Visible Light-Responsive N-Doped TiO2 Photocatalysis: Synthesis, Characterizations, and Applications

Photocatalysis based on semiconductors has recently been receiving considerable research interest because of its extensive applications in environmental remediation and renewable energy generation. Various semiconductor-based materials that are vital to solar energy utilization have been extensively investigated, among which titanium oxide (TiO2) has attracted considerable attention because of its exceptional physicochemical characteristics. However, the sluggish responsiveness to visible light in the solar spectrum and the inefficient separation of photoinduced electron–hole pairs hamper the practical application of TiO2 materials. To overcome the aforementioned serious drawbacks of TiO2, numerous strategies, such as doping with foreign atoms, particularly nitrogen (N), have been improved in the past few decades. This review aims to provide a comprehensive update and description of the recent developments of N-doped TiO2 materials for visible light-responsive photocatalysis, such as (1) the preparation of N-doped/co-doped TiO2 photocatalysts and (2) mechanistic studies on the reasons for visible light response. Furthermore, the most recent and significant advances in the field of solar energy applications of modified N-doped TiO2 are summarized. The analysis indicated the critical need for further development of these types of materials for the solar-to-energy conversion, particularly for water splitting purposes.


Introduction
Developing advanced photocatalysts for large-scale solar energy conversion systems is an effective method to alleviate the increasing energy demand and environmental degradation issues. Since the pioneering work conducted by Fujishima and Honda in 1972 [1], who developed n-type rutile titanium oxide (TiO 2 ) for photoelectrochemical water splitting under light illumination, the use of semiconductorbased materials for photocatalysis has become a promising technology to convert inexhaustible solar energy into storable chemical energy. In the past few decades, numerous candidate semiconductor materials, such as TiO 2 , CdS, and C 3 N 4 , have been developed and utilized as photocatalysts and extensively investigated. Among these numerous photocatalysts, TiO 2 has attracted considerable attention because of its high photochemical stability, environmental friendliness, cost-effectiveness, and other excellent properties [2,3].
In nature, TiO 2 has eight types of crystal structures, namely, anatase, rutile, brookite, TiO 2 -H, TiO 2 -II, TiO 2 -III, TiO 2 -R, and TiO 2 -B [4]. Among them, only the first three types of crystal phases have been investigated, mostly for application purposes, because these titanium oxides can be naturally formed at atmospheric pressure. The structural parameters and schematic structures of the three structures are illustrated in Fig. 1 and Table 1, respectively. Figure 1 shows that, in all of the three crystalline structures of TiO 2 , titanium (Ti) atoms are sixfold coordinated to oxygen (O) atoms, forming the distorted TiO 6 octahedral structure [5]. Despite having similar constituent structural units, their physical properties are different: Rutile TiO 2 is the most thermodynamically stable phase, whereas the metastable phase of anatase and brookite TiO 2 can be transformed into the rutile phase after calcination at a certain temperature. In general, both rutile and anatase TiO 2 have a tetragonal structure, whereas brookite TiO 2 has an orthorhombic structure (as shown in Table 1). These differences in the crystallite parameters of TiO 2 with different crystal phases can result 1 3 in different electronic structures and physicochemical properties of TiO 2 , which further leads to different activities in photocatalytic reactions. For instance, anatase TiO 2 shows higher water reduction activity than rutile TiO 2 , whereas rutile TiO 2 exhibits better water oxidation activity than anatase TiO 2 [6]. At the same time, compared with anatase and rutile TiO 2 , brookite TiO 2 shows better performance in several photocatalytic reactions [7]. These differences in the application of photocatalysis originate from not only the differences in physicochemical properties caused by physical structures but also the differences in electronic structures.
Density functional theory simulations revealed that the calculated bandgap for anatase, rutile, and brookite TiO 2 (as shown in Fig. 2) ranges from 1.86 to 2.38 eV [9], which are smaller than the experimental values. These underestimations can be attributed to the fact that the generalized gradient approximation function usually underestimates the bandgap of semiconductors [10]. Moreover, anatase TiO 2 can be considered an indirect bandgap material, whereas rutile and brookite TiO 2 can be considered direct bandgap materials. In the electronic structures of the three types of TiO 2 , the top of the valence band (VB) is primarily composed of O 2p orbitals and a few Ti 3d orbitals, whereas the bottom of the conduction band (CB) is primarily composed of Ti 3d orbitals and a few hybrid O 2p orbitals. Therefore, under light irradiation, the electrons are excited from the O 2p orbitals in the VB to the Ti 3d orbitals in the CB of the TiO 2 semiconductor, leaving the holes in the VB. Subsequently, these charge carriers migrate to the surface of the catalyst and participate in the corresponding redox reactions.
Extensive works on metal cation doping of TiO 2 photocatalysts have been reported. These reports show that metal cation doping can create an energy level of donor level (i.e., the level higher than the original VB maximum) or acceptor level (i.e., the level lower than the original CB minimum) in the forbidden bandgap of TiO 2 , hence decreasing the bandgap and increasing the visible light sensitivity of TiO 2 . However, doping of metal cation can decrease the carrier mobility, which is the result of the generation of strongly localized d states in the bandgap [55], and generate the electron-hole recombination centers derived from the newly inserted energy level [56], thus leading to the significantly  reduced the photocatalytic performance. Moreover, the thermodynamic instability of the resulting cation-doped structure can cause photocorrosion in the photocatalysis process [57], which also hinders the practical application of TiO 2 photocatalysts. By contrast, the use of nonmetal elements instead of metal elements to dope the TiO 2 semiconductor is a feasible pathway to develop its visible light activity. Since the breakthrough work conducted by Asahi et al. [58], who first investigated the visible light-driven catalytic performance of N-doped TiO 2 for the photodegradation of methylene blue and gaseous acetaldehyde, the nonmetal elements that act as dopants have been extensively investigated. Among these various nonmetal dopants (e.g., B, C, N, F, S, and Cl), the synthesis of N-doped TiO 2 has become the focus of considerable research because of the similar structural characteristics of O and N, such as polarizability, electronegativity, and ionic radii [56].
