*Review* **Titanium-Dioxide-Based Visible-Light-Sensitive Photocatalysis: Mechanistic Insight and Applications**

### **Shinya Higashimoto**

Department of Applied Chemistry, Faculty of Engineering, Osaka Institute of Technology, 5-16-1 Omiya, Asahi-ku, Osaka 535-8585, Japan; shinya.higashimoto@oit.ac.jp; Tel.: +81-(0)6-6954-4283

Received: 15 January 2019; Accepted: 14 February 2019; Published: 22 February 2019

**Abstract:** Titanium dioxide (TiO2) is one of the most practical and prevalent photo-functional materials. Many researchers have endeavored to design several types of visible-light-responsive photocatalysts. In particular, TiO2-based photocatalysts operating under visible light should be urgently designed and developed, in order to take advantage of the unlimited solar light available. Herein, we review recent advances of TiO2-based visible-light-sensitive photocatalysts, classified by the origins of charge separation photo-induced in (1) bulk impurity (N-doping), (2) hetero-junction of metal (Au NPs), and (3) interfacial surface complexes (ISC) and their related photocatalysts. These photocatalysts have demonstrated useful applications, such as photocatalytic mineralization of toxic agents in the polluted atmosphere and water, photocatalytic organic synthesis, and artificial photosynthesis. We wish to provide comprehension and enlightenment of modification strategies and mechanistic insight, and to inspire future work.

**Keywords:** Titanium dioxide (TiO2); visible-light-sensitive photocatalyst; N-doped TiO2; plasmonic Au NPs; interfacial surface complex (ISC); selective oxidation; decomposition of VOC; carbon nitride (C3N4); alkoxide; ligand to metal charge transfer (LMCT)

### **1. Introduction**

Titanium dioxide (TiO2) is one of the most practical and prevalent photo-functional materials, since it is chemically stable, abundant (Ti: 10th highest Clarke number), nontoxic, and cost-effective. In recent years, a great deal of attention has been directed towards TiO2 photocatalysis for useful applications such as photocatalytic mineralization of toxic agents in the polluted atmosphere and water, photocatalytic organic synthesis, and artificial photosynthesis [1–20].

The TiO2 involving Ti3+ sites that are oxygen-deficient at the impurity level exhibits n-type semiconductor. The photocatalytic activities of TiO2 strongly depend on crystal structures (anatase, brookite, and rutile), crystallinity, crystalline plane, morphology, particle sizes, defective sites, and surface OH groups. The valence band (V.B.) and conduction band (C.B.) of TiO2 consist of O 2p and Ti 3d orbitals, respectively, and their band gap (forbidden band) is circa ~3.0–3.2 eV (~410–380 nm). Photo-irradiation (*hv* > 3.2 eV) of the TiO2 photocatalyst leads to band gap excitation, resulting in charge separation of electrons into the C.B. and the holes in the V.B. These photo-formed electrons and holes simultaneously work as electron donors and acceptors, respectively, on the photocatalyst surface, thus enabling the photocatalytic reactions. Details are given in other articles and reviews [21–25]. UV light reaching the earth surface represents only a very small fraction (4%) of the solar energy available. Therefore, many researchers have endeavored to design several types of visible-light-responsive photocatalyst. In particular, TiO2-based photocatalysts operating under visible light should be urgently designed and developed, in order to take advantage of the unlimited solar light available.

In the late 1990s, Anpo et al. first reported that TiO2 doped with Cr, V, and Fe cations by ion implantation operates under visible light irradiation. They exhibited red shift of the band-edge of the TiO2, resulting in decomposition of NO into N2, O2, and N2O [26]. This work accelerated subsequent works for the design and development of visible-light-responsive photocatalysts. Recently, much attention has been paid to visible-light-responsive TiO2 prepared by: doping with nitrogen (N), carbon (C), and sulfur (S) ions etc.; surface plasmonic effects with Au or Ag nanoparticles (NPs); the interfacial surface complex (ISC); coupling with visible-light-sensitive hetero-semiconductors (cadmium sulfide, carbon nitride etc.); and dye-sensitized photocatalysts. In fact, some photocatalysts are considered to work under similar principles.

Along these backgrounds, this review focuses on the recent advances of the visible-light-sensitive TiO2 photocatalyst. These advances have been classified by the origin of charge separation photo-induced in (1) the bulk impurity (N-doping), (2) hetero-junction of metal (Au NPs), and (3) the interfacial surface complex (ISC) (See Figure 1). They have been well characterized by several spectroscopic techniques, and applied for mineralization of volatile organic compounds (VOC), water splitting to produce H2, and fine organic synthesis.

**Figure 1.** Visible-light-sensitive TiO2 photocatalyst modified by (**1**) nitrogen-doping, (**2**) plasmonic Au nanoparticles (NPs), and (**3**) interfacial surface complex (ISC).

### **2. Nitrogen-doped TiO2 Photocatalysts**

In 1986, Sato and co-workers first explored the photocatalytic activity of nitrogen-doped TiO2 (N-doped TiO2) photocatalysts for the oxidation of gaseous ethane and carbon monoxide [27]. They found that N-doped TiO2 photocatalyst exhibited a superior photocatalytic activity to pure TiO2 under visible light irradiation. Later, in 2001, Asahi et al. demonstrated visible-light-induced complete photo-oxidation of gaseous CH3CHO (one of VOCs) to CO2 with an N-doped TiO2 photocatalyst [28]. In this section, fundamental synthetic routes, characterizations, and application of photocatalytic reactions are highlighted.

### *2.1. Synthesis of N-doped TiO2 Photocatalyst*

N-doped TiO2 was prepared by employing several procedures and materials. Details are given in Reference [13]. Preparation methods for N-doped TiO2 photocatalysts can be classified into two categories: dry processes and wet processes.

### 2.1.1. Dry Processes

Typically, N-doped TiO2 powder can be prepared by the nitrification of TiO2 in an ammonia (NH3) gas flow at high temperature [28,29]. The amount of N doping into the TiO2 can be controlled by annealing temperatures in the range of 550−600 ◦C under an NH3 flow. However, a large number of O vacancies are introduced into the N-doped TiO2 with increasing annealing temperature, since the NH3 decomposes into N2 and H2 at high temperature, and TiO2 is simultaneously reduced by H2 [30]. Figure 2 shows schematics of N-doping into TiO2, accompanied by the formation of oxygen vacancies to exhibit the n-type semiconductor.

**Figure 2.** When the N3<sup>−</sup> is replaced with lattice O2<sup>−</sup> ions in the TiO2 lattice, the hole (h+) is formed in order to compensate for the charge balance (p-type semiconductor) (a). However, an oxygen vacancy is produced by the reduction with H2, which is formed by the decomposition of NH3 to produce an oxygen vacancy and excess electrons (b). As a consequence, N3<sup>−</sup> doped into TiO2 (N-doped TiO2) involves electrons located at N 2p and Ti 3d sites at impurity levels (n-type semiconductor) (c).

#### 2.1.2. Wet Processes

A sol-gel method can be employed for the preparation of N-doped TiO2 powder. Typically, NH3 aq. (NH4OH) is added to a solution of titanium (IV) isopropoxide (TTIP) [31–33] to form titanium hydroxide involving N-species. The precipitate was dried, followed by calcination at ~400–450 ◦C in air to obtain a yellowish TiO2 powder.

### *2.2. N-states in N-doped TiO2*

One of the major concerns is to understand the physico-chemical nature of the N species in N-doped TiO2, which are responsible for the visible light sensitivity. They were characterized by density functional theory (DFT) calculations, X-rap photoelectron spectroscopy (XPS), Ultraviolet-visible (UV-vis) and electron paramagnetic resonance (EPR) spectroscopy.

### 2.2.1. DFT Calculations

DFT calculations demonstrated the electronic structures of the N-doped TiO2 photocatalyst (see Figure 3). The substitution of N with lattice O of the N-doped TiO2 exhibits band gap narrowing (circa 0.1 eV) caused by mixing orbitals of N 2p with O 2p, resulting in the negative shift of the valence band edge. On the other hand, the interstitial N is localized to impurity states (N 2p levels) above the V.B. (circa 0.7 eV) in the mid-band gap. Therefore, the oxidation power of photo-induced holes on the N 2p is lower than on the O 2p in the TiO2 lattice.

**Figure 3.** Schematic illustration of structures and their corresponding energy bands for substitutional and interstitial N species in the N-doped TiO2, together with photo-induced electronic processes.

### 2.2.2. XPS Spectra

XPS analysis can confirm the oxidative states of the N species and bonding states in the N-doped TiO2 (See Figure 4I). N 1s XPS peaks at a binding energy in the range of ~396–400 eV showed different oxidative states of the N species. By the combination of the DFT calculations [31], it was identified that the N 1s XPS peaks at ~396–397 eV are due to the substitution of N with the lattice O of TiO2 [13,34], while those at ~399–400 eV are due to the interstitial N in the form of NO*x* or NH*x* [13,31,35,36].

**Figure 4.** XPS [I] and UV-vis absorption spectra [II] of (a) N-doped TiO2 nanoball film [34], and (b) N-doped TiO2 prepared by the sol-gel method [36].

### 2.2.3. Optical Properties

The UV-vis absorption spectra of the N-doped TiO2 are shown in Figure 4II. The N-doped TiO2 with the substitution of N exhibited band gap narrowing from 3.1 to 2.8 eV. On the other hand, the N-doped TiO2 prepared by the sol-gel method exhibited visible light absorption up to 540 nm (2.3 eV), due to the electronic transition from localized N doping level to the C.B. of the TiO2, while band-narrowing was not observed. These results are in good agreement with the DFT calculations.

