Photocatalytic fixation of nitrogen to ammonia: state-of-the-art advancements and future prospects

Abstract

The burgeoning development of ammonia (NH₃) synthesis technology addresses the urgency of food intake required to sustain the population growth of the last 100 years. To date, NH₃ has mostly been synthesized by the Haber–Bosch process in industry. Under the ever-increasing pressure of the fossil fuel depletion crisis and anthropogenic global climate change with continuous CO₂ emission in the 21st century, research targeting the synthesis of NH₃ under mild conditions in a sustainable and environment friendly manner is vigorous and thriving. Therefore, the focus of this review is the state-of-the-art engineering of efficient photocatalysts for dinitrogen (N₂) fixation toward NH₃ synthesis. Strenuous efforts have been devoted to modifying the intrinsic properties of semiconductors (i.e. poor electron transport, rapid electron–hole recombination and sluggish reaction kinetics), including nanoarchitecture design, crystal facet engineering, doping and heterostructuring. Herein, this review provides insights into the most recent advancements in understanding the charge carrier kinetics of photocatalysts with respect to charge transfer, migration and separation, which are of fundamental significance to photocatalytic N₂ fixation. Subsequently, the challenges, outlooks and future prospects at the forefront of this research platform are presented. As such, it is anticipated that this review will shed new light on photocatalytic N₂ fixation and NH₃ synthesis and will also provide a blueprint for further investigations and momentous breakthroughs in next-generation catalyst design.

---

Xingzhu Chen

Xingzhu Chen received her BS in Materials Science and Engineering from Wuhan University of Technology in 2016. She is presently working on her PhD at Wuhan University of Technology. Her research interests include photoelectrochemistry, photocatalysis, and the development of photocatalysts, related materials and devices.

Neng Li

Neng Li received his PhD degree in condensed matter physics from Huahzong University of Science and Technology in 2011. He then joined the Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences as a Research Fellow in 2011. He worked at the Department of Physics and Astronomy, University of Missouri-Kansas City as a postdoctoral researcher from 2012 to 2014. In 2015, he joined Wuhan University of Technology as a Full Professor. Dr Li was a Visiting Research Fellow at the Department of Materials Science & Metallurgy, University of Cambridge, UK from 2016 to 2017. His research interests focus on energy and environmental applications such as photocatalysis, photocatalytic hydrogen production, CO2 reduction, N2 fixation, and the development of hybrid functional materials and related devices.

Wee-Jun Ong

Wee-Jun Ong received his BE and PhD degrees in Chemical Engineering from Monash University. In 2015, he was a Visiting Research Fellow at the University of New South Wales and Monash University Clayton Campus, Australia. He is currently a Research Scientist at IMRE, A*STAR in Singapore. At present, he serves as the Associate Editor of Frontiers in Materials, an Editorial Board Member of Scientific Reports, Nanotechnology and Nano Futures, and a Community Board Member of Materials Horizons. He has also been the Lead Guest Editor of several Special Issues. His research interests primarily focus on photocatalytic, photoelectrochemical and electrochemical H2O splitting, CO2 reduction and N2 fixation for energy conversion and storage using two-dimensional (2D)-based nanomaterials, transition metal dichalcogenides/phosphides and carbonaceous-based hybrid nanocomposites. For more details, see http://https://sites.google.com/site/wjongresearch/.

1. Introduction

It is a well known fact that nitrogen is one of the starting building blocks for the synthesis of amino acids, nucleotides and other major biological compounds in all organisms.^1,2 The main source of nitrogen, dinitrogen (N₂), is the largest single component of the Earth's atmosphere (around 78% by volume). As an important precursor for nitrogen-containing compounds, ammonia (NH₃) could be directly synthesized from dinitrogen by a small group of organisms named diazotrophs in nature.³ However, this could scarcely meet the large needs of the fertilizer industry today.

The fixation of N₂ to NH₃ is thermodynamically accessible: N_2(g) + 3H_2(g) → 2NH_3(g), ΔH_298K = −92.2 kJ mol⁻¹. However, this reaction cannot occur spontaneously under ambient conditions. The reduction reactions of N₂ to its partial reduzates, such as diazene (N₂H₂) or hydrazine (N₂H₄), are also non-spontaneous due to the positive standard enthalpies of formation of +212.9 and +95.35 kJ mol⁻¹ for N₂H₂ and N₂H₄, respectively.⁴ There are many factors hindering the cleavage and hydrogenation of dinitrogen in nature. First, the thermodynamically strong cleavage energy of the first bond in N₂ (410 kJ mol⁻¹) manifests the critical challenge in the full dissociation of N=N, which explains the chemical inactivity of N₂ compared to other triple-bonded molecules.⁵ Additionally, from the aspect of kinetics, the large energy gap between the HOMO (the σ_g2p bonding orbital) and LUMO (the π_g*2p anti-bonding orbital) explains the high chemical stability of N₂. Other factors, such as low proton affinity, also account for the difficulty of direct protonation of N₂. Detailed descriptions can be found in the review article by Jia and Quadrelli.⁴

The Haber–Bosch process, which involves the reaction of N₂ and hydrogen (H₂) over iron or ruthenium-based catalysts, has been estimated to be one of the most intriguing discoveries of the last century.^6,7 After 100 years of development, the temperature is controlled to near 450 °C and the reaction pressure has decreased to 15 to 30 MPa from the original 50 to 100 MPa, while the efficiency is limited to 10% to 15%.^8,9 Inevitably, this process accounts for 1.6% of total global CO₂ emissions; this is attributed to the presence of hydrocarbons in the raw materials (as the main hydrogen source and energy supplier in this process).¹⁰

Parallel to the optimization of the thermocatalytic process,^11–14 several other methods have aroused widespread attention in the synthesis of ammonia to date. Before we engage in our topic of photocatalysis for the fixation of nitrogen to ammonia, we would like to briefly mention the progress of the other two traditional approaches. Owing to the variable coordination number of transition metals, organometallic complexes can participate in the activation of N₂. The ground-breaking work of the first nitrogen complex [Ru(NH₃)₅(N₂)]⁺ in 1965 challenged the traditional belief that dinitrogen cannot form complexes.¹⁵ Since then, nitrogen complexes have been developed to generate ammonia under particular conditions, in which central elements of various transition metals, such as molybdenum,^16–18 iron and titanium,^19–22 were tested as the sites of dinitrogen coordination and reduction. Furthermore, the model and action mechanism of iron molybdenum complexes as active sites in nitrogenase enzymes have greatly fascinated many chemists and biologists.^23–25 On the other hand, inspired by electron and proton transfer processes, numerous researchers have carried out investigations of electrochemical and electrocatalytic approaches.²⁶ Since a solid electrolyte cell was first used in the electrochemical synthesis of ammonia in 1998,²⁷ various conducting ceramic membranes have been tested to produce NH₃ from N₂.^28–31 Since then, polymer electrolytes such as Nafion and sulfonated polysulfone have been deemed to provide excellent proton conductivity under low temperature.^32–34 Later, molten salts such as chlorides,³⁵ oxide-carbonates or hydroxides were employed as electrolytes.^36–39 Very recently, Chen et al. reported an electrolyte-less cell with the highest faradaic efficiency of 95.1% and a relatively low rate of 3.60 × 10⁻¹² mol s⁻¹ cm⁻² at room temperature, wherein the electrocatalyst was based on Fe₂O₃ nanoparticles supported on conductive carbon nanotubes (Fe₂O₃-CNT).⁴⁰

With regard to the inexhaustible and clean solar energy from sunlight,^41–47 photocatalysis has been thrust into the limelight in the realms of energy and chemical fuel production; thus, blossoming interest in the design of myriad semiconductors has been witnessed in recent years.^48–60 Generally, the photocatalytic process makes use of photons as the driving force to propel the activation of N₂. Compared with electrochemical approaches, photon-driven fixation of N₂ has reached a stumbling block stemming from the difficulties of N₂ chemisorption and activation.⁶¹ The magnitudes of production rates for most studied photocatalysts remain below par,⁶² distinctly hindering the success of the solar fixation of N₂. In the initial studies, N₂ photoreduction was considered to take place in nature over abundant minerals on the surface of the earth; thus, earlier studies on the photocatalytic fixation of N₂ mainly focused on soil minerals and sand in nature.^63–65 This was introduced in the previous work of Schrauzer.⁶⁶ Since the first exploratory study on as-synthesized TiO₂-based photocatalysts for N₂ fixation under UV light in 1977,⁶⁷ a flurry of research activities have been devoted to the development of photocatalytic N₂ fixation, especially in the 21st century, due to the fact that the driving force (light) and ingredients (water and air) of this process are relatively clean, cheap and accessible, as shown in the following equation.

The thermodynamic non-spontaneous reaction, which can be regarded as a combination of water splitting and N₂ fixation, can be accomplished with the introduction of solar energy.

To the best of our knowledge, there is no literature pertaining to the latest research progress of photocatalytic N₂ fixation with respect to the classification of photocatalysts. In this review article, we will present and update the state-of-the-art research advancements in this growing field of N₂ fixation, as depicted in Fig. 1. Briefly, the classification of pristine photocatalysts is firstly discussed, followed by the rational development of hybrid heterojunction nanocomposites via various synthesis strategies for application in the fixation of N₂ to NH₃. Finally, the review is concluded with a summary, invigorating perspectives and future prospects on this research horizon. As such, it is projected that this article will markedly promote the importance of this research platform and help to provide new insights for the exploration of future investigations toward a sustainable future.

Fig. 1 Illustration of the main topics covered in this review article, exemplifying the latest research in the development of photocatalysts for N₂ reduction.

2. Fundamental process of N₂ photoreduction

2.1. Basic photocatalytic principles over semiconductors

The photocatalytic process of N₂ fixation can be divided into several steps. First, the photogenerated electrons are promoted to the conduction band, leaving vacant holes in the valance band. Afterwards, some of the electrons and holes recombine with each other; meanwhile, others migrate to the surface of the catalyst and participate in the redox reaction. Specifically, H₂O can be oxidized to O₂ by the holes, whereas N₂ is reduced to NH₃ after a series of multi-step injections of photogenerated electrons and water-derived protons.

The hydrogenation reactions related to this process and the corresponding reduction potentials are summarized in Table 1.^68–71 As is known to all, whether the photocatalytic redox reaction can occur largely depends on the reduction potential of the adsorbate and the position of the energy band in the semiconductor (Fig. 2).^72–85 For example, the position of the semiconductor conduction band should be higher (more negative) than the reduction potential of the N₂ hydrogenation, while the valance band should be lower (more positive) than the oxygen evolution potential. The maximum energy transition state is located in the very first electron transfer (−4.16 V vs. NHE) and proton-coupled electron transfer (−3.2 V vs. NHE) processes (Table 1), hampering the overall kinetic reaction.⁸⁶ On this basis, these are the main two limitations that must be overcome to activate N₂ molecules for NH₃ formation. It is of utmost importance to maintain a small band gap of the semiconductor, preferably in the visible light region, which can still satisfy the thermodynamic reduction potentials of N₂ to NH₃. Additionally, it is necessary to suppress the recombination of charge carriers in semiconductor photocatalysts in order to enhance the solar conversion efficiency and apparent quantum yield (AQY) of the reaction.

Table 1 Reduction potentials (vs. NHE at pH 0) of typical hydrogenation reactions related to the reduction of N₂ to NH₃^68–71

Reaction	E ^0 (V)
a Reduction potential E^0vs. NHE at pH 7. b Reduction potential E^0vs. RHE.
H2O → ½O2 + 2H^+ + 2e^−	0.81^a
2H^+ + 2e^− → H2	−0.42^a
N2 + e^− → N2^−	−4.16
N2 + H^+ + e^− → N2H	−3.2
N2 + 2H^+ + 2e^− → N2H2	−1.10^b
N2 + 4H^+ + 4e^− → N2H4	−0.36
N2 + 5H^+ + 4e^− → N2H5^+	−0.23
N2 + 6H^+ + 6e^− → 2NH3	0.55
N2 + 8H^+ + 8e^− → 2NH4^+	0.27

---

Fig. 2 Schematic of semiconductor-based photocatalysts used for the conversion of N₂ to NH₃. The redox potentials (V vs. NHE at pH = 0) of water splitting and dinitrogen hydrogenation are marked on the left.

2.2. N₂ reduction mechanism

Recently, a number of investigations from the viewpoint of experimental and density functional theory (DFT) calculations have been carried out to demystify N₂ reduction mechanisms over various catalysts.^87–89 As is mentioned above, in the case of photo-driven NH₃ synthesis, elementary electrochemical reactions can be employed to help decipher the hydrogenation process in greater detail. For example, at the very first proton-coupled electron transfer, the N₂ adsorbed on the catalyst surface (*N≡N) gains one proton (H⁺) from the environment and one photo-generated electron (e⁻) from the catalyst to generate an adsorbed chemical species (*N=NH) as shown: *N≡N + H⁺ + e⁻ → *N=NH˙

For the following transfer of H⁺/e⁻ pairs, 5 routes were proposed by Azofra et al. in their DFT study (Fig. 3).⁸⁷ There are two widely accepted mechanisms for the conversion of N₂ to NH₃. These are commonly called distal and alternating mechanisms.¹⁹ In the distal mechanism, H⁺/e⁻ pairs are proposed to consecutively attach to one N atom of N₂ to form a terminal nitride intermediate, liberating the first NH₃ and leaving a single N, which finally converts into another NH₃ (Fig. 3, path 1). In contrast, the alternating mechanism postulates that H⁺/e⁻ pairs occur alternately on the two N atoms of N₂ (Fig. 3, path 2).

Fig. 3 Five proposed routes for the conversion of N₂ to NH₃. Reproduced with permission.⁸⁷ Copyright 2016, Royal Society of Chemistry.

In the study of the N₂ fixation mechanism, the combination of experimental investigation with computational simulation will help us to further understand the reaction routes: the computational simulation gives a prospective direction for guidance of further experiments, while the experimental research provides feedback and confirmation of the optimization of the theoretical model.

3. Classification of photocatalysts for N₂ fixation to NH₃

Most traditional unmodified semiconductors cannot meet the energy standards for the reduction potentials of the intermediate reactions. Hitherto, many concerted efforts have been made regarding the choice and modification of semiconductors, such as doping, introducing vacancies, plasmon induction, facet tailoring and heterostructure assembly, to improve the photocatalytic performance. In the subsequent sections, the photocatalysts are classified based on their elemental compositions.

