Recent Advances in CuInS2-Based Photocathodes for Photoelectrochemical H2 Evolution

Photoelectrochemical (PEC) H2 production from water using solar energy is an ideal and environmentally friendly process. CuInS2 is a p-type semiconductor that offers many advantages for PEC H2 production. Therefore, this review summarizes studies on CuInS2-based PEC cells designed for H2 production. The theoretical background of PEC H2 evolution and properties of the CuInS2 semiconductor are initially explored. Subsequently, certain important strategies that have been executed to improve the activity and charge-separation characteristics of CuInS2 photoelectrodes are examined; these include CuInS2 synthesis methods, nanostructure development, heterojunction construction, and cocatalyst design. This review helps enhance the understanding of state-of-the-art CuInS2-based photocathodes to enable the development of superior equivalents for efficient PEC H2 production.


Introduction
The use of renewable energy to reduce greenhouse gas emissions and environmental problems caused by fossil fuels is attracting considerable attention [1]. However, the intermittent nature of renewable energy necessitates the use of diverse energy storage systems. In particular, hydrogen is a promising, environmentally friendly, and carbonneutral energy resource that emits only water while providing energy [2]. Hydrogen can be utilized in various systems, such as gas turbines, combustors, internal combustion engines, and fuel cells. However, CO 2 -emitting fossil fuels are primarily being used to produce hydrogen owing to the economic feasibility of this strategy, thereby preventing the realization of a carbon-neutral society. In contrast, the production of hydrogen and oxygen through water splitting using renewable solar energy is an ideal method for storing solar energy in chemical form. Photoelectrochemical (PEC) water splitting is a representative method for converting solar light into chemical energy. PEC cells comprise a semiconductor photoelectrode, a counter electrode, and an electrolyte. When photons are absorbed by the photocatalytic electrode, photo-excited electron-hole pairs drive the water-splitting reaction (H 2 O(l) → H 2 (g) + 1/2O 2 (g), ∆H 0 f ,298 K = −285.8 kJ/mol). However, the commercialization of PEC hydrogen production technology is hindered by its low solar-to-hydrogen conversion efficiency [3]. Therefore, the efficiency of PEC energy conversion must be improved.
A PEC cell is an ideal and environmentally friendly system for producing hydrogen via water photolysis, similar to photosynthesis in nature. Unlike photocatalytic water splitting, PEC water splitting does not require the separation of the generated hydrogen A PEC cell is an ideal and environmentally friendly system for producing hydrog via water photolysis, similar to photosynthesis in nature. Unlike photocatalytic wa splitting, PEC water splitting does not require the separation of the generated hydrog and oxygen. Moreover, PEC cells have theoretically been predicted to achieve high so to-hydrogen conversion efficiencies depending on a single photo absorber or tandem c figuration system (10-30%) [4]. In particular, the semiconductor photoelectrode is most important part of PEC cells, as it interacts with photons and converts them into ot forms of energy via charge-carrier excitation. Therefore, semiconductor materials suita for photoelectrodes must be developed to increase the efficiency of the PEC reaction [5 When semiconductor electrodes are in contact with electrolytes, the semiconduc experiences band bending owing to equilibration between the Fermi level of the semic ductor and the redox potential of the liquid. Under this band-bending condition, a dep tion layer is formed, and a built-in potential is generated. When a semiconductor-liq interface is irradiated with light, photons are absorbed by the semiconductor, and exci electron-hole pairs are subsequently generated. The electron-hole pairs are separated the built-in potential; the electrons reduce the chemical species, and the holes oxidize chemical species separately. In the case of n-type semiconductors, holes are transpor into the semiconductor-liquid interface, producing oxygen, whereas in p-type semic ductors, electrons are transported into the semiconductor-liquid interface, producing drogen ( Figure 1).

Figure 1.
