Recent Progress of the Design and Engineering of Bismuth Oxyhalides for Photocatalytic Nitrogen Fixation

Photocatalytic nitrogen ﬁ xation represents an effective technology for the arti ﬁ cial production of ammonia from atmospheric nitrogen, a critical step toward a sustainable economy. Bismuth oxyhalides (BiOX, X ¼ Cl, Br, and I) have emerged as viable catalysts for photocatalytic reduction of nitrogen into ammonia, due to their unique electronic structures and optical properties. Herein, the recent progress of BiOX-based photocatalysts for nitrogen ﬁ xation, with a focus on the reaction mechanism and pathways, materials preparation, and strategies of structural engineering for enhanced performance, is summarized. The article is concluded with a perspective where the promises and challenges of bismuth-based photocatalysts for nitrogen reduction to ammonia are highlighted, along with possible future research directions.


Introduction
Ammonia (NH 3 ) has been used extensively across the globe as an effective plant fertilizer in agricultural production, [1][2][3] and recently also emerged as a new clean energy and fuel. [4,5] Traditionally, ammonia is produced predominantly by the enzymatic processes in plants and microorganisms  Tg yearly). [6,7] Yet, such a productivity cannot meet the ever increasing demand of ammonia with the rapid expansion of human population. [7][8][9][10] In 1909, the invention of the Haber-Bosch method led to a significant leap in NH 3 production, [2,11,12] which is depicted below and represents a significant breakthrough in the history of artificial nitrogen fixation. [13][14][15] In fact, this reaction has thus far been the subject of three chemistry Nobel prizes, one for the discovery of the reaction, [16] a second for the development of high-pressure industrial processes, [17] and a third for the study of the fundamental surface chemistry on the catalysts. [18] In fact, the Haber-Bosch process has been heralded as the most important invention of the 20th century, which revolutionizes the nitrogen conversion to ammonia. [14,19] However, due to the high stability of N 2 (bond energy 941.3 kJ mol À1 ), [20,21] nitrogen fixation by the Haber-Bosch process necessitates operation with catalysts under high temperatures and high pressures. [22,23] This consumes a substantial amount of energy, greatly increases the cost of NH 3 production, [24,25] and causes serious environmental pollution. [21] In practice, the reactant gases need to be heated to 400-450 C at a pressure of 150-200 atm, [26,27] with the addition of Fe-based catalysts. [27,28] Meanwhile, a significant amount of energy needs to be invested to remove as much oxygen as possible from air to minimize catalyst oxidation. [29] After the deaeration step, the dissociated mixture of nitrogen and hydrogen forms ammonia on the iron-based catalyst. [30] Thus, this process is rather costly. [31] In a practical operation, the fixation of nitrogen to ammonia is completed in a high-pressure synthetic tower. The nitrogen and hydrogen gases are first mixed and compressed into the converter chamber from the top, [32,33] pass the heat exchanger at the bottom to increase the gas temperature, and then enter the contact chamber where Fe-based catalysts drives the production of ammonia. Note that only part of the nitrogen reacts with hydrogen to produce ammonia. [34] The mixture of nitrogen, hydrogen, and ammonia then leaves the converter through a heat exchanger, [35] where ammonia will be liquefied through a condenser, and separated from the mixture. [36] The remaining nitrogen and hydrogen is again compressed and fed into the synthetic tower to start a second cycle. In this operation, raw materials are saved and recycled to maximize the fixation efficiency. [35] Yet, many issues remain, in particular, low conversion rate of nitrogen to ammonia, high energy consumption, and serious pollution. [37] In fact, the Haber-Bosch process accounts for nearly 2% of global energy consumption, and is responsible for %1% of the global greenhouse gas emissions. Therefore, the development of new, effective nitrogen fixation technologies is highly desired that are lower-cost, high-performance, and environment-friendly. [7,10,12,38] In 1967, Fujishima and Honda accidentally discovered that under ultraviolet irradiation, titanium dioxide (TiO 2 ) single crystal can split water into oxygen and hydrogen. [39] Such a photocatalytic technology has since been exploited for a wide range of applications. For instance, in 1977, Schrauzer et al. demonstrated, for the first time ever, that TiO 2 can also be used as a photocatalyst for nitrogen fixation, [40] and in 1980, Schrauzer and Guth reported the first rigorous study of photocatalytic nitrogen fixation by natural materials. [41] These are just among the numerous studies on photocatalytic nitrogen fixation, a field that is booming rapidly. [42,43] In photocatalysis, design and engineering of effective semiconductor catalysts is a critical first step. Among these, bismuth-based materials have been attracting extensive attention, due to their low bandgap energy and high electrical conductivity. [44] Harriman et al. [45] were the first to report the photocatalytic activity of Bi 2 O 3 . In recent years, bismuth oxyhalides (BiOX, X ¼ Cl, Br, and I) have emerged as a unique class of photocatalysts due to their high photocatalytic activity and stability. For instance, Li et al. [46] in 2015 reported that BiOBr nanosheets exhibited efficient nitrogen fixation under visible light irradiation, where oxygen vacancies (OVs) on the exposed surface significantly enhanced the adsorption and activation of N 2 . Indeed, the photocatalytic performance can be significantly improved with an increase in the OV concentration and separation rate of photogenerated electron-hole pairs. [12,32] This important breakthrough essentially opens a door of photocatalytic nitrogen fixation based on BiOX materials. [47,48] Thus far, it has been well-known that adsorption of N 2 onto the catalyst surface plays a key role in determining the photocatalytic efficiency of nitrogen fixation, [43,49] which can be facilitated by the formation of OVs. [1,[50][51][52][53] Note that OVs are the most prevalent and widely studied anion defects with a relatively low formation energy on oxide surfaces, [13,54,55] and can act as electron traps to effectively capture and activate inert gas molecules, such as O 2 , CO 2 , and N 2 . [46,56] However, surface OVs introduced artificially can be easily oxidized, leading to cessation of N 2 activation in the fixation process, [12,57] and a large surface area is needed to maximize the production of OVs. [54] Therefore, substantial efforts have been devoted to the structural engineering of BiOX for the formation of abundant and stable OVs so as to enhance the performance of photocatalytic nitrogen fixation. [48,58] In this review article, we summarize recent progress in the design and engineering of BiOX-based materials for photocatalytic nitrogen fixation, within the context of sample synthesis, structural characterization, and OV formation. The corresponding structure-activity correlation is then exploited for the establishment of the reaction mechanism and pathway. We conclude the review with a perspective where the challenges of photocatalytic nitrogen fixation are highlighted, along with possible further research directions.