3.1. Metal oxide-based materials

3.1.1. Titanium oxide-based materials. The excellent performance of the photolysis of water on TiO₂ created great awareness of the application of TiO₂-based nitrogen photofixation catalysts.⁹⁰ In 1988, Bourgeois and his partners reported that unmodified TiO₂ exhibited N₂ photocatalytic activity after annealing in air because the thermal pretreatment generated surface defects which introduced defect or impurity states in the band gap of the semiconductor.⁹¹ Most recently, the photocatalytic conversion of N₂ to NH₃ on the surface oxygen vacancies of TiO₂ was systematically studied by Shiraishi's group.⁹² Among the tested catalysts, JRC-TIO-6 (rutile phase) exhibited the highest N₂ fixation activity, with a 2.7-fold enhancement using 2-PrOH as a sacrificial electron donor for 12 h of reaction. Further research corroborated that the loading of Ru, Pt or Pd particles would not enhance the yield of NH₃ because these particles covered the Ti³⁺ induced by surface oxygen defects. As revealed in Fig. 4a, the superficial Ti³⁺ provided abundant active sites for N₂ fixation by acting as an electron donor, leading to relative ease of dissociation of the N≡N bond. With several transformations of surficial Ti from Ti³⁺ to Ti⁴⁺, the electrons were naturally injected into N₂. Meanwhile, Ti³⁺ could be regenerated under UV irradiation. The groundbreaking exploration on TiO₂-based materials dates back to 1977. Schrauzer and Guth synthesized iron-doped TiO₂ by heating iron(III) sulfate-impregnated anatase TiO₂; they testified to its great ability to reduce N₂ under UV irradiation.⁶⁷ In their work, ammonia and a small quantity of hydrazine were detected, with the iron content in TiO₂ varying from 0% to 1%. Interestingly, the NH₃ yields reached a maximum value over 0.2% Fe₂O₃-doped TiO₂ under UV. The facile impregnation method described in this research has thus enlightened and inspired research endeavors by multitudinous worldwide researchers.

Fig. 4 (a) Mechanism for N₂ conversion to NH₃ over the surface oxygen vacancies of TiO₂. (b) A photocatalytic cycle for N₂ reduction at the surface of rutile TiO₂(110). Reproduced with permission.⁹² Copyright 2017, American Chemical Society. (c) Transmission electron microscopy (TEM) and (d) high-resolution TEM (HRTEM) images of Fe³⁺-doped TiO₂. Reproduced with permission.⁹⁹ Copyright 2014, Elsevier.

After that, Augugliaro et al. utilized an identical impregnation technique to prepare Fe₂O₃-hybridized TiO₂ and supported it on Al₂O₃. An improved NH₃ production rate was gained in gas–solid fluidized bed reactors.⁹³ According to literature reports, the effect of iron ions facilitated the transformation of the crystallinity of TiO₂, which noticeably affected the photocatalytic efficiency. Notably, the incorporation of Fe accelerated the phase transformation of anatase into rutile and also promoted grain growth during calcination. In this regard, the ratio of anatase to rutile played a prevailing role in the efficiency of the catalysts by maintaining a constant concentration of iron.^94,95 After a prolonged calcination process, the increased content and growing grain size of rutile phase as well as the loss of surface Ti–OH groups led to a decrease of the active surface area, resulting in inferior photocatalytic performance.⁶⁷

Several authors have employed diversified methods in the preparation of iron-doped TiO₂. In Radford and Francis's research, both the anatase and rutile phase of TiO₂ were sampled for doping with iron in metal vapor without high-heat treatment.⁹⁶ Aqueous suspensions of samples were irradiated under UV light. Undoped samples demonstrated minimal or no activity under light irradiation. It should be noted that the activity of doped anatase was higher than that of doped rutile, which can be attributed to the more negative flat band potential of anatase.⁹⁷ The author investigated whether iron was the sole active species involved in the photoactivity because the yield of ammonia over iron-doped anatase was far beyond their anticipation. However, Shrauzer indicated later that the NH₃ may have formed on a photo-assisted organoiron compound.⁶⁶ As for the role of Fe³⁺, Soria et al. demonstrated that iron ions could temporarily trap photogenerated electrons and promote the separation of charge carriers.⁹⁸ They coprecipitated an aqueous solution of TiCl₃, which contained Fe³⁺ ions, in an ammonia solution; they then heated the resulting solid for a day. The residual chloride ions were noted to exhibit poor activity of the coprecipitation specimens. Recently, Fe-doped TiO₂ with highly exposed (101) facets obtained via a hydrothermal method by adsorbing Fe³⁺ on a precursor of hydrogen titanate nanotubes was reported by Zhao et al. (Fig. 4c and d).⁹⁹ The N₂ photoreduction performance was enhanced by exposing the (101) facets of Fe-doped TiO₂ with ethanol as the scavenger. A low doping concentration of Fe³⁺ on the TiO₂ surface enhanced the trapping of electrons and holes by forming Fe²⁺ and Fe⁴⁺ to inhibit charge recombination. The unstable Fe²⁺ and Fe⁴⁺ transferred electrons and holes to Ti⁴⁺ and OH⁻, generating Ti³⁺ (as mentioned in the preceding section, Ti³⁺ plays an essential role in the photocatalytic activation)⁹² and ˙OH, respectively. Similar to the findings by Soria et al.,⁹⁸ the authors mentioned that no nitrite was formed in the presence of ethanol solution because ethanol functioned as a scavenger of ˙OH to prevent the oxidation of ammonia.

Apart from iron-doped TiO₂, other transition metals have been used as dopants in TiO₂.^100–103 The reports of these investigations, while significant, were popular in the last century; however, they are receiving little attention at present because there have been few breakthroughs in their activity in N₂ photoreduction.^100,101 In the original study, Schrauzer and Guth tested eleven metal dopants in addition to iron, including Co, Mo, Ni, Pd, V, Cr, Cu and the noble metals Pt, Ag, Au, and Pb.⁶⁷ Compared to the control group, the Co, Mo and Ni dopants contributed to enhancement of the NH₃ yield of TiO₂. However, eight other metals presented inhibitive effects on NH₃ formation. This phenomenon was attributed to the different influences of the metal dopants on the phase transformation of TiO₂. Apparently, the samples containing Co, Mo and Ni accelerated the anatase to rutile transformation during heat treatment; meanwhile, no acceleration of phase transformation was readily observed on doping with eight other metals. In contrast, a different phenomenon was manifested by Palmisano and his co-workers, where chromium-ion-doped TiO₂ was found to be effective in N₂ photoreduction.¹⁰⁰ They clarified that compared with iron-ion-doped TiO₂, the chromium-ion-doped TiO₂ demonstrated shorter diffusion lengths of minority carriers, accounting for its higher rate of hole–electron recombination. Another study suggested that chromium ions were mainly dispersed on the surface in the formation of the solid solution. The increase of Cr ions reduced the OH groups on the surface, which explains its lower activity at higher contents.¹⁰¹ Dopants such as Mg, Ce and V have also been investigated in TiO₂-based photocatalysts by Ileperuma et al.^102,103 In their work, catalysts were suspended in double-distilled water under irradiation. The results implied that the ammonia yield increased with increasing pH value. At the same time, higher nitrate content was detected in production.¹⁰² Similar studies involved doping TiO₂ with a relatively high concentration of 10 wt% Ce or V. The V-doped TiO₂ catalysts were shown to possess n-type semiconductor behavior at pH = 3, while the Ce-doped TiO₂ catalysts displayed p-type behaviour at pH = 12.5.¹⁰³ However, there was no substantial difference in the NH₃ yields of both systems.

In addition to the modification of metal doping, transition metals can also be loaded on the TiO₂ surface as a co-catalyst. For example, Miyami et al. have reported the enhancement of N₂ photoreduction using Ru-loaded TiO₂.¹⁰⁴ Khan et al. used a Ru(III) complex as a photosensitizer in a Pt–TiO₂ semiconductor.¹⁰⁵ Ru(III) complex could be excited and subsequently inject electrons into the conduction band of TiO₂ to form a Ru(IV) complex. A systematic study on noble metal-loaded TiO₂ was investigated by Ranjit et al.¹⁰⁶ By comparing four noble metal catalysts as co-catalysts of TiO₂, they found the order of photoactivity was Ru > Rh > Pd > Pt, which is closely associated with the strength of the M–H bond (M = noble metal). In other words, the incorporation of noble metals, which exhibit a high barrier for H₂ evolution, resulted in high NH₃ yields. In 2008, Linnik and Kisch designed a Ru-modified TiO₂ film with conducting glass as the substrate and humic acid as the sacrificial agent.¹⁰⁷ However, the activity of the catalysts was not greatly improved compared to other doping systems.

In 2000, Hoshino et al. designed a TiO₂/conducting polymer system with an organic/inorganic heterojunction.¹⁰⁸ The photogenerated carriers at the organic/inorganic interface led to the synthesis of ammonia. In their work, a composite of poly(3-methylthiophene) (P3MeT) and TiO₂ was irradiated under a fluorescent lamp, and NH₄⁺ClO₄⁻ needle crystals were obtained. A rigorous comparison with Schrauzer's method was performed in the following year, and a comparable ammonia production rate was obtained.¹⁰⁹ It was deduced that the separation of photogenerated charge carriers at the interface and the presence of P3MeT inhibited the accumulation of NH₃ in TiO_x. A further study was conducted by substituting TiO₂ crystal with its amorphous phase in 2007.¹¹⁰ These investigations undoubtedly provide a new concept and future direction of catalyst modification.

3.1.2. Other binary metal oxides. Many unremitting efforts have been devoted to the study of other metal oxides, such as iron oxide,^111–113 tungsten oxide,¹¹⁴ gallium oxide and bismuth monoxide.^115,116 Recent achievements have provided unprecedented ideas and vistas for investigations toward the application of metal oxides in N₂ photoreduction.

Most earlier studies suggested that bare Fe₂O₃ has no photocatalytic activity for N₂ cleavage.^67,98,117 It is worth noting that iron is useful in the Haber–Bosch process and also functions similarly to a nitrogen enzyme because of its good interaction with dinitrogen.^6,19,112 As expected, later studies have proven that ferric oxide is capable of photofixating nitrogen to ammonia.^111,113 However, in a pure form, neither Fe₂O₃ nor its reduced product Fe₃O₄ had N₂ photocatalytic activity. To overcome this bottleneck, Khader et al. successfully designed α-Fe₂O₃, which was effective in photo-activating nitrogen by partially reducing α-Fe₂O₃ to Fe₃O₄.¹¹¹ Fascinatingly, in the presence of 3 to 5 at% iron in the form of Fe(II) in the partially reduced Fe₂O₃, NH₃ was detected in an aqueous slurry of the catalyst under UV irradiation. Most recently, another comparison of Fe₂O₃ and TiO₂ was performed by Lashgari and Zeinalkhani.¹¹³ The 1 : 1 ratio nanocomposites of Fe₂O₃–TiO₂ and Pd-loaded Fe₂O₃–TiO₂ were taken into consideration. It is noteworthy that the rate of NH₃ production was in the order of TiO₂ < Fe₂O₃–TiO₂ < Pd/Fe₂O₃–TiO₂. Therefore, with the incorporation of Fe₂O₃ into TiO₂, the coupled heterojunction nanocomposites played a profound role in the N₂ photoreduction process owing to the extended photoabsorption of visible light in Fe₂O₃.

Furthermore, tungsten oxide with oxygen vacancies (WO_3−x) was reported to have poor activity in the gas phase.¹¹⁴ ZnO was reported to have photocatalytic activity for nitrogen fixation.¹¹⁸ In the case of commercial ZnO, the ammonia yield of unmodified ZnO was greater than that of Pt-loaded ZnO, which is consistent with a previous report by Miyama in 1980.¹⁰⁴ However, for ZnO prepared by means of wet etching or precipitation methods (Fig. 5a and b), the situation was reversed. For a wide band gap semiconductor, mesoporous β-Ga₂O₃ nanorods were employed in N₂ photoreduction by Zhao et al. in 2015.¹¹⁵ The photoactivity of β-Ga₂O₃ was ameliorated in the presence of different alcohols, including methanol, ethanol, and tert-butanol (TBA), which formed ˙CO₂⁻in situ. The reducing ability of ˙CO₂⁻ dramatically enhanced the reduction of N₂ to NH₃. Moreover, the presence of O₂ in the N₂ facilitated the formation of ˙CO₂⁻, which enhanced the reducing power (Fig. 5c). Very recently, a low-valent bismuth monoxide (BiO) without additional reducing agents was employed for N₂ photoreduction because the low-valent bismuth with empty 6d orbitals provided excellent N₂ chemisorption and activation centres (the synthesis procedure is shown in Fig. 5d). The N₂ was activated by three arranged Bi atoms by donating electrons to the 6d orbitals of Bi and accepting lone pairs of electrons from the three Bi atoms to its anti-bonding orbitals (σ*2p_x, π*2p_y and π*2p_z) (Fig. 3e).¹¹⁶

Fig. 5 HRTEM images of (a) the fusion of three prominent lattice planes along the wet etching ZnO grain boundaries and (b) a Pt cluster at the grain boundary. Reproduced with permission.¹¹⁸ Copyright 2010, American Chemical Society. (c) Possible electron transfer pathways for the conversion of N₂ to NH₃ over β-Ga₂O₃. Reproduced with permission.¹¹⁵ Copyright 2015, Royal Society of Chemistry. (d) Schematic of the synthesis process for BiO quantum dots. (e) A 1N₂–3Bi(II) side-on bond structure formed by electron sharing. Reproduced with permission.¹¹⁶ Copyright 2017, Royal Society of Chemistry.

In addition to the above-discussed TiO₂ photocatalyst, as reported in Section 3.1.1., ferric oxide has been shown to be a compelling and effective photocatalyst for N₂ photoreduction. Importantly, the narrow band gap of Fe₂O₃ guarantees its photocatalytic activity even in the visible light region.¹¹⁹ Although semiconductors with wide band gaps are favorable in the activation of N₂ as a result of their high redox abilities, they remain limited in their UV response. It is worth mentioning that bismuth-containing systems may be useful in N₂ fixation; this will be further corroborated in Section 3.3.