### 2.2.4. Electron Paramagnetic Resonance (EPR) Spectra

N species in the N-doped TiO2 are present at either diamagnetic (N−) or paramagnetic (N•) bulk centers, which are responsible for the visible light sensitivity [31,37]. The EPR measurements can detect the paramagnetic (N•) bulk centers (see Figure 5). One type, of three lines with a hyperfine tensor (*g* = 2.006 and *A* = 32.0 G) splitting by nuclear spin of nitrogen (*I* = 1), was observed. The signal intensity of N• radicals increased when the light was turned on, while the signal intensity significantly decreased when the light was turned off. In general, the paramagnetic interaction between N species and O2 makes EPR signals disappear. However, they were remarkably enhanced in the presence of O2 under λ > 420 nm, while its signal intensity still remained to some extent even after the light was turned off. These results suggest that N-species are located in bulk inside the TiO2, and visible light irradiation of the N-doped TiO2 exhibits effective charge separation to form holes (N• radicals) and electrons, which participate in the oxidation and reduction of reactant molecules, respectively.

**Figure 5.** Schematic illustration of [I] formation of paramagnetic ·N by the excitation of diamagnetic N<sup>−</sup> species. Electron paramagnetic resonance (EPR) signal [II] of ·N radicals on N–TiO2, and the relative signal intensity of *I*N/*I*N0 [III] under vacuum, in the presence of argon (Ar) or O2 (400 Pa) [37]. *I*N0 and *I*<sup>N</sup> show the intensity due to ·N radicals at the initial and measured time, respectively.

### 2.2.5. Photo-Electrochemical Properties

Nakamura et al. investigated the photo-electrochemical oxidation power of the N-doped TiO2 by employing several electron donors [38]. Figure 6 shows that the photo-induced hole on the N 2p level can directly oxidize only I− ions under visible light illumination, while I−, SCN−, Br−, and H2O are oxidized by the hole on the V.B. under UV light illumination. Therefore, the oxidation power of the holes induced on the N 2p level is lower than that of those on the O 2p on the V.B. Tang et al. studied the dynamics of photogenerated electrons and holes on the N-doped TiO2 using transient absorption spectroscopy [39]. They concluded that the lack of activity of nanocrystalline N-doped TiO2 film for photocatalytic water oxidation is due to rapid electron–hole recombination.

On the other hand, Higashimoto et al. investigated the photo-electrochemical reduction power of the N-doped TiO2 (see Figure 7) [33]. When the N-doped TiO2 was photo-excited under visible light irradiation, the photo-induced electrons were accumulated on the oxygen vacancies of TiO2. Subsequently, when various kinds of redox species as electron acceptors were introduced into the photo-charged N–TiO2, the accumulated electrons could reduce O2 molecules, Pt4+, Ag+, and Au3+ ions, but not MV2+, H+, and Cu2+ ions. In principle, the N-doped TiO2 has the potential to reduce H+/H2, but many oxygen vacancies involved in the bulk TiO2 could influence the drastic charge recombination. In particular, photo-induced electrons trapped at the oxygen vacancies (mainly γ region) could reduce O2 molecules to form such active oxygen species as hydrogen peroxide (H2O2), resulting in further oxidation of organic substrates.

**Figure 6.** Schematic illustration of proposed energy bands for the N-doped TiO2, together with some photo-induced electronic processes. *E*: equilibrium redox potentials for one electron transfer [38].

**Figure 7.** Energy levels for sub-band structures of N-doped TiO2 and photo-induced charge transfer into various kinds of redox species under visible light irradiation. The energy levels of sub-bands at the α, β, and γ potential regions (oxygen vacancies) and N-doping levels are also shown. Oxygen vacancies were estimated from the photo-electrochemical measurements. Signs of circle and cross stand for energetically favorable and unfavorable electron transfers, respectively [33].

### *2.3. Application to Photocatalytic Decomposition of Volatile Organic Compounds (VOC)*

Time profile for the photocatalytic decomposition of gaseous acetaldehyde on the N-doped TiO2 is shown in Figure 8. The N-doped TiO2 exhibited photocatalytic activity 5 times greater than TiO2 under visible light irradiation, while they exhibited similar activities under UV light irradiation [28].

**Figure 8.** Photocatalytic decomposition of gaseous acetaldehyde on the N-doped TiO2 photocatalyst. Evolved CO2 concentration (-, •, N-doped TiO2; -, , TiO2) [28].

Table 1 shows that the N-doped TiO2 exhibited photocatalytic activity for the decomposition of several kinds of VOC into CO2 under visible light irradiation (λ > 420 nm). It was observed that the N-doped TiO2 exhibited photocatalytic activity for the decomposition of aldehydes, but little activity for alcohol, acid, ketone, and halogene compounds. The vanadium species was deposited on the N-doped TiO2 (VCl3/N-doped TiO2) by impregnation method. As shown in Table 1, VCl3/N-doped TiO2 showed higher photocatalytic activity for the decomposition of all VOC, in particular, acetic acid or acetone by ~13–16 times more than N-doped TiO2. Therefore, it was confirmed that vanadium species worked as the effective co-catalyst.


**Table 1.** Yields of CO2 for the photocatalytic decomposition of various kinds of volatile organic compounds (VOC) in aqueous solutions with N–TiO2 and VCl3/N-doped TiO2 under visible light irradiation (λ > 420 nm) for 3 h [40].

Concentrations of VOC are (a) 0.5 M and (b) 50 mM.

Furthermore, effects of co-catalysts (48 metal ions using nitrate, sulfate, chloride, acetate, and oxide precursors) deposited on the N-doped TiO2 for the photocatalytic activities were examined (See Figure 9) [40]. The bars marked in yellow exhibited higher photocatalytic activities than the N-doped TiO2 by itself. In particular, N-doped-TiO2-deposited Cu, Fe, V, and Pt oxides exhibited high photocatalytic activities. The local structures of the co-catalysts were characterized by XPS. It was observed that Cu loaded N-doped TiO2 involves cuprous oxide (Cu2O) or Cu hydroxides, Fe loaded N-doped TiO2 involves clusters containing Fe–O bonds or Fe2+ hydroxide [41], and Pt loaded N-doped TiO2 involves Pt4+/Pt2+ species [36]. The redox potentials of co-catalysts such as V (+IV/+V), Fe (+II/+III), Cu (+I/+II), and Pt (+III/+IV) were in the range of circa +0.6 to +1.0 V vs. SHE, while the multi-electron reduction of O2 leads to the formation of active oxygen species via O2 + 2H+ + 2e<sup>−</sup> / H2O2 (*E*<sup>0</sup> = +0.687 V vs. SHE). Therefore, the co-catalysts, such as Pt, Fe, Cu, and V species, enhance the photocatalytic activity due to the effective electron transfer to O2 (O2 reduction), resulting in the formation of active oxygen species.

6HYHUDONLQGVRIPHWDOOLFVDOWVIRUFRFDWDO\VW

**Figure 9.** Photocatalytic activities for the decomposition of acetic acid under visible light irradiation (λ > 420 nm) on N-doped TiO2, modified by various kinds of metal species as co-catalysts. Each metal salt used in this study is shown [40].

### *2.4. C3N4-Modified TiO2 Compared with N-doped TiO2*

Several nitrogen sources such as urea, cyanamid, cyanuric acid, and melamine were employed for the preparation of N-containing TiO2 photocatalyst, i.e., the TiO2 surface is modified with polymerized carbon nitride (C3N4) [42–51]. The structures of the C, N-species strongly depend on their concentrations. If the C, N species are present in only a small amount, they act as a molecular photosensitizer. At higher amounts they form a C3N4 crystalline semiconductor, which chemically binds to TiO2. The C3N4–TiO2 was systematically synthesized by thermal condensation of cyanuric acid on the TiO2 surface [51]. In fact, H2 was evolved from TEA aq. on the C3N4–TiO2 photocatalyst under visible light irradiation, while the N-doped TiO2 did not exhibit H2 production. From characterization of C3N4–TiO2 by Fourier transformed-infrared (FT-IR), XPS, electrochemical measurements, and DFT calculations, the band structures and photo-induced charge separation mechanisms were demonstrated (Figure 10). The C3N4–TiO2 was found to exhibit photo-induced charge separation through the hetero-coupling of semiconductors between C3N4 and TiO2 on the surface. On the other hand, N-doped TiO2 was photo-sensitized by bulk impurity of the N-doping. It can be assumed that many oxygen vacancies promoted the charge recombination, resulting in weak reduction power in the N-doped TiO2.

**Figure 10.** Photo-induced charge separation on the C3N4 deposited TiO2 surface [51].

### **3. Plasmonic Au NPs Modified TiO2**

### *3.1. What Is Localized Surface Plasmon Resonance (LSPR)?*

Localized surface plasmon resonance (LSPR) is an optical phenomenon generated by light when it interacts with conductive nanoparticles (NPs) that are smaller than the incident wavelength. The LSPR is induced by the collective oscillations of delocalized electrons in response to an external electric field. The resonance wavelength strongly depends on the size and shape of the NPs, the interparticle distance, and the dielectric property of the surrounding medium. The Au and Ag NPs exhibit unique plasmon absorption [52,53]. The plasmonic Ag NPs are considered to be unstable under illumination, and could be applicable to multi-colored rewritable devices. In this section, we focused on stable plasmonic Au NPs exploited for a visible-light-sensitive photovoltaic fuel cell or photocatalyst [54,55].

### *3.2. Preparation and Characterization of Au–TiO2 Photocatalyst*

### 3.2.1. Photodeposition (PD) Methods

By using the photocatalysis of TiO2, metallic Au was deposited on the TiO2 surface, accompanied by the oxidation of methanol [56,57] or ethanol [58]. Typically, TiO2 powder was suspended in a 50 vol. % aqueous methanol in the presence of HAuCl4·6H2O, purged of air with argon. The suspension was photoirradiated with UV light under magnetic stirring. The temperature of the suspension during photoirradiation was maintained at 298K. The Au/TiO2 photocatalyst was centrifuged, washed with distilled water, dried at 393K, and ground in an agate mortar.

### 3.2.2. Colloid Photodeposition Operated in the Presence of a Hole Scavenger (CPH)

Colloidal Au NPs were prepared using the method reported by Frens [59]. In brief, mixtures of an aqueous tetrachloroauric acid (HAuCl4) solution and sodium citrate were heated and boiled for 1 h. The color of the solution changed from deep blue to deep red. The citrate plays a role in the reduction of Au ions, and the capping agent in suppressing the aggregation of Au NPs. The suspension of TiO2 in an aqueous solution of colloidal Au NPs and oxalic acid was then photo-irradiated at λ > 300 nm at 298 K under argon (Ar). The solids were recovered, washed, and dried to produce Au–TiO2. Details are given in Reference [60].