Bismuth-Based Photocatalysts for Nitrogen Fixation
Photocatalytic nitrogen fixation started as early as in the 1940s ( Figure 1). The purpose of the early work was to study the natural nitrogen fixation of soil, rather than artificial fixation by metals and metal oxides. [19,45] For instance, Dhar et al. [59] disinfected soil obtained from the field, and observed apparent nitrogen fixation both under photoirradiation and in the dark, with a higher rate for the former than for the latter. However, the results were not reproducible later on, likely because of the limited conditions at the time, such that the sterilization procedure used did not deactivate all nitrogen fixing bacteria. In addition, the fact that nonzero production of ammonia was observed even in the control experiment suggested that the photolysis of amino acids in dead microorganisms might be responsible for part of the ammonia produced. Although there was no clear concept of photocatalytic nitrogen fixation at that time and the conclusions obtained from the research were somewhat skeptical, it is nonetheless the first demonstration of photocatalytic nitrogen fixation. Later in 1977, Schrauzer et al. [40] reported the first photocatalytic nitrogen fixation by Fe-doped TiO 2 . Experimentally, they observed that in an argon atmosphere, the photolysis of chemisorbed water on partially outgassed TiO 2 powders produced H 2 and O 2 at a molar ratio of 2:1; yet, in the presence of molecular nitrogen, O 2 can still be formed, but the formation of H 2 was inhibited because chemically adsorbed nitrogen was reduced to NH 3 and trace N 2 H 4 , The doping of iron improved the photocatalytic activity of rutile TiO 2 , which was used in a prototype solar cell for the photochemical synthesis of ammonia with N 2 and H 2 O.
Since then, semiconductor photocatalysts have been used extensively in photocatalytic nitrogen fixation. Among these, bismuth-based materials have emerged as a new kind of effective photocatalysts because of their suitable bandgap and high separation rate of photogenerated electrons and holes. In 2015, Li et al. [46] adopted a simple solvothermal method to synthesize BiOBr nanosheets with abundant OVs exposed on the {001} facet (BOB-001-OV), and observed apparent photocatalytic activity toward nitrogen fixation even under visible light irradiation (λ > 420 nm), with the amount of NH 3 produced quantified by the Nessler reagent. This was a breakthrough in photocatalytic nitrogen fixation by semiconductor materials. However, with its 2D sheet-like structure and small specific surface area, the nitrogen fixation performance of BOB-001-OV was limited. In a later study in 2017, [52] Wang et al. successfully synthesized 1D ultrafine Bi 5 O 7 Br nanotubes by a water-guided self-assembly method, which greatly increased the specific surface area, improved the nitrogen fixation performance, and broke the bottleneck that limited the performance of the 2D counterparts. Importantly, such a performance can be further enhanced by an increase in surface OVs, as demonstrated in a more recent study by Li et al. [60] It should be noted that it is the strong interaction between the Bi 6p electrons and N 2p electrons that facilitates the adsorption of N 2 molecules onto the surface of the Bi-based catalysts, especially those with OVs. OVs can accumulate photogenerated electrons on the catalyst surface, which is conducive to N 2 adsorption. This leads to a weakened N≡N bond, [61] and the resulting activation of N 2 is crucial to the eventual reduction to NH 3 . [22]

Bismuth Oxyhalides
Because of their unique crystal and electronic structures, bismuth-based oxyhalides exhibit interesting photocatalytic activity and stability, and have been used as effective catalysts for N 2 fixation. [62] As shown in Figure 2, BiOX displays a layer structure consisting of a [Bi 2 O 2 ] plane and two planes of halogen atoms, and belongs to the tetragonal system. The formation of such a layered structure is due to strong covalent bonding interactions within the [Bi 2 O 2 ] layers and superposition of weak van der Waals forces (nonbond interactions) between the X atoms of the [X] layers along the C axis. [63,64] Importantly, the Bi─O bond has a much lower bond energy and a much longer bond length than other metal-oxygen bonds, which renders it easy to break the Bi─O bond, and for O 2 to escape from the catalyst surface creating OVs. [46] Notably, whereas the weak interaction between the surface of bismuth oxyhalide and N 2 molecules makes it difficult for N 2 to be chemically adsorbed, the OVs generated by the cleavage of the Bi─O bond greatly facilitate the adsorption of N 2 and increase the yield of N 2 conversion to NH 3 . [22] In typical metal oxide photocatalysts, the valence band (VB) usually consists of the O 2p orbital. However, for bismuth oxyhalides, both the O 2p and X np (n ¼ 3, 4, and 5) orbitals contribute to the VB, whereas the conduction band (CB) is dominated by the Bi 6p orbital. [47,65] This suggests that X can be exploited as a unique variable for the manipulation of the materials bandgap and therefore the photo absorption range. Note that the bandgap of the BiOX compounds decreases with increasing atomic number of the halogen (Figure 3a,b), [66,67] in the order of BiOCl (3.50 eV) > BiOBr (2.88 ev) > BiOI (1.79 eV). [67,68] In addition, one can see from Figure 3a that the CB is situated at À1.10 eV for BiOCl, À0.53 eV for BiOBr, and þ0.60 eV for BiOI, in comparison to the reduction potential (À0.28 eV) of N 2 /NH 3 . This suggests that both BiOCl and BiOBr are photocatalytically active toward the reduction of N 2 to NH 3 , and BiOBr is the optimal catalyst among the series for nitrogen fixation (whereas the pristine form is inactive, doped and/or defective BiOI also exhibits apparent photocatalytic activity toward nitrogen fixation, vide infra).
