1 Introduction
Over the past two and half centuries, the exploitation of fossil fuels has promoted the unprecedented advances of human civilization yet created a number of tremendous issues, e.g., energy crisis, environmental pollution, and climate change [1–3]. It is particularly noted that the current level of carbon dioxide emission will probably result in a terrible global temperature elevation of about 3.2 °C by 2100. To stop the devastating trend, as claimed by the Intergovernmental Panel on Climate Change (IPCC), the human community should reach carbon neutrality by around 2050 [4]. Since fossil fuels contribute to most of the carbon dioxide emissions, it is highly imperative to explore green energies as the alternative to fossil fuels for supporting the increasing population and the growing economy. Hydrogen is well recognized as a future energy carrier because of the following unrivaled advantages. First, the gravimetric energy of H2 is as high as 14230 kJ/kg, which is 3, 4.5, and nearly 100 times of gasoline, coal, and lithium-ion batteries, respectively [5,6]. Meanwhile, hydrogen is a substantially carbon-neutral energy vector as water is the exclusive product of its combustion. In practice, liquid hydrogen has been successfully applied as green fuel for generating power by fuel cell and/or airspace engine [7]. In addition to being a green fuel, hydrogen also demonstrates great potential in medical health, farming, and metal smelting, etc. However, hydrogen is now mainly derived from steam methane reforming and coking or gasification of coal [6]. These routes suffer from high temperatures or pressures, massive thermal energy input, and non-renewable fossil fuel consumption, which are not sustainable. Moreover, the current hydrogen supply is far below the vast demand by a carbon-neutral economy. It is predicted that the global market of hydrogen will increase from 80 million tons in 2022 to 536 million tons in 2050 [8]. Hence, exploring a green manner for large-scale hydrogen production by utilizing renewable resources is critically important. Sunlight and water are the two most abundant, clean, and renewable resources on the earth. Specifically, the energy delivered by sun irradiation per second is equivalent to 5 million tons of standard coal in China [9]. The energy offered by two hours of sunlight irradiation is sufficient to meet the annual global energy consumption in 2022. Water is widely distributed and highly accessible on the earth. Sunlight-powered water splitting thus offers a green approach for distributed H2 production and intermittent solar energy storage, known as the Holy Grail of chemistry, materials, and energy [10,11].
As briefly shown in Fig.1, green H2 produced from sunlight and water is an ideal energy carrier to drive a carbon-neutral economy. First, it can be directly applied for fuel vehicles or power plants e.g., internal combustion engine to generate power, and for combustion to generate heat for the resident living. As the only product of H2 combustion, H2O can be recycled to produce hydrogen again. In addition, as an ideal alternative to fossil fuels, hydrogen can be used for carbon dioxide hydrogenation to produce transportation fuels and value-added chemicals. Therefore, solar to hydrogen from water splitting is of great importance for achieving carbon neutrality. At present, there are three major types of sunlight-powered water splitting systems: photovoltaic-assisted electrocatalysis (PV-EC), photocatalysis (PC), and photo-electrocatalysis (PEC) [12]. A PV-EC system mainly composes of a photovoltaic cell with an electrolyzer. To date, for a state-of-the-art PV-EC system, the solar-to-hydrogen (STH) efficiency has reached 30% [13]. Nevertheless, the complex configuration and high operation/maintenance cost make PV-EC H2 not commercially viable. The cost of PV-EC H2 is estimated to be at least $8 per kg [14]. The current research is thereby concentrated on developing efficient, lost-cost, and stable PC and/or PEC systems, which is associated with their simplicity and low investment, operation, and maintenance cost. Compared to PC systems with light as the only driving force, PEC systems require bias to suppress the recombination of photogenerated carriers and to overcome the high energy barrier of the reaction, which is an electrochemical reaction under light illumination. In stark contrast, with the only inputs of sunlight and water, PC water splitting presents the simplest and lowest-cost configuration for hydrogen production without wires, electrodes, electrolyzes, and cells, etc. As a result, it is predicted that the cost of PC-H2 can be reduced to ($1.60–3.20)/kg, when a similar STH value to that of PV-EC systems is realized [15]. It shows a great potential to meet the target price of ($2.00–4.00)/kg set by the United States Department of Energy, which is commercially competitive with methane steam reforming [16].
In principle, both PC and PEC for water splitting involve three primary steps: photon absorption by light-harvesting semiconductor, electron-hole pairs generation and separation, and surface chemical reactions. For photocatalysis, the semiconductor can be excited by photons with energies larger than its energy bandgap (Fig.2(a)) [11,17]. The electrons are then pumped to the empty/unoccupied conduction band (CB), leaving behind the holes in the valence band (VB). Subsequently, the photogenerated electrons and holes enable hydrogen evolution reaction (HER) and oxygen evolution reaction (OER) simultaneously at the suitable reaction sites on the photocatalyst surface. In most cases, although the band structure of the semiconductors can straddle the redox potentials of water splitting, the realization of overall water splitting by photocatalysis without any external energy input except for light posed a great challenge because of the sluggish kinetics of surface chemical reactions [18–21]. Moreover, photogenerated electrons and holes can be recombined easily without the assistance of applied bias. Therefore, even for state-of-the-art photocatalysts, the STH is far below the commercial requirement (10%) [22]. In contrast, a typical PEC system works in an electrolytic cell, consisting of a photoanode and a photocathode in tandem configuration (Fig.2(b)). For a photoanode, photoexcited holes usually accumulate on the surface of the semiconductor and are subsequently consumed in oxidation reactions. In this context, the top of the valence band of the semiconductor must be more positive than the oxygen evolution potential to allow a photoanode to generate oxygen. In contrast, a p-type semiconductor works as a photocathode for hydrogen evolution when the conduction band edge is more negative than the hydrogen evolution potential. In general, only a minor proportion of photoexcited charge carriers can participate in PEC reactions on photoelectrodes in both cases because of the fierce electron-hole recombination [23]. The potential of electrons on the counter electrode is identical to the Fermi-level of the photoelectrode under photoexcitation. An external voltage can be applied between a photoelectrode and a counter electrode to compensate for the potential deficiency in order to drive redox reactions on a counter electrode even if the Fermi-level of the photoelectrode is positioned at an undesirable potential. During photoelectrocatalysis, applied bias can provide an external force for the system to overcome the energy barriers of HER and OER and to promote the separation of carriers [24]. They work together to improve STH efficiency. Moreover, hydrogen and oxygen from PEC overall water splitting can be spatially separated without product separation issues.