Thus far, an increasing number of literature reviews on the utilization of TiO 2 for solar energy utilization have been reported [59][60][61][62][63][64]. For example, Ma et al. [59] focused on the strategies to improve the generation of H 2 and the photoreduction of CO 2 with H 2 O to other energy resources on TiO 2 -based catalysts. They also summarized the studies of the charge carriers properties of TiO 2 to better understand the dynamic characteristics of TiO 2 photocatalysts with different phases. Ullattil et al. [60] focused on the various synthetic routes, structures, morphological changes, and electronic structures of black TiO 2 nanomaterials and provided readers an overview of the different applications of black TiO 2 nanomaterials in the field of environmental technology. Along similar lines, Naldoni et al. [64] illustrated the basic concepts of effective design of reduced TiO 2 photocatalysts for H 2 generation. Kumaravel et al. [61] and Chen et al. [62] discussed six unique challenges that inhibit the application of TiO 2 in organic effluents. Nonetheless, these reviews ignored the important role and the recent progress of visible light-responsive TiO 2 in practical applications in indoor environments, particularly the N-doped TiO 2 , which has attracted considerable attention in visible light applications. Therefore, this review critically analyzes the improvement of visible light-sensitive N-doped TiO 2 photocatalysts. Moreover, the application prospect of these photocatalysts under visible light illumination, particularly the overall photocatalytic water splitting reaction, is reviewed.

Synthesis of N-Doped TiO 2
The key to the successful utilization of N-doped TiO 2 is to rationally select the synthesis strategies to prepare the target photocatalysts. Figure 3 shows the common methods involved in the synthesis of N-doped TiO 2 material. Based on the phase of the reactants, the strategies to prepare N-doped TiO 2 can be coarsely split into the wet chemistry and solid-phase methods.

Wet Chemistry Method
The wet chemistry method is a kind of strategy to prepare the target material through a chemical reaction with the participation of liquid, including the precipitation, hydrothermal, sol-gel, and electrochemical deposition methods. The sol-gel and hydrothermal methods are usually described as soft methods for doping N and/or other dopants into the TiO 2 crystal lattice, followed by calcination of the sample at high temperature. These methods usually have the advantages of being cleanroom-free, cost-efficient, and widely applicable. For this purpose, in a typical procedure, the doped N usually originates from nitric acid (HNO 3 ) [65,66], ammonium hydroxide (NH 4 OH) [67], urea (CH 4 N 2 O) [68,69], ammonium chloride (NH 4 Cl) [70,71], and triethylamine (TEA) [72,73] and the Ti precursors usually include titanium trichloride (TiCl 3 ) [74], titanium tetrachloride (TiCl 4 ) [75], titanium sulfate (Ti(SO 4 ) 2 ) [76], and titanium (IV) isopropoxide (TTIP) [77]. For example, Zhang et al. [65] selected HNO 3 as the N source and prepared nanocrystalline N-doped TiO 2 via a one-step solvothermal route with a low temperature (180 °C). In another effort, Panda's group [67] prepared N-doped yellow TiO 2 hollow spheres through a simple solvothermal method using aqueous titanium peroxocarbonate complex solution and NH 4 OH as doping N sources. Notably, the catalytic activity of the obtained N-TiO 2 hollow spheres was better than that of other reported N-TiO 2 samples with different morphologies, which may be owing to the increased surface area and microreactor environment and the decreased recombination for light scattering caused by the inner hollow structure. By adding ammonia water, Cheng et al. [78] synthesized N-doped TiO 2 through a simple solution-based strategy. Their results showed that N doping could not only effectively inhibit the phase transition of TiO 2 from anatase to brookite but also help enhance the visible light absorption. Marques et al. [68] noted that the specific surface area and photocatalytic performance of N-doped TiO 2 could be influenced by different synthesis methods and parameters, e.g., doping temperature and molar ratio of urea to TiO 2 . Di Camillo et al. [70] reported that N-doped TiO 2 nanofibers could be easily deposited via the electrospinning method. Their results showed that the light absorption properties of these nanofibers could be modified even with a small amount of N, which can result in good photocatalytic performance even when illuminated with visible light. Generally, the catalysts obtained by these methods have specific sizes and morphologies and usually exhibit a higher purity and uniformity. However, the extensive applications of the catalysts may be limited by the multistep synthesis process and inevitable use of organic solvents.