### 3.2.3. Deposition Precipitation (DP) Method

Deposition–precipitation (DP) methods were employed for the deposition of a gold (III) species on the TiO2 surface [61,62]. The [AuCl(OH)3] −, main species present at pH 8, adjusted by NaOH aq., reacts with hydroxyl groups of the TiO2 surface to form a grafted hydroxyl–gold compound. The catalyst was then recovered, filtered, washed with deionized water, and dried. Finally, the powder was calcined at ~473–673 K in air.

### 3.2.4. Characterization of the Au–TiO2 Photocatalyst

The Au–TiO2 photocatalysts were typically characterized by the transmittance electron microscope (TEM) for the particle sizes, and UV-vis absorption for optical properties (See Table 2).


**Table 2.** Particle sizes of Au nanoparticles (NPs) and optical properties of the Au–TiO2 prepared by several techniques.

Kowalska et al. [56,57] reported that Au–TiO2 photocatalysts with different Au particle sizes (~10–60 nm) were prepared by photo-deposition (entry 1). The particle sizes of Au strongly depend on the particle sizes of the TiO2 polycrystalline structure. The top peak of plasmonic absorption was in the range of ~530–610 nm, depending on the particle sizes of the Au NPs. Tanaka and Kominami et al. [60,63–66] reported unique CPH methods for the preparation of Au-TiO2 (entry 2). The particle sizes were uniformed to be ~12–14 nm, which exhibits plasmonic absorption at ~550–560 nm. Thus, colloidal Au NPs were successfully loaded onto TiO2 without change in the original particle size. Furthermore, the top peak of Au plasmon absorption was found to extend towards 620 nm by simple calcinations of the samples. This phenomenon is due to high contact area between TiO2 and Au NPs without change of particle size [66]. Additionally, Naya et al. [67,68] and Shiraishi et al. [69] employed precipitation deposition methods to deposit small Au NPs (~2–6 nm) on TiO2 (entry 3).

### *3.3. Application of LSPR of Au–TiO2 to Several Photocatalytic Reactions*

Au NPs deposited on TiO2 have been used as visible-light-responsive photocatalysts for several chemical reactions: decomposition of VOCs, selective oxidation of an aromatic alcohol, direct water splitting, H2 formation from sacrificial aqueous solutions, and reduction of organic compounds (see Table 3). Several research groups concluded that photocatalytic activities are induced by LSPR of the Au NPs. Some research indicates that small Au NPs (~5 nm) effectively work for the reactions [61,69]. Tanaka and Kominami et al. suggest that two types of Au particles of different sizes loaded onto TiO2 exhibit different functionalities. That is, the larger Au particles contribute to strong light absorption, and the smaller Au particles act as a co-catalyst for H2 evolution [63].


**Table 3.** Applications to several photocatalytic reactions on the Au–TiO2 photocatalyst.

#### *3.4. Application to a Photovoltaic Fuel Cell Operating under Visible Light Irradiation*

The Au–TiO2 films were found to exhibit the behavior of a photovoltaic fuel cell [54,55]. An anodic photocurrent was yielded on the Au−TiO2 film as the visible light was irradiated, while the current was observed neither on a TiO2 film under visible light irradiation, nor on the Au−TiO2 film when the light was turned off. The short-circuit photocurrent density (*J*sc) was strongly influenced by kinds of donors, and the photocurrent efficiency was maximized in the presence of Fe2+ ions. Furthermore, the photocurrent action spectra were closely fitted with the absorption spectrum of the Au NPs deposited on the TiO2 film (See Figure 11).

**Figure 11.** Short-circuit photocurrent densities [I] vs. apparent formal potential of different donors on the Au−TiO2 photoanode in acetonitrile/ethylene glycol (v/v 60/40) containing 0.1 M LiNO3 and 0.1 M donors; IPCE [II] of the Au−TiO2 film in a N2-saturated acetonitrile and ethylene glycol (v/v: 60/40) solution containing 0.1 M FeCl2 and 0.05 M FeCl3 [55].

#### *3.5. Mechanisms of Charge Separation*

The mechanism for the Au plasmon-induced charge separation is shown in Figure 12. Visible light irradiation generates the photo-excited state of the Au NPs by LSPR. The photo-excited electrons are injected into the C.B. of TiO2, while the holes abstracted electrons from a donor in the solution. The Au NPs behave like an intrinsic semiconductor, and the Fermi levels of Au NPs and TiO2 are leveled out, resulting in the formation of Schottky barrier at Au–TiO2 junctions. This band model seems to be similar with dye-sensitized photo-anodic electrodes.

**Figure 12.** Schematic illustration [I] and its energy band levels [II] for the photo-induced charge separation on the Au–TiO2 in the presence of donors [55].

Recently, Furube et al. studied the plasmon-induced charge transfer mechanisms between Au NPs and TiO2 by means of femtosecond visible pump/infrared probe transient absorption spectroscopy [73]. The electron transfer from the Au NPs to the C.B. of TiO2 was confirmed to occur within 50 fs, and that the electron injection yielded 20–50% upon 550 nm laser excitation.

### **4. Photo-Induced Interfacial Charge Transfer**

### *4.1. Dye-Sensitized TiO2 Photocatalysis*

Dye sensitized TiO2 photocatalysis was studied in the late 1990s. The Ru complex, [Ru(bipy)3] 2+ grafted on the TiO2 surface exhibits visible light absorption [74,75]. In this system, the excitation of the Ru complex induces electron transfer via metal–ligand charge transfer (MLCT). The photo-induced electrons are then transferred onto TiO2, resulting in photocatalytic water splitting to produce H2. The platinum-chloride-modified TiO2 system was reported by Kisch et al. [76,77]. Photo-irradiation of Pt(IV) chloride exhibits visible-light absorption to generate the active center, (Pt4+(Cl−)4 <sup>+</sup> *<sup>h</sup>*<sup>ν</sup> → Pt3+Cl0(Cl–)3. The photo-induced electrons are transferred from Pt3+ to C.B. of TiO2 as reductive sites, while the Cl<sup>0</sup> work as the oxidative sites, resulting in the redox photocatalytic reactions. Important strategies to develop these types of photocatalysts are to design robust sensitizers adjusted with HOMO-LUMO levels.

### *4.2. Visible-Light-Responsive TiO2 Photocatalyst Modified by Phenolic Organic Compounds*

Strong interaction of phenolic groups in organic compounds with Ti–OH of the TiO2 surface probably forms two types of interfacial surface complexes (ISC, Figure 13I), which exhibits visible light absorption via LMCT. The photocatalysis of the ISC is strongly influenced by the electronic structures of the ISC (Figure 13II): the ISC with EWG exhibits strong oxidizability under visible light irradiation, and it can favorably oxidize the TEA, together with H2 evolution from deaerated TEA aqueous solutions [78]. The visible light response of the ISC is attributed to electronic excitation from the donor levels (0.7 V above V.B.) to the C.B. of TiO2 (see Figure 14). Therefore, the electronic structures of sensitizers strongly influence the photocatalytic activities. Ikeda et al. [79] demonstrated that a TiO2 photocatalyst modified with 1,1 -binaphthalene-2,2 -diol (bn(OH)2) exhibited photocatalytic H2 evolution from deaerated TEA aq. under visible light irradiation. Kamegawa et al. [80] designed a 2, 3-dihydroxynaphthalene (2,3-DN)-modified TiO2 photocatalyst for the reduction of nitrobenzene to aminobenzene under visible light irradiation.

**Figure 13.** Schematic illustration for the formation of two types of ISCs [I], and photocatalytic H2 evolution [II] from aq. TEA (10 vol. %) on (a) BC/TiO2, (b) MC/TiO2, (c) CA/TiO2, (d) BA/TiO2, (e) BN/TiO2, and (f) TN/TiO2 [78]. BC: 4-*t*-butyl catechol, MC: 3-methoxy catechol, CA: catecol, BA: 2,3-dihydroxy benzoic acid; BN: 3,4-dihydroxy benzonitrile; TN: tiron.

**Figure 14.** Schematic illustration of photo-induced charge separation on the BN/TiO2 for H2 evolution from TEA aq. in the presence of Pt as co-catalyst under visible light irradiation [78].

On the other hand, the phenolic compounds were degraded on the TiO2 in the presence of O2 under visible light (λ > 420 nm) illumination, producing Cl– and CO2 [81]. The ISC formed by the interaction of phenolic compounds with TiO2 exhibited self-degradation. It was proposed that an electronic transition occurs from the ISC to the C.B. of TiO2 to form active oxygen species, which also participate in the oxidative degradation of phenolic compounds.

### *4.3. Interfacial-Surface-Complex-Mediated Visible-Light-Sensitive TiO2 Photocatalysts*

The interfacial surface complex (ISC)-mediated visible-light-sensitive TiO2 photocatalyst was applied to selective oxidation of several aromatic alcohols [82–88]. Unlike to the ISC in Figure 14, reactant molecules adsorbed onto the TiO2 surface (ISC) is activated under visible-light irradiation, and they are converted into products. Figure 15 shows reaction time profiles for the oxidation of

benzyl alcohol in an acetonitrile solution suspended with TiO2 photocatalyst in the presence of O2 under visible light irradiation (λ > 420 nm). This reaction does not proceed without TiO2 or irradiation. It was found that the amount of benzyl alcohol decreased with an increase in the irradiation time, while the amount of benzaldehyde increased. Neither benzoic acid nor CO2 were formed as oxidative products. The yield of benzaldehyde reached circa 95%, and the carbon balance in the liquid phase was circa 95% after photo-irradiation for 4 h.

**Figure 15.** Selective oxidation of benzyl alcohol on TiO2 (50 mg) under visible light irradiation [82]. The initial amount of benzyl alcohol was 50 μmol. Amounts of: benzyl alcohol (a); benzaldehyde (b); benzoic acid (c); CO2 (d); and percentage of total organic compounds in solution (e).