Semiconductors and co-catalysts that determine photons absorption, charge behavior, and surface chemical reactions are vital to water splitting. The first priority of the research in this field is the exploration of appropriate semiconductor materials and efficient co-catalysts and integrating them in a rational configuration. Since Fujishima and Honda reported TiO2-catalyzed water splitting in 1972 [25], TiO2 has attracted immense research interest. As the most studied semiconductor, TiO2 exhibits admirable stability, low cost, and nontoxicity. However, like most of the reported metal oxide photocatalysts e.g., WO3, ZrO2, TaO3, and Ga2O3 with d0 or d10 electronic configurations, TiO2 is only responsive to ultraviolet light due to its wide bandgap (approximately 3.2 eV), which occupies 4% of the sunlight spectrum [26–28]. Over the past decades, a range of semiconductors such as metal oxides, metal sulfides, metal salts, and emerging materials like metal-organic frameworks (MOFs), covalent-organic frameworks (COFs), and their coupled hetero-structures have been developed for sunlight-powered water splitting [29–33]. In parallel, various approaches including surface engineering, nanostructure engineering, defect engineering, co-catalysts decoration, band structure engineering, and built-in electrical field engineering, have also been attempted to improve the performance of the systems [34–39]. However, despite considerable efforts, there has been no substantial success in sunlight-powered water splitting toward hydrogen. Many critical issues must be solved including low STH efficiency, poor stability, and high cost of the system prior to practical applications [40]. The exploration of a disruptive semiconductor platform to address the critical issues above is the crux of this grand topic. In recent years, Ga(X)N nanowires vertically aligned on silicon substrate, namely Ga(X)N/Si nanoarchitecture, has emerged as a semiconductor platform due to its unique structural, optoelectronic, and catalytic properties.
This paper will shed light on this promising platform for solar-powered water splitting toward hydrogen by both photocatalysis and photoelectrocatalysis. It primarily focuses on introducing the unique structure and optoelectronic properties of Ga(X)N/Si nanoarchitecture, and on explaining its superiority as an ideal platform for constructing next-generation artificial photosynthesis integrated device and system (APIDS). Moreover, it demonstrates the recent advances in photocatalytic/photoelectrocatalytic water splitting based on Ga(X)N/Si nanoarchitecture. Finally, it envisages the challenges and prospects of Ga(X)N/Si-based APIDS for large-scale solar water splitting hydrogen.
2 Principle of water splitting
The reaction of sunlight-powered overall water splitting is shown as the following equation [41]:
where h represents the Planck constant; υ presents the frequency of the incident photons (one mole of liquid water can be stoichiometrically decomposed into 1/2 mole of gaseous O2 and 1 mole of H2). Sunlight-powered water splitting toward hydrogen is both thermodynamically and kinetically challenging. First of all, overall water splitting toward hydrogen and oxygen is an uphill chemical reaction with an associated increase in the Gibbs free energy of 237 kJ/mol (as shown in Eq. (1)), corresponding to the redox potential of 1.23 eV [42]. That means external energy is essentially required to drive this reaction. Fig.2(a) shows the schematic diagram of the basic process of overall water splitting on a photocatalyst. Illuminated by photons with energy greater than the bandgap (Eg), the electrons (e–) of the semiconductor are excited into the CB while the holes (h+) are left in the VB. The photogenerated electrons and holes then migrate toward the surface of the photocatalyst, and redox reactions will subsequently occur.
Overall, water splitting composes of 2-electron hydrogen evolution and 4-electron water oxidation, corresponding to HER and OER as Eqs. (2) and (3) describe, respectively [16].
To drive these two redox reactions without applied bias, the conduction band minimum (CBM) of the semiconductor must be at a more negative potential than the H+/H2 level (0 V versus normal hydrogen electrode (NHE) pH = 0), while the valence band maximum (VBM) should be more positive than the O2/H2O level (1.23 V versus NHE) [23]. According to this theoretical value, only when the photon energy is greater than 1.23 eV, photocatalytic water splitting can be realized without applied bias. As a matter of fact, the photon energy required for solar water splitting is usually larger than 1.23 eV because of the overpotentials caused by the resistance of charge carriers migration, reaction energy barriers, and so on [43]. It is widely recognized that semiconductors play a decisive role in PC and PEC water-splitting systems. As standard, overall water-splitting semiconductors, transition metal oxides, whose CBM typically have d0 and d10 electronic structures, have a strong d and sp orbital hybridization effect and contain O2p orbitals in VBM [44]. Therefore, if CBM is more negative than the H+/H2 level, the band gap energy of these oxides will exceed 3 V (versus NHE, at pH = 0). Consequently, for OER, both d0-type and d10-type oxides have a surplus potential in their VBM compared to their CBM [45]. In addition, the OER performance is often restricted by the photooxidation induced by holes, poor carriers mobility, and so on. The behavior of electrons and holes is also closely related to the type, structure, and electronic properties of the co-catalysts that are utilized for decorating semiconductors. A rational cocatalyst is highly favorable for enhancing charge carriers separation/migration.
In addition, the primary function of the co-catalysts is to provide redox reaction sites which can lower the reaction energy barriers. Furthermore, co-catalysts can also improve the durability of the catalysts against photodegradation by efficiently consuming the photoinduced charge carriers, particularly by protecting the semiconductor surface from being oxidized by holes. Loading appropriate co-catalysts onto the semiconductor surface facilitates redox reactions on its surface, which will be discussed in Section 5.1.1 in detail.
3 Ga(X)N/Si nanoarchitecture
At present, silicon is the exclusive commercial material for converting sunlight into electricity by photovoltaic effect, due to its narrow bandgap, earth abundance, and matured fabrication technology. Benefiting from these merits, silicon has also been widely employed to build photocathodes for water splitting, and a batch of important progress have been made [46]. However, as discussed above, owing to the narrow bandgap of 1.1 eV, silicon cannot realize photocatalytic overall water splitting by itself. If being used as a photocathode, the reaction kinetics of water splitting over bare silicon is sluggish even with the aid of applied potential, due to the lack of sufficient active sites.
Ga(X)N has emerged as a rising star material for constructing next-generation APIDS because of its admirable optical, electronic, and catalytic properties [47–49]. When the dopant X is a homologous-group element e.g., In, Ga(In)N (usually written as InGaN) is usually generated in the form of mixed crystals (or alloys). In contrast, when the dopant X introduced is different group elements such as Si and Mg, n-type and p-type doped GaN can be formed without forming crystals or alloys because of the insignificant concentration of X, respectively. Typically, when decorating with suitable dopants at desired concentration, the band structure of Ga(X)N can be regulated for efficient solar water splitting by epitaxial growth without affecting lattice parameters and high-quality crystals [50–52]. As a matter of fact, epitaxial growth of Ga(X)N nanowires (NWs) on a silicon wafer assembles two outstanding semiconductors working in tandem. Such superior performance of these devices mainly benefits from the unique structural characteristics and optoelectronic properties of Ga(X)N and silicon as well as their rational configuration. In this section, state-of-the-art fabrication methods and equipments, as well as the outstanding properties of Ga(X)N/Si nanoarchitecture will be elaborated.
3.1 Fabrication of Ga(X)N/Si nanoarchitecture
Ga(X)N, with a hexagonal, wurtzite crystal structure, usually has stable physical properties and is difficult to be corroded [53]. Over the past decades, significant advance in science and technology has offered powerful tools for deliberately assembling the Ga(X)N/Si nanoarchitecture. Metal-organic chemical vapor deposition (MOCVD), magnetron sputtering (MS), and molecular beam epitaxy (MBE) are currently the three most utilized methods, which are attributed to chemical vapor deposition (MOCVD) and physical vapor deposition (MS and MBE), respectively. However, there exist several differences although both methods can grow a high-quality Ga(X)N/Si nanoarchitecture.
For the MOCVD process, precursors experienced pyrolytic reactions in a furnace, where the metallic atoms were deposited on the substrate at about 800–1000 °C while the organic compounds were removed [54]. In general, large-scale growth of Ga(X)N nanostructure can be achieved with the aid of a large injection of precursor gas and a large substrate, but this process still has some limitations. The growth rate of the MOCVD method is faster than that of MBE, which is one of the commercial fabrication methods of Ga(X)N [55,56]. However, this method has a poor interface control ability between two heterogeneous bands during the period of growth. Besides, highly required temperature and toxic gases from the preparation of precursors further increase the costs.