The photocatalytic activity toward nitrogen fixation can be further manipulated by the elemental composition of BiOX. For instance, Bi-rich Bi a O b X c has stood out as a unique family of photocatalysts, which also exhibit a layered structure, again, due to strong intralayer bonding and weak interlayer nonbonding (van der Waals force) interactions. [ Figure 3c, [70] where [Bi─O] layers are sandwiched between two layers of chloride ions, forming a self-induced electric field, similar to BiOCl. [71] Experimentally, the unique band structures of these Bi-rich  [46] Copyright 2015, American Chemical Society. Reproduced with permission. [52] Copyright 2017, John Wiley & Sons. Reproduced with permission. [60] Copyright 2020, American Chemical Society. materials are found to be conducive for effective photocatalytic nitrogen fixation (Figure 3a). [72] Specifically, in Bi a O b Cl c and Bi a O b Br c , the higher Bi content results in a downward shift of the CB minimum and an upward shift of the VB maximum edge, and therefore, a narrower bandgap than that of BiOCl and BiOBr, whereas a wider bandgap is observed with Bi a O b I c than with BiOI. [73,74] The internal electric field (IEF) may also contribute to the photocatalytic activity. [64] Since the layered crystal structure provides an open space to polarize related atoms and orbitals, the IEF is induced perpendicular to the [Bi 2 O 2 ] plane and the halogen anion planes in BiOX ( Figure 2). Such an inherent IEF can improve the diffusion of electron-hole pairs, thereby increasing the carrier mobility and reducing the recombination rate, which leads to improved efficiency of photocatalytic nitrogen fixation. [48,60,75] For instance, Li et al. [70] demonsrated that manipulation of the IEF amplitude can be exploited as a strategy to improve the activity of the photocatalysts. They prepared single crystal Bi 3 O 4 Cl nanosheets with rich {001} plane exposure (Figure 3d). When the Bi 3 O 4 Cl nanosheets were under visible light photoirradiation, electrons were excited from the VB to CB, leaving holes in the VB. The polarizationinduced IEF along the {001} crystal orientation of the nanosheets was found to promote the charge separation and facilitate the transfer of charge carriers from the bulk to the surface for subsequent photocatalytic reactions. Thus, an increasing amount of the {001} facet exposure led to improved separation and transfer efficiency of photogenerated electronhole pairs and eventual photocatalytic activity, due to an enhanced IEF.

Preparation of Bismuth Oxyhalides
As shown in Table 1, a range of methods have been reported for the preparation of BiOX, [7,8,47,52,60, which mainly include solvothermal process, precipitation process, reverse microemulsion, low-temperature chemical vapor transport, water guided self-assembly, and so on. The morphologies of the products vary accordingly, including nanospheres, nanowires, nanotubes, nanosheets, and other 2D and 3D nanostructures.
Chemical precipitation method has been widely used for the synthesis of BiOX photocatalysts due to its simple synthetic conditions, high efficiency, and low energy consumption. [76,77] Li et al. used Bi(NO 3 ) 3 ·5H 2 O as the bismuth source and KI as the halogen source to successfully synthesize a 3D hierarchical structure composed of BiOI nanoplates by chemical precipitation at room temperature. [78] The obtained BiOI showed effective photocatalytic activity under visible light for the degradation of model pollutants, such as methyl orange and phenol. Despite the apparent advantages, chemical precipitation also has its shortcomings, such as particle agglomeration, small surface area, impurity formation, uncontrollable morphology, and so on.
Hydrolysis is also a common method for the preparation of BiOX, which is based on the reaction between a Bi salt (e.g., Bi(NO 3 ) 3 and BiX 3 ) and oxyhalide or water. [48,77,79] The advantage of this method is that BiOX of various sizes can be easily synthesized in a simple reactor. For instance, Wang et al. used a hydrolysis method to prepare ultrafine Bi 5 O 7 Br nanotubes with a diameter of 5 nm, [52] which involved two major steps: 1) a bismuth-bromo-oleylamine (Bi-Br-OA) complex was formed by mixing Bi and Br ions in oleylamine and 2) with the gradual addition of water, the complex was hydrolyzed producing ultrafine nanotubes. However, the catalysts synthesized by the hydrolysis method in general have poor dispersibility, and surfactants are usually added to make them dispersible. Therefore, the hydrolysis method is solvent dependent.
In addition to chemical precipitation and hydrolysis, hydrothermal method is also commonly used to synthesize BiOX photocatalysts, where the crystallinity, morphology, and particle size are controllable at low temperatures. [48,79] A range of hydrothermal reaction parameters, including reaction duration, temperature, surfactant, precursor concentration, and pH, have been known to play a critical role in defining the physical and chemical properties and ultimately the photocatalytic performance of BiOX. For instance, Jiang et al. successfully synthesized flake-like BiOBr by hydrothermal treatment of Bi(NO 3 ) 3 ·5H 2 O and KBr precursors in acetic acid at 150 C for 18 h, which showed excellent photocatalytic activity toward the degradation for methyl orange. [80] Nevertheless, one can see that since the  reaction can be impacted by a number of experimental parameters, it is tedious to identify optimal reaction conditions. [79] Such BiOX-based catalysts exhibit apparent photocatalytic activity toward nitrogen fixation. [99][100][101][102][103][104][105][106][107][108][109][110][111][112][113][114] As depicted in Table 2, in a previous study, Li et al. [46] prepared BiOBr nanosheets with abundant OVs on the exposed {001} surface and observed a photocatalytic nitrogen fixation rate of 104.2 μmol h À1 g À1 under visible light photoirradiation. In a later study, Wang et al. [52] prepared Bi 5 O 7 Br nanotubes, and observed an improved photocatalytic nitrogen fixation rate of 1.38 mmol h À1 g À1 . Bi et al. [99] found that the visible light nitrogen fixation rate of H-BiOBr (BiOBr hollow microspheres prepared by the same calcination method but in a H 2 atmosphere) can reach 50.8 μmol g À1 h À1 . Di et al. [100] studied the promotion effect of structural defects in facilitating electron-hole separation in Bi 3 O 4 Br nanosheets and observed a photocatalytic hydrogen evolution rate of 25.4 μmol L À1 . Liu et al. [101] reported photocatalytic conversion of nitrogen and water to ammonia at the Bi 4 O 5 Br 2 /ZIF-8 three-phase interface at a rate of 327 μmol L À1 h À1 g À1 . In addition, Li et al. [60] showed that the nitrogen fixation rate of Bi 5 O 7 Br can be increased to 12.72 mmol g À1 h À1 by an increase in the OV concentration. The application of BiOCl in photocatalytic nitrogen fixation was first reported in 2016 by Li et al., [102] where Reproduced with permission. [92] Copyright 2011, Elsevier. BiOCl nanosheets were synthesized and involved in the protonassisted electron-transfer pathway. Rong et al. [68] reported the application of Bi 2 Te 3 /BiOCl in photocatalytic nitrogen fixation in 2018, with a nitrogen fixation rate of 315.9 μmol L À1 h À1 . Guo et al. [67] prepared a 2D/2D ZnIn 2 S 4 /BiOCl heterostructure for N 2 fixation at a rate of 14.6 μmol g À1 h À1 . Wu et al. [103] reported that the abundant OVs and dominant {001} surface of Br-doped porous BiOCl microchips led to an improved performance of nitrogen fixation. Doped/defective BiOI has also been used for photocatalytic nitrogen fixation. In 2019, Zeng et al. [5] reported the preparation of interstitial carbon-doped BiOI, which exhibited a photocatalytic nitrogen fixation rate of 311 μmol g À1 h À1 . In 2020, Lan et al. [8] prepared H-Bi 5 O 7 I microspheres with rich OVs for visible light nitrogen fixation, and observed a rate of 162.48 μmol g À1 h À1 . In general, the apparent quantum efficiency (AQE) of photocatalytic nitrogen fixation is low under ambient temperature and pressure. As shown in Table 2, Wang et al. [52] synthesized Bi 5 O 7 Br nanotubes and observed an AQE of only 2.3% at 420 nm, the highest ever reported. But, this is still far lower than actual industrial application requirements. [104] Nevertheless, such bismuth oxyhalides have been found to exhibit good stability and reusability (Table 2).