Solid-Phase Method
The solid-phase method is the thermal treatment process after mixing the N and TiO 2 sources under a given atmosphere. The solid-phase method is an economical and effective strategy for the large-scale synthesis of N-doped TiO 2 . In addition to the ball milling, atomic layer deposition (ALD), and chemical vapor deposition methods (as shown in Fig. 3), the most common and efficient method used to prepare the target N-TiO 2 catalyst is to oxidize TiN powder under O 2 gas flow or nitrate TiO 2 powder under NH 3 gas flow [79]. A previous study reported that, by annealing at 200-550 °C for 1.5 h, the TiN sample could be converted into N-doped TiO 2 under an O 2 gas atmosphere. In their study, Morikawa et al. [80] determined that the calcination temperature can influence the crystal phase of the obtained sample. For instance, a homogeneous rutile phase could be obtained under the calcination temperature of 550 °C, whereas a small amount of TiN remained in the rutile phase of the TiO 2 sample under the calcination temperature of 400 °C. In another effort, Zhao's group [81] prepared N-doped TiO 2 samples with a mixed phase of anatase and rutile by the incomplete oxidation of TiN at various temperatures. Then, they investigated the photocatalytic performances of the obtained samples by photocatalytic degradation of gas-phase toluene under visible light illumination and achieved improved photocatalytic performances. Recently, Xia et al. [82] prepared highly ordered porous nanotube arrays of N-doped TiO 2 through oxidation of the preprepared porous TiN nanotube arrays in an air atmosphere. Photocatalytic investigations indicated that the obtained porous N-doped TiO 2 nanotube arrays had a visible light photocatalytic kinetic rate constant that was threefold larger than that of pure anatase TiO 2 nanotube arrays, which could be attributed to its higher visible light responsiveness and larger surface area.
As for the nitrification of the TiO 2 sample under an NH 3 gas atmosphere, the synthesis conditions have obvious effects on the morphology of the N-doped TiO 2 photocatalyst. Hashimoto's group [83] processed TiO 2 with anatase phase at different temperatures for 3 h under a mixed NH 3 / Ar (67%/33%) gas flow to obtain the anatase phase of the N-doped TiO 2 material. The resulting samples with different colors signified that the change in calcination temperature and NH 3 gas flow resulted in different amounts of N in the N-doped TiO 2 material, which would result in different quantum yields because of the varying visible light absorption capabilities. In another effort, Pulgarin's group [84] used thiourea as the N source and mixed it with commercial anatase TiO 2 powder, followed by annealing at 400 °C and 500 °C, which resulted in visible light absorption of TiO 2 . Fang et al. [85] prepared N-doped TiO 2 by annealing P25 or TiO 2 xerogel powder under NH 3 /Ar atmosphere at different temperatures. They determined that, compared with the N-doped TiO 2 prepared by annealing P25, the N-doped xerogel powder had a higher concentration of substituted N and exhibited a more distinct visible light absorption. Another study of N-doped TiO 2 with abundant OVs reported by Huang et al. [86], which was prepared by a facile hydrothermal method with a baking process in an NH 3 atmosphere, showed that more OVs could be generated by doping N into TiO 2 , consequently ameliorating their catalytic behavior, particularly that of the OVs derived from visible light illumination.
Meanwhile, different N-containing molecules, such as urea and hexamethylenetetramine, were utilized for the doping process using a solid-phase (mechanochemical) method [68,87]. In this process, the mixture containing TiO 2 and urea/hexamethylenetetramine was ground using a planetary ball mill and calcined at 400 °C to remove the residual organic substances. Notably, the obtained N-doped powder mainly consists of rutile TiO 2 , indicating that high mechanical energy accelerates the phase transformation of anatase into rutile.
Furthermore, the sputtering process, ALD, and pulsed laser deposition are the most advanced techniques for the fabrication of crystalline N-doped TiO 2 thin films. For example, via the ALD operation, Vasu et al. [88] grew p-type epitaxial N-doped anatase TiO 2 (001) thin films on c-axis Al 2 O 3 substrate, employing TiCl 4 , double-distilled water, and NH 3 as Ti, O, and N sources, respectively. Their results showed the enhanced hole concentration and mobility of the obtained samples.
In each case, the presence of N within the lattice of TiO 2 has been shown to have an obvious effect on photocatalysis. As exhibited in Fig. 4 (Curve a), the pure anatase TiO 2 shows a single sharp edge, whereas the N-doped TiO 2 (Curve b in Fig. 4) shows two absorption edges: One is the main edge that originated from the oxide at approximately 390 nm, which does not change significantly compared with that of the undoped sample, and the other is the weak shoulder absorption at 400-550 nm wavelength. This noticeable shift of absorption to the visible light region corresponds to the yellow characteristic observed in the obtained N-doped samples [89]. Furthermore, with the increase in N content in the N-doped TiO 2 , the shoulder absorption initially shifts to the lower energy region, indicating that the N-doped samples are sensitive to visible light. Although the shoulder absorption expands the light absorption of the obtained TiO 2 semiconductor to a certain extent, it cannot extend its intrinsic band absorption to the visible light range, which will result in the low visible light absorption efficiency of the prepared N-doped TiO 2 . More importantly, this N substitution in TiO 2 usually leads to the generation of OVs, which results in a decrease in photocatalytic activity [90]. Hence, finding a simple and superior method for the scalable preparation of N-doped TiO 2 material with high visible light absorption efficiency is still a challenge.