Magnetron sputtering is also a technology suitable for the growth of Ga(X)N. The sputtered particles are deposited on the substrate by bombarding the target materials with high-energy particles [57]. The biggest advantage of this technology is its low cost and low temperature compared with the MBE and MOCVD methods. High-quality GaN films prepared by magnetron sputtering showed promising potential for some sensor applications and photovoltaic devices. However, due to the uncontrollable growth process, it is difficult for the magnetron sputtering technology to control the composition and growth of nanowires at the atomic scale, which is extremely limited for regulation of microstructure and improvement of photoelectric properties [58,59]. In general, as bottom-up growth modes, Ga(X)N nanowires grown by the above methods exhibits few crystal defects and an excellent carrier transport ability compared with the powder chemically synthesizing method. However, the limitations and shortcomings cannot be ignored. For example, the MOCVD method generally uses foreign metal catalysts to promote the growth of nanowires. Due to the crystal structure difference between metal and Ga(X)N, it will inevitably cause the formation of defects and transportation of carriers. Besides, III-nitrides nanowires grown by the magnetron sputtering method can easily generate SiNx with a high resistivity on the interface between the Si substrate and Ga(X)N [60,61], which affects the catalytic performance.
MBE is an epitaxial growth technique for compound semiconductors involving the reaction of molecular beams on a heated substrate under ultra-high vacuum conditions [62]. During the Ga(X)N NWs MBE growth process, adatoms move along with the substrate surface toward the non-polar plane of the nanostructures and then assemble at the top polar plane due to the lower chemical potential that promotes the axial growth of defect-free nanowires [63]. Therefore, although the growth rate of nanowires is slow, the atomically tunable characteristic enables the optoelectronic properties of Ga(X)N NWs to be flexibly adjusted in this process. More importantly, binary/ternary Ga(X)N nanostructures such as GaN/InGaN dots-in-a-wire, GaN/InGaN core-shell NWs, etc. were designed and fabricated successfully by utilizing the MBE method [64–66]. Overall, MBE-grown Ga(X)N NWs exhibit a more precise control than those grown by MOCVD during the growth process [67,68], which are more suitable for assembling high quality multi-band Ga(X)N/Si-based APIDS with superb carriers separation and tunable surface properties for mediating solar water splitting toward hydrogen [69].
3.2 Properties of InGaN/Si nanoarchitecture
In general, the semiconductors employed for water splitting were mainly fabricated by solvothermal, precipitation, calcination, etc. Despite the high accessibility and low cost, it is very challenging to tailor the structure and properties of semiconductor materials at the atomic scale by these approaches. For example, the semiconductors prepared by traditional synthetic methods possess many defects and mismatches due to the uncontrollable growth process. The resultant defects generally behave as charge carrier traps, thus unfavoring their separation and migration. In sharp contrast, as mentioned above, the recent development of the MBE technology offers a powerful tool for developing emerging semiconductor materials to promote the advance of artificial photosynthesis. Benefiting from this technology, the epitaxially grown InGaN/Si nanoarchitecture shows obvious advantages to breaking the bottleneck of sunlight-powered water splitting. Because of the efficient surface strain relaxation and precise atomic-scale control during the MBE process, the as-grown InGaN NWs owned nearly a dislocation-free structure which is beneficial for the mobility of photogeneratedcharges with significantly suppressed electron-hole recombination, especially if coupled with appropriate cocatalysts [70]. Additionally, the energy band structure of InGaN can be controllably tuned from 3.4 to 0.65 eV by alloying with dopants. Moreover, InGaN exhibits a different electron affinity according to the difference in electronegativity between metal atoms and nitrogen atoms [71,72]. Such a spontaneous polarization field effect often has a serious influence on the band edge. It has been shown that electron affinity (χ) of GaN and InN is 3.2–3.4 eV and 0.6–0.7 eV, respectively [70,71]. It is noted that the edges of both CB and VB of InGaN can be flexibly regulated for water splitting without compromising light absorption by varying the types and concentrations of X dopants (Fig.3(a)) [72]. More importantly, the N-rich non-polar surface (m-plane) of InGaN grown by MBE can effectively protect the architecture from photo-corrosion and therefore greatly improve the durability for H2 production from water splitting [73].
It is worth mentioning that the non-polar surface of GaN () also plays a significant role in the adsorption and activation of water molecules because of its tunable features [74]. The behavior of water molecules and/or reaction intermediates could be facilely tailored by adjusting the surface properties of GaN, especially if coupling with suitable co-catalysts. By ab initio molecular dynamics theoretical calculations, it was discovered that hydroxide and hydrogen ions tend to bind on the surface of Ga and N atoms (Fig.3(b)), respectively [75,76]. In addition, the electronic properties of GaN m-plane adsorption with hydroxyl and water molecules are quite different from those of other semiconductors such as TiO2. Such a chemical feature thus holds a grand promise for surface protonation and deprotonation of water molecules adsorbed on the surface. It is beneficial for decreasing the energy barrier of proton diffusion from the oxygen evolution reaction site to the hydrogen evolution site on the GaN surface, thus promoting the reaction.
4 Approaches for improving the performance of Ga(X)N/Si-based APIDS
Overall, the Ga(X)N/Si nanoarchitecture presents an appropriate semiconductor platform for solar water splitting by addressing the tremendous challenges of low light harvesting, fierce charge carriers recombination, as well as sluggish reaction kinetics [77]. In particular, the surface of Ga(X)N can be deliberately engineered to be polar or non-polar by terminating with different atoms. According to previous studies, the N-rich non-polar surface of GaN could act as an effective protective layer against photo-corrosion, thus facilitating the achievement of the long-term stability of APIDS [78]. On the other hand, silicon is regarded as a suitable light collector to absorb a large portion of sunlight with wavelengths up to 1100 nm owing to its low band gap (1.1 eV). Meanwhile, silicon is an abundant raw material in the crust of the earth with a low cost. From the viewpoint of practical applications, silicon and Ga(X)N are the two most produced semiconductors in industry. In addition, silicon has been widely used as a perfect substrate for epitaxial growth of Ga(X) NWs. Coupling Ga(X)N NWs with silicon substrate can offer a competitive platform for solar-driven water splitting toward hydrogen. Despite the grand prospective, considerable efforts are required to improve overall water splitting by utilization of the emerging Ga(X)N/Si nanoarchitecture. The following is the three popular strategies to enhance the performance of Ga(X)N/Si nanoarchitecture for sunlight-powered water splitting toward hydrogen.