Notably, the formation of OVs in BiOX facilitates oxygen production, [105] nitrogen fixation, [1,52,106] CO 2 reduction, [107,108] and organic pollutant degradations. [108,109] Despite the apparent advantages, limitations of the photocatalytic performance of BiOX remain, such as relatively large bandgaps that limit the solar light utilization efficiency, high recombination of photogenerated electron-hole pairs, and limited surface active centers. Further structural engineering is needed to optimize the performance, as detailed below.

Sacrificial Agents and Nitrogen Sources
Solar ammonia synthesis is usually conducted in two ways. The first is in ultrapure water, [46,52,60] and the other is in the presence of sacrificial agents (such as methanol, ethanol, or isopropanol) in water. [31,98,110] In the latter, a sacrificial agent is added to the photocatalytic reaction system (aqueous solution) to consume photogenerated holes and prevent photogenerated electrons from recombining with holes. For instance, Gao et al. [111] used the Nessler method and cation exchange chromatography (CEC) to quantify and compare ammonia production with and without sacrificial agents. For alcohol sacrificial agents, they may be oxidized into carbonyl-containing compounds, such as aldehydes and ketones, and form complexes with NH 3 , leading to an increase in the absorbance of the Nessler reagent. As the amount of ammonia produced in the traditional photocatalytic ammonia production experiment is extremely small (100-1000 μmol g À1 h À1 , Table 2), the coexistence of sacrificial agents and the Nessler reagent may compromise the rationality and accuracy of the positive detection of ammonia production. By contrast, CEC is far more reliable and accurate, as these organic compounds do not interfere with CEC analysis. [111] Furthermore, in recent years, tracking the source of nitrogen in ammonia production has also been the focus of extensive www.advancedsciencenews.com www.advenergysustres.com studies. Strict experimental protocols need to be followed to ensure the credibility of the results and minimize pollution of the photocatalysts. For instance, Zhang et al. [112] found that the ammonia concentration in tap water (349 μg L À1 ) was actually much higher than that of stale ultrapure water (31 μg L À1 ), stale redistilled water (52 μg L À1 ), and deionized water (48 μg L À1 ) (Figure 4a), whereas no ammonia was detected in fresh ultrapure water or fresh redistilled water. Thus, fresh ultrapure water or fresh redistilled water must be used in the photocatalytic experiments. In addition, prior to actual reactions, ammonia or amino groups may chemisorb or physisorb onto the photocatalyst surface, and such pollution may lead to interference with the detection of ammonia, and promote reactions that would not otherwise occur on the surface of clean catalysts, as observed with a series of photocatalysts (Figure 4b-d). Note that adsorbed ammonia can be easily removed within five washing cycles. Therefore, before the catalyst is loaded into the aqueous reaction, it should be rinsed with fresh ultrapure water until no ammonium is detected. In addition, it is recommended to conduct a control experiment in the presence of Ar and 15 N 2 to prove that NH 3 is indeed produced from the provided N 2 rather that from other contaminations (e.g., air, gloves, chemical reagents, glasswares, etc).

Mechanism of Nitrogen Fixation
Through the analysis of OVs and different nitrogen fixation pathways, the mechanism of photocatalytic nitrogen fixation is rationalized and summarized in the following section.

Role of OVs
OVs can be produced by oxygen detachment from the surfaces of transition metal oxides, affect the materials physical and chemical properties with undercoordinated atoms, and promote the formation of catalytic active sites. [51,115] In photocatalytic nitrogen fixation, one major challenge is the adsorption and activation of inert nitrogen molecules under ambient conditions. [73,116] This can be facilitated by surface OVs. In fact, vacancies or defects on the surface of photocatalysts can act as trapping sites for photogenerated electrons or protons to inhibit charge recombination, [117,118] and the surviving charges can participate in the adsorption of important reaction intermediates and boost the photocatalytic performance. [107] Note that the residual electrons are usually selectively localized near the Bi atoms around the OVs, rather than uniformly distributed in the entire crystal, leading to the coexistence of OVs and low-valence Bi. Thus, the OVs on the BiOX surface can serve as active sites for nitrogen fixation, where molecular nitrogen is adsorbed and activated, and photogenerated electrons are transferred to the surface OVs to achieve efficient nitrogen reduction. In addition, in the band structure, the impurity energy level composed of Bi p orbital and O p orbital typically appears below the bottom of the CB, which inhibits the recombination of photogenerated electrons and holes. [46] Indeed, extensively studies have shown that the OVs in oxide-based photocatalysts directly affect the nitrogen fixation efficiency. [119] In an early study, Di et al. [100] calculated the formation energy of OVs by using two models, one with a Bi atom defect and the other without a Bi atom defect. It was found that the Bi www.advancedsciencenews.com www.advenergysustres.com atom defect can effectively reduce the oxygen hole formation energy, so that oxygen-rich materials can be easily obtained (Figure 4e-g). [120] On the basis of these calculations, it was argued that by controlling the surface bismuth defect concentration, a tunable OV can be achieved to tailor the electronic structure of Bi 3 O 4 Br and serve as surface charge separation centers to further promote the photocatalytic activity. [5,73] In another study, Lan et al. [8] reported that the abundant OVs in bismuth-rich BiOI microspheres contributed to efficient nitrogen fixation under visible light irradiation. Experimentally, the Bi 5 O 7 I powders were annealed in a H 2 atmosphere at 300 C for 4 h to produce hydrogenated Bi 5 O 7 I (denoted as H-Bi 5 O 7 I). Both Bi 5 O 7 I and H-Bi 5 O 7 I displayed electron paramagnetic resonance (EPR) signals at g ¼ 2.0 (Figure 5a), suggesting the formation of OVs, and the OV concentration was higher in H-Bi 5 O 7 I than in Bi 5 O 7 I. This is likely because a large number of oxygen atoms were detached from the Bi 5 O 7 I surface during hydrogenation. [46,120,121] In photocatalytic nitrogen fixation tests (Figure 5b), a large amount of NH 3 was produced with H-Bi 5 O 7 I after 180 min of visible light irradiation at a nitrogen fixation rate of 162.48 μmol g À1 h À1 , and the amount of NH 3 produced increased approximately linearly with the illumination time. Such a performance was markedly better than that of pristine Bi 5 O 7 I, suggesting the significant role of OVs in the adsorption and activation of N 2 molecules. Notably, the OVs can also impede the recombination of charge carriers and facilitate interfacial charge transfer from the semiconductor to N 2 . In addition, H-Bi 5 O 7 I exhibited excellent stability in nitrogen fixation (Figure 5c), where no significant attenuation of the photocatalytic activity was observed even after four cycles, and no obvious change was detected in X-ray diffraction (XRD) measurements of the used H-Bi 5 O 7 I sample (Figure 5d).