Band bending of p-type semiconductors in electrolytes. When the semiconductor and e trolyte are contacted, the Fermi level of the semiconductor and the chemical potential of the elec lyte become equilibrated and lead to a 'band-bending energy' structure. Under the light illuminat condition, the excess photo-excited charge carrier breaks the equilibrium state and leads to cha separation via built-in potential at the semiconductor-electrolyte interface. A schematic of a pho electrochemical (PEC) cell with a p-type semiconductor photocathode is also depicted for w Figure 1. Band bending of p-type semiconductors in electrolytes. When the semiconductor and electrolyte are contacted, the Fermi level of the semiconductor and the chemical potential of the electrolyte become equilibrated and lead to a 'band-bending energy' structure. Under the light illumination condition, the excess photo-excited charge carrier breaks the equilibrium state and leads to charge separation via built-in potential at the semiconductor-electrolyte interface. A schematic of a photoelectrochemical (PEC) cell with a p-type semiconductor photocathode is also depicted for water splitting. Photo-excited electron is transported to the semiconductor/electrolyte interface, while the photo-excited hole is transported to the counter electrode through electrical wire and then to the counter electrode-electrolyte interface. Efficient PEC water-splitting reactions with semiconductor materials can occur under various conditions. First, the bandgap of semiconductor materials should be smaller than 3.0 eV for the efficient utilization of visible solar light. Second, the bandgap must be higher than 1.23 eV to generate a sufficient driving force for the water-splitting reaction. Third, the band position (that is, the energy levels of the conduction and valence bands) should be located at a suitable level compared to the redox potential of water. The conduction band of the semiconductor should have a higher energy level than that of HER, while the valence band of the semiconductor should have a lower energy level than that of OER. Additionally, the semiconductor material should be free from photo-corrosion during the PEC reactions, or the semiconductor should be protected by a passive layer from corrosion [6]. Figure 2 shows the bandgaps and band positions of commonly investigated semiconductors. However, semiconductors that can utilize visible solar light, exhibit PEC stability, and spontaneously decompose water without an external potential have not yet been reported. Therefore, various strategies have been attempted to resolve these shortcomings of semiconductor photoelectrodes. splitting. Photo-excited electron is transported to the semiconductor/electrolyte interface, while the photo-excited hole is transported to the counter electrode through electrical wire and then to the counter electrode-electrolyte interface.
Efficient PEC water-splitting reactions with semiconductor materials can occur under various conditions. First, the bandgap of semiconductor materials should be smaller than 3.0 eV for the efficient utilization of visible solar light. Second, the bandgap must be higher than 1.23 eV to generate a sufficient driving force for the water-splitting reaction. Third, the band position (that is, the energy levels of the conduction and valence bands) should be located at a suitable level compared to the redox potential of water. The conduction band of the semiconductor should have a higher energy level than that of HER, while the valence band of the semiconductor should have a lower energy level than that of OER. Additionally, the semiconductor material should be free from photo-corrosion during the PEC reactions, or the semiconductor should be protected by a passive layer from corrosion [6]. Figure 2 shows the bandgaps and band positions of commonly investigated semiconductors. However, semiconductors that can utilize visible solar light, exhibit PEC stability, and spontaneously decompose water without an external potential have not yet been reported. Therefore, various strategies have been attempted to resolve these shortcomings of semiconductor photoelectrodes. Among the reported semiconductor materials, chalcopyrites are attractive photocatalytic substances appropriate for PEC hydrogen production. Chalcopyrites are semiconductors with the chemical composition ABX2 and have a high absorption coefficient. Additionally, their band position is suitable for the hydrogen evolution reaction (HER). In particular, Cu-based chalcopyrite composites (A = Cu or Ag; B = In, Ga, or Al; X = S or Se) have a significantly high absorption coefficient (~10 5 cm −1 ) owing to their direct bandgap structure, resulting in superior light-absorption characteristics compared to those of other semiconductor materials (Figure 3) [7]. Moreover, the bandgap of chalcopyrites can be easily adjusted (1.1-2.5 eV) by forming a solid solution via the replacement of the samegroup atoms in the crystal lattice [8]. Furthermore, the higher level of the chalcopyrite conduction band allows excited electrons to have sufficient energy for driving the HER. Among the reported semiconductor materials, chalcopyrites are attractive photocatalytic substances appropriate for PEC hydrogen production. Chalcopyrites are semiconductors with the chemical composition ABX 2 and have a high absorption coefficient. Additionally, their band position is suitable for the hydrogen evolution reaction (HER). In particular, Cu-based chalcopyrite composites (A = Cu or Ag; B = In, Ga, or Al; X = S or Se) have a significantly high absorption coefficient (~10 5 cm −1 ) owing to their direct bandgap structure, resulting in superior light-absorption characteristics compared to those of other semiconductor materials ( Figure 3) [7]. Moreover, the bandgap of chalcopyrites can be easily adjusted (1.1-2.5 eV) by forming a solid solution via the replacement of the same-group atoms in the crystal lattice [8]. Furthermore, the higher level of the chalcopyrite conduction band allows excited electrons to have sufficient energy for driving the HER.  In this review, research trends in the use of photocathodes based on CuInS2-a not Reproduced from ref. [9] with permission of the American Physical Society.
In this review, research trends in the use of photocathodes based on CuInS 2 -a noteworthy chalcopyrite semiconductor-for PEC hydrogen evolution are surveyed. CuInS 2 has attractive properties, such as a suitable band position and bandgap for photocatalytic electrodes. Moreover, it is an economical chalcopyrite because it does not comprise relatively expensive elements, such as Ag, Ga, and Se. Additionally, it can be synthesized using economical methods, such as non-vacuum-based electrodeposition, solvothermal processes, and screen-printing techniques. Although concerns related to indium resource depletion exist, the amount of indium in the Earth's crust is sufficient for solar energy conversion and industrial applications [10].