Photocatalytic oxidation of benzyl alcohol and its derivatives into corresponding aldehydes was carried out with TiO2 under visible light irradiation. Benzyl alcohol and its derivatives substituted by –OCH3, –Cl, –NO2, –CH3, –CF3, and –C(CH3)3 groups were successfully converted to corresponding aldehydes with a high conversion and high selectivity on TiO2, while no other products were observed (See Table 4). However, the phenolic compound (entry 9) was deeply oxidized, since it strongly adsorbed on the TiO2 surface [82].


**Table 4.** Chemoselective photocatalytic oxidation of different kinds of benzylic alcohols on TiO2 [82].

4.3.1. What Is the Origin of the Visible Light Response?

The interaction of benzyl alcohol with TiO2 was analyzed by FT-IR spectroscopy (See Figure 16). Characteristic features of the ISC are as follows: (i) a remarkable downward negative band at 3715 cm−<sup>1</sup> attributed to the O–H stretching of the terminal OH group; (ii) a new band appeared at circa 1100 cm<sup>−</sup>1,

which is attributed to the C–O stretching of the alkoxide species formed by the interaction of benzyl alcohol with TiO2, while that of benzyl alcohol by itself is 1020 cm<sup>−</sup>1.

**Figure 16.** FT-IR spectra [I] of benzyl alcohol by itself and benzyl alcohol adsorbed on TiO2; and [II] their peak identification [82].

When the TiO2 was treated by diluted HF (aq), the IR band at 3715 cm−<sup>1</sup> on the HF–TiO2 drastically decreased, while the photocatalytic activity significantly decreased. The active sites were confirmed to be alkoxide by the interaction of benzyl alcohol with the terminal OH groups of TiO2.

TiO2 by itself exhibited absorption only in the UV region, which is attributed to the charge transition from V.B. to C.B. When the benzyl alcohol was adsorbed on TiO2, absorption in the visible region could be observed. This absorption in the visible light region is assignable to the ISC through the LMCT (See Figure 17). The action spectra of apparent quantum yield (AQY) plots were fitted with the photo-absorption of TiO2-adsorbed benzyl alcohol, suggesting that visible light absorption directly participated in the photocatalytic reactions.

**Figure 17.** UV-vis absorption spectra of TiO2 (a), TiO2 adsorbed with benzyl alcohol (b), and apparent quantum yield (AQY) for the formation of benzaldehyde (c); and schematic illustration of photo-induced charge transfer through LMCT in the alkoxide [82].

DFT calculations [87] indicated the interaction of benzyl alcohol with surface hydroxyl groups on the TiO2 surface, resulting in the formation of alkoxide species. The electron density contour maps for the alkoxide species are shown in Figure 18. The orbital #212 at −0.80 eV forms the V.B. of TiO2, while #218 at +2.25 eV forms the C.B. One type of surface state consisting of the orbital (#215) originates with the alkoxide species ([Ti]–O–CH2–ph) hybridized with the O2p AOs in the V.B. of the TiO2. The energy gap between #215 and #218 (2.8 eV) was confirmed to be the origin for the visible light response.

**Figure 18.** Photo-induced electron transfer from the hybridized orbital to the C.B. of TiO2 under visible light irradiation [87]. Density maps of V.B., C.B., and hybridized orbital are shown here.

### 4.3.2. What Makes the High Selectivity for the Photocatalytic Reactions?

It was observed that benzyl alcohol is adsorbed on TiO2 more favorably than benzaldehyde in a mixture of benzyl alcohol and benzaldehyde under dark conditions. This result indicates that the interaction between benzaldehyde and TiO2 is fairly weak. According to DFT calculations [87,88], the interaction of benzyl alcohol with the TiO2 surface formed a hybridized orbital, while benzaldehyde did not form orbital mixing. Therefore, once benzaldehyde was produced by the oxidation of benzyl alcohol, benzaldehyde was immediately released into the bulk solution, and was not oxidized further to benzoic acid or CO2.

### 4.3.3. Reaction Mechanisms behind the Selective Photocatalytic Oxidation of Benzyl Alcohol

The photocatalytic activities for the oxidation of benzyl alcohol or α, α-d2 benzyl alcohol were investigated. The kinetic isotope effect (KIE) [=*k*C-H/*k*CD] was estimated to be 3.9 at 295 K. This result suggests that the process for the α-deprotonation is the rate determining step (RDS) for the overall reaction. From the experimental and theoretical studies by DFT calculations, one of the favorable reaction paths is depicted in Figure 19. When benzyl alcohol interacts with Ti–OH of the TiO2, the alkoxide species (ISC) is formed on a Ti site (3). The ISC was photo-excited under visible light irradiation via LMCT of the ISC, which induces holes (h+) and electrons (e−). Subsequently, the electrons are transferred to O2 to form superoxide anions (the bonding distance between O–O becomes longer), which induces α-deprotonation of the benzyl alcohol (4-5TS). Such hydro-peroxide species would further induce the de-protonation from another benzyl alcohol to form benzaldehyde (7-8TS), resulting in regeneration of the surface terminal OH groups. The consecutive generation of the terminal OH groups would, thus, be one of the key factors for the photocatalytic reactions.

**Figure 19.** Possible reaction path for the selective oxidation of benzyl alcohol in the presence of O2 on the TiO2 under visible light irradiation [87].

#### *4.4. Photocatalytic Oxidation of Benzyl Amine into Imine*

Imines are important intermediates for the synthesis of pharmaceuticals and agricultural chemicals. Selective photocatalytic oxidation of benzyl amine into N-benzylidenebenzylamine takes place in the presence of O2 on the TiO2 at room temperature (Scheme 1) [89,90]. Several kinds of benzylic amines were examined, and they were converted into the corresponding imines, yielding circa 38–94% [89]. The origin of the visible light response is due to formation of amine oxide (ISC) through the interaction of benzylic amine onto the surface of TiO2, and the ISC exhibits electronic transition from the localized N 2p orbitals of the amine oxide (ISC) to the C.B. of TiO2. The photo-induced redox catalysis produces benzaldehyde in the presence of O2. Subsequently, the condensation reaction of benzaldehyde with another benzyl amine forms N-benzylidenebenzylamine under dark conditions.

**Scheme 1.** Selective oxidation of benzyl amine into N-benzylidenebenzylamine on the TiO2 photocatalyst under visible light irradiation.

#### **5. Conclusions**

This review focused on some fundamental issues behind the visible-light-sensitive TiO2 photocatalysts, highlighting the bulk and/or surface electronic structures modified by doping with nitrogen anions; plasmonic Au NPs, and interfacial surface complexes (ISC) and their related photocatalysts. Tailoring the interface and bulk properties, including surface band bending, sub-band structure, surface state distribution, and charge separation, significantly reflects on the photocatalysis. We hope that this review has provided some useful contributions for the future design and development of novel photocatalytic systems employing TiO2 as well as non-TiO2 semiconductor materials with nanoscale levels. The applications of such photocatalytic systems could not only convert

unlimited solar energy into chemical energy, but also protect our environment, leading to sustainable green chemistry.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **Abbreviations**


### **References**


© 2019 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Review* **Titanium Dioxide: From Engineering to Applications**

### **Xiaolan Kang †, Sihang Liu †, Zideng Dai, Yunping He, Xuezhi Song and Zhenquan Tan \***

School of Petroleum and Chemical Engineering, Dalian University of Technology, No. 2 Dagong Road, New District of Liaodong Bay, Panjin, Liaoning 124221, China; kxl@mail.dlut.edu.cn (X.K.); wdlsd@mail.dlut.edu.cn (S.L.); xiaodai@mail.dlut.edu.cn (Z.D.); yphe\_04@mail.dlut.edu.cn (Y.H.); songxz@dlut.edu.cn (X.S.)

**\*** Correspondence: tanzq@dlut.edu.cn; Tel.: +86-427-263-1808

† These authors contributed equally to this work.

Received: 17 January 2019; Accepted: 10 February 2019; Published: 19 February 2019

**Abstract:** Titanium dioxide (TiO2) nanomaterials have garnered extensive scientific interest since 1972 and have been widely used in many areas, such as sustainable energy generation and the removal of environmental pollutants. Although TiO2 possesses the desired performance in utilizing ultraviolet light, its overall solar activity is still very limited because of a wide bandgap (3.0–3.2 eV) that cannot make use of visible light or light of longer wavelength. This phenomenon is a deficiency for TiO2 with respect to its potential application in visible light photocatalysis and photoelectrochemical devices, as well as photovoltaics and sensors. The high overpotential, sluggish migration, and rapid recombination of photogenerated electron/hole pairs are crucial factors that restrict further application of TiO2. Recently, a broad range of research efforts has been devoted to enhancing the optical and electrical properties of TiO2, resulting in improved photocatalytic activity. This review mainly outlines state-of-the-art modification strategies in optimizing the photocatalytic performance of TiO2, including the introduction of intrinsic defects and foreign species into the TiO2 lattice, morphology and crystal facet control, and the development of unique mesocrystal structures. The band structures, electronic properties, and chemical features of the modified TiO2 nanomaterials are clarified in detail along with details regarding their photocatalytic performance and various applications.

**Keywords:** TiO2; energy band engineering; morphology modification; mesocrystals; applications

### **1. Introduction**

Over the past several decades, the increasing severe energy shortages and environmental pollution have caused great concern worldwide. To achieve sustainable development of society, there is an urgent need to explore environmentally friendly technologies applicable to pollutant recovery and clean energy supplies. In the long-term, solar energy is an inexhaustible source of renewable energy; therefore, developing technologies and materials to enhance solar energy utilization is central to both energy security and environmental stewardship. In 1972, Fujishima and Honda first published a study for producing hydrogen on titanium dioxide (TiO2) photoelectrodes under ultraviolet light illumination, which garnered worldwide attention [1,2]. From then on, semiconductor photocatalysis has been considered one of the most promising pathways to address both hydrogen production and pollution abatement. Photocatalysis can be widely used anywhere in the world, providing natural solar light or artificial indoor illumination is available [3].