In another study, Li et al. [60] examined the effect of OV concentration on the performance of visible light-driven nitrogen fixation catalyzed by Bi (Figure 6a). The corresponding N 2 adsorption energies were also calculated. From the reaction energy diagram, a nitrogen fixation pathway was proposed, *N 2 ! *NNH ! *NNH 2 ! *NNH 3 ! *N þ NH 3 ! *NH! *NH 2 ! *NH 3 ! NH 3 (Figure 6b and c), where one can see that N 2 adsorption on Bi 5 O 7 BrÀO was indeed favored with the most negative adsorption free energy (À0.017 eV) for the initial N 2 activation (*N 2 ), in comparison to À0.008 eV for pristine Bi 5 O 7 Br and À0.001 eV for Bi 5 O 7 BrþO (Figure 6d). The *N À NH 2 ! *N À NH 3 step exhibits the highest energy barrier (Figure 6b), suggesting that this is the rate-determining step in this reaction pathway. [122] It can be confirmed from these calculations that OVs indeed play an important role in determining the nitrogen fixation performance. The higher the OV concentration, the more favorable the reduction of N 2 to NH 3 . [123]

Mechanism and Pathways of Nitrogen Fixation
As shown in Figure 7a, the catalytic activity of semiconductor photocatalysts arises from photogenerated carriers (i.e., electronhole pairs), [124,125] where holes reside in the VB and electrons in www.advancedsciencenews.com www.advenergysustres.com the CB. [126] Migration of the these carriers to the catalysts surface renders it possible to undergo redox reactions with N 2 , H 2 O, etc. [127] Therefore, the mechanism of photocatalytic nitrogen fixation is similar to that of water decomposition and CO 2 reduction. [128] First, some oxygen atoms on the catalyst surface are separated from the holes by photoirradiation, forming oxygen holes at the top of the VB [129] ; concurrently, the electrons become excited to the CB. [130] When N 2 and H 2 O molecules are adsorbed to the holes, the N≡N bond and H─O bond become activated and broken by the photogenerated electrons, reducing N 2 to NH 3 . [131] In this reaction, H 2 O loses electrons on the VB and is oxidized to form O 2 and H þ , and N 2 and protons accept electrons on the CB to produce NH 3 . [100,132] Note that nitrogen adsorption is facilitated by the formation of σ and π bonds by the d orbitals of transition metals and p orbitals of N 2 . The π back bonding from the transition metals to N 2 weakens the N≡N triple bond and enhances the strength of metal─N bond. [133] In addition, DFT calculations show that the {001} plane OVs tend to adsorb N 2 in the end-on configuration, with only a single N atom bonded to two Bi atoms, whereas for OVs on the {010} plane, N 2 adsorption adopts a side-on form, with the two N atoms bonded to three adjacent Bi atoms (Figure 7b-d). In the latter, because of simultaneous adsorption and activation of both N atoms, the N─N bond length was elongated to 1.198 Å from 1.09 Å, in comparison to only 1.137 Å in the former. Thus, the {010} plane OVs are more effective in activating molecular nitrogen than those on the {001} plane. Consistent results were obtained in the calculations of the energy of N 2 adsorption. [102] Figure 7e shows the mechanism of photocatalytic nitrogen fixation by Bi 5 O 7 Br. Under photo irradiation, the nitrogen fixation reaction on the Bi 5 O 7 Br surface can be divided into two processes: 1) water splitting and 2) reduction of N 2 to NH 3 . In general, catalytic nitrogen fixation at a heterogeneous surface can proceed by two recognized mechanisms: associative mechanism and dissociative mechanism. [8,119,121,134,135] Theoretically, the Haber-Bosch process follows the dissociative mechanism (Figure 7f ). [24,136] In this process, NH 3 is formed by the combination of N and H þ after the bond breaking of N 2 . Because of the high bond energy of N≡N, a significant energy input is required in this initial step. [137] Two different associative N 2 reduction pathways have been considered, i.e., distal and alternating pathways, invoking distinctly different intermediates. As shown in Figure 7f, in the alternating pathway, two N atoms are hydrogenated alternately, forming hydrazine intermediates after four hydrogenation steps, and only the first NH 3 is released in the fifth step. In comparison, in the distal pathway, a single N atom of N 2 is hydrogenated in three steps until the first NH 3 is released, and then, the remaining nitride-N is hydrogenated three more times to produce a second NH 3 . [138] Protonation tends to take place on the nitrogen atom far away from the catalyst surface in the distal associative mechanism. However, in the alternating pathway, protons are in turn added onto the two Reproduced with permission. [60] Copyright 2020, American Chemical Society.
www.advancedsciencenews.com www.advenergysustres.com nitrogen atoms of N 2 before one of the nitrogen atoms is converted into NH 3 by the N─N bond cleavage. However, DFT calculations indicate that the detailed pathways of heterogeneous catalytic nitrogen reduction reaction in aqueous solution vary with the catalytic systems and catalyst materials. [139]

Strategies for Activity Enhancement of BiOX Photocatalysts
As shown earlier, bismuth oxyhalides have been proved to be promising photocatalysts toward N 2 reduction to NH 3 . However, due to the large energy bandgap, most Bi-based photocatalysts exhibit only a low utilization rate of solar energy and a high recombination rate of photoexcited charge carriers, which compsomises the photocatalytic nitrogen fixation efficiency. [5,8,24,140] To mitigate these issues, [53] a variety of strategies have been developed, such as elemental doping [141] and formation of heterostructures, [53,140] whereby the bandgap and electronic structure of the materials can be readily manipulated.