3.1.3. Ternary metal oxides. The development of iron-doped TiO₂ in N₂ photoreduction has invigorated interest in ternary metal oxides, such as iron titanate. At a high concentration of Fe, the excess Fe³⁺ will form pseudobrookite (Fe₂TiO₅) phase. This multiphasic compound was shown to have no photocatalytic activity.^117,120 Soria et al. also indicated that no NH₃ production was detected over pure Fe₂TiO₅ synthesized by a coprecipitation method.⁹⁸

However, there are some controversial opinions on the possible use of iron titanate in N₂ fixation, as reported in 2001.¹²¹ The dominant hypothesis for the conflicting results in Fe systems is mainly accredited to the different approaches to catalyst preparation.¹²² Using the sol–gel method, Rusina and his coworkers prepared iron titanate films with relatively high Fe contents by reacting iron chloride and tetraisopropyl titanate, leading to the successful development of a new Fe₂Ti₂O₇ phase.¹²¹ After 90 min of light irradiation (λ ≥ 320 nm), a maximum ammonia yield of 17 μM was obtained at 75% alcohol content when Fe/Ti was fixed at 1 : 1. The produced ammonia was subsequently oxidized to nitrite in the presence of ethanol. Meanwhile, in another report, Zhao et al. revealed that ethanol could prevent the oxidation of ammonia.⁹⁹ In a continuation of their work, further investigations on iron titanate films were conducted by Kisch and Linnik.¹²³ The electron transfer mechanism of N₂ fixation on the Fe₂Ti₂O₇ thin films involved a series of photochemical processes of nitrogen → diazene → hydrazine → ammonia → nitrate. In addition, the essential role of chloride ions was discussed. Although Schrauzer and Soria have noted that the generation of TiO₂ using chloride ions is unfavorable,^98,124 both the inhibiting and accelerating effects of iron chloride as the precursor were corroborated.

In addition to iron titanate systems, titanate-like SrTiO₃ and BaTiO₃ have also been investigated under UV irradiation for N₂ photoreduction. Judging from the results of the experiments, undoped SrTiO₃ and BaTiO₃ produced minimal NH₃ yields (0.82 and 1.74 μmol g_cat⁻¹ after 2 h of reaction, respectively). Their yields of NH₃ reached 5 to 5.2 μmol g_cat⁻¹ in 2 h when doped with RuO₂ and NiO.¹²⁵ In another paper, plasmonic-metal/semiconductor photocatalysts elicited great attention for their high light-harvesting properties.¹²⁶ Using gold nanoparticles (Au-NPs) as the plasmonic nanostructure, Oshikiri and his co-workers designed Au-NPs/Nb-SrTiO₃/Ru (Fig. 6a) and Au-NPs/Nb-SrTiO₃/Zr/ZrO_x (Fig. 6b–d) semiconductor photoelectrodes in 2014 and 2016, respectively.^127,128 Au-NPs were loaded on a 0.05 wt% niobium-doped strontium titanate (Nb-SrTiO₃) single crystalline substrate, and Ru or Zr film as a co-catalyst was deposited on the opposite side of Nb-SrTiO₃ (Fig. 6c and d). Zr was oxidized to ZrO_x. Ethanol was added as a sacrificial electron donor in the anodic chamber. In later reports, the production rates of Zr/ZrO_x and Ru systems were compared to evaluate the production selectivity. Interestingly, H₂ was the main product in the case of Ru systems, indicating that proton reduction is the primary reaction on the Ru co-catalyst. In contrast, in the case of Au-NPs/Nb-SrTiO₃/Zr/ZrO_x, the production rate of NH₃ was conspicuously higher because Zr and ZrO_x bind N more favorably than H.¹²⁹ However, its NH₃ production rate was still obviously lower than that of the electrosynthesis of ammonia.¹³⁰ Recently, sparked by the unique features of Mo element in the Mo-cofactor of nitrogenase enzymes, Hao et al. fabricated hydrogenated bismuth molybdate (H-Bi₂MoO₆) which was capable of reducing N₂ under sunlight.¹³¹ In this context, the hydrogenation reaction induced oxygen vacancies on Bi₂MoO₆ (Fig. 6e–h), endowing H-Bi₂MoO₆ with activity under visible light for N₂ fixation. Using air instead of pure nitrogen, the NH₃ production rate under sunlight was as high as 1.3 mmol g⁻¹ h⁻¹. This high nitrogen activation could not be attained if molybdenum was replaced with tungsten as a counterpart or if the hydrogenation process was omitted.

Fig. 6 (a) Schematic of the NH₃ photosynthesis device using an Au-NPs/SrTiO₃ photoelectrode. Reproduced with permission.¹²⁷ Copyright 2014, Wiley-VCH. (b) Diagram of NH₃ synthesis on the Au-NPs/Nb-SrTiO₃/Zr/ZrO_x photoelectrode. Cross-sectional view of the bright field scanning TEM (BF-STEM) images of (c) the Au-NPs/Nb-SrTiO₃ interface and (d) the Zr film deposited onto Nb-SrTiO₃. Reproduced with permission.¹²⁸ Copyright 2016, Wiley-VCH. (e–h) HRTEM images of H-Bi₂MoO₆ (the superficial and lattice disorders induced by defects are marked by parallel dashed lines and circles, respectively). Reproduced with permission.¹³¹ Copyright 2016, Wiley-VCH.

Judging from the existing literature, we can conclude that most of the investigations of ternary metal oxides are focused on titanate systems.^{121–123,125,127,128} Iron titanate exhibits relatively high photoactivity in N₂ fixation under UV irradiation, in a pure nitrogen flow, and in the presence of organic scavengers. On the other hand, bismuth molybdate has been reported to enable impressive NH₃ production rates after performing the hydrogenation process under mild conditions. However, for N₂ activation, strontium titanate does not reach the general level of other catalysts even after modification by doping or plasmon induction.

3.1.4. Hydrous oxides. As early as 1987, Tennakone and his co-workers discovered hydrous ferric oxide photocatalysts for N₂ reduction under visible light.¹³² They attested that hydrous ferric oxide was superior to TiO₂ in the case of N₂ fixation because of its exceptional nitrogen chemisorption and highly negative flat band potential.¹³³ Since then, hydrous oxides such as Cu₂O·xH₂O,¹³⁴ Sm₂O₃·xH₂O/V₂O₃·xH₂O and WO₃·H₂O have been systematically investigated as photocatalysts for N₂ reduction.^135,136

In the initial study by Tennakone et al., amorphous Fe₂O₃ (H₂O)_n was prepared by gradually adding KOH solution to FeCl₃ solution. After saturation with N₂, the solution was irradiated under a 100 W tungsten lamp (IR and UV filtered off). A maximum NH₃ yield of ca. 4 μM was detected after 40 min.¹³² In 1991, ultrafine particles of Fe(O)OH were prepared from photohydrolysis of iron(II) bicarbonate by the same author. Nitrate was also detected in the solution after prolonged irradiation. The initial quantum yield of NH₃ was about 10⁻², which is higher than that of hydrous ferric oxide (∼10⁻³).¹³³ Ileperuma et al. claimed that a hydrous ferric oxide-loaded bentonite system produced more NH₃ and less nitrate than the parent hydrous ferric oxide under UV irradiation.¹³⁷ In contrast with pure hydrous ferric oxide, a system of coprecipitating Fe(III) and Ti(IV) hydrous oxides (Ti to Fe ratio = ∼8%) was found to be more effective in the generation of NH₃, while the rate of nitrate production using this composite was comparatively low.¹³⁸ Hydrous ferric oxide was believed to play roles in nitrogen chemisorption and reduction, while TiO₂ provided hole accumulation centres. Although the authors revealed the predominant effect of the heterojunction interface in the mechanism of electron–hole separation, they still questioned the meaning of these concepts in a colloidal system. In 1992, the yield of vanadium-substituted hydrous ferric oxides was ameliorated under UV irradiation. V³⁺ sites demonstrated distinctive attributes in capturing photogenerated holes in the mixed hydroxide.¹³⁹ With the incorporation of V³⁺, a maximum NH₃ yield of about 200 μM was reached in about 24 h when the vanadium-to-iron ratio was at unity. The subsequent decrease of NH₃ concentration was ascribed to the inactivation of catalyst followed by the photocatalytic decomposition of NH₃. This reason for the decrease of ammonia was also suggested in 1987 because of the continuous generation of H₂ and the unchanged content of O₂.¹³² Conversely, Ileperuma et al. pointed out that NH₃ was oxidized to nitrate because NO₂⁻ and NO₃⁻ were detected in the final products along with a decrease in NH₃ yield.¹³⁷

Additionally, hydrous cuprous oxide was able to photoreduce N₂ to NH₃ after impregnation with cuprous chloride, which underwent sacrificial oxidation to form cupric chloride. The high ammonia yield presumably resulted from the separation of oxidation and reduction sites as well as the chemisorption of N₂ on Cu₂O·xH₂O, which served as reduction sites.¹³⁴ Pursuing the exciting idea in a previous study of coprecipitated hydrous oxides of Fe and Ti,¹³⁸ a coprecipitate of samarium(III) and vanadium(III) was subsequently explored.¹³⁵ The maximum NH₃ yield was obtained within 1 h when the V-to-Sm ratio was ∼1. When employing this composite, the ammonia production rate was strikingly higher than that of previously reported vanadium-substituted hydrous ferric oxide.¹³⁹ Very recently, highly efficient N₂ photoreduction was achieved over carbon-WO₃·H₂O (HWO-C) in water.¹³⁶ WO₃·H₂O was selected as the photocatalyst due to its extraordinary electron and proton conductivity. As a result, the rapid transfer of photogenerated electrons and water-derived protons to N₂ was enhanced. In this circumstance, carbon modification prominently enhanced the NH₃ production rate of HWO by increasing its surface activation and simultaneously promoting the separation and transport of charge carriers to retard the electron–hole recombination rate. A similar phenomenon in the use of other carbon-modified photocatalysts for potential energy applications has also been reported by our research group.¹⁴⁰ Therefore, it is evident that the modification of carbon plays a decisive role in suppressing the recombination of electron–hole pairs by prolonging the lifetime of charge carriers for enhanced photoactivity.

3.2. Metal sulfide-based materials

In addition to the large family of metal oxide-based photocatalysts, the exploration of metal sulfides has also experienced a tremendous upsurge in the scientific community, especially in the field of photocatalysis.^141–143 The narrow band gap of metal sulfides is conducive to strong absorption of visible light, which results in highly efficient solar utilization efficiency.

In 1980, Miyama et al. employed CdS semiconductors for the fixation of nitrogen under UV irradiation; the recorded yield of NH₃ in 5 h was 10.67 μmol g_cat⁻¹.¹⁰⁴ They constructed CdS/Pt binary wafered catalysts whose NH₃ yield was drastically higher than that of pristine CdS. The remarkable performance of the loading of Pt co-catalysts has heightened research interest in other potential modifications of CdS photocatalysts. In 1988, Khan et al. reported the fixation of N₂ with a CdS/Pt/RuO₂ semiconductor particulate system under the illumination of visible light (λ = 505 nm).¹⁴⁴ Dinitrogen was activated to react with [Ru(Hedta)H₂O]⁻ to form the Ru(II) dinitrogen complex. The holes in the valence band trapped the electrons released from RuO₂. Two years later, a similar research team conducted the stepwise reduction of dinitrogen to ammonia on visible-light-responsive Pt/CdS·Ag₂S/RuO₂ photocatalysts.¹⁴⁵ The addition of silver facilitated the migration of electrons from the conduction band to minimize the photocorrosion of CdS. The production rate of NH₃ was twice that in the absence of silver. In view of the obtained results, it is clear that noble metals such as Pt, Ag and Ru work well with CdS. The decrease in the yield of NH₃ over time may originate from the photocorrosion of CdS to S and Cd²⁺.¹⁴⁴ To retain the high photocatalytic activity and stability of CdS for the reduction of N₂ to NH₃, in 2017, Ye et al. demonstrated the fixation of N₂ using Cd_0.5Zn_0.5S solid solution for the first time.¹⁴⁶ To enhance the photocatalytic performance, a transition metal phosphide (Ni₂P) was employed as a co-catalyst to rapidly transfer the photoinduced electrons to Ni₂P via well-contacted heterointerfaces to diminish the charge recombination.

On the other hand, owing to the recent interdisciplinary interest in material science and biology, organic-sulfide catalysts have also been designed for enhanced N₂ fixation activity. In 2016, Brown et al. reported nitrogen reduction by the MoFe protein, which is the active site of nitrogenase, adsorbed onto CdS nanorods to form biohybrid complexes.¹⁴⁷ The NH₃ production rate reached 315 nmol per mg MoFe protein per min under visible light. In this regard, CdS nanocrystals were used to photosensitize the MoFe protein so that the photon energy was derived from ATP (Fig. 7a). Furthermore, the rates were comparable to that of physiological NH₃ production by nitrogenase with energy provided by ATP. Because the active sites of nitrogenase contain the elements Fe, Mo and S, a synthetic complex of Fe, Mo and S could be taken into account. By exploiting the advantage of the FeMo cofactor in nitrogenases, Banergee et al. inferred that biomimetic chalcogels comprising FeMoS inorganic clusters could reduce N₂ to NH₃ in aqueous media under light.¹⁴⁸ The Fe₂Mo₆S₈ chalcogel was an analog to the MoFe active site in the enzyme and was linked by Sn₂S₆ ligands to form an amorphous network (Fig. 7b). This study proved that structural analogues of nitrogenase can be functional and can even confer better properties than nitrogenase. In 2016, to make a better mimic of nitrogenases, not only MoFe protein but also Fe protein was considered by Liu et al.¹⁴⁹ They employed a chalcogel system consisting of Fe₂Mo₆S₈(SPh)₃ and Fe₃S₄ biomimetic clusters linked by Sn₂S₆. Iron was believed to be more vital than molybdenum for the light-driven reduction of nitrogen due to the fact that a weak bonding orbital between nitrogen and iron was manifested through localized orbital analysis. More importantly, their conclusion that iron is a better active site for N₂ binding than Mo has also been revealed by recent biochemical and spectroscopic data.^19,150

Fig. 7 (a) Reaction scheme for N₂ reduction to NH₃ by CdS:MoFe protein biohybrids. Reproduced with permission.¹⁴⁷ Copyright 2016, American Association for the Advancement of Science. (b) Schematic of the composition of Mo₂Fe₆S₈–Sn₂S₆ biomimetic chalcogel. Reproduced with permission.¹⁴⁸ Copyright 2015, American Chemical Society. (c) HRTEM image of g-C₃N₄/ZnMoCdS. (d) Schematic of the charge transfer process at the heterojunction interface of g-C₃N₄/ZnMoCdS. Reproduced with permission.¹⁵⁵ Copyright 2016, Royal Society of Chemistry. (e) Schematic of the trion-induced multi-electron N₂ reduction process. Reproduced with permission.⁷¹ Copyright 2017, Elsevier.