Co-incorporation of N and other nonmetal anions has been considered a new potential pathway to construct the energy band structure of TiO 2 . The remarkable synergetic effects in N,F-co-doped TiO 2 have been reported by some interesting works. For instance, Domen's group [91] utilized SiO 2 powder as the O source in their investigation of N,F co-doping into anatase TiO 2 by nitriding an (NH 4 ) 2 TiF 6 precursor. Despite their results showing good absorption in the visible region (λ < 570 nm), which drives water oxidation for the N-doped catalyst, a shortcoming remains, i.e., under the preparation conditions, the N, O, and F contents of the product were uncontrollable. Afterward, Maeda et al. [92] prepared high-crystallinity TiN x O y F z material with an anatase crystal structure via nitridation of NH 4 TiOF 3 in an NH 3 atmosphere, which can be used as a visible lightresponsive photocatalyst for water oxidation at wavelengths of up to 540 nm. In another investigation, Zong et al. [93] reported that the visible light-responsive anatase TiO 2 platelets dominated by (001) facets were prepared by simple nitridation of the TiOF 2 precursor. The synthesized TiOFN materials exhibited excellent visible light-driven activity for water oxidation. Although the aforementioned studies achieved the absorption of visible light for N-doped TiO 2 , it was only used in water oxidation, and the performance of water reduction for H 2 evolution of the catalyst was not investigated. Recently, Maeda and colleagues [94] fabricated a novel N,F-co-doped rutile TiO 2 by nitridation of the rutile TiO 2 and (NH 4 ) 2 TiF 6 mixture and evaluated its photocatalytic oxidation activity of water. Their evaluation indicated that the N,F-co-doped samples exhibited water oxidation activity even in the presence of reversible electron acceptors under the action of the RuO 2 cocatalyst. Moreover, in In addition to the F dopant, B was selected as the codopant of N to optimize the band structure of the TiO 2 semiconductor. Liu et al. [95] reported the first example and demonstrated the substantial effect of B,N co-doping on improving the visible light response and charge carriers separation of TiO 2 . Their results also indicated that, during the N doping process, the presence of the B dopant stabilized the structure relative to undoped TiO 2 . Recently, in the study conducted by Hong et al. [96], a red anatase TiO 2 with a spatially homogeneous distribution of B and N dopants, which was sensitive to the sharp-edged strong visible light absorption spectrum over 680 nm, was obtained. The synthesized red TiO 2 sample can drive photocatalytic water oxidation for O 2 evolution under the illumination of visible light exceeding 550 nm and induce photocatalytic water reduction for H 2 evolution under visible light illumination. They concluded that the adsorption of HNCO and related derivatives on the surface of TiO 2 released by the thermal hydrolysis of urea significantly promoted the diffusion of the N dopant into the microspheres, thereby inducing uniform N doping with the help of the B dopant, resulting in a strong band-to-band absorption in the entire visible light range. Furthermore, several N-doped TiO 2 nanomaterials modified with other nonmetals (such as C, P, and S) were investigated for their visible light catalytic activities [97][98][99]. The results showed that, compared with undoped TiO 2 or mono-N-doped TiO 2 , the modified N-doped TiO 2 samples exhibited enhanced photocatalytic activities.

Mechanisms for Visible Light Response
Generally, electron-hole pairs generated via light illumination are the basic components for photocatalytic reactions. Since the pioneering work conducted by Asahi et al. [58] to synthesize N-doped TiO 2 material by sputter deposition, tremendous efforts have been exerted to investigate the potential mechanisms for the photocatalytic reactions of N-doped TiO 2 particles under visible light illumination and explore the electronic, optical, and structural characteristics of N-doped TiO 2 samples.
Even though the visible light absorption and enhanced photocatalytic activity of N-doped TiO 2 catalysts have been generally recognized, in theory, there are still open debates about the chemical properties, location of the photoactive N species involved, and origin of the visible light response. In general, the main proposed mechanisms for the improved visible light absorption of N-doped TiO 2 can be divided into three areas, which are discussed in the subsequent paragraphs.
Based on the first principle calculations, in 2001, Asahi et al. [58] investigated the contribution of N doping to the narrowed bandgap of TiO 2 and compared three N-doped TiO 2 systems, namely, substitutional N doping, interstitial N doping, and both types of doping into anatase TiO 2 . Their results indicated that the N 2p state could hybridize with the O 2p state (Fig. 5a), resulting in the narrowed bandgap and enhanced visible light-driven photoactivity. However, since 2003, this conclusion was challenged by other opinions, which tend to support that N doping will not result in the reduction of the TiO 2 semiconductor bandgap based on the experimental and theoretical studies but introduce impurity energy levels or induce the formation of OVs/Ti 3+ defects. Specifically, Irie et al. [83] evaluated the oxidization capability of the TiO 2-x N x powder by the decomposition of gaseous 2-propanol under the same illumination conditions (i.e., the absorbed photon number of visible or UV light was equal to each other). They determined that, regardless of the x, the quantum yield from visible light illumination was lower than that from UV light illumination. Moreover, the quantum yield decreased with the increase in the N concentration when irradiated with UV and visible light. Therefore, they concluded that the formation of the isolated and localized midgap state above the VB by N doping (as  Fig. 5b) was the cause of the visible light-responsive N-doped TiO 2 materials.
Another point is that the OVs caused by N doping contribute to the absorption and photocatalytic performances in the visible light wavelength range [101], as shown in Fig. 5c. One explanation is that NH 3 can decompose into N 2 and reducing gas H 2 at approximately 550 °C [102]. As a result, the decomposed H 2 can result in the formation of reduced TiO 2 when annealing TiO 2 powder in an NH 3 flow. For instance, Zhao et al. [103] prepared N-doped TiO 2 photocatalysts through a simple solvothermal strategy, followed by a solid-state chemical reduction process. Based on the analysis of the results of the control experiments, they proposed that the reason for the narrowed N-TiO 2-x bandgap is the synergetic effect of the defect energy states (including Ti 3+ and OVs) and N 2p midgap state. Another generally accepted explanation is that the OVs are derived from the charge imbalance (N 3− vs. O 2− ) after N doping [90,104]. Meanwhile, the formation of Ti 3+ is accompanied by the generation of OVs. This opinion has been verified by Serpone [105], who believe that co-doping N and other anions into the TiO 2 lattice will increase the number of OVs.
