heterojunctions composed of S-C 3 N 4 have been widely investigated by many researchers for optimizing single semiconductor photoelectronic conversion. C 3 N 4 is inexpensive, harmless, and has a narrow bandgap of 2.7 eV, qualifying it for consideration as a photocatalytic material for H 2 production, O 2 evolution, and CO 2 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 CO 2 to CH 3 OH on g-C 3 N 4 -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-C 3 N 4 resulting in enhanced photocurrent production as confirmed experimentally. Indeed, this g-C 3 N 4 -ZnO photocatalyst attains 2.3 times higher photocatalytic activity than does pure g-C 3 N 4 . [79] Li et al. synthesized a direct Z-scheme g-C 3 N 4 -TiO 2 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 TiO 2 layer and recombine with holes from g-C 3 N 4 ; this favors reduction at g-C 3 N 4 and oxidation at the TiO 2 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-C 3 N 4 -TiO 2 -400 and 30%g-C 3 N 4 -TiO 2 -600 -they showed that increased light absorption does not always lead to improved photocatalytic activity. Rather,  20%g-C 3 N 4 -TiO 2 -400, which possesses a larger specific surface area and better visible light absorption, results in lower photocatalytic activity than 30%g-C 3 N 4 -TiO 2 -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 (C 3 N 4 ) nanocomposite for photocatalytic water splitting. Such a CQDs-C 3 N 4 hybrid is made of low-cost and environmentally friendly materials. This device attained quantum efficiencies of 16% for wavelength λ ¼ 420 AE 20 nm, 6.29% for λ ¼ 580 AE 15 nm, and 4.42% for λ ¼ 600 AE 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 photocatalysta 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 C 3 N and the CQDs were calculated to be 3.25 and 4.65 eV, respectively, so that electron flow is driven from C 3 N to the CQDs. The computed CQD/C 3 N differential charges indicate that holes collect at C 3 N and electrons at the CQDs. The computed absorption coefficients, as shown in Figure 9b, reveal that while bare C 3 N absorbs mainly short-wavelength visible light (<600 nm) and CQD absorption extends to %700 nm, absorption of the CQDs/C 3 N composite extends to %900 nm.
Extended time-dependent density functional theory (TDDFT) calculations reveal that photoexcited electrons transfer from C 3 N to CQD, which is in good agreement with ultrafast charge evolution results, which favor oxidation at C 3 N and reduction at CQD sites. This CQDs/C 3 N hybrid photocatalyst design enables the full use of visible and IR light to generate and distribute high energy holes and electrons on the C 3 N and CQDs layers. Representative works on S-g-C 3 N 4 photocatalyst junctions are summarized in Table 4. [83] 8. Semiconductor-(RGO/Metal)-Graphitic Carbon Nitride (g-C 3

N 4 ) Heterojunctions
Recently, much effort has been devoted to the construction of all-solid-state Z-scheme semiconductor-metal-g-C 3 N 4 and semiconductor-RGO-g-C 3 N 4 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-C 3 N 4 ternary heterojunction by a facile two-step photoreduction method; this device exhibited enhanced visible light photocatalytic performance. When the CdS/Au/g-C 3 N 4 heterojunction was irradiated, photogenerated electrons from CdS and holes from g-C 3 N 4 transferred to the intermediate metal; this process favored oxidation and reduction separately at CdS and g-C 3 N 4 , respectively. Experimentally, the ternary CdS/Au/g-C 3 N 4 heterojunction exhibited better photocatalytic activity than did CdS/g-C 3 N 4 . 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-C 3 N 4 imparts greater redox ability to excited charges than in the binary Au/g-C 3 N 4 . [84] Lately, RGO has been investigated as a mediator in semiconductor-g-C 3 N 4 heterojunctions. Wu et al. demonstrated an allsolid-state g-C 3 N 4 -RGO-TiO 2 Z-scheme device that performs enhanced photocatalytic degradation of methylene blue. Upon light illumination photogenerated electrons from TiO 2 and holes from g-C 3 N 4 collect in RGO, enabling oxidation at TiO 2 and reduction at g-C 3 N 4 . This indirect g-C 3 N 4 -RGO-TiO 2 Z-scheme device exhibited a maximal degradation rate of 0.0137 min À1about 4.7 and 3.2 times greater than in either pure g-C 3 N 4 (0.0029 min À1 ) or direct Z-scheme g-C 3 N 4 -TiO 2 (0.0043 min À1 ), respectively. The improved charge transfer and separation due to incorporating RGO in this nanoheterojunction results in an enhanced reduction of oxygen at g-C 3 N 4 and oxidation of hydroxyl radicals at TiO 2 . [85] Figure 9. a) Differential charge distribution of the carbon quantum dot (CQD)/C 3 N structure. The yellow and blue regions represent electron and hole charge distributions, respectively, with an isosurface value of 0.0006 e Å À3 . b) Photoabsorption spectrum of the bare C 3 N (red), CQD (blue), and CQD/ C 3 N composite structures (black). c) PDOS of the CQDs/C 3 N hybrid. a-c) Reproduced with permission. [83] Copyright 2018, Royal Society of Chemistry.
www.advancedsciencenews.com www.small-science-journal.com Marchal et al. produced Au/TiO 2 -g-C 3 N 4 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 TiO 2 and g-C 3 N 4 migrated to and deposited in Au nanoparticles, favoring the production of H 2 . High energy holes in the TiO 2 and g-C 3 N 4 layers oxidize H 2 O and produce oxygen. This composite photocatalyst yields high H 2 production (350 μmol h À1 g À1 catalyst) using minimal amounts of sacrificial agent (≤1 vol%); its performance exceeds that of either binary Au/TiO 2 or Au/g-C 3 N 4 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-C 3 N 4 to the CB of TiO 2 . [86] Recently, an investigation on ternary rGO@g-C 3 N 4 /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-C 3 N 4 to ZnO by reducing the charge recombination. Representative works on S-(RGO/metal)-g-(C 3 N 4 ) are summarized in Table 5. WS 2 /g-C 3 N 4 g-C 3 N 4 WS 2 g-C 3 N 4 WS 2 H 2 production [331] g-C 3 N 4 /SnS g-C 3 N 4 SnS g-C 3 N 4 SnS, SnS g-C 3 N 4 Reduction of aqueous Cr(VI) [332] g-C 3 N 4 /BiOBr g-C 3 N 4 BiOBr g-C 3 N 4 BiOBr, BiOBr g-C 3 N 4 Oxidation of NO and reduction of CO 2 [333] www.advancedsciencenews.com www.small-science-journal.com

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, semiconductorgraphitic 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 Table 5. Summary of representative S-(RGO/metal)-g-(C 3 N 4 ) photocatalyst junction publications.

Model/sample
Oxidation site

Reduction site
Mediator Charge-transfer direction across the junction the efficiency of charge separation for improved optoelectronic conversion.
Also recently, heterojunction photocatalysis studies have been extended to consider semiconductor-RGO-graphene, graphenegraphitic 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-C 3 N 4 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.