Motivated by the successful fixation of N₂ under visible light with oxygen vacancies on the surface of BiOBr,¹⁵¹ a number of researchers wondered if sulfur vacancies could eventually promote photoactivity. Inspired by this idea, Hu et al. obtained Zn_0.1Sn_0.1Cd_0.8S with sulfur vacancies formed by multi-component metal sulfides as catalysts for reducing N₂ under visible light.¹⁵² Surprisingly, the sulfur vacancies introduced chemical adsorption sites on the surface, which aided the activation of N₂ molecules by extending the bond distance between the nitrogen atoms in N₂. It was also implied that the generation rate of NH₃ was linearly related to the concentration of sulfur vacancies. Additionally, the sulfur vacancies also trapped electrons, thus promoting the separation of photoinduced electrons and holes. In the same year, a similar quaternary system of Mo_0.1Ni_0.1Cd_0.8S, which contributed to the reduction of N₂ under visible light, was published by Cao et al.¹⁵³ At the same time, tri-component metal sulfides were studied by this group. Two similar heterostructures of g-C₃N₄/ZnSnCdS and g-C₃N₄/ZnMoCdS were employed for N₂ photoreduction.^154,155 In the hybrid of g-C₃N₄/ZnMoCdS (Fig. 7c), a tight heterojunction coupling between g-C₃N₄ and ZnMoCdS was pivotal for efficient charge transfer. The photogenerated electrons were transferred from g-C₃N₄ to the metal sulfide, while the holes were transported in the opposite direction (Fig. 7d).¹⁵⁵ This clearly accounted for the improved visible light absorption after hybridization. Furthermore, the redistribution of electrons and holes on each side of the heterojunction established an internal electric field to impede the recombination of charge carriers.

As a transition metal dichalcogenide (TMD), it is well-known that MoS₂ possesses excellent electrical,¹⁵⁶ optical,¹⁵⁷ and optoelectronic properties.¹⁵⁸ In 2017, Sun et al. reported photocatalytic N₂ reduction to NH₃ with ultrathin MoS₂.¹⁵⁹ In this paper, the photocatalytic performances of MoS₂ samples under different preparation conditions were studied; it was found that the sonicated ultrathin MoS₂ sample generated a large amount of NH₃ with favorably high stability. In ultrathin TMDs, the tightly bound excitons could facilely capture additional electrons to form charged excitons with more than two electrons in one bound state.¹⁶⁰ Thus, it is believed that these charged excitons functioned as electron-rich species to facilitate the multi-electron reduction process of molecular N₂ (Fig. 7e). Particularly, a six-electron transfer profess was responsible for the acceleration of photocatalytic N₂ reduction to NH₃.

3.3. Bismuth oxyhalides

Bismuth oxyhalides, BiOX (X = Cl, Br, and I), have recently become well known for their superior optical properties; they are also propitious for industrial applications, namely the photodecomposition of organic pollutants.¹⁶¹ The layered structure of BiOX provides adequate space for the polarization of atoms, and the as-formed internal electric field will contribute to the efficient separation and transfer of charge carriers.^162,163

In recent years, BiOX has also been reported for applications in solar nitrogen fixation.^{61,86,88,151,164} In 2015, Li et al. employed {001}-faceted BiOBr nanosheets with oxygen vacancies (OVs) to reduce N₂ under visible light.¹⁵¹ In this scenario, the N₂ fixation rate was estimated to be 104.2 μmol h⁻¹ (per gram of BOB-001-OV). Remarkably, it was a quantum leap for N₂ fixation under visible light in the absence of organic scavengers. Theoretical simulations revealed that the OVs of BiOBr could elongate the N≡N triple bond of adsorbed N₂ from 1.078 to 1.133 Å (Fig. 8a), which subsequently promoted the activation of N₂ molecules. Moreover, the large number of OVs on the surface of BiOBr formed a defect state lying near the bottom of the conduction band of the photocatalyst, hampering the recombination of electron–hole pairs. However, the surface OVs of BiOBr are naturally oxidized, leading to reduced activity. Li's group also examined the photocatalytic activity of BiOCl with OVs on different facets.⁸⁸ They claimed that the kinetics and mechanism of N₂ fixation varied due to the dissimilar facets of BiOCl nanosheets. They highlighted that nitrogen fixation with OVs on the {001} facets followed a distal pathway (N₂ → N-NH₃ → N + NH₃ → 2NH₃), whereas the reaction on the {010} facets followed an alternative pathway (N₂ → N₂H₃ → N₂H₄) (Fig. 8b). The quantum yields were 1.8% h⁻¹ and 4.3% h⁻¹ for OVs on the {001} and {010} facets of BiOCl, respectively, under UV light (λ = 254 nm). In the most recent work by Ye's group, ultrafine light-switchable OVs of Bi₅O₇Br nanotubes with a diameter of ca. 5 nm and large exposed surface sites were synthesized via a water-assisted low-temperature wet chemical approach (Fig. 8c–e).¹⁶⁴ The NH₃ generation rate of nanotubes (1.38 mmol h⁻¹ g⁻¹) was 2.5 times higher than that of nanosheets, with an apparent quantum efficiency (AQE) as high as 2.3% at 420 nm.

Fig. 8 (a) The adsorption geometry and the charge density difference of N₂ on the OVs of the BiOBr (001) surface (the yellow and blue isosurfaces represent charge accumulation and depletion, respectively). Reproduced with permission.¹⁵¹ Copyright 2015, American Chemical Society. (b) Free energy profiles of OV-mediated N₂ fixation on the (001) and (010) surfaces of BiOCl. Reproduced with permission.⁸⁸ Copyright 2016, Royal Society of Chemistry. (c and d) TEM images of Bi₅O₇Br nanotubes at different magnifications. (e) Schematic of the synthesis procedure of ultrafine Bi₅O₇Br nanotubes. Reproduced with permission.¹⁶⁴ Copyright 2017, Wiley-VCH.

As a continuation of the previous research by Zhang's and Ye's groups,^88,164 Bai et al. utilized bismuth-rich bismuth oxyhalide (Bi₅O₇I) nanosheets with different dominant {001} and {101} facets.⁸⁹ The N₂ fixation rates of Bi₅O₇I-001 and Bi₅O₇I-100 were 111.5 μmol L⁻¹ h⁻¹ and 47.6 μmol L⁻¹ h⁻¹, respectively. After 5 cycles, the photocatalytic properties and structures of Bi₅O₇I-001 and Bi₅O₇I-100 elucidated high stabilities for the N₂ fixation process. The more negative conduction band potential of Bi₅O₇I-001 (−1.45 V) compared to that of Bi₅O₇I-100 (−0.85 V), explains the higher NH₃ production rate of Bi₅O₇I-001. In short, although most literature studies place significant emphasis on the exposure of different facets, it is conceivable that the combined effects of OVs and facet-dependent studies will open a new outlook and provide inspiration for the development of advanced photocatalysts for N₂ photoreduction. Additionally, the combination of experimental results and theoretical simulations is highly necessary to fully elucidate the mechanism of N≡N triple bond activation and the NH₃ formation pathway. Thus, more emphasis on facet-controlled and vacancy-mediated bismuth oxyhalides should be devoted in the future to accentuate the scientific aspects and unravel the exact reaction steps for N₂ photofixation.

3.4. Carbonaceous materials

Metal-free semiconductor photocatalysts have triggered a renaissance of great interest since the advent of H-terminated boron-doped (B-doped) diamond as a solid-state source of electrons in water for photo-driven N₂ fixation in 2013.¹⁶⁵ It was demonstrated that the UV excitation of B-doped diamond induced electron emissions and their ejection into liquids to produce solvated electrons which reacted quickly with protons to form neutral atomic hydrogen, H˙, and finally facilitated the subsequent reaction with N₂ to yield NH₃. In this context, the photocatalytic activity decreased with time because the oxidation of the H-terminated surface to O-terminated resulted in a loss of electron affinity. The authors also revealed that photocatalytic behaviour could be observed using the natural diamond dispersed in water. Further study of the mechanism of action of aqueous solvated electrons was methodically reported in the following year.¹⁶⁶ Three possible reaction steps, including (1) electron transfer (N₂ + e⁻ → N₂⁻), (2) proton-coupled electron transfer (N₂ + e⁻ + H₂O → N₂H + OH⁻) and (3) hydrogen atom addition (N₂ + H˙ → N₂H), were proposed with the help of kinetic modelling. It was concluded that the reduction of N₂ to NH₃ involved hydrogen atom addition at the initial steps and protonation/electron transfer by solvated electrons at the later steps. In 2016, metal-diamond heterostructures (i.e. diamond thin films grown on Mo, Ni and Ti metal substrates) were reported. It was noted that some electrons excited from the metal substrates would be injected into the conduction band of the diamond film, followed by emission into water.¹⁶⁷

In the present literature, most research has focused a spotlight on carbonaceous nanomaterials. Benefiting from the rational significance of the oxygen vacancies in bismuth oxyhalides, as described in Section 3.3.,¹⁵¹ graphitic carbon nitride (g-C₃N₄) with nitrogen vacancies (NV-g-C₃N₄) has become a focal area in materials science since 2015.⁷³ Because they have the same shape and size as the nitrogen atoms in dinitrogen, nitrogen vacancies (NVs) are beneficial in the selective chemisorption and activation of N₂. This explains why the nitrogen fixation rate remained unchanged when N₂ was replaced by air as the N₂ source. In addition, NVs prominently improved the separation of charge carriers by trapping photogenerated electrons and promoting electron transfer to the adsorbed N₂.⁶² Following this astounding discovery, NV-g-C₃N₄ was extensively investigated based on manifold aspects, namely the modification of preparation routes for large-scale production and the enlargement of the specific surface area.^168–171 Furthermore, modifications such as iron doping and ruthenium loading have been exhaustively explored in recent years.^172,173 Similar to diamond, the H-termination of Ru-loaded g-C₃N₄ played an indispensable role in the activation of N₂.¹⁷⁰

Recently, g-C₃N₄ has been widely employed to construct heterostructure junctions with another component, including the hybridization of g-C₃N₄ with metal sulfides,^154,155 metal oxides,^174–176 and reduced graphene oxide (rGO).¹⁷⁷ As a proof of concept, multi-metal oxides with tunable band structures match favorably with other semiconductors when engineering intact heterojunction interfaces.¹⁷⁸ After assembly, the g-C₃N₄/MgAlFeO nanorod heterostructure was effective in N₂ photoreduction. In particular, a Z-scheme heterojunction was established, as depicted in Fig. 9a. In this instant, the photoexcited electrons in the conduction band of MgAlFeO were transferred to the valance band of g-C₃N₄. Due to this phenomenon, the electrons in the conduction band of g-C₃N₄ did not recombine with the holes, instead participating in the N₂ reduction.¹⁷⁴ A similar Z-scheme heterojunction mechanism for N₂ fixation was reported in other studies: (1) W₁₈O₄₉/g-C₃N₄ catalyst (Fig. 9b and c) was active under near-infrared (NIR) irradiation¹⁷⁵ because of the coherent oscillations of the surface free electrons induced by the oxygen vacancies on W₁₈O₄₉. (2) 3,4-Dihydroxybenzaldehyde-functionalized Ga₂O₃/graphitic carbon nitride (Ga₂O₃-DBD/g-C₃N₄), synthesized by incorporating Ga₂O₃-DBD NPs into g-C₃N₄ networks (Fig. 9d),¹⁷⁶ was reported to be capable of converting N₂ to NH₃ with the aid of the strong reducing power of ˙CO₂⁻, which was formed by the oxidation of methanol by active oxygen species (Fig. 9e). In essence, this is similar to the N₂ reduction mechanism of β-Ga₂O₃, as discussed earlier in Fig. 5c.

Fig. 9 (a) Tentative Z-scheme heterojunction interface of g-C₃N₄/MgAlFeO. Reproduced with permission.¹⁷⁴ Copyright 2017, Royal Society of Chemistry. (b) SEM and (c) HRTEM images of WO₁₈O₄₉/g-C₃N₄. Reproduced with permission.¹⁷⁵ Copyright 2017, Royal Society of Chemistry. (d) The formation process and (e) photocatalytic N₂ reduction mechanism of Ga₂O₃-DBD/g-C₃N₄. Reproduced with permission.¹⁷⁶ Copyright 2017, Elsevier. (f) Possible paths of photo-generated electrons from graphene for the activation of N₂ molecules. Reproduced with permission.¹⁸³ Copyright 2016, American Chemical Society. (g) Schematic of the synthesis procedure of the Fe-Al@3DG and (h) NH₃ synthesis over Fe-Al@3DG. Reproduced with permission.¹⁸⁴ Copyright 2017, Elsevier.

As for two-dimensional (2D) graphene, a metal-free 2D/2D hybrid heterostructure of g-C₃N₄/rGO was designed by coupling rGO and protonated g-C₃N₄, in which rGO served as a conducting substrate.¹⁷⁷ Protonated g-C₃N₄ exhibited 5 times the NH₃ production rate of pristine g-C₃N₄, and the activity was further ameliorated after assembly with rGO. It was deduced that the face-to-face contact area on the g-C₃N₄/rGO heterointerface for enhanced charge transfer substantially decreased the electron–hole recombination rate of g-C₃N₄.¹⁷⁹ In essence, this finding is analogous to our previous studies on metal-free 2D/2D graphene/g-C₃N₄ for CO₂ photoreduction, in which graphene acts as an excellent electron reservoir.^180–182 There is another report on the loading of iron oxide on bulk three-dimensional (3D) cross-linked graphene (Fe@3DG) for the reduction of N₂ to NH₃ under light irradiation.¹⁸³ Graphene played the role of generating highly energetic electrons. A significant enhancement of photocatalytic activity was observed by increasing the electron density of Fe due to its electron donation from the Fe d orbital to the anti-bonding orbital of N₂ (Fig. 9f).

Because the photoactivity of the Fe@3DG nanocomposite significantly decreased after 5 hours of reaction, in 2017, the same research group modified the existing Fe@3DG nanocomposite by incorporating Al₂O₃ to form Fe-Al@3DG photocatalysts (Fig. 9g).¹⁸⁴ Al₂O₃ functioned as an excellent dispersing agent and barrier to prevent the aggregation of Fe₂O₃ nanoparticles on the 3D graphene (Fig. 9h). Fascinatingly, the NH₃ production yield of 20Fe-2Al@3DG (20 wt% Fe and 2 wt% Al) remained steady even after 60 hours of light-driven reaction, which was more than two times that of the parent 20Fe@3DG (20 wt% Fe), demonstrating the salient role of Al₂O₃ as a structural promoter. In general, the employment of 2D/2D heterojunction interfaces and 3D nanoarchitecture design for photocatalysis based on the motivation of effective charge carrier transfer and separation is still in the infant stage. Furthermore, most existing literature studies focus on 0D/2D hybrid nanostructures. Therefore, extensive studies in this related field are imperative to achieve excellent prospects in addressing the limitations of this process for scientific merit and to enhance fundamental research knowledge toward realizing industrialization in the future.