4.1 Decoration with rational co-catalysts
Co-catalysts play a dominant role in integrated sunlight-powered water splitting systems except for the semiconductor platform. The development of rational cocatalysts is very critical for improving the overall performance of water splitting. For the bare Ga(X)N/Si nanoarchitecture, it is not efficient in catalyzing water splitting arising from the lack of appropriate active sites [79]. The decoration of the Ga(X)N/Si nanoarchitecture with rational co-catalysts can effectively lower the energy barrier of water splitting by creating redox reaction sites. Meanwhile, a rationally designed cocatalyst can also function as an effective trap to extract charge carriers with the assistance of a built-in electrical field at the cocatalyst/semiconductor interface. It thus inhibits the rapid recombination of photogenerated electrons and holes, and subsequently promotes surface chemical reactions [80]. The interfacial charge transfer influenced by cocatalysts is closely related to the types of metals, size, morphology as well as the cocatalysts/semiconductor contact configuration. In addition, the loaded co-catalysts can improve the stability of the integrated systems by protecting the semiconductor surface from photooxidation. It can be realized by the enhanced consumption of the photogenerated holes for water oxidation. Among various co-catalysts, noble metals, especially platinum, and rhodium, are known to be state-of-the-art co-catalysts for protons reduction toward H2 [81,82]. However, they are fundamentally limited by prohibitive prices and inadequate reserves. Therefore, it is highly imperative to explore cost-effective alternatives. To date, earth-abundant co-catalysts such as NiPx and MoSx have been tentatively explored as alternatives to noble metals for hydrogen production and a number of appreciable achievements have been made [83–85].
OER, the rate-limiting step of water splitting, also requires the participation of cocatalysts to promote the reaction since it is a sluggish process that involves a 4 electrons transfer process. Oxides such as CoOx, IrO2, RuO2, and Fe2O3 have been widely used as OER cocatalysts for promoting the reaction [86–89]. However, to be frank, in spite of the great progress, it remains a grand challenge to develop an efficient and durable cocatalysts for meeting the practical demand with a high STH efficiency and a low cost.
4.2 Incorporation of X dopants
The incorporation of X dopants is regarded as a key strategy to enhance the performance of the Ga(X)N/Si nanoarchitecture. Typically, the introduction of suitable X dopants not only changes the band structure of Ga(X)N but also influences the flow rate and direction of charge carriers, thus significantly affecting the efficiency of the Ga(X)N/Si nanoarchitecture for water splitting [90]. In principle, the large band gap of GaN (3.4 eV) makes the light absorption shorter than 365 nm. It is thus detrimental to achieve a high STH efficiency. The incorporation of appropriate dopants, e.g., In can directly reduce the band gap of GaN, which will vary from 2.48 to 1.65 eV while the indium content is increased from approximately 24% to approximately 51%. It is worth noting that both the valence band maximum and the conduction band minimum of Ga(In)N can be tailored to straddle the redox potentials of H2O, thermodynamically enabling overall water splitting.
Proper doping can also improve the activity of water splitting by reshaping the band structure because the band bending of Ga(X)N is directly related to the type of dopants. For example, an upward band bending can be achieved by using n-type dopants like Ge, Si while a p-type GaN doping with Mg exhibits a downward bending state [91]. In previous investigations, Kibria et al. found that the near-surface Fermi-level and band bending of GaN NWs can be controlled by a p-type dopant i.e., Mg [92]. When p-doped GaN NWs were in equilibrium with water, the downward bending was regulated by Mg doping, which resulted in orders of magnitude change of the STH efficiency. For the doping of In (InGaN), the band position of Ga(X)N can be flexibly adjusted to change the optical absorption characteristics of the semiconductor. Therefore, appropriate doping of inequivalent and III-group elements is viable for reshaping and changing the band structure, thus facilitating the overcoming of the energy barrier of water splitting of GaN NWs/Si-based APIDS.
4.3 Construction of tunnel junction
The tunnel junction is beneficial for the transportation of charge carriers due to the enhanced quantum tunneling possibility at an ultrathin barrier. As reported by Sharif and coworkers, the GaN/InGaN/GaN polarization engineered nanowire tunnel junction exhibited an enhanced light output efficiency of visible nanowire LEDs [93]. The construction of Ga(X)N tunnel junction thus facilitates the improvement of the light-to-fuel efficiency [51,94]. In this context, Fan et al. found that the connection of p-type InGaN NWs with n-type GaN NWs could form a low-resistance contact and could be further used as an electron-blocking layer to enhance the electron extraction behavior [94]. For this structure, there exists an n++-GaN/In0.4Ga0.6N/p++-GaN polarization-enhanced tunnel junction between n-GaN and p-InGaN. Both the top InGaN and the bottom GaN/Si light absorber can simultaneously drive proton reduction based on the lateral carrier extraction properties of the nanowires. The PEC performance of this tunnel-junction Ga(X)N/Si-based APIDS obviously outstrips the conventional tandem photoelectrodes, confirming the viability of building a Ga(X)N tunnel junction for highly efficient solar-driven water splitting toward hydrogen. This, however, still remains at the infant stage and considerable efforts are required to optimize the tunnel junction structure and reveal the underlying mechanism
5 Advances in Ga(X)N/Si-based APIDS for solar hydrogen production
Nanostructured Ga(X)N have unique advantages, such as the precise control of energy bandgap to meet the redox potentials of water splitting, tunable light harvesting ability, excellent charge carriers separation and extraction effect, as well as photo-corrosion resistance. Together with the well-known merits of silicon substrate, the coupled Ga(X)N/Si nanoarchitecture is one of the most suitable semiconductor platforms for building next generation APIDS. To date, Ga(X)N/Si-based APIDS have achieved remarkable achievements in hydrogen production both in PC and PEC water splitting.
5.1 Ga(X)N/Si-based APIDS for PC water splitting
As mentioned above, the energy band edges of the semiconductor must straddle the redox potentials of H+/H2 and O2/H2O for spontaneously splitting water into H2 and O2, respectively. Additionally, semiconductor materials should meet high stability requirements with significantly inhibited photo-corrosion, low cost, and a broad response range of sunlight. As one of the most promising materials to construct new-generation APIDS, Ga(X)N NWs have been tentatively applied for PC water splitting toward hydrogen, and a series of remarkable achievements have been made. According to previous studies, for photocatalysis, the research interest has been mainly focused on Ga(X)N NWs, and silicon was only deemed as a substrate. The three main strategies for improving the photocatalytic performance of Ga(X)N NWs are summarized below.
5.1.1 Co-catalysts loading
As a matter of fact, the reaction kinetics of water splitting over bare Ga(X)N is highly sluggish because of the deficiency of active sites although the process is thermodynamically feasible. It is noted that based on a previous study by the first-principle density functional theory (DFT) calculation, the Ga-terminated surface of GaN is effective for H2O molecule adsorption and activation, thus presenting a promising scaffold for coupling with rational cocatalysts to drive water splitting [75]. For the first time, Wang et al. boldly used epitaxial 1-D structured GaN nanowires as photocatalysts to explore its viability for the overall water-splitting performance [95]. However, due to the existence of surface defects and n-type semiconductor doping, the emergence of Fermi-level pinning and the upward energy band gap reduce the rate of photogenerated electrons transfer to react with protons (Fig.4(a)). To improve the efficiency of H2 evolution, as mentioned above, it is very common to deposit active sites to promote charge carriers migration to the catalyst surface with a reduced energy barrier. In this regard, an efficient Rh/Cr2O3 core-shell cocatalyst was deposited on the surface of GaN nanowires (Fig.4(b) and Fig.4(c)). The amorphous layer Cr2O3/Rh forms a Schottky contact with GaN, thereby overcoming the exposed energy band gap pinning of the GaN surface, which facilitates the transfer of photoexcited electrons to H+. During the overall water splitting tests, the noble metal (Rh) core promotes the evolution of H2 while the Cr2O3 shell prevents the reverse reaction water splitting (H2 and O2 form water). Oxygen is also evolved from the surface of Rh/Cr2O3-decorated GaN NWs (Fig.4(b)). It can be observed that stable and nearly stoichiometric H2 and O2 are generated from pure water, and there is no significant decrease in the photocatalytic activity after 18 h (Fig.4(d)). This work has pioneered the creation of Rh/Cr2O3-decorated Ga(X)N/Si nanoarchitecture as an efficient and stable photocatalyst for unassisted overall water splitting, providing a far-reaching significance for the area of solar toward hydrogen. In this study, however, the mechanism of O2 yielded from the photocatalyst surface without oxidation active sites remained unknown, which is known to be the rate-determining step and thus deserves research efforts.