The advantageous physical properties of CuInS 2 permit PEC hydrogen production. Because CuInS 2 has a bandgap of~1.5 eV, which enables solar light absorption over an extremely wide range (from ultraviolet to near-infrared) and a high extinction coefficient (>10 5 cm −1 ) [11], it has been used in solar energy conversion systems, such as solar cells and photocatalysts, for a long time [12][13][14][15]. Although CuInS 2 generally exhibits p-type semiconductor characteristics, it can also attain n-type semiconductor attributes through the control of Cu vacancies or dopants [16][17][18]. However, in most studies on PEC cells, CuInS 2 is used as a p-type semiconductor and as a photocathode for hydrogen evolution. Despite having an excellent light-absorption ability, CuInS 2 -based photoelectrodes generate a photocurrent of~16 mA·cm −2 , which is approximately half of the theoretical photocurrent (~28.96 mA·cm −2 ) predicted from the bandgap (1.55 eV). Table 1 summarizes the activity of CuInS 2 photocathodes reported to date for PEC hydrogen evolution in terms of the photocurrent density, onset potential, and half-cell applied-bias photon-to-current conversion efficiency (HC-ABPE). The activity of CuInS 2 remains restricted by the high recombination of photogenerated electron-hole pairs, a surface state, inferior electrochemical catalytic properties, and low photochemical stability [19,20].   Therefore, methods for fabricating CuInS 2 photocathodes and strategies for improving their PEC performance are summarized and discussed in this review.

Synthesis Methods and Morphological Control Techniques for CuInS 2 Photocathodes
The synthesis methods for CuInS 2 thin films can be categorized into physical and chemical techniques. Sputtering and co-evaporation are representative physical methods that require ultrahigh-vacuum conditions, under which phase-pure chalcopyrite crystals with minimal impurities can be grown. Sputtering is a typical method for synthesizing CuInS 2 under vacuum conditions, in which Cu and In are sequentially deposited on substrates and converted into CuInS 2 through sulfurization. This method can produce thin films with excellent uniformity and readily controllable stoichiometries. However, the stipulated vacuum conditions lead to a remarkably high thin-film fabrication cost, increasing the unit cost of hydrogen production [52]. Therefore, wet chemical methods are being studied as an alternative CuInS 2 synthesis strategy ( Figure 4).   [33] and (b) electrodeposition/sulfurization [42]. (c) Linear sweep voltammetry (LSV) profiles and HC-ABPE data of CuInS2-based photocathodes fabricated by electrodeposition/sulfurization [42]. Reproduced from ref. [29] with permission of the Royal Society of Chemistry. Reproduced from ref. [39] with permission from Wiley-VCH.
CuInS2 can also be synthesized using solution-based approaches, such as chemical bath deposition, electrodeposition, spin coating, spray pyrolysis, and hydrothermal methods, which facilitate economical and large-scale preparation. Spin coating involves coating Cu/In precursor ink onto a rotating substrate. In this method, the Cu/In ratio can be easily controlled by adjusting the composition of the precursor solution; moreover, the thin-film thickness can be regulated by repetitive spin coating [53]. Spray pyrolysis, which involves spraying a precursor solution onto a high-temperature substrate, facilitates the preparation of large-area films. Here, the CuInS2 composition can be adjusted using the concentration of the spray solution [23]. Electrodeposition-sulfurization is predominantly used to synthesize CuInS2 photocathodes for PEC hydrogen evolution. Here, Cu and In are electrochemically deposited on conductive substrates in an electrolytic solution [42]. Co-deposition has also been performed to fabricate Cu-In alloy films [21]. Simultaneous electrodeposition of three elements can be performed in a Cu-In-S ternary electrolytic bath. Unlike other chalcopyrite materials, the CuInS2 prepared by electrodeposition exhibits similar or even higher performance than that of the CuInS2 photoelectrodes prepared by co-evaporation or sputtering. Therefore, electrodeposition is a promising technique for synthesizing high-performance CuInS2 thin films ( Figure 5).  [33] and (b) electrodeposition/sulfurization [42]. (c) Linear sweep voltammetry (LSV) profiles and HC-ABPE data of CuInS 2 -based photocathodes fabricated by electrodeposition/sulfurization [42]. Reproduced from ref. [29] with permission of the Royal Society of Chemistry. Reproduced from ref. [39] with permission from Wiley-VCH. CuInS 2 can also be synthesized using solution-based approaches, such as chemical bath deposition, electrodeposition, spin coating, spray pyrolysis, and hydrothermal methods, which facilitate economical and large-scale preparation. Spin coating involves coating Cu/In precursor ink onto a rotating substrate. In this method, the Cu/In ratio can be easily controlled by adjusting the composition of the precursor solution; moreover, the thin-film thickness can be regulated by repetitive spin coating [53]. Spray pyrolysis, which involves spraying a precursor solution onto a high-temperature substrate, facilitates the preparation of large-area films. Here, the CuInS 2 composition can be adjusted using the concentration of the spray solution [23]. Electrodeposition-sulfurization is predominantly used to synthesize CuInS 2 photocathodes for PEC hydrogen evolution. Here, Cu and In are electrochemically deposited on conductive substrates in an electrolytic solution [42]. Co-deposition has also been performed to fabricate Cu-In alloy films [21]. Simultaneous electrodeposition of three elements can be performed in a Cu-In-S ternary electrolytic bath. Unlike other chalcopyrite materials, the CuInS 2 prepared by electrodeposition exhibits similar or even higher performance than that of the CuInS 2 photoelectrodes prepared by co-evaporation or sputtering. Therefore, electrodeposition is a promising technique for synthesizing high-performance CuInS 2 thin films ( Figure 5).  [28], (b) photoelectrode based on 3D micro bowl arrays using CuInS2 nanosheet array grown using solvothermal method as photocatalyst [26], and (c) CuInS2 nanosheet arrays fabricated from In2S3 self-sacrificial templates using solvothermal and sputtering [25]. Reproduced from ref. [28] and ref. [26] with permission from Elsevier. Reproduced from ref. [25] with permission from Wiley-VCH.
However, the aforementioned methods are limited in terms of permitting nanostructural control to achieve high PEC activities. The multidimensional structure of semiconductor thin films is often critical to determining the PEC activity. One-dimensional (1D) nanostructures allow efficient PEC reactions owing to the improved charge separation/transport, and their unique optical and electrical properties lead to enhanced light absorption [31]. Additionally, most nanostructures exhibit higher light-trapping efficiencies and photocurrent densities than those of thin films of various materials. For example, photoelectrodes with structures such as nanowire arrays (NWAs), nanorod arrays, and nanotubes have the advantage of charge separation owing to the improved charge-carrier collection, rapid minority-carrier diffusion, and robustness of the electrode-electrolyte interface [31].
Three-dimensional branched nanoarray structures can improve the PEC performance of CuInS2 photocathodes by enhancing their light-absorption characteristics and surface area. Moreover, two-dimensional structures can improve the PEC performance through enhancements in the electronic and optical properties and surface area. In particular, layered bonding is beneficial in terms of promoting photocarrier separation and improving the PEC performance by shortening the charge-transport duration and distance  [28], (b) photoelectrode based on 3D micro bowl arrays using CuInS 2 nanosheet array grown using solvothermal method as photocatalyst [26], and (c) CuInS 2 nanosheet arrays fabricated from In 2 S 3 self-sacrificial templates using solvothermal and sputtering [25]. Reproduced from ref. [28] and ref. [26] with permission from Elsevier. Reproduced from ref. [25] with permission from Wiley-VCH.
However, the aforementioned methods are limited in terms of permitting nanostructural control to achieve high PEC activities. The multidimensional structure of semiconductor thin films is often critical to determining the PEC activity. One-dimensional (1D) nanostructures allow efficient PEC reactions owing to the improved charge separation/transport, and their unique optical and electrical properties lead to enhanced light absorption [31]. Additionally, most nanostructures exhibit higher light-trapping efficiencies and photocurrent densities than those of thin films of various materials. For example, photoelectrodes with structures such as nanowire arrays (NWAs), nanorod arrays, and nanotubes have the advantage of charge separation owing to the improved charge-carrier collection, rapid minority-carrier diffusion, and robustness of the electrode-electrolyte interface [31].