Although the aforementioned ideas can explain the reasons for the visible light absorption of N-doped TiO 2 material, there is still no consensus on whether the bandgap can be narrowed and its working mechanism. Therefore, this issue is still open for discussion and in-depth research needs to be conducted.

Characterizations of N-Doped TiO 2 Photocatalysts
The photocatalysis process has been well known to involve three crucial stages, namely, light absorption, charge carriers separation, and surface reaction [106]. Apart from the energy states, which are highly related to the light absorption stage, the investigation of the behavior of photoinduced charge carriers is essential to determine photocatalytic performance. The photoinduced charge carriers are fleetly trapped at the surface states of the photocatalyst particle within sub-picoseconds or a few picoseconds after excitation [107]. Although the lifetime is transitory, the measurement of the transfer kinetic of these charge carriers can provide important information for researchers and reveal the main process in the photocatalytic reaction. Thus far, time-resolved spectroscopy has been proven to be a powerful technique to analyze the charge carriers dynamics of photocatalysts, particularly the photocatalytic processes.
Time-resolved microwave conductivity (TRMC) is one of the powerful technologies employed to investigate the charge carriers dynamics. The obtained TRMC signal decays for TiO 2 -based material reflect the change in conductivity caused by the variations of the number of generated electrons. Consequently, the recombination and capture processes of electrons in TiO 2 samples can be determined based on the variations of these signals. Through this technology, Katoh et al. [108] investigated the charge carriers separation and recombination processes in the N-doped TiO 2 photocatalyst. Their results showed that the electronic behavior of N-doped TiO 2 excited by UV light (Ti 3d ← O 2p transition) was similar to that excited by visible light (450 nm, Ti 3d ← N 2p transition). However, the trapping rate increased (i.e., the charge carriers separation efficiency of N-doped TiO 2 excited by visible light excitation was one-third of that of pure TiO 2 excited by UV light) in the N-doped sample relative to the undoped sample, which meant that N doping will seriously affect the concentration of OVs.
Yamanaka and Morikawa [109] used femtosecond timeresolved diffuse reflectance (TDR) spectroscopy to elucidate the kinetics of charge carriers separation and trapping in N-doped TiO 2 powder under weak excitation conditions. As shown in Fig. 6a, the TDR spectra of N-doped TiO 2 revealed that the electrons and holes captured on the surface were produced fleetly after excitation, which was similar to that of TiO 2 under 360 nm light excitation (Ti 3d ← O 2p transition). The additional OVs lead to deep trapping (the time constant is approximately 300 ps); thus, the number of trapped electrons on the surface was lower than that of pure TiO 2 . Under 450 nm excitation, the TDR spectra of N-doped TiO 2 showed the signals derived from the Ti 3d ← N 2p transition. Furthermore, compared with the 360 nm excitation, the 450 nm excitation exhibited two differences in time evolution, i.e., it significantly decreased immediately after excitation and the electron depth was trapped in 1 ps.
In another study, Chen et al. [110] investigated the decay kinetics of photogenerated electrons that result in recombination or carrier reactions in the synthesized N-doped TiO 2 samples using time-resolved infrared (IR) spectroscopy. The observed transient absorbance spectrum at 1910 cm −1 shown in Fig. 7a revealed that the MTiO 2 /Ala-DDA samples exhibited the strongest initial absorption, indicating the existence of a large number of charge carriers. The pure anatase MTiO 2 /DDA exhibited weaker absorption intensity, indicating that the formation of OVs on tetrahedral Ti 4+ sites created a shallow donor state below the CB of the MTiO 2 /Ala-DDA samples. As shown in Fig. 7b, the normalized absorbance decay revealed that the obtained MTiO 2 / Ala-DDA exhibited a slower decay rate than the two other MTiO 2 samples, indicating that the separation of electrons and holes was effectively realized. This may be benefited from the shallow donor state produced by the OVs on tetrahedral Ti 4+ sites can more effectively promote charge carriers separation and electron capture in MTiO 2 /Ala-DDA.
Time-resolved absorption (TA) spectroscopy is another powerful and delightful technique to investigate the nature of photogenerated charge carriers. Tang et al. [111] explored the nature of charge carriers in N-doped films. They explained that the reason for the poor photocatalytic O 2 production from water oxidation of nanocrystalline N-doped TiO 2 film under visible light illumination was mainly the rapid recombination of charge carriers at doping-induced states rather than the deep trapping of photogenerated holes in the bulk material (as shown in Fig. 8).
Moreover, they concluded that a long lifetime of no less than 0.4 s for the photogenerated holes was essential to the evolution of O 2 on the nanocrystalline N-doped TiO 2 . Therefore, to achieve an improved O 2 evolution efficiency for this doped material, the key strategy is to enhance the charge carriers separation with the aid of cocatalysts, such as Pt cocatalyst.

Main Applications
The aim of engineering functional N-doped TiO 2 samples with well-defined morphologies and crystal phases is to optimize their electronic properties, thereby achieving better performance for photocatalytic applications. In general, a semiconductor-based photocatalytic process usually includes three independent and indispensable steps (Fig. 9) [112], namely, absorption of photons and generation of electron-hole pairs, separation and migration of charge carriers from the matrix to the surface, and surface reactions (i.e., reduction and oxidation reactions). Thus far, tremendous efforts ( Table 2) have been exerted to enable the practical applications of N-doped TiO 2 in photooxidation reactions (e.g., photodegradation, photocatalytic disinfection, and O 2 generation from water splitting) or photoreduction reactions (e.g., CO 2 reduction and H 2 generation from water splitting).