3.5 Other potential materials

Gallium phosphide is a typical semiconductor that has been researched in the photocatalytic field for a long time.^185,186 In 1978, Dickson et al. employed a photoelectrochemical cell to perform N₂ fixation.¹⁸⁷ The system consisted of a p-GaP cathode and an aluminum metal anode immersed in a non-aqueous electrolyte of titanium tetraisopropoxide and AlCl₃ dissolved in glyme. In this system, the p-GaP electrode absorbed light and provided the activation energy for the photoenhanced reduction of N₂. In 1980, Miyama et al. reported the performance of GaP and GaP/Pt for the reduction of N₂. In fact, a few researchers have already implemented the reduction of N₂ on GaP nanoparticles without light.¹⁰⁴

Zeolites have also been applied in the photochemical synthesis of NH₃ from N₂. Khan et al. reported the production of ammonia with titanium-exchanged zeolites under visible light.¹⁸⁸ Two years later, they tested the exchange of different types of zeolites with titanium ions. It is noteworthy that the Ti³⁺ ions contributed to the reduction of N₂ for the oxidation of Ti³⁺ to Ti⁴⁺.¹⁸⁹ Furthermore, the NH₃ production was reported to increase with increasing re-exchange time of the zeolites by Ti³⁺.

As mentioned above,^127,128 localized surface plasmon resonance (LSPR) of metal particles has also been utilized to drive the photoabsorption of other catalysts. In 2014, Zeng et al. applied catalytically active metal (Os) directly to LSPR Au particles (Os–Au nanocomposites) for the NH₃ synthesis reaction.¹⁹⁰ It was found that the Au nanoparticles with superior LSPR absorbance and photoabsorption did not show any catalytic activity, while Os nanoparticles with high catalytic activity could not absorb photoenergy. By allying the concept of LSPR Au particles with the active metal (Os), the Os–Au nanohybrids outperformed the individual components, thus improving the solar light utilization for the chemical synthesis of NH₃. In 2016, a solar-driven photoelectrochemical cell based on plasmon-enhanced black silicon (bSi) was documented by MacFarlane's group.¹⁹¹ The plasmon-enhanced bSi was decorated with Au nanoparticles (GNPs) and a layer of Cr (Fig. 10a–c). In the GNP/bSi/Cr photoelectrochemical cell, bSi, GNPs and Cr performed as photoabsorber, reduction catalysis sites and hole-sink layer, respectively. The bSi ameliorated the scattering and absorption of light and provided a sufficient surface area for the decoration of a large amount of GNPs. From the charge transfer mechanism, the GNPs accumulated photogenerated electrons, whereas Cr acted as a sacrificial anode to gather holes. This research resulted in hundreds of times improvement in the NH₃ production rate compared with plasmon-induced photoelectrodes (Au NPs/Nb-SrTiO₃/Ru and Au NPs/Nb-SrTiO₃/Zr/ZrO_x), as mentioned in Section 3.1.3.^127,128

Fig. 10 (a) TEM and (b) HRTEM images of a GNPs-coated p-type silicon nanowire. (c) Schematic of the fabrication of the GNP/bSi/Cr photoelectric cell. Reproduced with permission.¹⁹¹ Copyright 2016, Nature Publishing Group. (d) TEM image of the 5 wt% Ru-modified GaN NWs (inset plot shows the diameter distributions of the Ru sub-nanoclusters). (e) SEM image of the developed InGaN/GaN nanowires on the Si (111) substrate. (f) Schematic of the InGaN/GaN nanowire structure. (g) Rate of NH₃ generation over various III-nitride semiconductors under visible light irradiation (λ > 400 nm). Reproduced with permission.¹⁹² Copyright 2017, Wiley-VCH.

A similar one-dimensional (1D) structure of gallium nitride nanowires (GaN NWs) grown on a silicon (Si) substrate was very recently designed by Li and co-workers.¹⁹² The Ge-doped n-type GaN NWs exhibited improved N₂ photofixation activity under UV irradiation with the aid of ultrasmall Ru clusters of ca. 0.8 nm (Fig. 10d). After the incorporation of Ru, the photoexcited electrons were facilely transferred from the III-nitride semiconductors to Ru as a result of an effective interfacial metal/semiconductor Schottky junction with a barrier height of 0.94 eV, thus demonstrating the role of Ru as an electron reservoir to facilitate the cleavage of N≡N triple bonds. Furthermore, the band gap was shifted to the visible light region (2.34 eV) by introducing In (ca. 25% composition) into GaN NWs, forming 5 segments of n-type InGaN NWs on the GaN NW template (Fig. 10e and f). Assisted by Ru sub-nanoclusters, the photoactivity of n-type InGaN/n-GaN toward NH₃ synthesis was the highest compared to other reference samples under visible light irradiation (λ > 400 nm) (Fig. 10g). For ease of comparison of the selected literature studies, Tables 2 and 3 summarize most of the photocatalysts, including metal oxide-based, metal sulfide-based, bismuth oxyhalides (BiOX), and metal-free carbonaceous nanomaterials.

Table 2 Summary of metal oxide-based photocatalysts for the reduction of N₂ to NH₃(I)

	Year	Catalyst	T (K)	Light source	Nitrogen source	Organic scavenger	NH3 average rNH3	Author	Ref.
a The unit mM h^−1 cm^−2 is not standardized because the volume of the solution was not given in the literature. b The catalyst cannot be weighed because it denatures upon drying. c rt denotes room temperature. d ϕ = apparent quantum yield. e 2-PrOH denotes 2-propanol. f EDTA denotes ethylenediaminetetraacetic acid. g TBA denotes tert-butanol.
TiO2-Based Materials	1997	0.2 wt% Fe-doped TiO2	313	390–420 nm	N2	None	10 μmol g^−1 h^−1	Schrauzer	67
	1988	TiO2	500	UV	N2	None	0.83 μmol g^−1 h^−1	Bourgeois	91
	1991	0.5 wt% Fe-doped TiO2	353	UV	N2	None	6 μmol g^−1 h^−1	Soria	98
	2014	Fe-doped TiO2 (Fe/Ti = 10^−4)	298	λ = 254 nm	Air	Ethanol	1 gcat L^−1; 400 μM h^−1	Zhao	99
	2017	JRC-TIO-6 (rutile)	313	λ > 254 nm	N2	2-PrOH^e	1 gcat L^−1; 2.5 μM h^−1	Hirakawa	92
	1977	0.4 wt% Co-doped TiO2	313	UV	N2	None	6.3 μmol g^−1 h^−1	Schrauzer	67
	1977	0.4 wt% Mo-doped TiO2	313	UV	N2	None	6.7 μmol g^−1 h^−1	Schrauzer	67
	1977	0.4 wt% Cr-doped TiO2	313	UV	N2	None	0.37 μmol g^−1 h^−1	Schrauzer	67
	1988	0.5 wt% Cr-doped TiO2	353	UV	N2	None	2.6 μmol g^−1 h^−1	Palmisano	100
	1990	2 wt% Mg-doped TiO2	—	UV	N2	None	0.67 gcat L^−1; 6.9 μM h^−1	Ileperuma	102
	1993	10 wt% V-doped TiO2	—	UV	N2	None	0.8 gcat L^−1; 4.9 μM h^−1	Ileperuma	103
	1993	10 wt% Ce-doped TiO2	—	UV	N2	None	0.8 gcat L^−1; 3.4 μM h^−1	Ileperuma	103
	1996	0.24 wt% Ru-loaded TiO2	—	UV-vis	N2	Ethanol	22.7 μmol g^−1 h^−1	Ranjit	106
	2001	TiO2/P3MeT	293	UV	N2	None	—	Hoshino	109
	2008	5% RuCl3-modified TiO2	rt^c	UV-vis	N2	Humic acid	4 μM h^−1 cm^−2^a	Linnik	109
Other binary oxides	1987	Partially reduced Fe2O3	303	UV-vis	N2	None	10 μmol g^−1 h^−1	Khader	111
	2010	Pt-loaded ZnO	—	UV	N2	Na-EDTA^f	860 μmol g^−1 h^−1	Janet	118
	2015	Mesoporous β-Ga2O3 nanorods	298	λ = 254 nm	N2/O2	TBA^g	ϕ = 36.1% ^d	Zhao	115
	2017	Fe2O3	298	UV-vis	N2	Ethanol	0.5 gcat L^−1; 1362.5 μM h^−1	Lashgari	113
	2017	BiO quantum dots	298	UV-vis	N2	None	0.25 gcat L^−1; 50.5 μM h^−1	Sun	116
Ternary metal oxides	1983	BaTiO3	313	UV	N2	None	0.87 μmol g^−1 h^−1	Li	125
	1983	RuO2–NiO–BaTiO3	313	UV	N2	None	2.6 μmol g^−1 h^−1	Li	125
	2001	Iron titanate films (Fe/Ti = 1 : 1)	—	λ ≥ 320 nm	N2	Ethanol	0.57 μM h^−1 cm^−2^a	Rusina	121
	2014	Au NPs/Nb-SrTiO3/Ru	rt^c	550–800 nm	N2	Ethanol	1.1 nmol h^−1 cm^−2	Oshikiri	127
	2016	Au NPs/Nb-SrTiO3/Zr/ZrOx	rt^c	550–800 nm	N2	Ethanol	6.5 nmol h^−1 cm^−2	Oshikiri	128
	2016	H-Bi2MoO6	rt^c	Sunlight	Air	None	1300 μmol g^−1 h^−1	Hao	131
Hydrous oxides	1987	Fe2O3(H2O)n	—	Vis	N2	None	6 μM h^−1^b	Tennakone	132
	1988	Hydrous oxides of Fe(III) and Ti(IV) (Ti : Fe = 8%)	—	Vis	N2	None	22 μM h^−1^b	Tennakone	138
	1989	Cu2O·xH2O·CuCl	—	UV	N2	None	70 μM h^−1^b	Tennakone	134
	1991	Fe(O)OH	—	UV	N2	None	9.25 μM h^−1^b	Tennakone	133
	1991	Bentonite-hydrous ferric oxide	—	UV	N2	None	2 gcat L^−1; 1.33 μM h^−1	Ilepruma	137
	1992	V(III)-substituted hydrous ferric oxide	299	UV	N2	None	8.33 μM h^−1^b	Tennakone	139
	1993	Sm2O3·xH2O/V2O3·xH2O	299	UV	N2	None	100 μM h^−1^b	Tennakone	135
	2016	Carbon-tungstic acid hybrids	—	UV-vis	N2	None	205 μmol g^−1 h^−1^b	Li	136

---

Table 3 Summary of other types of photocatalysts for the reduction of N₂ to NH₃(II)

	Year	Catalyst	T (K)	Light source	Nitrogen source	Organic scavenger	NH3 average rNH3	Author	Ref.
a The catalyst cannot be weighed because it denatures upon drying. b rt denotes room temperature. c EDTA denotes ethylenediaminetetraacetic acid. d Using optical filters, light of λ > 800 nm and λ < 800 nm was removed for the 300 W Xenon lamp and 200 W infrared light source, respectively, to simulate full-spectrum light.
BiOX	2015	BiOCl	298	UV-vis	N2	Methanol	0.67 gcat L^−1; 46.2 μM h^−1	Li	88
	2015	BiOBr	298	Vis	N2	None	0.5 gcat L^−1; 1042 μM h^−1	Li	151
	2016	Bi5O7I	293	UV-vis	N2	Methanol	0.5 gcat L^−1; 111.5 μM h^−1	Bai	89
	2017	Bi5O7Br nanotubes	rt^b	λ > 400 nm	N2	None	1380 μmol g^−1 h^−1	Wang	164
Metal sulfides	1980	CdS/Pt	311	UV	N2	None	3.26 μmol g^−1 h^−1	Miyama	104
	1988	CdS/Pt/RuO2	303	λ ≥ 505 nm	N2	None	4 gcat L^−1; 620 μm h^−1	Khan	144
	1990	Pt/CdS·Ag2S/RuO2	303	UV-vis	N2	None	2 gcat L^−1; 1260 μM h^−1	Khan	145
	2015	[Mo2Fe6S8(SPh)3]^3+-[Sn2S6]^4−	rt^b	UV-vis	N2	—	10.1 μM h^−1^a	Banerjee	148
	2016	Mo2Fe6S8(SPh)3-Fe4S4-[Sn2S6]^4−	rt^b	UV-vis	N2	—	18.82 μM h^−1^a	Liu	149
	2016	CdS/MoFe protein	—	λ = 405 nm	—	None	315 μmol gprotein^−1 min^−1	Brown	147
	2016	Zn0.1Sn0.1Cd0.8S	303	400–800 nm	N2	Ethanol	0.4 gcat L^−1; 105.2 μM h^−1	Hu	152
	2016	Mo0.1Ni0.1Cd0.8S	303	400–800 nm	N2	Ethanol	0.4 gcat L^−1; 71.2 μM h^−1	Cao	153
	2016	g-C3N4/ZnSnCdS	303	400–800 nm	N2	Ethanol	0.4 gcat L^−1; 167.6 μM h^−1	Hu	154
	2016	g-C3N4/ZnMoCdS	298	400–800 nm	Air	Ethanol	0.4 gcat L^−1; 77.6 μM h^−1	Zhang	155
	2016	Ni2P/Cd0.5Zn0.5S	293	λ ≥ 400 nm	N2	None	0.4 gcat L^−1; 101.5 μM h^−1	Ye	146
	2016	MoS2	298	λ ≥ 420 nm	N2	None	325 μmol g^−1 h^−1	Sun	159
Carbonaceous materials	2013	Boron-doped diamond	—	UV	N2	None	—	Zhu	165
	2015	g-C3N4	—	λ ≥ 420 nm	Air	Methanol	1 gcat L^−1; 160 μM h^−1	Dong	62
	2016	g-C3N4/rGO	303	400–800 nm	Air	Na-EDTA^c	0.4 gcat L^−1; 206 μM h^−1	Hu	177
	2016	Fe-loaded 3D graphene	473	UV	N2/H2	None	24 μmol g^−1 h^−1	Lu	183
	2017	Fe-Al-loaded 3D graphene	473	UV	N2/H2	None	25.3 μmol g^−1 h^−1	Yang	184
	2017	Fe-doped g-C3N4	303	400–800 nm	N2	Ethanol	0.4 gcat L^−1; 120 μM h^−1	Hu	172
	2017	g-C3N4/MgAlFeO	303	400–800 nm	N2	Ethanol	0.4 gcat L^−1; 166.8 μM h^−1	Wang	174
	2017	W18O49/g-C3N4	—	UV-vis-NIR^d	N2	Ethanol	0.4 gcat L^−1; 57.8 μM h^−1	Liang	175
	2017	Ga2O3-DBD/g-C3N4	—	UV-vis	N2	Ethanol	0.4 gcat L^−1; 112.5 μM h^−1	Cao	176
Other system	1980	GaP/Pt	311	UV	N2	None	5 μmol g^−1 h^−1	Miyama	104
	1981	Ti^3+-exchanged zeolites	—	Vis	N2	None	20.6 μmol g^−1 h^−1	Khan	188
	1983	Ti^3+-exchanged zeolites	∼305	Vis	N2	None	32.9 μmol g^−1 h^−1	Khan	189
	2015	Cs2O-promoted Os–Au	333	λ ≥ 450 nm	N2/H2	None	2685 μmol g^−1 h^−1	Zeng	190
	2016	GNP/bSi/Cr	—	UV-vis	N2	None	0.78 μmol h^−1 cm^−2	Ali	191
	2017	5% Ru@n-GaN NWs	283	290–380 nm	N2	None	120 μmol g^−1 h^−1	Li	192

---

4. Concluding remarks

The photoreduction of N₂ to NH₃ is regarded as a scientifically challenging yet environmentally friendly technology for the sustainable growth of the human population. NH₃ has elicited much research fascination and broad interdisciplinary attention as a hydrogen carrier in addition to its wide use as an industrial raw material and fertilizer.^193,194 In this review article, we have thoroughly presented and classified a diverse selection of photocatalysts for N₂ fixation. The selected literature reports are summarized in Tables 2 and 3. Overall, we genuinely envision that this review will establish benchmarks and mitigate the present obstacles in N₂ photoreduction as well as broaden new inroads at the forefront of this research hotspot. In order to optimize the practical schemes, several concluding remarks, prospects and suggestions are elucidated.