Three-dimensional branched nanoarray structures can improve the PEC performance of CuInS 2 photocathodes by enhancing their light-absorption characteristics and surface area. Moreover, two-dimensional structures can improve the PEC performance through enhancements in the electronic and optical properties and surface area. In particular, layered bonding is beneficial in terms of promoting photocarrier separation and improving the PEC performance by shortening the charge-transport duration and distance [28]. However, the complex nature of multinary chalcogenide materials hinders the development of practical techniques for synthesizing vertically aligned structures. Most previously reported copper chalcopyrite nanostructures have been synthesized using hard nanoscale templates, such as anodic aluminum oxide (AAO). Yang et al. prepared CuInS 2 photocathodes with nanorod structures by impregnating AAO templates with a Cu/In precursor solution and used them for PEC hydrogen production [31]. The charge-carrier collection was found to be inefficient when the nanowire length was greater than the carrier-diffusion length (~1 µm). Additionally, the transport of photogenerated electrons to carrier collectors was suppressed as the surface area of CuInS 2 decreased with increasing nanowire diameter. Therefore, the 1D nanostructures had to be optimized in terms of diameter and length to maximize their advantages. Moreover, the control of the pores and depth of AAO templates have been reported to be key to preparing CuInS 2 nanorod arrays of an appropriate length. Hydrothermal/solvothermal approaches have also been implemented to adjust the CuInS 2 nanostructures [24,25,36]. Tu et al. deposited In 2 S 3 nanosheet arrays on a fluorine-doped tin oxide (FTO) substrate using a solvothermal method and used them as self-sacrificing templates. Substantial CuInS 2 deposition was performed on the 2D structure, yielding 2D CuInS 2 nanosheets exhibiting strong excitonic effects, which led to an enhanced optoelectronic response and reduced radiative lifetime. This nanosheet array has the advantages of improved light collection and photogenerated charge pair creation [25].

Heterojunction Interface for CuInS 2
The low photovoltage generated by the rapid recombination of electron-hole pairs is a major factor that impedes photoelectrode activity [20]. In particular, the rate of hydrogen generation from CuInS 2 is significantly reduced, owing to its low photovoltage. However, the hydrogen production efficiency can be improved by creating a p-n junction between p-type CuInS 2 and n-type semiconductors [54].
Two types of p-n junctions exist-homojunctions and heterojunctions-which comprise the same and two different semiconductor materials, respectively. As they permit adjustment of the energy level that is matched with the bandgap between CuInS 2 and other semiconductor materials, p-n-type heterojunctions have been extensively used to suppress the recombination of electron-hole pairs in CuInS 2 photocathodes [34]. Among them, CdS is predominantly used for chalcopyrites and different p-type photoelectrodes (p-SnS, CdTe, and CuZnSnS 4 ) and is known to significantly improve photovoltage [54].
The difference in energy levels between two materials is particularly critical for enabling charge transfer at a heterojunction interface. The CuInS 2 /CdS interface has a cliff-type heterojunction with a difference between conduction band levels of −0.47 eV. When excited electrons move from CuInS 2 to CdS through the conduction band, the energy difference (1.05 eV) between the conduction band of CdS and the valence band of CuInS 2 increases the recombination chance of electron-hole pairs at the interface. However, the CuInS 2 -In 2 S 3 interface has a notch-type energy structure when the heterojunction is formed. Moreover, the energy difference (1.79 eV) between the conduction band of In 2 S 3 and the valence band of CuInS 2 is relatively larger than that of CuInS 2 /CdS. Therefore, the electron-hole recombination is suppressed, and charges can be efficiently separated and transported to the reactant [39]. Owing to this difference in the energy structure, the CuInS 2 /In 2 S 3 photoelectrode (−18 mA·cm −2 ) generates a higher photocurrent than that of CuInS 2 /CdS ( Figure 6). Nanomaterials 2023, 13, x FOR PEER REVIEW 11 of  [27]. (d) Proposed energy band diagram of the CuInS2/C3N4 [33]. (e) Sc matic energy diagrams of CuInS2/PNDI3OT-Se1-polymer heterojunctions, indicating charge tra fer between semiconductors [48]. Reproduced from ref. [39] with permission from the Ameri Chemical Society. Reproduced from ref. [20,33,48] with permission from the Royal Society of Che istry. Reproduced from ref. [27] with permission from Elsevier.
MoSx, Ni-MoSx, SnS2, Sb2S3, and TaOx can also be used as n-type semiconductors heterojunction formation with CuInS2 photocathodes. MoSx electrodeposited on CuI was found to generate an additional photovoltage of ~300 mV owing to p-n junction f mation [35]. Moreover, Chae et al. found that the in situ transformation of crystalline M into amorphous MoSx enhanced the charge transfer at the MoSx/CuInS2 interface, wh improved the PEC HER activity [47]. Zhao et al. reported that Ni doping of MoSx increas the PEC activity owing to the improved hydrogen evolution kinetics of MoSx. The pho cathode with the Ni-MoSx/CuInS2 heterojunction structure exhibited a higher onset pot tial (0.5 VRHE) than that of bare CuInS2 photocathodes [34].