Elemental Doping
Elemental doping is one of the most important means of structural engineering. [142] Indeed, a number of studies have been conducted where doping is used to manipulate the photocatalytic nitrogen fixation performance of BiOX. For instance, Liu et al. [106] studied the performance of Fe-doped BiOBr in photocatalytic nitrogen fixation.  Figure 8a,b shows the high-resolution transmission electron microscopy (HRTEM) images of BiOBr and Fe-BiOBr, respectively. From the lattice fringes in Figure 8a, the interplanar distance of the BiOBr(101) facets is estimated to be 0.344 nm, which diminishes slightly to 0.340 nm in Fe-BiOBr (Figure 8b), suggesting that the incorporation of Fe into BiOBr leads to lattice compression. [143] Low-temperature electron paramagnetic resonance (EPR) measurements were then conducted to evaluate and compare the OV concentration in Fe-BiOBr, and BiOBr. From Figure 8c, one can see that BiOBr exhibits a clearly defined signal at 3400 G with g ¼ 1.999 that is characteristic of OVs. [46,52] After Fe doping, the signal became intensified significantly (Fe-BiOBr), whereas the signal Reproduced with permission. [102] Copyright 2016, Royal Society of Chemistry. e) Schematic diagram of photocatalytic nitrogen fixation catalyzed by Bi 5 O 7 Br. Reproduced with permission. [60] Copyright 2020, American Chemical Society. f ) Traditionally accepted mechanisms of nitrogen reduction to ammonia. Reproduced with permission. [24] Copyright 2020, Royal Society of Chemistry.
www.advancedsciencenews.com www.advenergysustres.com vanished altogether after the sample was calcined in air at 300 C for 5 h, suggesting the disappearance of OVs (OV-free Fe-BiOBr). This indicates that Fe doping promoted the generation of OVs on the BiOBr surface. The material structures were further characterized by X-ray photoelectron spectroscopy (XPS) measurements. In the Fe 2p spectrum of Fe-BiOBr (Figure 8d), two peaks can be resolved at 713.5/727.4 eV and 710.6/723.4 eV, corresponding to the 2p 3/2 electrons of Fe(III) and Fe(II) species, respectively. [144] In the Bi 4f spectra of Fe-BiOBr and pristine BiOBr (Figure 8e), two main peaks can be identified at 164.0 eV and 159.5 eV, due to the 4f 5/2 and 4f 7/2 electrons of Bi(III), respectively. Two additional peaks are observed in the spectrum of BiOBr at somewhat lower binding energies of 163.8 and 158.4 eV, corresponding to the OV-bound Bi atoms, [52,145] which have been known to be able to activate the N≡N triple bond. [52] In the Bi 4f spectrum of Fe-BiOBr, the two additional peaks at the higher binding energies of 166.4 and 161.1 eV are indicative of Bi at a higher valence state, [146] likely due to electron transfer to Fe through the Bi─O─Fe bond. That is, Fe withdrew electrons from the nearby O and Bi atoms to form electron-rich Fe(II), which is prone to donate the excess electrons to produce more stable halffilled 3 d orbitals of Fe(III). [147] Therefore, the OV-bound Fe(II) atoms in Fe-BiOBr are believed to be the active sites for N 2 photo fixation. Consistent results were obtained in DFT calculations. In the charge density map of Fe-BiOBr (Figure 8f ), the electron clouds of the O atoms around Fe are deformed and withdrawn by Fe, and the electron-rich Fe binds with N 2 and activates the triple bond by injecting the localized electrons (Figure 8g). [148] The N 2 photo fixation efficiency was evaluated under visible light without any sacrificial agent. As shown in Figure 9a, the performance of Fe-BiOBr (382.6 μmol À1 g À1 h À1 ) was eight times better than that of pristine BiOBr (51.68 μmol À1 g À1 h À1 ), and there was virtually no decline of the activity after four cycles of test (Figure 9b).
In another study, Zeng et al. [5] dissolved 1.94 g of Bi(NO 3 ) 3 ·5H 2 O in 40 mL of 1 M HNO 3 under stirring for 30 min. Then, 0.005 g of glucose was added under sonication for 15 min. Separately, 0.664 g of KI was dissolved into 20 mL of deionized water under stirring for 30 min, and gradually added to the Bi(NO 3 ) 3 solution under strong agitation. After the pH was adjusted to 7 with an NH 3 solution, the mixture was transferred to a 100 mL Teflon-lined stainless steel autoclave, and heated at 160 C for 24 h to obtain C-BiOI powders, which were collected by centrifugation, rinsed with water and ethanol four times, and dried at 60 C for 18 h. Variable carbon doping was achieved by changing the initial feed of glucose. Four C-BiOI samples were prepared at the C to Bi molar ratios of 1:30, 1:15, 1:6, and 1:2 (denoted as C-BiOI-1, C-BiOI-2, CBiOI-3, and C-BiOI-4, respectively). From the transient photocurrent responses of the BiOI and C-BiOI samples (Figure 10a), one can see that under simulated sunlight, all C-BiOI photocatalysts show stronger photocurrent responses than pristine BiOI. This is precisely because carbon doping destroyed the periodicity of the crystal lattice and promoted the generation of OVs, which changed the charge distribution in the catalyst and enhanced the IEF, leading to enhanced separation of charge carriers and improved photocatalytic activity. [150]  Reproduced with permission. [106] Copyright 2020, American Chemical Society. Interestingly, carbon doping also increased the specific surface area, another contribution to the improved photocatalytic nitrogen fixation performance. From the scanning electron microscopy (SEM) images in Figure 10b and c, C-BiOI can be seen to display a morphology similar to that of BiOI (Figure 10d and e), but the average size of the crystals is smaller, corresponding to a larger specific surface area, which is beneficial to the photocatalytic reaction. This was indeed manifested in photocatalytic nitrogen fixation experiments, where ethanol was used as a hole scavenger and water as the proton source. [151] From Figure 10f, it can be seen that after carbon was incorporated into the BiOI lattice, the nitrogen fixation rate varied significantly, first increasing and then decreasing with the increase in carbon content. C-BiOI-3 was found to exhibit the highest nitrogen fixation rate at 311 μmol L À1 h À1 , 3.7 times that of pristine BiOI. Furthermore, one can see from Figure 10g that after five test cycles, the photocatalytic performance of C-BiOI-3 only decreased slightly, indicating good stability.