(I) Pre-treatment

In the study of iron-doped TiO₂, it is apparent that the N₂ reduction photocatalytic activity is highly dependent on the pre-treatment method of the as-developed catalysts. The different preparation techniques are presented in Table 4. On the one hand, the synthetic routes markedly influence the photocatalytic activity based on a plethora of aspects, including chemical compounds, defects, crystal morphologies and particle sizes. Additionally, heat treatment plays a central role in the physicochemical properties of the nanomaterials. Notably, annealing at a high temperature introduces defect states, which enhance the photoreduction ability.⁹¹ In another aspect, prolonged heat treatment adversely affects the phase transformation and the amount of surface ˙OH groups, which is detrimental to the activity of photocatalysts.⁶⁷ As described earlier, the treatment of catalysts in H₂ atmosphere will contribute to an increase of H_ads on the surface or even form an H-terminated surface, thus facilitating the hydrogenation of nitrogen. However, the amount of surficial H will decrease with time, leading to suppressed photoactivity.

Table 4 Typical pre-treatment methods for iron-doped TiO₂^67,98,99,121

Year	Author	Preparation procedures
1977	Schrauzer	Calcine iron sulfate impregnated with anatase TiO2 at 1273 K for 1 h in air
1991	Soria	Coprecipitate solution of TiCl3 and iron ions in an NH3 solution and anneal the resulting solid at 823 K for 24 h
2001	Rusina	React iron chloride with tetraisopropyl titanate in a sol–gel method and calcine the resulting solid at 923 K for 20 min
2014	Zhao	Incorporate Fe^3+ into hydrogen titanate phase nanotubes via a hydrothermal method followed by performing calcination at 773 K for 3 h

---

(II) Reaction mixture

In addition to the gas-phase reaction, aqueous-phase photoreaction has also been widely employed via the suspension of photocatalysts in liquid. An aqueous slurry with an appropriate concentration of photocatalyst enhances the photoreaction by accelerating the dispersion of the photocatalyst for improved mass transfer. However, high concentrations lead to poor penetration of photons.⁹⁶ It has also been reported that in the presence of photocatalysts, the ˙OH radicals in water oxidize ammonia to nitrate.¹²⁵ As such, ammonia and nitrate are detected in the final products of photoreaction.¹³³

The presence of organic scavengers such as ethanol and methanol can obviously increase the NH₃ yield by providing sacrificial electron donors to the holes in the semiconductor.¹²³ As for the side reaction of the sacrificial agent, Rusina et al. inferred that ethanol facilitates the formation of nitrite and nitrate,¹²¹ while Zhao and Lashgari claimed that ethanol serves as a superior ˙OH scavenger which protects the NH₃ yield from further oxidation to its higher oxidation states of nitrite or nitrate (Fig. 11).^99,113 Usually, holes react with hydroxide ions to form hydroxyl radicals (OH⁻ + h_VB⁺ → ˙OH), which can further oxidize NH₃. In this regard, when ethanol is present, it can be oxidized to CH₃CHO and C₂H₅OC₂H₅.⁹⁹ Along with the consumption of photo-generated holes in an oxidation process, the N₂ reduction rates can be markedly increased due to the availability of abundant photo-generated electrons for the reduction reactions. Although ethanol is generally utilized as a typical organic scavenger, it is crucial to comprehensively study different alcohols in order to determine suitable sacrificial agents for specific light-driven catalytic systems.

Fig. 11 Schematic of the reaction mechanism of N₂ fixation at the semiconductor/solution interface in the presence of alcohol as a sacrificial agent. Reproduced with permission.¹¹³ Copyright 2017, Elsevier.

(III) Modification of semiconductors

Transition metals are widely used as dopants, co-catalysts and plasmonic nanostructures with the aim of enhancing photocatalytic efficiency. Among these, earth-abundant iron is the most dominant as an efficient metal dopant. In contrast, it has been reported that noble metals as dopants in TiO₂ demonstrate inhibiting effects for N₂ photoreduction.⁶⁷ Moreover, this inhibiting effect has been observed on Pt-loaded ZnO compared with pure ZnO.¹⁰⁴ However, the incorporation of Ru as a co-catalyst on semiconductors has exhibited an evident positive influence on N₂ fixation activity.¹⁰⁶ Recently, plasmonic metals were reported to enhance photoabsorption via LSPR. To date, Au has been selected to couple with myriad catalysts, such as SrTiO₃,^127,128 Os and black silicon,^190,191 because its light-harvesting effects can efficiently utilize the visible light region of the solar spectrum.

Additionally, the flourishing investigation of oxygen vacancies induced in BiOX has inspired many researchers due to their unique characteristics for the efficient separation of charge carriers. As illustrated in Fig. 12,¹⁵¹ OVs suppress the recombination of electrons and holes by directly trapping electrons from the conduction band and then transferring them to the anti-bonding orbitals of adsorbed N₂. Since the first seminal report on oxygen vacancies in BiOBr for N₂ fixation in 2015, a number of studies on nitrogen, sulfur and oxygen vacancies have been reported to date.^62,152,154 As a proof of concept, a hybrid structure is always desirable for modulating band structures for ameliorated solar energy conversion.^195–203 Importantly, the heterostructure will spatially accumulate the separation and transfer of photogenerated charge carriers upon their interactions at the contacted interface.

Fig. 12 Schematic of the enhancement of OVs-induced photocatalysts for interfacial electron transfer processes. Reproduced with permission.¹⁵¹ Copyright 2015, American Chemical Society.

(IV) Suggestions, outlook and future prospects

As shown in Table 2, the assessments of photocatalytic activities for N₂ reduction are not unified. The standard NH₃ production rate can be expressed in units of μmol h⁻¹ g⁻¹ for the gas-phase reaction, μM h⁻¹ for the reaction in solution and μmol h⁻¹ cm⁻² for photoelectrochemical cells. Therefore, apparent quantum yield (AQY) should be generalized as a scientific tool to account for dissimilar experimental conditions, such as light source, catalyst loading, reaction duration and illumination area.

Most previous studies have focused on metal oxide-based nanomaterials, especially in the case of TiO₂. However, TiO₂ is greatly restricted by its UV excitation, which constitutes less than 5% of the solar light spectrum. Additionally, Cd-containing dichalcogenides are mostly in the form of sulfides, while the stability of CdS urgently requires improvement by forming multi-metal systems. Furthermore, air and visible light are gradually being adopted as nitrogen and light sources instead of pure nitrogen and UV irradiation. In order to use air as the N₂ source, the selectivity of the desired products should be taken into consideration to avoid the side product of H₂ evolution and the decomposition of NH₃. In contrast, for the aqueous-phase reaction, more attention should be paid to the nitrite and nitrate contents in the products. In addition, research on metal-free systems has been primarily dominated by g-C₃N₄ to date. Although in-depth studies on band structures and charge carrier dynamics are necessary, the exploration of new potential photocatalysts is gaining in momentum to surmount the bottlenecks of the fundamental research and applications of N₂ fixation. It is our aim to extend the utilization and optical absorption of solar light up to the NIR in addition to UV and visible light to cater to the entire spectrum of solar light.

In other aspect, a robust device for assembling different components is overwhelmingly necessary to develop a multifunctional platform for artificial photosynthesis – N₂ fixation. The engineering of an intact “machine” should comprise a N₂ reduction catalyst, an oxidation catalyst, a light-harvesting absorber and an electron-transporting bridge to attain high photocatalytic efficiency of N₂ fixation. Therefore, the synergetic interactions of various moieties within an individual matrix must be taken into consideration. Good examples of such versatile “machines” include porous metal–organic frameworks (MOFs) and covalent organic frameworks (COFs), to which further exploration should be devoted in the future. Apart from photon-driven catalysis, with the aid of external bias, a photoelectrocatalytic system is another smart avenue to separate the redox reactions (i.e. N₂ reduction and H₂O oxidation) at the photocathode and photoanode, respectively. Unequivocally, there is an extreme paucity of literature reports on photoelectrochemical N₂ fixation at the moment. Thus, a combined light-driven and electro-driven reaction would undeniably greatly facilitate N₂ fixation activity. To this end, dissimilar photocatalysts at the cathode and anode sides could be designed. Essentially, with the present knowledge on the electrochemical reduction of N₂ based on a wealth of research activities to date, this provides a scientific guiding star for photoelectrochemical N₂ reduction systems.

For heterogeneous photocatalytic systems, insights into their reaction pathways are essential, despite the complexity of the N₂ fixation reaction. It is prominent that tuning the size of photocatalysts can remarkably alter the ratio of exposed atoms and active sites on the corners, edges and surfaces with different coordination numbers. As a result, this influences the adsorption ability of the reactant N₂ molecules as well as the intermediates during the photoreduction process, giving rise to dissimilar performance and activity of the shape- and size-controlled photocatalysts. To comprehend the atomic insights in reactivity, advanced in situ or operando characterization techniques are indispensable. For example, in situ FTIR analysis is beneficial to examine the adsorbed nitrogen species on the surface of photocatalysts. Meanwhile, environmental/in situ TEM and X-ray absorption spectroscopy can provide information on the structure evolution and the modification of the surface coordination number during the photocatalytic reaction process. Equally importantly, time-resolved spectroscopy, femtosecond transient absorption spectroscopy and terahertz time-domain spectroscopy would be helpful to study the excited-state dynamics in a typical photoinduced electron-transfer pathway. Additionally, computer-aided catalyst design via DFT calculations for N₂ fixation would be a robust tool for simulating band structures and reaction mechanisms to achieve excellent sunlight absorption and unprecedented catalysis performance. This could in turn provide auspicious references for the rational experimental design of photocatalysts. Therefore, the innovative engineering of photocatalysts from experimental and theoretical aspects is crucial to systematically tune the band gap for effective absorption of solar photon flux as well as high photoactivity for solar energy conversion.

In summary, it is evident that the product yield of N₂ photoreduction has gradually increased from a magnitude of μmol g⁻¹ to mmol g⁻¹. Without a doubt, the goal of achieving a magnitude of mol g⁻¹ for the product yield will no longer be a dream; instead, this dream will be transformed into reality and practicality. Judging from the amount of existing literature on this research platform at this juncture, N₂ photoreduction has ignited interest and attention in the research community in recent years. However, it should be emphasized that the quantity of samples obtained from each batch is still at the laboratory scale. To divert from the laboratory-scale level to industrial applications, amplifying the yield of catalysts while retaining their intrinsic original structures is a global challenge. More thought should be given to translating these studies from academic research to practicality. Despite the aforementioned impediments to the photoactivation of N₂, the future direction of this area is indisputably promising and prospective with the synergetic cooperation of international researchers from diversified disciplines, such as computational scientists, chemists, materials scientists and physicists. Overall, we strongly believe that solar energy conversion for N₂ fixation will be enriched to begin a revolution of renewable energy for practical benefits and future commercialization.

Regarding the long-term prospects, there is an infinite scope of daunting challenges and opportunities for global researchers to expand the fundamental science at the forefront of this research hotspot. We are certain that this review article will provide an excellent basis for the next research era not only in photocatalytic N₂ fixation specifically, but also in the interdisciplinary fields of chemistry, materials science, energy conversion and energy storage. On the whole, with ceaseless determination and efforts from a diverse range of scientific communities and accomplishments in years to come, the research that has been comprehensively discussed in this review article will unquestionably be surpassed.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

N. Li is grateful for financial support from China Scholarship Council (CSC) with No. 201606955033, National Natural Science Foundation of China (NSFC) with No. 51461135004, Natural Science Foundation of Hubei Province with No. 2015CFB227, Fundamental Research Funds for the Central Universities, and the research board of the State Key Laboratory of Silicate Materials for Architectures. W.-J. Ong is also thankful for support from the Institute of Materials Research and Engineering (IMRE), Agency for Science, Technology and Research (A*STAR) in Singapore.