The introduction of n-SnS2 with a bandgap of 1.62 eV into CuInS2 photoelectrod can permit effective charge transfer owing to the energy-level difference between the t  [20]. (c) Schematic diagram of the CuInS 2 /Sb 2 S 3 /Pt photoelectrodes. This band alignment indicates the charge-transfer process [27]. (d) Proposed energy band diagram of the CuInS 2 /C 3 N 4 [33]. (e) Schematic energy diagrams of CuInS 2 /PNDI3OT-Se1-polymer heterojunctions, indicating charge transfer between semiconductors [48]. Reproduced from ref. [39] with permission from the American Chemical Society. Reproduced from ref. [20,33,48] with permission from the Royal Society of Chemistry. Reproduced from ref. [27] with permission from Elsevier. MoS x , Ni-MoS x , SnS 2 , Sb 2 S 3 , and TaO x can also be used as n-type semiconductors for heterojunction formation with CuInS 2 photocathodes. MoS x electrodeposited on CuInS 2 was found to generate an additional photovoltage of~300 mV owing to p-n junction formation [35]. Moreover, Chae et al. found that the in situ transformation of crystalline MoS x into amorphous MoS x enhanced the charge transfer at the MoS x /CuInS 2 interface, which improved the PEC HER activity [47]. Zhao et al. reported that Ni doping of MoS x increased the PEC activity owing to the improved hydrogen evolution kinetics of MoS x . The photocathode with the Ni-MoS x /CuInS 2 heterojunction structure exhibited a higher onset potential (0.5 V RHE ) than that of bare CuInS 2 photocathodes [34].
The introduction of n-SnS 2 with a bandgap of 1.62 eV into CuInS 2 photoelectrodes can permit effective charge transfer owing to the energy-level difference between the two materials [36]. Similarly, at the interface between CuInS 2 and TaO x , which is a multifunctional passive layer with lower conduction and valence bands than those of CuInS 2 , electrons and holes are efficiently transferred by a cliff-like energy-level structure similar to that of the CuInS 2 /SnS 2 heterojunction. In particular, TaO x has been observed to act as a multifunctional passivation layer that not only suppresses the recombination of electron-hole pairs but also increases the activity and stability for hydrogen production. Therefore, the incident-photon-to-electron conversion efficiency (IPCE) has been found to increase in longer-wavelength regions owing to the prolonged charge-carrier lifetime, resulting in a three-fold higher photocurrent than that of the CuInS 2 photoelectrodes [4].
Recently, atomically graded passivation layers were developed for CuInS 2 photocathodes. First, Mo:TaO x layers were prepared by implanting Mo atoms into TaO x layers. The resulting Mo:TaO x film was then sulfurized, yielding a (Ta,Mo) x (O,S) y layer. The graded distribution of Mo and S in the TaO x matrix led to the synergetic PEC activity of CuInS 2 , similar to that of CuInS 2 photocathodes decorated with Pt cocatalysts [32]. n-Type polymer semiconductors could also be used to form p-n heterojunctions with CuInS 2 . When the n-type polymer semiconductor PNDI3OT-Se was deposited on CuInS 2 , the onset potential shifted, and the photocurrent increased, similar to that in an inorganic p-n junction, such as the CdS/CuInS 2 heterojunction [48]. The band position of PNDI3OT was adjusted using the monomer structure. Moreover, the valence band position of the polymer and CuInS 2 were critical for enabling efficient charge transfer at the polymer/CuInS 2 interface. This result highlights the potential of polymer/inorganic hybrid photoelectrodes for realizing efficient hydrogen production.

Cocatalyst Decoration for Hydrogen Evolution
Even if photogenerated charges have a sufficient thermodynamic potential, they can easily recombine if the HER on the photoelectrode surface is kinetically unfavored. Poor kinetics at the semiconductor surface leads to a high overpotential; however, cocatalysts can reduce the activation energy, boosting the HER rate. Therefore, the HER rate of photocathodes can be improved by applying suitable cocatalysts onto the surface of semiconductor photoelectrodes. Noble metals are considered excellent electrocatalysts because of their high conductivity and low overpotential for the HER. Among noble and transition metals, Pt is known to be one of the best electrocatalysts for the HER. The activity trends can be explained using a 'volcano plot' showing the adsorption/desorption energy of hydrogen ions on the catalyst surface. Therefore, noble metals are extensively used as cocatalysts for photocathodes; however, their expensiveness necessitates the development of economical cocatalyst alternatives. For instance, relatively inexpensive transition metal cocatalysts or protective layers have been used to enhance the activity and stability of Cu(In,Ga)Se 2 (CIGS) photoelectrodes. Moreover, Ni-Mo cocatalysts have improved the stability and activity of CIGS photoelectrodes under neutral electrolyte conditions [55]. When a Ti-Mo layer was introduced between Pt cocatalysts and CIGS/CdS electrodes, the stability and activity were significantly increased in neutral electrolytes [56]. However, several transition metals are unstable under low-pH conditions, which limits the use of strongly acidic electrolytes. Transition metal-sulfide cocatalysts (MoS x , CoS x , CoSe x , and NiS x ) have exhibited high hydrogen-generation activity in chalcopyrite and Si photoelectrodes [57][58][59][60][61][62]. However, the development of cocatalysts for CuInS 2 photoelectrodes is lagging behind that of other chalcopyrite-based photoelectrodes, with Pt cocatalysts being used in most cases. Therefore, various HER cocatalysts must be applied to CuInS 2 -based photoelectrodes, and their activity and mechanism should be investigated (Figure 7).