Semiconductor materials are often used as photocatalysts [4]. According to band energy theory, the discontinuous band structure of semiconductors is composed of low energy valence bands filled with electrons, high-energy conduction bands, and band gaps. When the energy of the incident photons equals or exceeds the bandgap, the photoexcitation of electron–hole pairs and the consequential

photocatalytic redox reaction take place [5]. The photocatalytic process mainly involves the steps of generation, separation, recombination, and surface capture of photogenerated electrons and hole pairs. Photochemical reactions occur on the surface of a solid catalyst, which includes two half-reaction oxidation reactions of photogenerated holes and reduction reactions of photogenerated electrons [6]. The specific process that occurs in semiconductors is described in Figure 1. During this process, a large proportion of charge carriers (e−/h+ pairs) recombine quickly at the surface and interior of the bulk material, leading to the dissipation of absorbed energy in the form of light (photon generation) or heat (lattice vibration). Therefore, these charge carriers cannot participate in the subsequent photocatalytic reactions, which is detrimental to the whole process [7].

**Figure 1.** Photocatalytic process in semiconductor.

The electrons and holes that successfully migrate to the surface of the semiconductor without recombining can be involved in the reduction and oxidation reactions, respectively, which are the bases for photodegradation of organic pollutants and photocatalytic water splitting to produce H2 [8]. As excellent oxidizers, the photogenerated holes can mineralize organic pollutants directly. In addition, the holes can also form hydroxyl radicals (•OH) with strong oxidizing properties. Photoexcited electrons, on the other hand, can produce superoxide radicals (O2•−) and •OH. These free radicals and e−/h<sup>+</sup> pairs are highly reactive and can induce a series of redox reactions. In addition, with respect to water splitting, photogenerated electrons can be captured by H+ in water to generate hydrogen, while holes will oxidize H2O to form O2 [9–11].

In general, to increase the activity of photocatalysts and utilize visible light more effectively, several requirements need to be satisfied. First, the light absorption process determines the amount of excited charges, which means that more charge carriers are likely to be accumulated on the surface if more light can be absorbed by the photocatalyst. Additionally, considering that ultraviolet (UV) light occupies less than 4% of sunlight's emission spectrum, while visible light accounts for approximately 40%, a smaller bandgap is necessary for a semiconductor to absorb solar energy across a broad range of spectra. Therefore, improving the optical absorption properties has become a common purpose for photocatalyst design to enhance their overall activity [12]. In addition, the position of conduction bands (CBs) and valence bands (VBs) is critical, which are responsible for the production of active species, such as •OH, HO2•, H2O2, and O2•−. Furthermore, the photogenerated electrons and holes should be transported and separated efficiently in the photocatalyst because the fast recombination of charge carriers will otherwise result in low reactivity. Finally, the as-prepared photocatalytic materials and their modification processes should be environmentally friendly and economical [13].

Since 1972, TiO2 has been intensively investigated due to its thermal and chemical stability, superhydrophilicity, low toxicity, and natural geologic abundance. Compared with other semiconductor materials, TiO2 is of ubiquitous interest across many research fields and for many applications [14], such as photodegradation of pollutants and hazardous materials, photolysis (splitting) of water to yield H2, artificial photosynthesis, etc. Nevertheless, the poor visible light absorption and fast electron–hole recombination, as well as the sluggish transfer kinetics of the charge carriers to the surrounding media, considerably limit the photocatalytic activities of TiO2. Hence, during the past few decades, much effort has been devoted to overcoming these problems by, for example, reducing e−/h+ pair recombination and improving the optical absorption properties by energy band regulation, morphology control, and the construction of heterogeneous junctions [15].

In this review, we mainly focus on the regulation of the electronic structure and modification of the micromorphology of TiO2 nanomaterials to achieve property enhancements that could be applicable to a variety of potential applications.

### **2. Energy Band Engineering of TiO2**

The absorption of incident light and redox potential of TiO2 mainly depend on its energy band configuration [16]. To utilize solar energy more effectively, it is necessary to explore and develop longwave-light-sensitive TiO2 photocatalysts with excellent performance on the basis of energy band engineering [17]. A better understanding of the electronic structure of TiO2 is important for band gap modification. The molecular orbital bonding energy diagram in Figure 2 clearly shows the fundamental features of anatase TiO2 [18]. The chemical bonding of anatase TiO2 can be deconstructed into Ti, e.g., Ti *t*2g (*dyz*, *dxz*, and *dxy*), O *p<sup>σ</sup>* (in the Ti3O cluster plane), and O *p<sup>π</sup>* (out of the Ti3O cluster plane). The upper valence bands include three main regions: the *σ* bonding, which is located at the bottom, is the most stable bond type, and arises from the hybridization of Ti, e.g., O *pσ*; the hybridization of the O *p<sup>π</sup>* and Ti *dyz* (or *dxz*) orbitals constitutes the middle energy region of π bonding; and the higher energy region in the top of the valence bands, which is dominated by the O *pπ* orbitals. The conduction band is composed of Ti 3d and 4s, and the bottom of the conduction bands is composed of the isolated Ti *dxy* orbitals [19,20]. For the purpose of narrowing the bandgap of TiO2, three basic approaches of adjusting the VBs or CBs or the continuous modification of the VBs and CBs of the anatase are shown in Figure 3.

**Figure 2.** (**a**) Total and projected densities of states (DOS) of the anatase TiO2 structure and (**b**) molecular orbital bonding structure for anatase TiO2 [18]. Copyright 2004 The American Physical Society.

**Figure 3.** Three schemes of the band gap modifications of TiO2 match the solar spectrum: (**a**) a higher shift in valence band maximum (VBM); (**b**) a lower shift in conduction band minimum (CBM); and (**c**) continuous modification of both VBM and CBM.

### *2.1. Doping of TiO2*

To extend the visible light response of TiO2 and improve its photocatalytic activities, various modification strategies, such as dye sensitization, impurity or intrinsic doping or semiconductor coupling, have been developed [21–23]. Among them, introducing impurity ions into the TiO2 crystal lattice to substitute the host anions and/or cations has earned much attention in the past decade.

By means of physical or chemical methods, researchers have been able to introduce a variety of ions into the TiO2 matrix, where they change the band structure of TiO2 by inducing impurity states within the bandgap [2], as shown in Figure 4. In general, ion doping contributes to the improved activities of TiO2 in three ways: (1) by narrowing the bandgap and promoting the adsorption of the main region of the solar spectrum, such as doping with N, S, C, B, etc. [24,25]; (2) by improving the conductivity of TiO2 and the mobility of charge carriers, the increased charge traps can reduce bulk recombination and separate photogenerated electrons and holes more efficiently (e.g., Zn, Fe, and Y) [26]; and (3) by altering the conduction band position of TiO2 with certain metal ion dopants, such as Zr4+, Nb5+, and W6+, which further affects the carrier transfer properties [27].

**Figure 4.** TiO2 nanoparticles with different doping elements [2]. Copyright 2014 American Chemical Society.

TiO2 doping can be doped with a variety of metal ions, including transition metal and rare earth metal ions. For transition metal dopants, such as Fe, Mn, V, Cu, and Cr, both delocalized and localized impurity states will be created within the band gap of TiO2 along the crystal field splitting of metal 3d orbitals [28–30]. Mizushima et al. determined impurity levels of 1.9 to 3.0 eV below CBM by doping V, Cr, Mn, and Fe based on a large number of experimental results, and they suggested that cation vacancies may lead to these impurity states [31]. An early work by Borgarello et al. in 1982 reported that Cr3+-doped TiO2 nanoparticles (investigated for properties of photocatalytic hydrogen evolution) exhibit excellent absorption of visible light in the range of 400 to 550 nm. They believed that the 3d electrons of Cr3+ were excited into the conduction band of TiO2, thus inducing a visible light response [32]. Doping TiO2 with certain earth rare metal ions represents another promising method to prolong the recombination time of charge carriers and improve their separation efficiency. The 4f electrons in most rare earth elements can give rise to the formation of a multielectron configuration, which acts as a shallow trap for photogenerated electrons and holes [33]. Furthermore, the use of rare earth metal ion dopants in TiO2 tends to facilitate the utilization of solar light from ultraviolet to infrared light regions. Li et al. prepared a series of Ce-doped TiO2 nanoparticles by the sol–gel method. The characterization results showed that Ce ions entered the TiO2 matrix at Ti sites, leading to the formation of impurity states, as shown in Figure 5. In addition, enhanced separation of the photogenerated charge carriers was also realized due to the coexistence of Ce3+ and Ce4+ dopant ions [34].

**Figure 5.** Band energy structure and charge transfer [34]. Copyright 2017 American Chemical Society.

Anandan et al. studied the photodegradation of monocrotophos under visible light irradiation with La-doped TiO2. They associated rapid mineralization with the enhanced separation of electrons and holes by doping La3+ into the TiO2 matrix, which subsequently generated a large number of •OH radicals along with the trapping of excess holes at the surface [35]. In contrast, based on the density functional theory calculation method, Sun et al. worked extensively on the changes of the electronic structure and the photocatalytic activity of TiO2 after introducing substitutional La dopants. Their calculations demonstrate that the enhanced visible light absorption of La–TiO2 mainly arises from adsorbed La on the TiO2 surface rather than from substitutional La doping [36]. Notably, not all kinds of dopants give rise to positive consequences. Chio et al. systematically studied 21 kinds of metal ion-doped TiO2 materials and their application with respect to various photocatalytic reactions [37]. The results associated with model reactions for the photocatalytic reduction of carbon tetrachloride and the photodegradation of chloroform indicated that only the doping of certain ions, such as Fe3+, Ru3+, Re5+, V4+, and Mo5+, increased reactivity. In addition, the study demonstrated that optimizing the content and placement of the dopant ions content play a positive role in affecting photocatalytic activity. Despite the robust photoactivity of certain metal ion-doped TiO2 catalysts, some inevitable problems remain and need to be considered. The metal-doped nanomaterials have been shown to suffer from unstable optical properties and thermal instability, in addition to the need to use expensive ion implantation equipment to produce these enhanced materials [38]. Furthermore, the localized *d*-electron state formed in the band gap of TiO2 may become the recombination center of photogenerated electron–hole pairs, thereby leading to a decline in the photocatalytic activity.