References

1. V. Rosca, M. Duca, M. T. de Groot and M. T. M. Koper, Chem. Rev., 2009, 109, 2209–2244.
2. D. E. Canfield, A. N. Glazer and P. G. Falkowski, Science, 2010, 330, 192.
3. B. K. Burgess and D. J. Lowe, Chem. Rev., 1996, 96, 2983–3012.
4. H.-P. Jia and E. A. Quadrelli, Chem. Soc. Rev., 2014, 43, 547–564.
5. A. Shilov, Russ. Chem. Bull., 2003, 52, 2555–2562.
6. V. Smil, Nature, 1999, 400, 415.
7. B. M. Hoffman, D. Lukoyanov, Z.-Y. Yang, D. R. Dean and L. C. Seefeldt, Chem. Rev., 2014, 114, 4041–4062.
8. C. N. Satterfield, Heterogeneous Catalysis in Practice, McGraw-Hill Companies, 1980.
9. R. Schlögl, Angew. Chem., Int. Ed., 2003, 42, 2004–2008.
10. H. Liu, Chin. J. Catal., 2014, 35, 1619–1640.
11. M. Temkin and V. Pyzhev, Acta Physiochim. U. R. S. S., 1940, 12, 217–222.
12. C. Liang, Z. Wei, Q. Xin and C. Li, Appl. Catal., A, 2001, 208, 193–201.
13. A. Vojvodic, A. J. Medford, F. Studt, F. Abild-Pedersen, T. S. Khan, T. Bligaard and J. K. Nørskov, Chem. Phys. Lett., 2014, 598, 108–112.
14. Y. Inoue, M. Kitano, K. Kishida, H. Abe, Y. Niwa, M. Sasase, Y. Fujita, H. Ishikawa, T. Yokoyama and M. Hara, ACS Catal., 2016, 6, 7577–7584.
15. A. D. Allen and C. V. Senoff, Chem. Commun., 1965, 621–622.
16. D. V. Yandulov and R. R. Schrock, Science, 2003, 301, 76–78.
17. G. Schwarz, R. R. Mendel and M. W. Ribbe, Nature, 2009, 460, 839–847.
18. K. Arashiba, Y. Miyake and Y. Nishibayashi, Nat. Chem., 2011, 3, 120–125.
19. J. S. Anderson, J. Rittle and J. C. Peters, Nature, 2013, 501, 84–87.
20. S. E. Creutz and J. C. Peters, J. Am. Chem. Soc., 2014, 136, 1105–1115.
21. J. Rittle and J. C. Peters, J. Am. Chem. Soc., 2016, 138, 4243–4248.
22. T. Shima, S. Hu, G. Luo, X. Kang, Y. Luo and Z. Hou, Science, 2013, 340, 1549–1552.
23. B. M. Hoffman, D. R. Dean and L. C. Seefeldt, Acc. Chem. Res., 2009, 42, 609–619.
24. R. Bjornsson, F. Neese, R. R. Schrock, O. Einsle and S. DeBeer, J. Biol. Inorg. Chem., 2015, 20, 447–460.
25. K. C. MacLeod and P. L. Holland, Nat. Chem., 2013, 5, 559–565.
26. V. Kyriakou, I. Garagounis, E. Vasileiou, A. Vourros and M. Stoukides, Catal. Today, 2017, 286, 2–13.
27. G. Marnellos and M. Stoukides, Science, 1998, 282, 98–100.
28. Z. Li, R. Liu, Y. Xie, S. Feng and J. Wang, Solid State Ionics, 2005, 176, 1063–1066.
29. E. Vasileiou, V. Kyriakou, I. Garagounis, A. Vourros, A. Manerbino, W. G. Coors and M. Stoukides, Top. Catal., 2015, 58, 1193–1201.
30. E. Vasileiou, V. Kyriakou, I. Garagounis, A. Vourros and M. Stoukides, Solid State Ionics, 2015, 275, 110–116.
31. Y. Kobayashi, N. Shimoda, Y. Kimura and S. Satokawa, ECS Trans., 2017, 75, 43–52.
32. V. Kordali, G. Kyriacou and C. Lambrou, Chem. Commun., 2000, 1673–1674.
33. G. Xu, R. Liu and J. Wang, Sci. China, Ser. B: Chem., 2009, 52, 1171–1175.
34. R. Lan and S. Tao, RSC Adv., 2013, 3, 18016–18021.
35. T. Murakami, T. Nishikiori, T. Nohira and Y. Ito, J. Am. Chem. Soc., 2003, 125, 334–335.
36. T. Murakami, T. Nohira, Y. Araki, T. Goto, R. Hagiwara and Y. H. Ogata, Electrochem. Solid-State Lett., 2007, 10, E4–E6.
37. I. A. Amar, R. Lan and S. Tao, RSC Adv., 2015, 5, 38977–38983.
38. S. Licht, B. Cui, B. Wang, F.-F. Li, J. Lau and S. Liu, Science, 2014, 345, 637–640.
39. F.-F. Li and S. Licht, Inorg. Chem., 2014, 53, 10042–10044.
40. S. Chen, S. Perathoner, C. Ampelli, C. Mebrahtu, D. Su and G. Centi, Angew. Chem., Int. Ed., 2017, 56, 2699–2703.
41. A. B. Djurisic, Y. H. Leung and A. M. Ching Ng, Mater. Horiz., 2014, 1, 400–410.
42. M. Marszewski, S. Cao, J. Yu and M. Jaroniec, Mater. Horiz., 2015, 2, 261–278.
43. Y. Ma, X. Wang, Y. Jia, X. Chen, H. Han and C. Li, Chem. Rev., 2014, 114, 9987–10043.
44. J. L. White, M. F. Baruch, J. E. Pander Iii, Y. Hu, I. C. Fortmeyer, J. E. Park, T. Zhang, K. Liao, J. Gu, Y. Yan, T. W. Shaw, E. Abelev and A. B. Bocarsly, Chem. Rev., 2015, 115, 12888–12935.
45. K. Wenderich and G. Mul, Chem. Rev., 2016, 116, 14587–14619.
46. Y. Zhao, X. Jia, G. I. N. Waterhouse, L.-Z. Wu, C.-H. Tung, D. O'Hare and T. Zhang, Adv. Energy Mater., 2016, 6, 1501974.
47. X. Liu, J. Iocozzia, Y. Wang, X. Cui, Y. Chen, S. Zhao, Z. Li and Z. Lin, Energy Environ. Sci., 2017, 10, 402–434.
48. A. Kudo and Y. Miseki, Chem. Soc. Rev., 2009, 38, 253–278.
49. L. K. Putri, B.-J. Ng, W.-J. Ong, H. W. Lee, W. S. Chang and S.-P. Chai, ACS Appl. Mater. Interfaces, 2017, 9, 4558–4569.
50. W.-J. Ong, L.-L. Tan, S.-P. Chai, S.-T. Yong and A. R. Mohamed, Nanoscale, 2014, 6, 1946–2008.
51. J. Di, J. Xia, H. Li and Z. Liu, Nano Energy, 2017, 35, 79–91.
52. J. Yang, D. Wang, H. Han and C. Li, Acc. Chem. Res., 2013, 46, 1900–1909.
53. X. Chen, S. Shen, L. Guo and S. S. Mao, Chem. Rev., 2010, 110, 6503–6570.
54. S. Chen, T. Takata and K. Domen, Nat. Rev. Mater., 2017, 2, 17050.
55. Q. Lu, Y. Yu, Q. Ma, B. Chen and H. Zhang, Adv. Mater., 2016, 28, 1917–1933.
56. G. A. M. Hutton, B. C. M. Martindale and E. Reisner, Chem. Soc. Rev., 2017 10.1039/c7cs00235a.
57. G. Zhang, Z.-A. Lan and X. Wang, Chem. Sci., 2017, 8, 5261–5274.
58. X. Chang, T. Wang and J. Gong, Energy Environ. Sci., 2016, 9, 2177–2196.
59. S. J. A. Moniz, S. A. Shevlin, D. J. Martin, Z.-X. Guo and J. Tang, Energy Environ. Sci., 2015, 8, 731–759.
60. X. Meng, L. Liu, S. Ouyang, H. Xu, D. Wang, N. Zhao and J. Ye, Adv. Mater., 2016, 28, 6781–6803.
61. J. Li, H. Li, G. Zhan and L. Zhang, Acc. Chem. Res., 2017, 50, 112–121.
62. G. Dong, W. Ho and C. Wang, J. Mater. Chem. A, 2015, 3, 23435–23441.
63. N. Dhar and S. Chowdhry, Proc. Natl. Acad. Sci., India, 1968, 38, 485–490.
64. G. N. Schrauzer, N. Strampach, L. N. Hui, M. R. Palmer and J. Salehi, Proc. Natl. Acad. Sci. U. S. A., 1983, 80, 3873–3876.
65. A. J. Medford and M. C. Hatzell, ACS Catal., 2017, 7, 2624–2643.
66. G. N. Schrauzer, Energy Efficiency and Renewable Energy Through Nanotechnology, Springer, 2011, pp. 601–623.
67. G. N. Schrauzer and T. D. Guth, J. Am. Chem. Soc., 1977, 99, 7189–7193.
68. T. Bazhenova and A. Shilov, Coord. Chem. Rev., 1995, 144, 69–145.
69. C. J. van der Ham, M. T. Koper and D. G. Hetterscheid, Chem. Soc. Rev., 2014, 43, 5183–5191.
70. L.-j. Guo, Y.-j. Wang and T. He, Chem. Rec., 2016, 16, 1918–1933.
71. S. Sun, X. Li, W. Wang, L. Zhang and X. Sun, Appl. Catal., B, 2017, 200, 323–329.
72. W. J. Ong, L. L. Tan, S. P. Chai, S. T. Yong and A. R. Mohamed, ChemSusChem, 2014, 7, 690–719.
73. W.-J. Ong, L.-L. Tan, Y. H. Ng, S.-T. Yong and S.-P. Chai, Chem. Rev., 2016, 116, 7159–7329.
74. L. K. Putri, L.-L. Tan, W.-J. Ong, W. S. Chang and S.-P. Chai, Appl. Mater. Today, 2016, 4, 9–16.
75. A. Fuertes, Mater. Horiz., 2015, 2, 453–461.
76. F. K. Kessler, Y. Zheng, D. Schwarz, C. Merschjann, W. Schnick, X. Wang and M. J. Bojdys, Nat. Rev. Mater., 2017, 2, 17030.
77. K. Sivula and R. van de Krol, Nat. Rev. Mater., 2016, 1, 16010.
78. I. Roger, M. A. Shipman and M. D. Symes, Nat. Rev. Chem., 2017, 1, 0003.
79. Y. Tachibana, L. Vayssieres and J. R. Durrant, Nat. Photonics, 2012, 6, 511–518.
80. J. Low, J. Yu, M. Jaroniec, S. Wageh and A. A. Al-Ghamdi, Adv. Mater., 2017, 29, 1601694.
81. S. Cao, J. Low, J. Yu and M. Jaroniec, Adv. Mater., 2015, 27, 2150–2176.
82. H. Li, Y. Zhou, W. Tu, J. Ye and Z. Zou, Adv. Funct. Mater., 2015, 25, 998–1013.
83. L. K. Putri, W.-J. Ong, W. S. Chang and S.-P. Chai, Appl. Surf. Sci., 2015, 358(part A), 2–14.
84. S. Hoang and P.-X. Gao, Adv. Energy Mater., 2016, 6, 1600683.
85. X. Li, J. Yu and M. Jaroniec, Chem. Soc. Rev., 2016, 45, 2603–2636.
86. H. Li, J. Li, Z. Ai, F. Jia and L. Zhang, Angew. Chem., 2017 DOI:10.1002/ange.201705628.
87. L. M. Azofra, N. Li, D. R. MacFarlane and C. Sun, Energy Environ. Sci., 2016, 9, 2545–2549.
88. H. Li, J. Shang, J. Shi, K. Zhao and L. Zhang, Nanoscale, 2016, 8, 1986–1993.
89. Y. Bai, L. Ye, T. Chen, L. Wang, X. Shi, X. Zhang and D. Chen, ACS Appl. Mater. Interfaces, 2016, 8, 27661–27668.
90. A. Fujishima and K. Honda, Nature, 1972, 238, 37–38.
91. S. Bourgeois, D. Diakite and M. Perdereau, React. Solids, 1988, 6, 95–104.
92. H. Hirakawa, M. Hashimoto, Y. Shiraishi and T. Hirai, J. Am. Chem. Soc., 2017, 139, 10929–10936.
93. V. Augugliaro, A. Lauricella, L. Rizzuti, M. Schiavello and A. Sclafani, Int. J. Hydrogen Energy, 1982, 7, 845–849.
94. Y. Iida and S. Ozaki, J. Am. Ceram. Soc., 1961, 44, 120–127.
95. J. Navio, M. Macias, M. Gonzalez-Catalan and A. Justo, J. Mater. Sci., 1992, 27, 3036–3042.
96. P. P. Radford and C. G. Francis, J. Chem. Soc., Chem. Commun., 1983, 1520–1521.
97. M. Rao, K. Rajeshwar, V. P. Verneker and J. DuBow, J. Phys. Chem., 1980, 84, 1987–1991.
98. J. Soria, J. Conesa, V. Augugliaro, L. Palmisano, M. Schiavello and A. Sclafani, J. Phys. Chem., 1991, 95, 274–282.
99. W. Zhao, J. Zhang, X. Zhu, M. Zhang, J. Tang, M. Tan and Y. Wang, Appl. Catal., B, 2014, 144, 468–477.
100. L. Palmisano, V. Augugliaro, A. Sclafani and M. Schiavello, J. Phys. Chem., 1988, 92, 6710–6713.
101. C. Martin, I. Martin, V. Rives, L. Palmisano and M. Schiavello, J. Catal., 1992, 134, 434–444.
102. O. Ileperuma, K. Tennakone and W. Dissanayake, Appl. Catal., 1990, 62, L1–L5.
103. O. Ileperuma, C. Thaminimulla and W. Kiridena, Sol. Energy Mater. Sol. Cells, 1993, 28, 335–343.
104. H. Miyama, N. Fujii and Y. Nagae, Chem. Phys. Lett., 1980, 74, 523–524.
105. M. M. Taqui Khan, D. Chatterjee and M. Bala, J. Photochem. Photobiol., A, 1992, 67, 349–352.
106. K. T. Ranjit, T. K. Varadarajan and B. Viswanathan, J. Photochem. Photobiol., A, 1996, 96, 181–185.
107. O. P. Linnik and H. Kisch, Mendeleev Commun., 2008, 18, 10–11.
108. K. Hoshino, M. Inui, T. Kitamura and H. Kokado, Angew. Chem., Int. Ed., 2000, 39, 2509–2512.
109. K. Hoshino, Chem. – Eur. J., 2001, 7, 2727–2731.