In the work above, introducing a core-shell structure Rh/Cr2O3 co-catalyst can significantly improve the photocatalytic activity of hydrogen production from water splitting. Guan et al, on this basis, further improved the catalytic performance by employing a supplementary Co3O4 co-catalyst and by doping indium [96]. The dual co-catalysts decorated double-bands GaN/InGaN NWs, as schematically illustrated in Fig.5(a) and Fig.5(b), exhibited an STH efficiency of approximately 2.7% for unassisted overall pure water splitting under concentrated sunlight on a device surface area of approximately 3 cm2. H2 and O2 were produced in a stoichiometric ratio of 2:1, and the H2 evolution rate reached 1.14 mmol/(cm2∙h) after modifying with Co3O4 and Rh/Cr2O3 as dual co-catalysts. The apparent quantum efficiency (AQE) is increased by 37% in the wavelength range of 200–490 nm, corresponding to the energy conversion efficiency (ECE) of 14% in this range (Fig.5(c)). It should be noted that the GaN/InGaN NWs loaded with Co3O4 and Rh/Cr2O3 can operate stably for 580 h in pure water without obviously varied H2/O2 ratio, showing an impressive long-term overall water splitting (Fig.5(d)). The loading of the suitable co-catalyst Co3O4 nanoparticles could enable the accumulation of photogenerated holes from the surface and then promote water oxidation rather than cause the self-oxidation of the semiconductor, thus contributing to impressive stability. This work also presents an inspiration that the appropriate integration of reduction cocatalyst with oxidation active sites do a great favor to the improvement of the performance of Ga(X)N/Si-based APIDS for overall water splitting. It not only facilitates the achievement of enhanced activity by lowering the energy barrier but also benefits the stability of the device by accelerating the consumption of photogenerated holes for water oxidation.
5.1.2 Photoexcited charge carriers separationtuning
N-type and p-type semiconductor photocatalysts exhibit an upward and downward energy band bending state, respectively [92]. This phenomenon causes the problems of accumulation of photogenerated electrons and depletion of holes on the surface of the photocatalysts, which greatly influences the STH efficiency of unassisted water splitting. Near-surface tuning of energy band structure plays a key role in altering photocatalytic performances. In this regard, Kibria et al. [92] suggested that the quantum efficiency of overall water splitting over Ga(In)N NWs can be enhanced by nearly two orders of magnitude by tuning the near-surface band bending by doping. In this work, during the epitaxial growth, Mg was employed as a p-type dopant. It was observed that the downward bending of In0.26Ga0.74N NWs became gentle with the increasing Mg-dopant incorporation (Fig.6(a)). The large tuning range (approximately 1.7 eV) of the surface Fermi-level provides an opportunity for achieving nearly flat band conditions for InGaN NWs in equilibrium with water, thereby leading to the rapid diffusion of both electrons and holes to be balanced with the redox reactions. The rate of H2 evolution is thus obviously enhanced, reaching approximately 0.78 mol/(h·g) by the optimum Mg-doped In0.26Ga0.74N NWs (Fig.6(b)), which is more than 30 times higher than that of undoped samples. The H2/O2 ratio was nearly 2:1 at the irradiation of visible light, indicating a balanced oxidation and reduction reaction of water splitting (Fig.6(c)). By loading Rh/Cr2O3 co-catalyst onto p-GaN/p-In0.2Ga0.8N NWs, the generation rates of H2 and O2 remain stoichiometrically unvaried at a different wavelength of light (Fig.6(d)). This work proposes the concept of precise tuning of the surface band bending state by appropriately doping with various dopants (X). It successfully improves the rate of H2 evolution of overall water splitting. This work actually provides a new method for alleviating the potential barrier on the surface of nanostructured Ga(X)N, thus mediating the transfer/migration of charge carriers to participate in the reaction by doping X elements when required.
It is well known that separating photogenerated charge carriers and their transportation to the appropriate active sites are critical for efficient solar water splitting but poses a great challenge [97,98]. Moreover, the precise control of the flow of carriers from bulk toward the redox sites also remains elusive. The construction of a built-in electric field is considered an effective approach to achieve maximum carrier separation, because it can be used as the driving force for charge separation that drives the targeted surface reaction in the photocatalytic systems. On the other hand, the photogenerated electrons will reverse transfer under the driving of the electric field, thus drastically expediting the separation of electron-hole pairs. In this regard, Chowdhury et al. [99] subtly fabricated a dual-band InGaN nanosheet photochemical diode (PCD), which could spontaneously accelerate the separation of photogenerated electrons and holes to drive the reduction and oxidation reaction of water by constructing a built-in electric field via gradient-varied doping. Dual-band GaN/InGaN nanowires were vertically aligned on the Si substrate (111) with some HER cocatalysts randomly distributed on its surfaces (Fig.7(a)). In a PCD nanostructure, the water oxidation and proton reduction reaction sites of the nanosheet are located on two spatially separated transverse curved redox surfaces, coupled between the parallel anode and cathode surfaces (Fig.7(b)). The band bending can be reduced in the bulk, which can further lower the recombination probability by constructing the built-in electric field (Fig.7(c)). Importantly, the p-type dopant, i.e., Mg was precisely doped with a concentration gradient at the lateral scale of the nanowire, establishing a strong built-in electric field with controllable direction between the two parallel surfaces of bandgap lateral scale. Such an integrated APIDS achieved a superior STH efficiency of approximately 3.3% and it is observed that the evolution of stoichiometric H2 and O2 can be realized in neutral pH water by utilizing an AM 1.5G filter. Additionally, the activities of overall water splitting of photocatalysts with or without PCDs structure are quite different (Fig.7(b)). This work provides critical insight into achieving superior APIDS for efficient solar toward hydrogen and reveals a method to precisely tune the behavior of photogenerated carriers through the built-in electric field, which is profound for unassisted overall water splitting or PC conversion of other chemical fuels.