Recently, the non-metal doping of nitrogen (N), sulfur (S), carbon (C), fluorine (F), iodine (I), and phosphorus (P) has been extensively studied due to their relatively high photostability and photoelectric properties [39]. However, in comparison to metal-doped TiO2, the role of the non-metal dopants as recombination centers of charge carriers might be minimized. By replacing the oxygen atoms in the TiO2 lattice, the non-metal elements can significantly narrow the bandgap and thereby improve the visible light response of TiO2. In addition, impurity states can be formed near the valence band edge alone with non-metal doping, as displayed in Figure 6. Instead of acting as recombination centers, these occupied levels can be regarded as shallow traps that effectively separate photogenerated electron–hole pairs [40].

**Figure 6.** Comparison of atomic p levels among anions. The band gap of TiO2 is formed between the O 2pπ and Ti 3d states [39]. Copyright 2014 American Chemical Society.

In 2001, Asahi et al. first published research on N-doped TiO2 nanomaterials, which initiated a wave of studies related to non-metal-doped photocatalysts [41]. In a similar work, Zhao et al. reported highly active N-doped TiO2 nanotubes for CO2 reduction. Despite the tubular structure with a large surface area providing more surface active sites, the N dopants contributed more to the improved photocatalytic activity. It was found that a redshift of the light absorption and a color center were achieved with N-doped TiO2 nanotubes because N atoms can substitute for the lattice O atoms of TiO2, thereby reducing its bandgap and resulting in a ~4 times higher visible light photocatalytic CO2 reduction activity in comparison to pure TiO2 nanotubes [42]. Irie et al. prepared C-doped TiO2 nanoparticles by oxidizing TiC powder, and the efficiency of decomposing gaseous isopropanol under visible light was significantly improved [43]. S-doped anatase TiO2 with a high surface area was obtained by Li et al. They treated pure TiO2 using a supercritical strategy and used the materials for methylene blue degradation under visible light irradiation. S atoms with large diameters are difficult to dope into the TiO2 lattice, but X-ray photoelectron spectroscopy (XPS) detected the existence of S–Ti–O bonds, which introduced lattice defects, acting as shallow traps for electrons and reducing carrier recombination [44]. Li et al. mixed HIO3 with tetrabutyl titanate and hydrolyzed the samples directly to obtain I-doped TiO2, which significantly boosted its visible light performance [45].

Although various non-metal ions are used for doping modification of TiO2, N doping is still one of the most widely used methods to modify the electronic structure and to extend light absorption to the visible range [46]. However, researchers have not yet come to a complete agreement regarding the mechanisms associated with the N doping enhancements. In the literature, it is not difficult to find studies stating that it is not only the dopant concentration but also the dopant location in the TiO2 lattice (surface or bulk, substitutional, and interstitial) that ultimately determines the photocatalytic properties [17,47]. In the case of N-doped TiO2 nanomaterials, some researchers believe that only the substitution of O2<sup>−</sup> by N3<sup>−</sup> with high dopant concentrations can elevate the valence band edge, bringing about the desired band gap narrowing [48,49]. However, others suggest that the doping of N will induce oxygen vacancies in TiO2 and that the enhanced visible light adsorption is associated with the local state induced in the band gap, rather than the generally believed theory that the introduction of N into the TiO2 lattice can reduce its band gap, as shown in Figure 7 [50].

**Figure 7.** (**a**) Diffuse reflectance spectra of the anatase TiO2 nanobelts before and after heat treatment in ammonia gas flow at different temperatures and (**b**) the band structure of N-doped-TiO2 under visible and UV light irradiation [50]. Copyright © 2009 American Chemical Society.

As another widely studied non-metal-doped TiO2, F-doped TiO2 also shows promising potential for photocatalytic applications. Zhang et al. obtained F-doped TiO2 mesocrystals through the topological transformation of TiOF2 precursors. An in situ characterization technique was adopted to detect the doping process. The results showed that the doping of F was accompanied by the formation of oxygen defects, which ensured a higher visible light response [51]. Park et al. added sodium fluoride to aqueous TiO2 suspensions to obtain surface fluorinated TiO2, and a series of characterizations showed that neither an improvement in crystallinity nor a redshift of the band edge was achieved, but the photocatalytic oxidation of phenol and Acid Orange was considerably enhanced. They attributed such photocatalytic improvement to fluorine surface modification, which enhances free •OH radical-mediated oxidation pathways [19]. Similar to the doping of N, the reason for the observed high performance upon F doping is still undetermined. Some studies suggest that instead of entering the TiO2 lattice, fluorine ions adsorbed on the surface of TiO2 can increase the wettability and surface acidity, which is beneficial to the adsorptivity and e−/h<sup>+</sup> separation of the oxide [20]. Other researchers hold the opinion that a tail state in the band gap of TiO2 is formed by F doping, which favors the more efficient utilization of incident light. Recently, an increasing number of studies proposed that a charge compensation effect induced by F doping brings about the formation of a certain amount of oxygen vacancies and Ti3+ in TiO2, resulting in the enhanced absorption of visible light [52,53]. Although the principle of F doping is not very clear, the proper doping level of F can effectively improve the activity of TiO2.

#### *2.2. Intrinsic Defect Formation*

In 2011, a black TiO2 with a narrowed bandgap (approximately 1.5 eV) and fabricated by hydrogenation reduction was reported to achieve absorption of full spectrum sunlight and improved photocatalytic activity [54]. Unsurprisingly, this discovery has aroused worldwide scientific interest

and paved the way towards intrinsic defect modification. Creating intrinsic defects in the TiO2 lattice is a kind of self-structural modification that includes surface disorder layers, Ti3+/oxygen vacancy self-doping, formation of surface Ti–OH, and incorporation of doped-Consequentially, considerable changes in surface properties and electronic and crystal structures are often achieved in this process [55–57]. Furthermore, studies in terms of defect engineered TiO2 have confirmed that these intrinsic defects are emerging as a promising attribute for improving the separation of electrons and holes, outperforming, in some cases, other kinds of modified TiO2 nanomaterials [58].

Since the study by Chen et al., various methods have been developed to induce defects in TiO2, including direct reduction of TiO2; that is, the currently reported H2, Al, Na, Mg, NaBH4, hydrides, imidazoles, etc. can effectively transfer modify pure TiO2 nanomaterials into their defect engineered counterparts under certain conditions [59,60]. In addition, electrochemical reduction and high-energy particle bombardment (such as photon beam and H2 plasma or electron beam) are widely used to induce TiO2 defects. Partial oxidation from low-valence-state Ti species such as TiH2, TiO, TiCl3, TiN, and even Ti foil represents another promising approach, fulfilling the needs for highly active TiO2−<sup>x</sup> photocatalysts [61]. Liu et al. prepared rice-shaped Ti3+ self-doped TiO2−<sup>x</sup> nanoparticles through mild hydrothermal treatment of TiH2 in H2O2 aqueous solution, and proposed a unique "surface oxide-interface diffusion–redox mechanism" (as shown in Figure 8) to explain the formation process of TiO2−<sup>x</sup> [62]. The defect types and their formation mechanism in TiO2−<sup>x</sup> are closely related to the preparation methods. Generally, the Ti–H bond is present only in hydrogen-reduced TiO2−x, while the surface disorder layer causes severe damage to the TiO2 structure. Thus, relatively strong reduction conditions are required, such as high temperature/pressure hydrogen reduction, aluminothermic reduction, hydrogen plasma treatment, etc. Surface Ti–OH, Ti3+, and oxygen vacancies commonly exist in most defective TiO2 nanostructures [63].

**Figure 8.** (**A**) Schematic of the formation mechanisms for the rice-shaped Ti3+ self-doped TiO2−<sup>x</sup> nanoparticles. (**B**,**C**) The interface diffusion–redox diagram. The green arrows indicate ion diffusion [62]. Copyrighted 2014 The Royal Society of Chemistry.

The dominant mechanism involved in improving photocatalytic performance by inducing intrinsic defects into TiO2 can be explained, both experimentally and theoretically, to be the regulation of the band structure of TiO2 and boosted charge separation and transport. For black TiO2, band tail states and shallow dopant states can be formed to reduce its band gap and further increase its optical absorption properties. Chen et al. observed a disordered surface layer in black TiO2 nanocrystals after a hydrogenation treatment, as shown in Figure 9. From the high-resolution transmission electron microscopy (HRTEM) spectra, it can be readily observed that the straight lattice fringes are bent at the edge of the particles, and the lattice spacing is no longer uniform, indicating that the hydrotreated black TiO2 nanoparticles possess a "crystal-disordered" core–shell structure. Such a disordered layer is believed to facilitate the introduction of the tail state at the top of the valence band and the bottom of the conduction band, consequently yielding a redshift of the light absorption [54]. Moreover, because the disorder layer exhibits a set of properties that are distinct from those of their crystalline counterparts, rapid charge separation could be realized when the amorphous layer closely contacts crystalline TiO2. The lattice distortions tend to blueshift the VBM while having less impact on CBM. Therefore, the photogenerated holes accumulate in the thin disordered shell and participate in the photocatalytic reactions immediately; electrons are widely spread in both the shell and core regions. This result highlights the strong synergistic effect on charge transfer between the crystalline and disordered parts [64].

**Figure 9.** (**A**) Schematic illustration of the structure and electronic DOS of a semiconductor in the form of a disorder-engineered nanocrystal with dopant incorporation. (**B**) A photo comparing unmodified white and disorder-engineered black TiO2 nanocrystals. (**C**,**D**) HRTEM images of TiO2 nanocrystals before and after hydrogenation, respectively [54]. Copyright 2011 American Association for the Advancement of Science.