Degradation
N-doped TiO 2 material is an effective photocatalytic system for treating some harmful compounds in polluted water/ air, and it has been extensively investigated for water treatment and air purification. In the early 2000s, Asahi et al. [58] reported that, under visible light illumination (wavelength < 500 nm), N-doped TiO 2 exhibited better photocatalytic performance for the degradation of methylene blue and gaseous acetaldehyde than undoped TiO 2 . Salarian et al. [113] synthesized N-doped TiO 2 via a facile hydrothermal route and investigated the photodegradation performance of diazinon. Their results showed that the degradation and mineralization efficiencies of diazinon could reach 85% and 63%, respectively. Wang et al. [123] showed that N-doped TiO 2 nanoparticles prepared by a post-thermal treatment exhibited enhanced visible light photocatalytic activity for the degradation of methylene blue. Marques et al. [68] used urea as the N source and prepared N-doped TiO 2 nanoparticles with a large specific surface area using the modified sol-gel method. They determined that the doping calcination temperature and molar ratio of urea to TiO 2 have a considerable influence on the N content in N-doped TiO 2 powders. The results of their photocatalytic degradation experiments on methylene blue dye indicated that, under both UV-A light (365 nm) and visible light (400-700 nm) illumination, N-doped TiO 2 exhibited higher photocatalytic efficiency than the pristine TiO 2 . Furthermore, the molar ratio of N to TiO 2 used for doping strongly affects the photocatalytic performance of the obtained N-doped TiO 2 .
Moreover, the photocatalysis studies on N-doped TiO 2 with different exposed facets revealed different activities. Shi et al. [125] reported that N-TiO 2 plates with dominant facets [001] had superior photocatalytic activities for the degradation of methylene blue under visible light irradiation (λ > 420 nm) and concluded that the improved performance was attributed to the smaller particle size (approximately 25 nm) and larger specific surface area in comparison with micrometer-sized N-doped TiO 2 materials. Similarly, the [001]-dominated TiO 2 nanosheets co-doped with N and La reported by Wang's group [124] exhibited higher visible light photocatalytic activity for the decomposition of RhB than pure TiO 2 , N-TiO 2 , La-TiO 2 , and N,La-TiO 2 nanoparticles (Fig. 10). This may be because the N, La-TiO 2 nanosheets exhibited the strengthened absorption of visible light than other samples. Moreover, the synergistic effect between N and La, i.e., N doping narrowed the bandgap of TiO 2 , whereas La doping improved the separation efficiency of photoelectrons and holes, was attributed to the enhanced photodegradation performance.

Sterilization
Photocatalytic disinfection is becoming a promising process for water treatment because it is cost-effective and environmentally safe. Makropoulou et al. [114] used various N precursors (such as urea, TEA, and NH 3 ) to synthesize N-doped TiO 2 photocatalysts and employed these photocatalysts to inactivate bacteria in an aqueous substrate under simulated sunlight illumination. As shown in Fig. 11, under simulated sunlight illumination, the Gram-negative bacteria Escherichia coli and Pseudomonas aeruginosa dispersed in solution were successfully inactivated, and the highest reduction rate of 6log10 was reached within 60 min of treatment. In a recent report, Horovitz et al. [115] investigated the efficiency of N-doped TiO 2 -coated Al 2 O 3 photocatalytic membrane reactors to remove MS2 bacteriophage with different

Organic Transformation
Photocatalytic organic transformation as a green, sustainable, and environmentally friendly technique has received increasing attention, and many organic materials with special functions were synthesized via selective oxidation or reduction with the aid of photocatalysts in recent years. For instance, Japa et al. [126] investigated the selective transformation of benzyl alcohol and benzylamine into benzaldehyde and benzylamine under visible light illumination using N-doped TiO 2 as catalysts, which were prepared by thermal hydrolysis of TiOSO 4 using NH 4 OH as both precipitating agent and N source. As shown in Fig. 12, the Shoulder absorption 300 W Xe lamp (400 nm < λ < 800 nm) H 2 evolution from the reduction of water [122] optimized N-doped TiO 2 (T_400) exhibited superior photocatalytic activity for the selective oxidation of benzyl alcohol and benzylamine to benzaldehyde and N-benzylidenebenzylamine products, respectively, with > 85% conversion and > 95% selectivity, regardless of the substituents of the benzylamine molecule. The control sample (T_400_NaOH), prepared using NaOH (instead of NH 4 OH) as precipitating agent, was also analyzed, and the results revealed the beneficial nature of N doping for selective oxidation. Singha et al. [67] synthesized N-doped yellow TiO 2 hollow spheres using aqueous titanium peroxocarbonate complex solution as a precursor and NH 4 OH and used them to  promote the synthesis of active esters of N-hydroxyimide and alcohol through simultaneous selective oxidation of alcohol to aldehyde, followed by cross-dehydrogenative coupling via visible light illumination under ambient conditions. Xu et al. [127] fabricated N-doped anatase/rutile mixed-phase TiO 2 photocatalysts with carbonaceous species by pyrolyzing MIL-125 (Ti) and P25 composites as the precursor and subsequently doping N into the pyrolysis product using urea as the dopant. In this study, the photocatalysts showed excellent performances of photocatalytic cyclohexane oxidation under visible light illumination. The yield of KA oil (cyclohexanol and cyclohexanone) reached 112.44 µmol after 5-h illumination of the photocatalyst with the optimal N-doped anatase/rutile ratio, which was 5.66 and 39.87 times that of N-doped anatase TiO 2 and N-doped P25, respectively. In the meantime, the selectivity toward cyclohexanone could reach 100%. Cipagauta-Díaz's group [128] tested the photocatalytic activity of N-doped TiO 2 materials, which were synthesized by a low-temperature microwave-assisted sol-gel method, in the photoreduction of 4-nitrophenol (4-NP) to 4-aminophenol under a visible light source. Although the enhanced photoreduction of 4-NP for N-doped TiO 2 (TiEN * 0.5 and TiEN * 1) in comparison with bare TiO 2 was obtained (Fig. 13), the utilization of visible light has a low efficiency.