110. K. Hoshino, R. Kuchii and T. Ogawa, Appl. Catal., B, 2008, 79, 81–88.
111. M. M. Khader, N. N. Lichtin, G. H. Vurens, M. Salmeron and G. A. Somorjai, Langmuir, 1987, 3, 303–304.
112. C. J. Jacobsen, S. Dahl, B. S. Clausen, S. Bahn, A. Logadottir and J. K. Nørskov, J. Am. Chem. Soc., 2001, 123, 8404–8405.
113. M. Lashgari and P. Zeinalkhani, Appl. Catal., A, 2017, 529, 91–97.
114. E. Endoh, J. K. Leland and A. J. Bard, J. Phys. Chem., 1986, 90, 6223–6226.
115. W. Zhao, H. Xi, M. Zhang, Y. Li, J. Chen, J. Zhang and X. Zhu, Chem. Commun., 2015, 51, 4785–4788.
116. S. Sun, Q. An, W. Wang, L. Zhang, J. Liu and W. A. Goddard Iii, J. Mater. Chem. A, 2017, 5, 201–209.
117. V. Augugliaro, F. D'alba, L. Rizzuti, M. Schiavello and A. Sclafani, Int. J. Hydrogen Energy, 1982, 7, 851–855.
118. C. M. Janet, S. Navaladian, B. Viswanathan, T. K. Varadarajan and R. P. Viswanath, J. Phys. Chem. C, 2010, 114, 2622–2632.
119. H. Pan, X. Meng and G. Qin, Phys. Chem. Chem. Phys., 2014, 16, 25442–25448.
120. D. Cordischi, N. Burriesci, F. D'Alba, M. Petrera, G. Polizzotti and M. Schiavello, J. Solid State Chem., 1985, 56, 182–190.
121. O. Rusina, A. Eremenko, G. Frank, H. P. Strunk and H. Kisch, Angew. Chem., Int. Ed., 2001, 40, 3993–3995.
122. O. Rusina, O. Linnik, A. Eremenko and H. Kisch, Chem. – Eur. J., 2003, 9, 561–565.
123. O. Linnik and H. Kisch, Photochem. Photobiol. Sci., 2006, 5, 938–942.
124. G. N. Schrauzer, T. D. Guth, J. Salehi, N. Strampach, N. H. Liu and M. R. Palmer, in Homogeneous and Heterogeneous Photocatalysis, ed. E. Pelizzetti and N. Serpone, Reidel, Hingham, MA, 1986, vol. 174, pp. 509–518.
125. Q.-s. Li, K. Domen, S. Naito, T. Onishi and K. Tamaru, Chem. Lett., 1983, 321–324.
126. S. Linic, P. Christopher and D. B. Ingram, Nat. Mater., 2011, 10, 911–921.
127. T. Oshikiri, K. Ueno and H. Misawa, Angew. Chem., Int. Ed., 2014, 53, 9802–9805.
128. T. Oshikiri, K. Ueno and H. Misawa, Angew. Chem., 2016, 128, 4010–4014.
129. E. Skulason, T. Bligaard, S. Gudmundsdóttir, F. Studt, J. Rossmeisl, F. Abild-Pedersen, T. Vegge, H. Jónsson and J. K. Nørskov, Phys. Chem. Chem. Phys., 2012, 14, 1235–1245.
130. M. A. Shipman and M. D. Symes, Catal. Today, 2017, 286, 57–68.
131. Y. Hao, X. Dong, S. Zhai, H. Ma, X. Wang and X. Zhang, Chem. – Eur. J., 2016, 22, 18722–18728.
132. K. Tennakone, S. Wickramanayake, C. Fernando, O. Ileperuma and S. Punchihewa, J. Chem. Soc., Chem. Commun., 1987, 1078–1080.
133. K. Tennakone, J. Bandara, C. Thaminimulla, W. Jayatilake, U. Ketipearachchi, O. Ileperuma and M. Priyadarshana, Langmuir, 1991, 7, 2166–2168.
134. K. Tennakone, S. Punchihewa and R. Tantrigoda, Sol. Energy Mater., 1989, 18, 217–221.
135. K. Tennakone, C. Thaminimulla and W. Kiridena, Langmuir, 1993, 9, 723–726.
136. X. Li, W. Wang, D. Jiang, S. Sun, L. Zhang and X. Sun, Chem. – Eur. J., 2016, 22, 13819–13822.
137. O. Ileperuma, W. Kiridena and W. Dissanayake, J. Photochem. Photobiol., A, 1991, 59, 191–197.
138. K. Tennakone, C. Fernando, S. Wickramanayake, M. Damayanthi, L. Silva, W. Wijeratne, O. Illeperuma and S. Punchihewa, Sol. Energy Mater., 1988, 17, 47–53.
139. K. Tennakone, C. Thaminimulla and J. Bandara, J. Photochem. Photobiol., A, 1992, 68, 131–135.
140. W.-J. Ong, L. K. Putri, Y.-C. Tan, L.-L. Tan, N. Li, Y. H. Ng, X. Wen and S.-P. Chai, Nano Res., 2017, 10, 1673–1696.
141. X. Zong, H. Yan, G. Wu, G. Ma, F. Wen, L. Wang and C. Li, J. Am. Chem. Soc., 2008, 130, 7176–7177.
142. H. Yan, J. Yang, G. Ma, G. Wu, X. Zong, Z. Lei, J. Shi and C. Li, J. Catal., 2009, 266, 165–168.
143. S. Navalón, A. Dhakshinamoorthy, M. Álvaro and H. Garcia, ChemSusChem, 2013, 6, 562–577.
144. M. M. T. Khan, R. C. Bhardwaj and C. Bhardwaj, Angew. Chem., Int. Ed., 1988, 27, 923–925.
145. M. T. Khan and N. N. Rao, J. Mol. Catal. A: Chem., 1990, 58, 323–329.
146. L. Ye, C. Han, Z. Ma, Y. Leng, J. Li, X. Ji, D. Bi, H. Xie and Z. Huang, Chem. Eng. J., 2017, 307, 311–318.
147. K. A. Brown, D. F. Harris, M. B. Wilker, A. Rasmussen, N. Khadka, H. Hamby, S. Keable, G. Dukovic, J. W. Peters and L. C. Seefeldt, Science, 2016, 352, 448–450.
148. A. Banerjee, B. D. Yuhas, E. A. Margulies, Y. Zhang, Y. Shim, M. R. Wasielewski and M. G. Kanatzidis, J. Am. Chem. Soc., 2015, 137, 2030–2034.
149. J. Liu, M. S. Kelley, W. Wu, A. Banerjee, A. P. Douvalis, J. Wu, Y. Zhang, G. C. Schatz and M. G. Kanatzidis, Proc. Natl. Acad. Sci. U. S. A., 2016, 113, 5530–5535.
150. B. M. Hoffman, D. R. Dean and L. C. Seefeldt, Acc. Chem. Res., 2009, 42, 609–619.
151. H. Li, J. Shang, Z. Ai and L. Zhang, J. Am. Chem. Soc., 2015, 137, 6393–6399.
152. S. Hu, X. Chen, Q. Li, Y. Zhao and W. Mao, Catal. Sci. Technol., 2016, 6, 5884–5890.
153. Y. Cao, S. Hu, F. Li, Z. Fan, J. Bai, G. Lu and Q. Wang, RSC Adv., 2016, 6, 49862–49867.
154. S. Hu, Y. Li, F. Li, Z. Fan, H. Ma, W. Li and X. Kang, ACS Sustainable Chem. Eng., 2016, 4, 2269–2278.
155. Q. Zhang, S. Hu, Z. Fan, D. Liu, Y. Zhao, H. Ma and F. Li, Dalton Trans., 2016, 45, 3497–3505.
156. S. Wu, J. S. Ross, G.-B. Liu, G. Aivazian, A. Jones, Z. Fei, W. Zhu, D. Xiao, W. Yao, D. Cobden and X. Xu, Nat. Phys., 2013, 9, 149–153.
157. K. F. Mak, K. He, J. Shan and T. F. Heinz, Nat. Nanotechnol., 2012, 7, 494–498.
158. E. A. A. Pogna, M. Marsili, D. De Fazio, S. Dal Conte, C. Manzoni, D. Sangalli, D. Yoon, A. Lombardo, A. C. Ferrari, A. Marini, G. Cerullo and D. Prezzi, ACS Nano, 2016, 10, 1182–1188.
159. S. Sun, X. Li, W. Wang, L. Zhang and X. Sun, Appl. Catal., B, 2017, 200, 323–329.
160. C. H. Lui, A. J. Frenzel, D. V. Pilon, Y. H. Lee, X. Ling, G. M. Akselrod, J. Kong and N. Gedik, Phys. Rev. Lett., 2014, 113, 166801.
161. H. Cheng, B. Huang and Y. Dai, Nanoscale, 2014, 6, 2009–2026.
162. W. L. Huang and Q. Zhu, J. Comput. Chem., 2009, 30, 183–190.
163. X. Y. Kong, W. Lee, W. J. Ong, S. P. Chai and A. R. Mohamed, ChemCatChem, 2016, 8, 3074–3081.
164. S. Wang, X. Hai, X. Ding, K. Chang, Y. Xiang, X. Meng, Z. Yang, H. Chen and J. Ye, Adv. Mater., 2017, 29, 1701774.
165. D. Zhu, L. Zhang, R. E. Ruther and R. J. Hamers, Nat. Mater., 2013, 12, 836–841.
166. J. R. Christianson, D. Zhu, R. J. Hamers and J. R. Schmidt, J. Phys. Chem. B, 2014, 118, 195–203.
167. J. A. Bandy, D. Zhu and R. J. Hamers, Diamond Relat. Mater., 2016, 64, 34–41.
168. H. Ma, Z. Shi, S. Li and N. Liu, Appl. Surf. Sci., 2016, 379, 309–315.
169. S. Li, X. Chen, S. Hu, Q. Li, J. Bai and F. Wang, RSC Adv., 2016, 6, 45931–45937.
170. H. Ma, Z. Shi, Q. Li and S. Li, J. Phys. Chem. Solids, 2016, 99, 51–58.
171. G. Wu, Y. Gao and B. Zheng, Ceram. Int., 2016, 42, 6985–6992.
172. S. Hu, X. Chen, Q. Li, F. Li, Z. Fan, H. Wang, Y. Wang, B. Zheng and G. Wu, Appl. Catal., B, 2017, 201, 58–69.
173. Z. Ma, S. Zhao, X. Xiong, B. Hu and C. Song, Catal. Lett., 2016, 146, 2324–2329.
174. Y. Wang, W. Wei, M. Li, S. Hu, J. Zhang and R. Feng, RSC Adv., 2017, 7, 18099–18107.
175. H. Liang, H. Zou and S. Hu, New J. Chem., 2017, 41, 8920–8926.
176. S. Cao, N. Zhou, F. Gao, H. Chen and F. Jiang, Appl. Catal., B, 2017, 218, 600–610.
177. S. Hu, W. Zhang, J. Bai, G. Lu, L. Zhang and G. Wu, RSC Adv., 2016, 6, 25695–25702.
178. S. Hu, F. Li, Z. Fan, F. Wang, Y. Zhao and Z. Lv, Dalton Trans., 2015, 44, 1084–1092.
179. Q. Han, N. Chen, J. Zhang and L. Qu, Mater. Horiz., 2017, 4, 832–850.
180. W.-J. Ong, L.-L. Tan, S.-P. Chai and S.-T. Yong, Chem. Commun., 2015, 51, 858–861.
181. W.-J. Ong, L.-L. Tan, S.-P. Chai, S.-T. Yong and A. R. Mohamed, Nano Energy, 2015, 13, 757–770.
182. W.-J. Ong, Front. Mater., 2017, 4, 11,  DOI:10.3389/fmats.2017.00011.
183. Y. Lu, Y. Yang, T. Zhang, Z. Ge, H. Chang, P. Xiao, Y. Xie, L. Hua, Q. Li and H. Li, ACS Nano, 2016, 10, 10507–10515.
184. Y. Yang, T. Zhang, Z. Ge, Y. Lu, H. Chang, P. Xiao, R. Zhao, Y. Ma and Y. Chen, Carbon, 2017, 124, 72–78.
185. P. Dean, Phys. Rev., 1967, 157, 655.
186. B. Aurian-Blajeni, M. Halmann and J. Manassen, Sol. Energy Mater., 1983, 8, 425–440.
187. C. R. Dickson and A. J. Nozik, J. Am. Chem. Soc., 1978, 100, 8007–8009.
188. F. Khan, P.-L. Yue, L. Rizzuti, V. Augugliaro and M. Schiavello, J. Chem. Soc., Chem. Commun., 1981, 20, 1049–1050.
189. F. Khan, P. Yue, L. Rizzuti, V. Augugliaro and A. Brucato, Ind. Eng. Chem. Res., 1983, 22, 238–241.
190. H. Zeng, S. Terazono and T. Tanuma, Catal. Commun., 2015, 59, 40–44.
191. M. Ali, F. Zhou, K. Chen, C. Kotzur, C. Xiao, L. Bourgeois, X. Zhang and D. R. MacFarlane, Nat. Commun., 2016, 7, 11335.
192. L. Li, Y. Wang, S. Vanka, X. Mu, Z. Mi and C.-J. Li, Angew. Chem., 2017, 129, 8827–8831.
193. A. Klerke, C. H. Christensen, J. K. Norskov and T. Vegge, J. Mater. Chem., 2008, 18, 2304–2310.
194. R. Lan, J. T. S. Irvine and S. Tao, Int. J. Hydrogen Energy, 2012, 37, 1482–1494.
195. W.-J. Ong, L. K. Putri, L.-L. Tan, S.-P. Chai and S.-T. Yong, Appl. Catal., B, 2016, 180, 530–543.
196. D. Zeng, W.-J. Ong, H. Zheng, M. Wu, Y. Chen, D.-L. Peng and M. Han, J. Mater. Chem. A, 2017, 5, 16171–16178.
197. L. K. Putri, W.-J. Ong, W. S. Chang and S.-P. Chai, Catal. Sci. Technol., 2016, 6, 744–754.
198. D. Zeng, L. Xiao, W.-J. Ong, P. Wu, H. Zheng, Y. Chen and D.-L. Peng, ChemSusChem, 2017 DOI:10.1002/cssc.201701345.
199. W.-J. Ong, L.-L. Tan, S.-P. Chai and S.-T. Yong, Dalton Trans., 2015, 44, 1249–1257.
200. D. Zeng, W. Xu, W.-J. Ong, J. Xu, H. Ren, Y. Chen, H. Zheng and D.-L. Peng, Appl. Catal., B, 2017 DOI:10.1016/j.apcatb.2017.08.041.
201. W.-J. Ong, L.-L. Tan, S.-P. Chai, S.-T. Yong and A. R. Mohamed, Nano Res., 2014, 7, 1528–1547.
202. L. j. Guo, Y. j. Wang and T. He, Chem. Rec., 2016, 16, 1918–1933.
203. P. Qiu, C. Xu, N. Zhou, H. Chen and F. Jiang, Appl. Catal., B, 2017 DOI:10.1016/j.apcatb.2017.09.010.

---