In summary, the fact that the C-doped BiOI samples show significantly improved photocatalytic N 2 fixation, as compared to the pristine counterparts, indicates that elemental doping may be an effective strategy to improve the photocatalytic performance of bismuth oxyhalides toward nitrogen fixation. [5]

Heterojunction Composites
The photocatalytic nitrogen fixation performance can also be enhanced by manipulation of the bandgap structure through the formation of heterojunctions, i.e., the formation of composites by combining two semiconductors with different band structures. Figure 11 shows various types of heterostructures produced in bismuth-based photocatalysts reported so far toward N 2 fixation, such as cocatalyst loading, type I heterojunctions, type II heterojunctions, and Z-scheme heterojunctions. [152][153][154] The resultant structural changes lead to new properties that are not available in single phase materials. For example, if a Figure 9. a) Visible light-driven N 2 fixation on Fe-BiOBr and BiOBr, compared with the results of control experiments with Fe-BiOBr in the dark, Ar bubbling, and OV-free Fe-BiOBr. b) Cycling tests of N 2 fixation on Fe-BiOBr. Reproduced with permission. [106] Copyright 2020, American Chemical Society. c) Photocatalytic NH 3 production rate over the BiOCl NSs-Fe-x% in the first 1 h. d) Photocatalytic cycling tests for BiOCl NSs-Fe-5%. Reproduced with permission. [149] Copyright 2019, American Chemical Society. wide-bandgap semiconductor is modified by a narrow-bandgap semiconductor, or two kinds of narrow-bandgap semiconductors are combined, the range of photo response of the composite catalysts will be effectively expanded, leading to an enhanced photo energy utilization rate. Therefore, the construction of heterojunctions is effective to boost the performances of a range of photocatalysts. [24] Table 3 summarizes recent studies on photocatalytic nitrogen fixation by BiOX catalysts by forming heterojunctions with other materials. For instance, Xiao et al. [140] successfully constructed MoO 2 /BiOCl nanocomposites by electrostatic adsorption and observed an excellent photocatalytic nitrogen fixation efficiency when water was used as the proton source. This was due to the strong interactions between BiOCl nanoplates and MoO 2 nanosheets that manipulated the electronic structure of the interface and provided active sites for catalytic reactions. The Mo─O─Bi bond formed at the interface of MoO 2 and BiOCl acted as a bridge for electrons and promoted the separation and migration of photogenerated charge carriers (Figure 12a). Also, it can be seen from Figure 12b that in comparison with BiOCl, MoO 2 /BiOCl showed stronger adsorption of N 2 , which greatly increased the reduction rate of N 2 . Similarly, in MoO 2 / BiOCl, it can be seen that MoO 2 exhibited a nanosheet morphology, and was supported on the surface of BiOCl nanoplates, as evidenced in transmission electron microscopy (TEM) and SEM measurements (Figure 12c-e). From Figure 12f and g, the maximum rate of ammonia production was %35 μmol g À1 h À1 , eight times higher than that of the original BiOCl and six times higher than that of MoO 2 . Mechanistically, the introduction of MoO 2 enhanced the adsorption and activation of N 2 , due to the formation of Mo─O─Bi bonds at the MoO 2 /BiOCl interface. Moreover, MoO 2 nanosheets possessed metal-like properties that accelerated electron transfer of BiOCl, and the Mo atoms in MoO 2 can coordinate with N 2 , thereby improving the adsorption and activation of N 2 . Under photoirradiation, the photogenerated holes of BiOCl might transfer to the (110) crystal surface for water oxidation, [155] and the photogenerated electrons migrated Reproduced with permission. [5] Copyright 2018, Elsevier. to the interface where MoO 2 and BiOCl were situated for nitrogen fixation reaction (Figure 12a). In another study, Rong et al. [68] prepared a type I heterostructure composed of Bi 2 Te 3 and BiOCl. From the SEM and TEM images in Figure 13a-d, one can observe that both the pristine BiOCl and Bi 2 Te 3 /BiOCl composite exhibited a similar flowerlike morphology. Yet, the electronic structure and photo response varied markedly, as the CB and VB of the BiOCl sample was identified at -1.1 and þ2.4 eV, respectively, [156] whereas those of Bi 2 Te 3 at þ0.57 and þ0.72 eV, respectively. [157] From Figure 13e, one can see that because of a small bandgap of only 0.15 eV, Bi 2 Te 3 can absorb both UV and visible light, but the low CB (0.57 eV), as compared to that of N 2 /NH 3 (À0.092 eV, Figure 2), renders it inactive for N 2 reduction to NH 3 . [158] For BiOCl, the relatively large bandgap (3.5 eV) indicates that photo absorption is limited to the UV range, whereas the high CB potential (À1.1 eV, Figure 2) is appropriate for N 2 reduction to NH 3 , and the high recombination rate of photogenerated holes and charges in BiOCl limits the photocatalytic activity. For the Bi 2 Te 3 /BiOCl composite, the photogenerated charge carriers of Bi 2 Te 3 effectively suppress the recombination of photogenerated electrons and holes of BiOCl, which prolongs the lifetime of CB electrons in BiOCl and improves the photocatalytic nitrogen fixation performance. From Figure 13f, one can see that under UV radiation, Bi 2 Te 3 /BiOCl exhibits an effective ammonia release rate of 315.9 μmol L À1 h À1 , much higher than those of BiOCl (98.2 μmol L À1 h À1 ) and Bi 2 Te 3 /BiOCl (24.3 μmol L À1 h À1 ).
Xue et al. [53] studied the photocatalytic nitrogen fixation performance of Bi 2 MoO 6 /OV-BiOBr composites, where type II heterojunctions were formed between n-type Bi 2 MoO 6 nanorods and p-type OV-rich BiOBr nanosheetss. MoO 3 nanorods were first prepared by a simple hydrothermal method, and refluxed in a Bi(NO 3 ) 3 aqueous solution at 120 C for 4 h to generate Bi 2 MoO 6 nanoparticles. Finally, OV-BiOBr nanosheets were grown in situ onto the Bi 2 MoO 6 nanosheets by a solvothermal method to produce Bi 2 MoO 6 /OV-BiOBr composites, which effectively converted N 2 to NH 3 under ambient conditions without using ultrapure water.