Charge transfer at the photoelectrode-cocatalyst interface should also be the focus, along with the electrocatalytic activity of cocatalysts. Although Pt is experimentally and theoretically regarded as one of the best HER electrocatalysts, it may be unsuitable for application on photoelectrode surfaces. Patra et al. used Au nanoparticles as a cocatalyst with CuInS 2 particles, which significantly increased the hydrogen evolution activity owing to the smooth charge transport between CuInS 2 and Au through surface-Plasmon-resonance effects [37]. Chae et al. incorporated photoelectrodes with Ru, which has a lower production activity than that of Pt and achieved a high solar-to-hydrogen conversion efficiency [35].
The high work function of Ru induced effective charge transfer from CuInS 2 /MoS x along with ohmic junction formation, whereas the CuInS 2 /MoS x /Pt interface did not form an ohmic junction. Therefore, charge transfer at the cocatalyst-CuInS 2 interface must be considered for designing efficient cocatalysts.  [35]. Reproduced from ref. [31] with permission from Elsev Charge transfer at the photoelectrode-cocatalyst interface should also be the fo along with the electrocatalytic activity of cocatalysts. Although Pt is experimentally theoretically regarded as one of the best HER electrocatalysts, it may be unsuitable application on photoelectrode surfaces. Patra et al. used Au nanoparticles as a cocata with CuInS2 particles, which significantly increased the hydrogen evolution activity ing to the smooth charge transport between CuInS2 and Au through surface-Plasmon onance effects [37]. Chae et al. incorporated photoelectrodes with Ru, which has a lo production activity than that of Pt and achieved a high solar-to-hydrogen conversio ficiency [35]. The high work function of Ru induced effective charge transfer f CuInS2/MoSx along with ohmic junction formation, whereas the CuInS2/MoSx/Pt inter did not form an ohmic junction. Therefore, charge transfer at the cocatalyst-CuInS2 in face must be considered for designing efficient cocatalysts.

Hole Transport Layers
Generally, holes in the p-type chalcopyrite semiconductor CuInS2-which are a jority carrier-have higher mobility and longer lifetimes than those of electrons Therefore, improving the electron transport/transfer in the semiconductor layer or a semiconductor-electrolyte interface has received more attention than hole transporta However, the enhancement of hole transfer has been found to be crucial for impro the activity of CuInS2. In most cases, CuInS2 crystals have been grown on Mo foil or coated substrates because of the ohmic contact enabled between Mo and CuInS2 (or o chalcopyrite materials) [64,65]. However, transparent substrates can be effective dep ing on the direction of illumination for determining the charge-transport property of toelectrodes [66]. CuInS2 films have been grown on transparent conductive substr such as glass coated with indium tin oxide (ITO) or fluorine-doped tin oxide (FTO), w  [35]. Reproduced from ref. [31] with permission from Elsevier.

Hole Transport Layers
Generally, holes in the p-type chalcopyrite semiconductor CuInS 2 -which are a majority carrier-have higher mobility and longer lifetimes than those of electrons [63]. Therefore, improving the electron transport/transfer in the semiconductor layer or at the semiconductor-electrolyte interface has received more attention than hole transportation. However, the enhancement of hole transfer has been found to be crucial for improving the activity of CuInS 2 . In most cases, CuInS 2 crystals have been grown on Mo foil or Mocoated substrates because of the ohmic contact enabled between Mo and CuInS 2 (or other chalcopyrite materials) [64,65]. However, transparent substrates can be effective depending on the direction of illumination for determining the charge-transport property of photoelectrodes [66]. CuInS 2 films have been grown on transparent conductive substrates, such as glass coated with indium tin oxide (ITO) or fluorine-doped tin oxide (FTO), which have been extensively used for PEC cells. Liu et al. reported that FeOOH interlayers between CuInS 2 and ITO improved the hole transfer from CuInS 2 to the ITO layer, which led to the FTO/FeOOH/CuInS 2 /Pt photocathodes exhibiting a photoactivity 3.1 times higher than that of ITO/CuInS 2 photocathodes [50]. FeOOH effectively captured excited holes owing to its hole-storing tendency, and the electron-hole pair recombination of CuInS 2 was suppressed. Kumar et al. synthesized NaNbO 3 /CuInS 2 /In 2 S 3 and NaNbO 3 /In 2 S 3 /CuInS 2 core/shell/shell nanocubes for PEC applications [49]. In the NaNbO 3 /CuInS 2 /In 2 S 3 system, CuInS 2 and In 2 S 3 exhibited z-scheme behavior; however, NaNbO 3 /In 2 S 3 / CuInS 2 had a type II band structure. The z-scheme nanocube immobilized on ITO exhibited a considerably higher photocurrent than that of the electrodes immobilized with the type-II-structured nanocube. This type of energy-level design may facilitate hole transfer at a non-ohmic interface between CuInS 2 and the transparent, conductive metal-oxide layer (ITO or FTO).