5.1.3 Multi-band tunnel junction structure design
The wide energy bandgap of about 3.4 eV of GaN can result in poor absorption of visible light. Improving photocatalysts response to visible light is very important for Ga(X)N-based APIDS to achieve highly effective one-step overall water splitting. Compared with conventional oxide-based photocatalysts, InxGa1–xN can be tuned to cover almost the entire solar spectrum by varying the indium content. Kibria et al. [100] designed a type of multi-band GaN NWs without co-catalysts to further improve the light absorption, charge carriers separation, and surface chemical reactions. As the HAADF image of as grown InGaN/GaN nanowire, ten 3-nm self-organized InGaN/GaN quantum dots with the indium compositions in the range of 15%–50% were subsequently merged along the axial dimension of the GaN nanowire (Fig.8(a)). This triple-band structure (GaN, In0.11Ga0.89N, and In0.32Ga0.68N) improved the light-harvesting ability in UV, blue, and green spectral range according to the photoluminescence (PL) spectrum (Fig.8(b)). Three decreased energy bandgap structures, i.e., 3.40, 2.96, and 2.22 eV enable efficient utilization of light that cannot be absorbed by single-band GaN nanowires, significantly improving apparent quantum efficiency (AQE) in the range of visible light (Fig.8(c)). It is noted that in this architecture, the three segments with various bandgaps own sufficient redox potentials for overcoming the barrier of overall water splitting, exhibiting the great flexibility of Ga(X)N/Si-based platform, which can be subtly tailored for practically meeting the optical, electronic, and catalytic requirements on APIDS. In this study, the introduction of GaN/InGaN quantum structure may play a vital role in chemical reactions by influencing the behavior of reactants molecules, which remains unknown.
Similar to the above work, quadruple-band GaN/InGaN nanowire arrays were prepared by Wang et al. [101]. This monolithically integrated quadruple-band InGaN nanowire photocatalyst consists of Mg-doped (p-type) In0.35Ga0.65N (2.1 eV), In0.27Ga0.73N (2.4 eV), In0.2Ga0.8N (2.6 eV) and GaN (3.4 eV) segments, which were vertically grown on non-planar silicon wafers (Fig.8(d)). Upon light illumination, photons with an energy below 3.4 eV cannot be absorbed by GaN. However, this part of the light can be efficiently absorbed by InGaN with a smaller energy bandgap. The PL spectrum measured at room temperature shows that such a quadruple-band GaN/InGaN NW has the ability to absorb solar photons in the whole visible spectral range compared to single InGaN segments (Fig.8(e)). The quadruple-band GaN/InGaN NWs photocatalyst can further broaden the absorption range of visible light, resulting in an admirable STH efficiency of about 5.3%. Additionally, after several cycles, H2 and O2 with a ratio of 2:1 were stably produced (Fig.8(f)). Based on the discussion above, it can be seen that in sharp contrast with the conventional semiconductors, the Ga(X)N/Si nanoarchitecture provides a wide compositional and structural window for tuning the optical and electronic properties to improve the performance of water splitting toward hydrogen. Herein, it is worth noting that the critical function of the designed multi-band structured Ga(X)N/Si remains exclusive, which requires growing research efforts.
The advances above show the solid viability of employing the Ga(X)N/Si nanoarchitecture for unassisted photocatalytic water splitting. The diverse approaches for engineering the structural and optoelectronic properties of the nanoarchitecture are promising in breaking the limit of STH energy efficiency. Loading suitable co-catalysts and doping with different elements not only provides active sites but also improves the behavior of photogenerated carriers. In terms of reaction kinetics, with the assists of the advanced Ga(X)N/Si nanoarchitecture, the above method greatly reduces the energy barrier of water decomposition, which is more conducive to preventing the occurrence of reverse reactions. Rational construction of InGaN/GaN multi-band can effectively improve the light absorption efficiency and show unique photoelectric properties due to the unique structural properties of the Ga(X)N/Si nanoarchitecture. Of note, major attention has been devoted to nanostructured Ga(X)N. However, the function of the silicon substrate has not been revealed during water splitting, which also deserves more research efforts.
5.2 Ga(X)N/Si-based APIDS for PEC water splitting
Compared with the PC system, bias is applied for PEC systems by an external circuit, which not only favors overcoming the high-reaction barrier of water splitting but also facilitates the separation of electron-hole pairs. Meanwhile, electrolytes such as H2SO4 and NaOH, are usually employed for accelerating PEC water splitting while photocatalytic overall water splitting can be conducted in distilled water. One of the most outstanding features of PEC water splitting is that hydrogen and oxygen can be spatially separated on the cathode and the anode in an appropriate PEC cell, thus inhibiting the backward reaction of H2 and O2 toward H2O. The subsequent separation of the explosive H2/O2 mixture is not required for the PEC process. At present, the key issue which needs to be solved urgently is the design and fabrication of a rational photocathode. Nanostructured Ga(X)N/Si can serve as a proper semiconductor platform for building ideal photocathodes for water splitting toward hydrogen.
5.2.1 Engineering Ga(X)N/Si nanoarchitecture
The Ga(X)N/Si nanoarchitecture is an emerging platform for PEC water splitting. As a bold endeavor, AlOtaibi et al. [102] reported Si-doped n-type GaN nanowires grown on a Si (111) substrate as a photocathode for PEC water splitting, while IrOxand Ag/AgCl were employed as the counter electrode and reference electrode in a three-electrode system, respectively (Fig.9(a) and Fig.9(b)). Ga(Si)N/Si-based photocathode showed a saturated photocurrent density of 14 mA/cm2 in 1 mol/L HBr under AM 1.5G solar-simulated illumination (100 mW/cm2) (Fig.9(c)), which is higher than that of the undoped sample (10 mA/cm2). Herein, the surface resistance of GaN NWs was dramatically reduced by Si doping, which was due to the increased conduction path and the large carrier concentration. As a result, the incident-photon-to-current-conversion efficiency (IPCE) of Ga(Si)N/Si reached 18%, outperforming the undoped GaN/Si sample (15%) (Fig.9(d)). However, it is observed that below the saturation voltage, the current density of Si-doped nanowires under light illumination is smaller than that of undoped nanowires. This observation can be explained by the fact that by high-level doping of Si, the electron-hole recombination becomes more important in the natural region of GaN NWs, thus leading to a reduced current flow between the nanowire and electrolyte compared to the undoped ones (Fig.9(e)). To simulate the actual situation, the prepared photoelectrode was tested in a two-electrode configuration cell. H2 was steadily produced over 5 h with an average rate of approximately 38 µmol/h, suggesting a good corrosion-resistant property of GaN NWs in HBr solution (Fig.9(f)). However, the bare Ga(X)N/Si nanoarchitecture is essentially limited by sluggish reaction kinetics and severe photo corrosion. Further engineering Ga(X)N/Si nanoarchitecture is, therefore, expected.
Long-term stability is a key parameter of photoelectrodes for practical applications. The stability of GaN NWs/Si-based photocathodes was investigated under simulated sunlight irradiation in a three-electrode configuration using 0.5 mol/L H2SO4 as the electrolyte. In this work, using n+-GaN nanowires vertically aligned onto n+-p Si substrate as a light absorber platform, Pt nanoparticles, and state-of-the-art HER cocatalyst were uniformly deposited onto the platform to promote hydrogen evolution from water splitting (Fig.10(a), Fig.10(c) and Fig.10(d)). Surprisingly, GaN is able to protect the photocathode from photo-corrosion due to its N-rich surface. Nearly perfect CB edges alignment between GaN and Si enables the photogenerated electrons to be efficiently extracted from the underlying silicon substrate to GaN NWs and Pt nanoparticles to participate in HER (Fig.10(b)). Meanwhile, due to the large surface area of GaN nanowires, the transport resistance of charge carriers at the semiconductor/liquid interface is significantly reduced compared with that of a single silicon photocathode. In addition, light absorption can also be enhanced by nanostructured one-dimensional GaN arrays. Therefore, the J−V curve was obviously improved (Fig.10(e)). The maximum IPCE of the Pt-decorated-GaN NWs/Si photocathodes was approximately 80% at 620 nm when the photocathode was measured at 0 V versus reversible hydrogen electrode (RHE) in a 0.5 mol/L H2SO4 aqueous solution (Fig.10(f)). The saturated photocurrent density reached 38 mA/cm2 with a decent onset potential of approximately 0.5 V vs. RHE and a maximum applied bias photon-to-current efficiency (ABPE) of 10.5% was achieved at 0.32 V versus RHE under an AM 1.5G one-sun illumination (Fig.10(g)). Chronoamperometry analysis for this photocathode retains superior stability longer than 100 h with an appreciable current density of 38 mA/cm2 (Fig.10(h)). Herein, the impressive stability of the photocathode is mainly attributed to the N-rich surface of GaN NWs, which can act as a multifunctional protection layer against photo corrosion.