For Ti3+/oxygen vacancy incorporation and H-doping in reduced TiO2−x, the hybridization of Ti-3d, O-2p and H-1s orbitals results in the mid-gap states formation below the CBM and the Fermi level's upshift [65,66]. The extra electrons in either Ti3+ or oxygen vacancies are inclined to occupy the empty states of Ti ions, forming new Ti 3d bands below the CBM. With a further increase in defect concentration, the 3d band shifts deeper and finally results in multiple bands in the CBM. Moreover, the existence of multiple mid-gap states as well as the associated derivate (surface Ti–OH) can also function as extra carrier trap sites or carrier scavengers to prolong the lifetime of electrons and holes [67]. The high concentration of electron donors will greatly improve the conductivity of materials and promote the transfer of carriers [68]. Wang et al. treated pure white TiO2 with hydrogen plasma to fabricate H-doped black TiO2 for photodegradation of methyl orange under visible light irradiation. The as-prepared samples showed a degradation rate 2.5 times that of the white counterpart [69]. Sinhamahapatra et al. reported a novel controlled magnesiothermic reduction to synthesize reduced TiO2−<sup>x</sup> under 5% H2/Ar atmosphere [70]. During this process, the band position and band gap, surface defects and oxygen vacancies can be well regulated to maximize the optical adsorption in the visible and infrared regions and minimize the charge recombination centers. As shown in Figure 10, a new controlled magnesium thermal reduction method to synthesize and reduce black TiO2 under 5% H2/Ar atmosphere. The material has the best band gap and band position, oxygen vacancy, surface

defect, and charge recombination center, and the optical absorption in visible and infrared regions is improved obviously. These synergistic effects enable the defective TiO2−<sup>x</sup> with Pt as a co-catalyst to produce H2 at a rate of 43 mmol h−<sup>1</sup> g−<sup>1</sup> under the full solar wavelength light illumination, superior to other reported photocatalysts for hydrogen production.

**Figure 10.** (**a**) H2 generation profile, (**b**) rate (rH2) of hydrogen generation for different samples, and (**c**) the stability study of the sample BT-0.5 under the full solar wavelength range of light [70]. Copyright 2015 The Royal Society of Chemistry.

To date, numerous strategies, either common or uncommon, have been developed to introduce various kinds of dopants or defects into the TiO2 matrix. However, considering its highly stable nature, most methods are rigorous and energy-consuming, and are contrary to the sustainable and environmentally friendly development criteria. Therefore, an increasing number of studies are dedicated to seek convenient, economical, energy efficient, and environmentally friendly methods for the structural modification of TiO2 [71]. In our recent studies, we developed a facile photoreduction strategy to induce intrinsic defects into anatase TiO2 to modulate its band structure, thereby extending the absorption of incident light to the visible region. As shown in Figure 11, the band gap was narrowed to 2.7 eV, and the color changed to earth yellow after the photoreduction treatment. NH4TiOF3 mesocrystals were adopted as precursors, which can release fluorine and nitrogen ions during the topological transformation process. Thus, non-metal ion doping (i.e., F and N ions) was also achieved simultaneously, further improving the transport and separation of photogenerated charge carriers. The as-prepared NF–TiO2−<sup>x</sup> exhibited excellent photocatalytic degradation and photoelectrochemical efficiency under visible light irradiation compared to pristine TiO2 [72,73].

**Figure 11.** (**a**) UV–Vis diffuse reflectance spectra and (**b**) Tauc plot for band gap determination [73]. Copyright 2018 Springer Nature Publishing AG.

#### **3. Morphology Modification**

It is well known that the photocatalytic performance of semiconductors is closely related to their structural and morphological characteristics at the nanoscale, including their size, dimensionality, pore structure and volume, specific surface area, exposed surface facets, and crystalline phase content [74]. During the past few decades, numerous promising structure engineering strategies have been developed to fabricate highly active photocatalysts with the desired morphology and structure. Among them, particular emphasis has been placed on controlling and optimizing the structural dimensionality of a given semiconductor to improve its photocatalytic efficiency.

Zero-dimensional TiO2 nanospheres are the most widely studied TiO2-based materials because of their high specific surface area and attractive pore structures [75–77]. Figure 12 shows a classic ripening approach to synthesize hollow nanospheres [75]. As photocatalytic reactions take place on the surface of the photocatalyst, TiO2 nanoparticles with smaller sizes are inclined to provide more reactive sites, resulting in better photocatalytic performance. Moreover, due to the quantum size effect, the photogenerated electrons and holes in the bulk regions are able to migrate to the surface of TiO2 nanoparticles via shorter distances, thereby considerably reducing the carrier quench rate [78]. TiO2 nanospheres are also good candidates as light captors, and their structural features enable as much light as possible to access the interior, resulting in amazing light harvesting capabilities. However, it should be mentioned that the diffusion length of photogenerated electrons and holes must be longer than the particle size to avoid the recombination of the dominant carriers on the surface of the photocatalyst, which is very important for achieving efficient charge carrier dynamics [79].

One-dimensional (1D) nanostructures, including nanotube (NT), nanorod (NR), nanobelt (NB), and nanowire (NW), have become a popular research topic in recent years. They have been extensively studied because of their distinct optical, electronic and chemical properties. Despite some similar features with nanoparticles, such as quantum confinement effects and large surface area, 1D nanomaterials possess many unique properties, which are hard for other categories of structured materials to achieve. For example, 1D nanostructures restrict the migration of electrons and protons by allowing the lateral confinement of electrons/protons and guide their transport in the axial direction [80,81]. Furthermore, excellent flexibility and mechanical properties enable them to be easily used and recycled. In this regard, 1D TiO2-ordered nanostructures are promising not only for constructing highly active photocatalytic systems but also for building blocks for various (photo)electrochemical devices, such as batteries, fuel cells, solar cells, and photoelectrochemical cells. To further optimize the photocatalytic reactivity of 1D TiO2 nanomaterials, one can precisely regulate the aspect ratio (the ratio of length to diameter) or modify these 1D nanostructures with novel strategies to accelerate electron transport and separation processes, as well as to enhance the capture of incident light; TiO2 nanotubes are examples of these materials [82]. Through the electrochemical anodization process, it is possible to precisely control the tube crystal structure

(anatase, rutile, or amorphous) and tube geometry (diameter and length), as shown in Figure 13a, or direct the tube arrangements to obtain a defined tube-to-tube interspace (Figure 13b). For the sake of extending the scope of application, constructing flow through membranes with TiO2 nanotubes is a good choice (Figure 13c). Other modifications for minimizing charge carrier annihilation and boosting light harvesting are illustrated in Figure 13d–i, ranging from self-decoration to surface alterations to energy band engineering.

**Figure 12.** (**A**) Schematic illustration (cross-sectional views) of the ripening process and two types (i and ii) of hollow structures. Evolution (TEM images) of TiO2 nanospheres synthesized with 30 mL of TiF4 (1.33 mM) at 180 ◦C with different reaction times: (**B**) 2 h (scale bar = 200 nm), (**C**) 20 h (scale bar = 200 nm), and (**D**) 50 h (scale bar = 500 nm) [75]. Copyright 2004 American Chemical Society.

TiO2 nanosheets, nanoflakes, and thin films consist of titania-based two-dimensional nanomaterials, which have flat surfaces and high aspect ratios. The lateral size of some nanomaterials is controllable, ranging from the sub-micrometer or even nanometer level to several tens of micrometers with thicknesses of 1–10 nm. Such structures provide TiO2 nanomaterials with several unique characteristics, such as excellent adhesion to substrates, low turbidity and high smoothness [83]. Furthermore, when exposed to UV light irradiation, TiO2 2D nanomaterials exhibit superhydrophilicity, which leads to a variety of potential applications, such as self-cleaning coatings and electrodes in photoelectronic devices [84]. Notably, considering that photocatalytic reactions always occur on the surface of catalysts, the exposed crystal facets are of great importance in determining the photocatalytic performance. Accordingly, developing TiO2 crystals with different active facets is highly desirable in many applications. In general, TiO2 nanocrystals have three basic low-index exposed facets—{101}, {001}, and {010}—with surface energy relationships of {001}, 0.90 J m−<sup>2</sup> > {100}, 0.53 J m−<sup>2</sup> > {101}, 0.44 J m−<sup>2</sup> [85,86]. Therefore, as the most thermodynamically stable facets, the {001} crystal facet is dominant among most anatase TiO2 nanomaterials, reducing the overall surface energy of the material. In 2008, Yang et al. first reported TiO2 single crystals with 47% highly active {001} facets exposed to HF as capping agents [87]. This work has attracted considerable global attention. Since then, TiO2 with various ratios of exposed {001} facets have been successfully fabricated [88]. Meanwhile, other active planes, such as {010}, {111}, and {110}, have also been reported and widely used in water splitting, solar cells, artificial light synthesis and other fields, as shown in Figure 14 [89]. Zheng et al. obtained {001} facet-oriented anatase by facile heat treatment of a tetrabutyl titanate, absolute ethanol, and HF mixture. Such a material with 85% {001} facets exhibited much higher photocatalytic activity in comparison to commercial P25 materials [90].

**Figure 13.** Schematic drawing of (**a**,**b**) formation and (**c***−***i**) modification of anodic nanotube arrays (as discussed in the text) [82]. Copyright 2017 American Chemical Society.

During the process of photocatalytic reactions, oxidation predominantly occurs in the {001} facets, while reduction occurs in the {101} crystal plane of TiO2 because the {101} facet (with relatively low surface energy) tends to attract more electrons. Electron holes subsequently accumulate in the {001} plane, facilitating the space separation of electron–hole pairs [91]. In addition, Ti atoms of the {001} plane exist mainly in the form of 5-coordination, which can provide more active sites that more readily attract free reactant molecules than the {101} plane. Thus, when a certain proportion of {001} crystal facets are exposed, the photocatalytic activity increases rapidly. Nevertheless, it is not always the case that a higher {001} crystal face exposure ratio results in improved catalytic performance. Studies have reported that the photocatalytic activity is compromised when the proportion of {001} facets exceeds 71% [89]. In addition, faceted TiO2 photocatalysts suffer from weak visible light utilization due to their large band gap. Hence, the modification of the electronic structure of faceted TiO2 to fully utilize sunlight and promote the migration and separation of electron/hole pairs is highly desirable. Wang et al. prepared Ti3+ self-doped TiO2 mesoporous nanosheets dominated by {001} facets with supercritical technology. They associated the extended region of incident light absorption with the introduction of Ti3+ [91]. Using an ionic liquid as a surface control agent, Biplab et al. synthesized microporous TiO2 nanocrystals with exposed {001} facets. After depositing Pt on the surface, the hydrogen production rate in visible irradiation was greatly improved [92].