Solid Solutions of CuInS 2
Solid solutions of chalcopyrite CuInS 2 crystals and other atoms can be readily prepared. For example, various photocathode materials, such as Cu(In,Ga)Se 2 , CuGaSe 2 , (Ag,Cu)GaSe 2 , and CuIn(S,Se) 2 , have been studied for the HER [67][68][69][70]. However, the activity and physicochemical properties of this type of material are beyond the scope of this review. Nevertheless, the solid solution of chalcopyrite CuInS 2 and zincblende ZnS is worth mentioning. Zhao et al. prepared solid solution crystals of (CuInS 2 ) x (ZnS) 1-x using an electrodeposition-sulfurization method [41]. The enlarged bandgap of the solid solution (CuInS 2 ) x (ZnS) 1-x led to a significantly high onset potential (up to 0.8 V RHE ). The half-cell solar-to-hydrogen conversion efficiency of the (CuInS 2 ) x (ZnS) 1-x /CdS/Pt photocathodes was 5.6%, which is a remarkably high value among CuInS 2 -based photocathodes. This result emphasizes the benefits of solid solutions containing CuInS 2 or other materials over those of the simple bilayer-structured equivalents because of the graded band structure obtained by the atomic gradation of the semiconductor.

Summary and Outlook
CuInS 2 -based photocathodes for PEC hydrogen evolution are reviewed herein, with a focus on their synthesis methods, heterolayer construction, and cocatalyst design. CuInS 2 has been synthesized using physical (co-evaporation and sputtering) and chemical techniques (electrodeposition, spin coating, spray pyrolysis, and hydrothermal methods). Generally, the physical methods yield a dense, high-quality chalcopyrite thin film with large grains and high crystallinity, with the highly crystalline semiconductors exhibiting high PEC activity. Therefore, the physical methods enable higher PEC performance than that of the solution-based methods. However, unlike other quaternary or pentanary chalcopyrite materials, CuInS 2 synthesized using electrodeposition methods has similar or superior PEC activity compared to that of CuInS 2 films prepared using physical methods. Additionally, certain solution-based methods offer advantages in controlling the morphological features of CuInS 2 , such as dimensions and porosity. However, the complex nanostructure can also have more defect sites than those in a highly crystallized thin film. The activity of CuInS 2 -based photoelectrodes synthesized to date using solvo(hydro)thermal or other solution-based methods is considerably inferior to that of specimens prepared by electrodeposition. Therefore, nanostructural control remains insufficient for overcoming the rapid charge recombination at the defect or surface states of the CuInS 2 crystal. A new solution-based synthesis method for CuInS 2 with high PEC activity remains to be developed.
The use of heterojunctions to suppress electron-hole recombination has been extensively studied. Many sulfide materials (CdS, In 2 S 3 , ZnS, NiS x , and MoS x ) have been investigated for constructing a heterojunction or p-n junction with CuInS 2 . Moreover, certain metal oxides, atomic-graded metal oxides, and n-type polymer semiconductors have also been introduced. The bandgap alignment between CuInS 2 and heteromaterials is critical to enabling efficient charge transfer at the interface. Suitable band energy matching can successfully suppress the electron-hole recombination, boosting the photocurrent and improving the onset potential.
Cocatalysts that can facilitate hydrogen evolution from a photoelectrode surface are also important. The CuInS 2 surface is kinetically unfavorable for the HER and thereby requires a cocatalyst to reduce the overpotential. Pt, which has been widely used as a cocatalyst for CuInS 2 -based photocathodes, should be replaced with economical and earth-abundant materials. Although certain noble metals and metal sulfides have been investigated as cocatalysts for CuInS 2 photoelectrodes, more diverse cocatalyst materials should be examined. Additionally, when a cocatalyst is applied to a PEC system, the electrochemical activity of the cocatalyst, as well as the interfacial charge transfer between CuInS 2 and the cocatalyst, must be considered. Similarly, interfacial charge transfer between CuInS 2 and substrates can be controlled using hole-transporting interlayers. Finally, the use of a partial solid solution at the interface with CuInS 2 can be an effective strategy for enabling efficient charge transfer at several interfaces in a photoelectrode structure. All these strategies should be combined to achieve the theoretical limit of a PEC cell with a CuInS 2 photoelectrode.