5.2.2 Decoration of GaN/Si with earth-abundant cocatalysts
For the sake of practical applications, the exploration of earth-abundant alternatives to noble metals is very important for assembling an affordable Ga(X)N/Si-based APIDS. Molybdenum sulfide (MoSx), as an earth-abundant HER cocatalyst, was tentatively applied for decorating the GaN/Si nanoarchitecture to accelerate water splitting by photoelectrocatalysis (Fig.11(a) and Fig.11(b)) [103]. Owing to the well matched geometric and electronic properties between GaN and MoSx, nearly defect-free GaN NWs play an ideal electron-migration channel between Si wafers to the deposited HER cocatalyst of MoS2 (Fig.11(c), Fig.11(d), Fig.11(i)). In this work, a superior current density of approximately 40 mA/cm2 at 0 V versus RHE with an onset potential of 0.4 V versus RHE was obtained under simulated solar illumination (Fig.11(e)). The ABPE reached 5.0% at 0.16 V versus RHE (Fig.11(f)). Meanwhile, there was no significant drop in photocurrent within 10 h illumination, suggesting the appreciable stability of the device (Fig.11(g)). HR-TEM characterization and DFT calculation reveal that GaN and MoSx have excellent geometric matching in lattice parameters, which favors the migration of electrons, thus promoting water splitting. Besides, the one-dimensional structure of GaN NWs is beneficial for alleviating photon scattering and exposing catalytic sites with a high density. The combined effects above significantly improve the PEC water splitting performance for MoSx@GaN NWs/Si in contrast to MoSx/Si (Fig.11(h)). Overall, this study demonstrates that the high-quality GaN NWs can be used as ideal scaffolds for loading low-cost HER cocatalysts, e.g., MoSx to assemble an industry-friendly PEC device.
In addition, inspired by the natural [Fe-Fe]-hydrogenase, an efficient HER biocatalyst from green plants, Zhou and coworkers [104] developed few-atomic-layers iron (FeFAL) as promising alternatives to noble metals for improving the performance of GaN NWs/Si nanoarchitecture by a facile and controllable electrodeposition approach (Fig.12(a)). In this architecture, upon simulated solar irradiation, the photogenerated electrons are extracted from silicon by GaN NWs to the surface to participate in HER which is catalyzed by FeFAL. According to the SEM and bright-field scanning transmission electron microscopy (STEM-BF) images of FeFAL:GaN NWs/n+-p Si (Fig.12(b) and Fig.12(c)), only a few atomic layers of iron are anchored on the m-plane of GaN NWs to form a core-shell structure. A high current density of 30 mA/cm2 is achieved at a minor overpotential of 0.2 V, which is obviously superior to that of n+-p Si only, Fe/n+-p Si, and GaN NWs/n+-p Si (Fig.12(d)). DFT calculations reveal that the Gibbs free energy of HER is lowered by the decoration of FeFAL, thus energetically facilitating water splitting toward hydrogen (Fig.12(e)). Such a novel-metal-free device provides preferable technical support for green hydrogen production from PEC water splitting.
5.2.3 Self-improved effect of Ga(X)N/Si-based APIDS
According to previous knowledge, photoelectrodes easily suffer from deactivation because of the severe photo corrosion in the electrolyte under light illumination. GaN NWs were, therefore, employed as a special protective layer to improve the durability of the Si-based photocathode owing to its unique N-rich surface as discussed above. However, the fundamental understanding of the critical role of GaN NWs in PEC water splitting remains largely unknown. Recently, an interesting self-improved effect was reported for GaN NWs/Si-based photocathodes during HER [105]. In this work, the STEM-EELS and the XPS measurements were used to track whether the formation of new chemical species is the origin of PEC performance enhancement. It can be observed that a protection layer with a thickness of 1 nm mixed with Ga, N, and O was formed on the sidewalls of the grain, while the top was covered with a Ga- and O-containing layer. This leads to the conclusion that oxygen also has an orientation preference on the non-polar or semi-polar planes (Fig.13(a)).
Additionally, atomic force microscopy (AFM) analysis reveals that the influence of surface roughness changes for PEC performance as the reaction proceeds. The surface morphology and roughness of the CA-0 h and CA-10 h samples are similar. The effect of surface roughness variation on PEC performance can be ruled out (Fig.13(b)). Theoretical calculation results illustrate that the non-polar sidewall is conducive to mixing the Ga, O, and N. In addition, the most stable structure is obtained by partial oxygen substitution (50%) of nitrogen in both surface and subsurface bilayers (Fig.13(c)). Chronoamperometric (CA) tests were measured on different samples under one-sun illumination at an unvaried bias of –0.6 V vs. RHE in 0.5 mol/L H2SO4. During the linear sweep voltammetry (LSV) measurements, the onset potential was shifted from −0.46 V vs. RHE for the CA-0 h sample to –0.08 V versus RHE for the CA-10 h sample. Interestingly, the photocurrent density gradually increased, reaching a saturation state of 25 mA/cm2 after 3 h of illumination with a Faraday efficiency of nearly closing to 100% (Fig.13(d)). Benefiting from the self-improving characteristics of GaN, the loading amounts of noble metal co-catalysts (such as Pt) could be reduced. For example, the CA-0 h sample needs 4 times of Pt photo-deposition time to shift its turn-on voltage to 0.34 V vs. RHE compared with a CA-4 h sample. In this case, a thin layer of GaN oxide was found to be formed on the sidewalls of GaN nanowires by partial oxygen substitution at the nitrogen sites during the PEC water splitting reaction, which mainly contributed to the self-improved PEC performance. Briefly speaking, this study presents an unprecedented strategy to generate a self-improved photoelectrode by partial oxidation of GaN, showing the possibility to regulate the PEC performance via surface atom engineering.