From the SEM and TEM images in Figure 14a-c, one can see that the obtained Bi 2 MoO 6 /OV-BiOBr composite was composed of Bi 2 MoO 6 nanorods (about 400-600 nm in diameter) grafted with ultrathin OV-BiOBr nanopetals (10-20 nm in thickness). Electrochemical impedance spectroscopy (EIS) measurements were then conducted to examine the charge-transfer characteristics and interfacial reaction resistance of Bi 2 MoO 6 /OV-BiOBr composite, Bi 2 MoO 6 nanorods, and OV-BiOBr nanosheets both in the dark and under photoirradiation with a 300 W Xe lamp without water absorption. From the Nyquist plots in Figure 14d,e, one can see that in comparison with the original Bi 2 MoO 6 nanonodes and OV-BiOBr nanosheets, the Bi 2 MoO 6 /OV-BiOBr composite exhibited a markedly reduced semicircle radius (i.e., charge-transfer resistance, R CT ) with and without light illumination, suggesting that the heterostructure can promote interfacial charge transfer. Moreover, the R CT of the Bi 2 MoO 6 /OV-BiOBr composite was smaller under photoirradiation than that in the dark, likely due to the photo generation of an increasing number of charge carriers. From the UV-vis diffuse reflectance spectroscopy (DRS) measurements (Figure 14f ), the absorption threshold (485 nm) of Bi 2 MoO 6 / OV-BiOBr composite can be seen to be somewhat redshifted, Figure 11. Illustration of different heterostructures for photocatalytic nitrogen reduction to ammonia. Reproduced with permission. [24] Copyright 2020, Royal Society of Chemistry. in comparison to those of Bi 2 MoO 6 (481 nm) and OV-BiOBr (427 nm), suggesting enhanced efficiency of photo energy utilization and therefore the N 2 fixation performance. In addition, from Figure 14g-i, one can see that Bi 2 MoO 6 /OV-BiOBr exhibits a higher specific surface area than Bi 2 MoO 6 nanorods and BiOBr nanosheets, implying enhanced accessibility to the photocatalytic active sites. Moreover, the large number of OVs on the surface of the OV-BiOBr nanosheets is conducive to the adsorption and activation of N 2 molecules. [158] Figure 14j and k show the solar-driven N 2 fixation tests of Bi 2 MoO 6 /OV-BiOBr in pure water and under visible light irradiation (300 W Xe lamp with a 420 nm cutoff filter), which yielded an ammonia production rate of 90.7 μmol g À1 h À1 and 81.0 μmol g À1 h À1 , respectively, far exceeding those by the OV-BiOBr nanosheets (31.2 μmol g À1 h À1 ) and Bi 2 MoO 6 nanocrystals (3.0 μmol g À1 h À1 ) alone. Separately, Chen et al. [153] synthesized Ag/AgBr/Bi 4 O 5 Br 2 nanocomposites by a combination of hydrothermal and ion   (Figure 15b), the lattice fringes of both Bi 4 O 5 Br 2 and AgBr can be clearly identified. AgBr is a wellknown, highly efficient photocatalyst and can effectively prevent the recombination of electron-hole pairs. [159] Figure 15c shows    . Reproduced with permission. [53] Copyright2019, Royal Society of Chemistry.
www.advancedsciencenews.com www.advenergysustres.com Reproduced with permission. [153] Copyright2019, Royal Society of Chemistry. five cycles (Figure 15f ). These results indicate that the AgBr/ Bi 4 O 5 Br 2 composite might serve as viable catalysts in photocatalytic N 2 fixation, due to its high photoactivity and stability. This is largely ascribed to the formation of Z-type heterojunctions (Figure 15g), where electrons were accumulated in the Bi 4 O 5 Br 2 CB, holes remained on the AgBr VB, and Ag nanoparticles likely served as a bridge for charge transfer between the different semiconductors. [162] The photogenerated electrons of AgBr may migrate to Ag nanoparticles through the Schottky barrier, and react with h þ from Bi 4 O 5 Br 2 VB, thereby effectively promoting the separation of photoinduced electron-hole pairs for photocatalytic reactions.

Summary and Perspectives
For a sustainable future, photocatalytic nitrogen reduction provides a promising alternative to the traditional Haber-Bosch process in the production of NH 3 . Despite substantial progress in recent years, significant challenges remain in the development of efficient photocatalysts for artificial ammonia production. [163] Bismuth-based photocatalysts have been proved to be viable for photocatalytic conversion of nitrogen to ammonia, due to the ready formation of OVs that facilitate the adsorption and activation of N 2 , and the performance can be further enhanced by deliberate structural engineering, such as elemental doping and formation of heterojunctions. It should be recognized that the amount of ammonia produced over the reported photocatalysts has remained small so far. To render the technology feasible for practical applications, significant breakthroughs are needed, by taking advantage of recent progress in the studies of reaction mechanism and catalyst engineering. [126,156] In particular, as experimental artifacts may interfere with the reliable analysis of photocatalytic nitrogento-ammonia conversion, [24,164] new synthetic and testing protocols are urgently called for. Mechanistically, the introduction of OVs over bismuth oxyhalides has been known to not only enhance the adsorption and activation of N 2 but also expand the wavelength range of light absorption and electrical conductivity, thereby facilitating the generation and spatial separation of photogenerated electrons and holes. Nonetheless, a complete understanding of the reaction mechanism and pathways of photocatalytic nitrogen fixation remains lacking. [165] In fact, as multifarious factors can cooperatively contribute to the photocatalytic activity, it is challenging to pinpoint the kinetic bottleneck for photocatalytic nitrogen fixation to ammonia. [8] In this regard, ingenious design of the photocatalysts with spatially decoupled light absorption unit and surface active sites is highly desired, which would make it possible to deconvolute the contributions of catalytic activity and light absorption to the overall photocatalytic performance of N 2 -to-ammonia conversion. Thus, to differentiate the critical contributions of the varied factors to the photoactivity of N 2 fixation to ammonia, rational design of bismuth-based composite photocatalysts with integrated functional components at a system level is crucial to accelerate the progress of nitrogen photoreduction to ammonia. [134] In addition, most early studies are focused on improving the conversion efficiency of N 2 to ammonia in the presence of sacrificial agents, and the studies of photocatalysts for the conversion of N 2 to ammonia and molecular oxygen with pure water have been relatively scarce, which is the ultimate goal of photocatalytic utilization of solar energy and remains a challenge. [166] For effective reduction of N 2 , another key factor that affects the overall photocatalytic reaction efficiency is the oxidation of water to O 2 by photogenerated holes. Considering the slow transfer of photogenerated holes in most photocatalysts to the surface, the multicharge transfer process of water oxidation is a potential rate limiting step. Therefore, for N 2 reduction and water oxidation, deliberate control of the spatial distributions of isolated active centers is likely a viable strategy. [167]