CO 2 Reduction
The photocatalytic reduction of CO 2 to hydrocarbons has considerable potential to provide alternative fuel to the world and solve the CO 2 emission problem. Since the first report by Inoue et al. [129], tremendous efforts have been exerted to enhance the efficiency of TiO 2 -based catalysts for photocatalytic CO 2 reduction. For instance, Bjelajac et al. [116] synthesized N-doped TiO 2 nanoparticles and investigated their photocatalytic CO 2 reduction performance. The results shown in Fig. 14 indicated that the N-doped sample exhibited superior production rates of H 2 , CH 4 , and CO. By contrast, the P25 catalyst did not affect the productivity of CH 4 . One reason for this phenomenon is that the bandgap of the obtained N-doped TiO 2 is smaller; thus, the photon absorption is higher than that of P25. The other reason is the optimal grain sizes of the synthetic samples, as the particles of the P25 sample are generally larger. The larger the grain size is, the lower the charge carriers separation, thus leading to a higher electron consumption rate. To improve the separation efficiency of photoinduced charge carriers, Akple et al. [117] prepared visible lightresponsive N-doped anatase TiO 2 microsheets, which have a heterojunction structure with 65% (101) and 35% (001) exposed faces, through a hydrothermal route. Under visible light irradiation, the obtained N-doped TiO 2 exhibited better photocatalytic performance for CO 2 reduction than pure TiN and P25 because of the synergistic effect of surface heterojunctions to improve electron-hole separation and N doping, which is beneficial to visible light absorption. Apart from the pretreatments and surface heterojunctions, the selection of N sources also has an effect on the photocatalytic activity of the N-doped TiO 2 catalyst. Zhang et al. [118] demonstrated that the N-N group in the N source (N 2 H 4 or NH 3 ) plays a decisive role in the selectivity of the main product of photocatalytic reduction on the surface of the obtained N-doped TiO 2 catalysts. They determined that, under visible light illumination, the main products of N-TiO 2 doped with N 2 H 4 and NH 3 were CH 4 and CO, respectively. Therefore, they concluded that the presence of N-N groups on the surface of N 2 H 4 -doped N-TiO 2 provided a reducing environment, which caused CH 4 to become the main reduction product.

Water Splitting
From the perspective of converting solar energy into fuel energy, photocatalytic water splitting has attracted considerable attention. In terms of N-doped TiO 2 , scientists have exerted tremendous efforts to investigate the mechanisms and properties under visible light illumination. Yun et al. [119] evaluated the effects of the amount of N in N-doped TiO 2 nanoparticles on O 2 generation via visible light-driven water oxidation and claimed that a small amount of N dopant (up to 3 at%) added to the samples can accelerate the production of O 2 instead of the formation of OH • species. However, larger amounts of N dopant doped into TiO 2-x N x can reduce the rates of both O 2 production and OH • formation, which can be attributed to the low energy level of the VB and the short lifetime of the photoinduced charge carriers. Lee et al. [120] described the improvement of O 2 evolution from water oxidation on the N-doped TiO 2 nanospheres and nanorods prepared by calcination in NH 3 flow. The measurement of photocatalytic activities showed that the two N-doped samples exhibited enhanced O 2 evolution performance than the undoped samples. Recently, Liu's group [96] developed a new nitrogenation method to dope uniform B and N into Fig. 13 a C/C 0 plot as a function of time for 4-nitrophenol (4-NP) photoreduction; b zero-order kinetics plot of 4-NP photoreduction; c diffuse reflectance spectroscopy UV-Vis spectra; and d Kubelka-Munk modified spectra of the materials TiO 2 , TiEN*0.5, and TiEN*1. Reproduced from Ref. [128] with permission. Copyright 2020, Wiley-VCH TiO 2 microspheres with exposed facets [001], as shown in Fig. 15a-e. The obtained red TiO 2 did not show an increase in defects (such as Ti 3+ ), which usually exist in doped TiO 2 . This decrement may be the result of charge compensation between interstitial B 3+ and substituted N 3− , which can hinder the formation of new OVs in the uniform B,N-codoped TiO 2 . After loading the RuO 2 cocatalyst, the red TiO 2 exhibited a higher photocatalytic O 2 evolution activity and stability than the pristine TiO 2 (Fig. 15f, g). In their work, Wu et al. [97] reported that a C, N substitute was uniformly doped into anatase TiO 2 decahedral plates. Because of the uniform C,N co-doping to narrow the bandgap, the obtained TiO 2-x (CN) y sample had an ideal spectrum of strong bandto-band visible light absorption, which had the capability to induce water oxidation to generate O 2 with CoO x that used as cocatalyst under visible light illumination.