5.2.4 Ga(X)N-based tunnel junction photocathode
Although GaN has been successfully applied to construct suitable photocathodes, its large bandgap is not favored for effectively harvesting the solar spectrum at long wavelengths. As discussed above, the energy band gap of GaN can be reduced from 3.4 to 0.6 eV by indium incorporation. In this context, an In0.42Ga0.58 N tunnel-junction was coupled with Si wafer for broadening light absorption [90]. In particular, n-type GaN NWs as underlying light absorbers were grown on n-type silicon wafers, and an n++/p++ GaN tunnel junction was then built (Fig.13(a)). This design of tunnel junction can reduce the depletion depth and resistivity to improve the separation of charge carriers. When the indium content varies from approximately 24% to approximately 51%, the emission wavelength can be shifted from approximately 500 nm to approximately 750 nm, corresponding to the energy bandgap varying from 2.48 to 1.65 eV, respectively (Fig.14(b)). Of note, as the indium content increases, the emission peaks get broadened and weakened, suggesting an increasing number of defects. Consequently, the photocurrent density measured at 0 V versus RHE showed an increasing trend due to the reduced bandgap (Fig.14(c)). However, this value decreases significantly when the indium content is greater than 42% because of the increasing defects and the resultant enhanced non-radiative recombination of InGaN. The optimized In0.42Ga0.58N/GaN/Si photocathode exhibited a maximum APBE of 4% at 0.52 V versus RHE. It can maintain a photocurrent density of 12.3 mA/cm2 at 0 V versus RHE with an onset potential of 0.79 V vs. RHE in 0.5 mol/L H2SO4 under an AM 1.5G solar illumination (100 mW/cm2) (Fig.14(d) and Fig.14(e)). In this work, as discussed above, Ga(X)N can offer a broad and flexible window for tailoring the optical properties, thus being an ideal building block of suitable photocathodes by coupling with silicon. They work in synergy to maximize solar-to-fuel efficiency.
6 Conclusions and outlook
In conclusion, the Ga(X)N/Si nanoarchitecture has emerged as a promising semiconductor platform for sunlight-powered water splitting in either PC or PEC systems. This paper summarized the recent important progress in this field. These breakthroughs shed light on the availability of the Ga(X)N/Si nanoarchitecture for green hydrogen production with the inputs of sunlight and water. Especially, the rapid advances in the growth technologies like MOCVD, MBE, and MS enable the precise assembly of Ga(X)N and Si at the atomic scale, thus presenting one tunable platform for addressing the critical issues of solar-to-fuels conversion. First of all, the widely adjustable energy bandgap of Ga(X)N endows the platform with a large solar spectrum respond range and flexible redox potentials that can meet the requirements of water splitting. Secondly, the state-of-the-art growth technology enables the growth of nearly defect-free Ga(X)N with controllable carriers when preparing rationally designed multi-band or complex nanostructures by various strategies including doping. However, the resultant increasing defects and mismatches arising from the incorporation of X might be detrimental to charge carriers separation by serving recombination centers. The community should devote more efforts to improving the crystal quality, morphology, and size of Ga(X)N when X dopants are incorporated, and thus minimize the negative influence of defects and mismatches. From the viewpoint of chemical reaction, the introduced defects may be adverse to promoting the reaction by regulating the adsorption/activation of chemical reactants. The water molecules can be spontaneously dissociated into hydroxyl and hydrogen ions on the non-polar m-plane surface of GaN. Certainly, the reduction of the reaction energy barrier by using appropriate cocatalysts is also required. Moreover, as suggested by the previous studies, the N-rich surface can protect the Ga(X)N/Si nanoarchitecture from photooxidation, thus benefiting long-term stability. Up to date, state-of-the-art Ga(X)N/Si-based photocathode has been tested in acidic electrolyze under simulated solar illumination for over 3000 h. Thrillingly, the most recent discovery showed that during PEC water splitting, the partial oxidation of GaN into gallium oxynitride led to a self-improved behavior over Pt/GaN NWs/Si after a stability testing, which had rarely been reported for other material systems. Moreover, this platform mainly consists of the two most produced semiconductors, i.e., silicon, and Ga(X)N, which are widely used in industry. Together with the continued breakthroughs of the catalysts-developed community, the Ga(X)N/Si nanoarchitecture holds grand promise in breaking the bottlenecks of commercial applications of solar-to-hydrogen conversion.
Despite a series of important progresses, the research on Ga(X)N/Si nanoarchitecture for solar-to-hydrogen conversion remains highly infant. Moreover, owing to the unique surface and electronic properties, the Ga(X)N/Si nanoarchitecture also exhibits great potential in yielding high-value H2O2 rather than O2 from the half oxidation reaction of water splitting, thus making water splitting more economically favorable. It is worthy of devoting considerable efforts to exploring appropriate oxidation cocatalysts to catalyze H2O2 formation from water splitting by a 2-electron process. To achieve eventual industrial success, a number of issues are to be addressed. First of all, the device needs to be facilely produced with dramatically reduced cost. Specifically, noble metal co-catalysts such as Pt, Rh, IrOx, etc., need to be substituted by earth-abundant and inexpensive materials via highly controlled methods. Meanwhile, the community looks forward to seeing the advances in the grown technologies, e.g., MBE, MOCVD, and MS with economically competitive costs. Additionally, outstanding long-term stability is another essential factor of sunlight-powered water-spitting systems for practical applications. Strategies to improve the stability of APIDS are thus extremely desired. The introduction of highly dense protective layers by precise control methods such as atomic layer deposition without affecting the catalytic activities is one of the feasible pathways in the future. Moreover, the rational design and construction of suitable electron and hole transportation layers on the surface of semiconductor photocatalysts are also worth attempting to improve the stability of Ga(X)N/Si-based APIDS.
As a popular method for studying the chemical reaction at the molecular level, ab-initio calculations based on the density functional theory were employed to uncover the mechanism of water splitting catalyzed by Ga(X)N/Si-based APIDS and had made important advances. However, in most instances, this kind of calculation can only simulate the reaction process at a small scale of tens or hundreds of atoms. Simulating the large-volume polyatomic case is highly desirable yet challenging, especially if the computing resource is limited. More suitable computational methods such as high-throughput computing and machine learning based on big data can be used to simulate the situation at a large scale, which is more matched with the real situation. Advanced computation methods to study the behaviors of photons absorption, charge carriers separation, migration, as well as chemical species evolution over Ga(X)N/Si-based APIDS are also highly expected, especially in photoexcited states. Moreover, bold assumptions can be made that the artificial intelligence-assisted technology developed in the future may come into a major play in rational structure design, cocatalyst development, and reaction mechanism study. Combined with various advanced characterization techniques like in situ FTIR, in situ Raman, and STEM-XPS cascading system, a deep mechanistic understanding of water splitting over Ga(X)N/Si-based APIDS will be further clarified. It is highly critical for guiding the rational design of revolutionary APIDS and opening a new route toward zero carbon emission in the future.
Overall, for the sake of advancing toward practice, much more effort should be directed toward improving the activity, stability, and energy efficiency of Ga(X)N/Si-based APIDS with the assistance of various advanced technologies. On this basis, rational quantum structures are expected to be designed and fabricated to break the bottlenecks of STH efficiency and stability. In this context, advanced growth and characterization techniques are highly required.
In addition to water, biomass is the second most earth-abundant renewable hydrogen source with an annual yield of 200 billion tons. Therefore, it is also a promising reserve for green hydrogen production catalyzed by Ga(X)N/Si-based AIPDS using sunlight. In this case, the simultaneous generation of high-value chemicals and green hydrogen is feasible, which can make green hydrogen production from biomass more economically competitive. Of note, owing to the notable molecular discrepancy between biomass and water, the complex structure of biomass (C−C, C−H, CO, COOH, O−H, C−O, aromatic rings, etc.), the rational design and assembly of new Ga(X)N/Si-based APIDS with different approaches for efficient hydrogen generation are under demand.