In addition to photocatalytic O 2 evolution, N-doped TiO 2 materials are often used as photocatalysts for H 2 generation from water reduction. Shangguan and co-workers [121] prepared N-doped TiO 2 catalysts with large specific surface areas by calcining a mixture of TiO 2 and urea and investigated their photocatalytic H 2 generation activity. They confirmed the existence of N on TiO 2 as chemisorbed N 2 molecules and substitutional N on TiO 2 via XPS measurements and summarized that the crystalline phase of TiO 2 and the substitutional N were the main factors improving the photocatalytic performance of water reduction for H 2 generation. Sun et al. [122] prepared anatase N-doped TiO 2 nanobelts with a surface heterojunction of co-exposed facets (101) and (001) and measured their photocatalytic water splitting performance. Based on the research findings, they proposed that, during the catalytic process, the surface heterojunction not only promoted the disintegration of water molecules but also enhanced the space separation of charge carriers to effectively inhibit the recombination of the charge carriers (Fig. 16), thus achieving water reduction for H 2 production under visible light stimulation. The B,N-co-doped TiO 2 prepared by Liu et al. [97] also had the photocatalytic capability to convert water into H 2 when loaded with 0.5 wt% Pt cocatalyst and methanol utilized as a sacrificial agent under visible light illumination.
Although the investigation of N,F-co-doped anatase TiO 2 has been reported, only a few reports are available for the rutile form. In recent work, Maeda et al. [94] focused on N,F-co-doped rutile TiO 2 material (R-TiO 2 :N,F), which was synthesized via a nitridation process at 500 °C using . c Average yield rates of each product for all catalysts ("G" denotes that the gel was directly annealed in an NH 3 atmosphere in a tube furnace, whereas "P" denotes that the gel was pre-annealed in air at 500 °C for 3 h and annealed in an NH 3 atmosphere). Reproduced from Ref. [116] with permission. Copyright 2020, Royal Society of Chemistry rutile TiO 2 and (NH 4 ) 2 TiF 6 as raw materials, as O 2 generation photocatalyst for visible light-driven Z-scheme water splitting (Fig. 17a-d). Notably, the prepared samples showed clear absorption in the visible region, and the more pronounced visible light absorption in the present R-TiO 2 :N,F samples was attributed to the increased N content. This material exhibited water oxidation activity with the help of the RuO 2 cocatalyst even in the presence of a reversible electron acceptor, such as IO 3 − or Fe 3+ . However, water reduction activity was not observed because of the high Ti 3+ concentration, which was evidenced by the absorption at longer wavelengths (λ > 600 nm) [130].
By contrast, as shown in Fig. 17g, the combination of RuO 2 -loaded R-TiO 2 :N,F and Ru/SrTiO 3 :Rh photocatalysts could be used as a visible light-driven Z-scheme water splitting system with [Co(bpy) 3 ] 3+/2+ acts as a shuttle redox mediator and exhibited a stoichiometric photocatalytic overall water splitting activity with long-term stability.
Although the solar energy conversion efficiencies of these materials are still low, the methodologies proposed in these studies should contribute to the improvement of the photocatalytic performance of N-doped TiO 2 -based visible light-responsive materials.

Summary and Outlook
This review attempts to provide a comprehensive update and summary focusing on the synthesis, characterizations, and practical applications of visible light-responsive N-doped TiO 2 photocatalyst. N-doped TiO 2 not only effectively reduces the bandgap but also reduces the recombination efficiency of photoinduced charge carriers. Although the detailed source of photocatalysis of this material remains a debate, its basic principles and mechanisms have guided the search for other novel similar photocatalysis materials. Generally, the photocatalytic reaction rates of N-doped TiO 2 are still low. The findings can be attributed to the fact that the visible light absorption of N-doped samples is still poor, which cannot achieve the ideal bandto-band absorption. Moreover, with the increase in N content, the introduced OVs can act as recombination centers, thus resulting in a decreased photocatalytic reaction rate. Despite tremendous efforts that have been exerted in the area of enhancing the visible light absorption of TiO 2 -based materials, the following investigations need to be conducted in the future: (1) To explore the synergistic effect of various synthesis strategies, as no single solution can make materials suitable for different applications; (2) How co-doping of various N precursors and other nonmetals affects the band structure and photocatalytic performance of TiO 2 remains to be clarified; (3) Although most of these N-doped TiO 2 samples can individually produce H 2 or O 2 from H 2 O splitting in the presence of appropriate sacrificial agents under visible light illumination, which indicates that overall water splitting is possible in principle, it has not yet been realized. Further effort is required to realize the utilization of TiO 2 -based catalysts in photocatalytic overall water splitting activity under visible light irradiation; (4) The use of cocatalysts in photocatalytic water splitting systems can not only provide more active sites for redox reactions on the surface of the semiconductors but also accelerate the extraction and transfer of generated electrons for surface reactions. Recently, singleatom catalysts (SACs) with 100% atom utilization have shown superior performances in heterogeneous catalysis fields, such as photocatalytic reactions. In a sense, SACs could be employed as cocatalysts for N-doped TiO 2 materials to enhance their photocatalytic activities because of their fascinating strengths in enhancing light harvesting, charge transfer dynamics, and surface reactions of a photocatalytic system; (5) Conjugated polymers (CPs) have been used to produce solar energy conversion materials in photovoltaics because of their outstanding light-harvesting and low- cost processing. However, their photocatalytic activity has only recently been highlighted with the preparation of robust, metal-free, and visible or near-IR light-active photocatalysts. In the future, hybrid photocatalysts based on N-doped TiO 2 semiconductors and CPs can be fabricated and used as novel promising photoactive materials for pollutant degradation, energy conversion through water splitting, and/or CO 2 reduction using photocatalytic processes.