**Figure 14.** Summary of main shapes and applications (i.e., lithium ion batteries, photocatalytic hydrogen evolution, photodegradation, and solar cells) of anatase, rutile, and brookite TiO2 crystals with their surfaces consisting of different Facets [89]. Copyright 2014 American Chemical Society.

A three-dimensional TiO2 hierarchical structure based on intrinsic shape-dependent properties has been the central focus of many recent studies. Designed and fabricated 3D TiO2 nanomaterials commonly incorporate interconnected structures, hollow structures and hierarchical superstructures constructed from small dimensional building blocks [93]. Most of these novel structures include larger spatial dimensions and more varied morphologies. The high surface-to-volume ratio provides a more efficient diffusion path for reactant molecules, enabling the contaminant molecules to enter the framework of the photocatalyst for efficient purification, separation, and storage. In addition, the unique optical characteristic is of particular interest because many of these architectures have distinctive physicochemical properties favorable for incident light utilization. For example, when light is irradiated onto the surface of the TiO2 hierarchical structure, photons are scattered multiple times, so the probability of the catalyst absorbing photons is increased; this phenomenon is known as the "trapping effect" and is illustrated in Figure 15 [94].

**Figure 15.** Schematic diagram of the reflecting and scattering effects in hierarchical microspheres [94]. Copyright 2014 The Royal Society of Chemistry.

*Catalysts* **2019**, *9*, 191

The hollow structure TiO2 nanomaterials have attracted considerable attention due to their amazing light harvesting ability, low density, and large specific surface area. The hollow structure, on the one hand, is capable of providing a large amount of space to accommodate more reactant molecules, thereby increasing the effective contact between the catalyst and the reactants. On the other hand, incident light inside the cavity can undergo multiple reflections to capture more light, as shown in Figure 16 [95]. Kondo et al. obtained TiO2 hollow nanospheres through hydrothermal and calcination processes with polymer polyethylene cationic balls as templates. The as-prepared photocatalyst had more favorable activity than its commercial counterparts with respect to decomposing isopropanol [96]. In the following work, an ultrathin TiO2 shell-like structure was prepared in a similar manner with a shell thickness of approximately 5 nm. The morphology of the TiO2 hollow materials prepared by the hard template method is relatively uniform, and the composition and thickness of the shells are adjustable. However, the preparation process is complicated and requires multiple execution steps to be realized. Moreover, the hollow structure may be destroyed when the template is removed. Therefore, alternative strategies, including soft templates and non-template methods, have played an increasingly important role in the development of hollow structure TiO2 nanomaterials in recent years.

**Figure 16.** Comparison of photocatalytic activities of titania spheres with solid, sphere-in-sphere, and hollow structures [95]. Copyright 2007 American Chemical Society.

Li et al. prepared hollow TiO2 nanospheres with high photocatalytic activity by a template-free process. The increased catalytic activity is mainly due to the multiple reflections of incident light inside the TiO2 sphere, which extends the optical path [97]. Multichannel TiO2 hollow nanofibers were constructed by Zhao et al. for degrading gaseous acetaldehyde, and the specific surface area of this material increased rapidly as the number of channels increased. They proposed that the multichannel hollow structures induced both an inner trap effect on gaseous molecules and a multiple-reflection effect on incident light, which were the main reasons for the improved photocatalytic activity of TiO2 hollow fibers [98]. Shang et al. synthesized submicron-sized TiO2 hollow spheres from a mixture of TiCl4, alcohols, and acetone by a template-free solvothermal method. Control of the sphere size was achieved by adjusting the ratio of ethanol to acetone. Based on a series of characterizations, they suggested a possible formation mechanism for the hollow structure: the tiny anatase phase TiO2 nanoparticles with poor crystallinity form through a hydrolysis reaction, due to the very high surface energy, and then quickly aggregate to form spheres. The increased water promotes the crystallinity of particles in the spherical shell, while the internal particles dissolve and migrate to the spherical shell, leading to the formation of highly crystalline TiO2 hollow spheres [99]. An intriguing work carried out by Kang et al. to establish hierarchical anatase TiO2 nanocubes with hollow structures has been reported recently. Instead of seeking complicated templates or surfactants, they directly converted NH4TiOF3 mesocrystals to hollow spiny TiO2 with a high specific area and photodegradation activity [73].

### **4. TiO2 Mesocrystals**

It is widely accepted that for TiO2-based photocatalytic materials, large crystallites result in high structural coherence, which benefits the transfer and separation of electron–hole pair, while the availability of plentiful reaction sites is dependent on obtaining large specific surface areas. However, producing a structure that simultaneously satisfies the requirements of large crystallites and high surface area is extremely challenging. Fortunately, the advent of mesocrystals is a promising material that may meet the challenge [100]. Mesocrystals were first proposed by Cölfen and Antonietti in 2005, and since then have received increased attention [101]. Different from the classical single crystals in which the crystal lattice of the entire sample is continuous with no grain boundaries and polycrystals whose units do not have the same orientation, mesocrystals are a new kind of superstructure material that follow a nonclassical crystallization process involving crystallographically ordered assemblies of nanocrystal building blocks. The relevant formation mechanisms of TiO2 mesocrystals reported thus far mainly include topotactic transformation, mineral bridges, nanoparticle alignment with organic matrices, physical ordering, space constraints, and self-similar growth [100]. Different methods may give rise to different structures and morphologies, but the as-prepared TiO2 mesocrystals are usually single-crystal-like structures with high porosity, surface area, and crystallinity; they are considered periodically hierarchical structures that are similar to sophisticated biominerals. All of these features pave the way for a wide range of applications, such as catalysis and energy storage and conversion [102].

Fabrication and modification strategies for TiO2 mesocrystals have developed rapidly in recent years. Due to the similar structure between NH4TiOF3 and TiO2, preparing TiO2 mesocrystals through topotactic transformation from NH4TiOF3 represents an innovative process. As illustrated in Figure 17, the critical parameters in the {001} facets of both NH4TiOF3 and TiO2 are quite similar, with an average lattice mismatch of 0.02%. The position of titanium atoms in the {001} plane of TiO2 is similar to NH4TiOF3, but in NH4TiOF3, these are separated by ammonium ions in a lamellar structure. Hence, it is reasonable to use NH4TiOF3 as a starting material, transforming it into TiO2 mesocrystals by thermal decomposition or aqueous hydrolysis with H3BO3 [71].

Based on this mechanism, Majima et al. performed extensive studies on tailoring TiO2 mesocrystals with versatile structures and morphologies, as well as postmodifications to further improve their photocatalytic efficiency. For example, to investigate the anisotropic electron flow in different facets and to maximize their separation during the photocatalytic reaction, Zhang et al. controllably synthesized a specific facet-dominated TiO2 superstructure with NH4F as an orientation-directing agent [103]. Under UV light irradiation, mesocrystals with different facet ratios showed different reactivity orders in the photooxidation of 4-chlorophenol, i.e., {001} > {101} (by 1.7 times), and photoreduction, i.e., {101} > {001} (by 2–3 times).

Moreover, constructing the composite of MoS2 and TiO2 mesocrystals, as well as the co-catalyst selective modification on TiO2, also showed the desired separation of photogenerated charge carriers during the hydrogen evolution reaction [104]. In terms of extending the absorption of incident light to the visible region, Zhang et al. tried doping or codoping non-mental elements into the TiO2 matrix to examine the effects on its electronic structure and band gap. An in situ fluorine-doped TiO2 superstructure was recently realized. F doping into TiO2 mesocrystals for the incorporation of active color centers facilitates visible light harvesting and accelerates charge separation for hydrogen generation [51]. They further introduced nitrogen and fluorine codopants into {001} facet-oriented TiO2 mesocrystals during topochemical transformation for photoreduction of Cr(VI) under visible light illumination. The extended optical light absorption could be attributed to doped nitrogen, which introduces the isolated mid-gap state. The high yield of hydroxyl radicals and preferential adsorption are correlated with fluorine doping, as confirmed by the comparison between untreated TiO2 with TiO2 washed in NaOH aqueous solution. The synergistic effect on charge separation and trapping was suggested through a femtosecond time-resolved diffused reflectance (TDR) measurement [105]. As shown in Figure 18, the g-C3N4 nanosheet/TiO2 mesocrystal metal-free composite was successfully constructed by Elbanna et al. [106]. The as-prepared sample exhibited an excellent hydrogen evolution rate under visible light irradiation without any noble metal co-catalyst. Then, they further broadened the light capture of the TiO2 mesocrystals to include near-infrared regions. Au nanorods (NRs) with various aspect ratios were loaded onto the surface of TiO2 by the ligand exchange method. Different aspect ratios resulted in different incident light absorption and photogenerated electron transfer. The highest photocatalytic activity of Au NRs and TMC composites reached 924 μmol h−<sup>1</sup> g−<sup>1</sup> under visible-near-infrared (NIR) light irradiation [107].

**Figure 17.** Illustration of the oriented transformation of NH4TiOF3 mesophyte to TiO2 (anatase) mesocrystal [71]. Copyright 2008 American Chemical Society.

**Figure 18.** Representative scheme of electron injection and movement in g-C3N4 NS (31 wt %)/TMC during visible-light irradiation [106]. Copyright 2017 American Chemical Society.

Considering the aforementioned merits of mesocrystal nanomaterials, we recently tried different approaches to further improve the optical absorption properties of TiO2 mesocrystals, in addition to their enhanced transfer and separation properties. Oxygen vacancies and N dopants were successfully introduced into the TiO2 lattice with a facile low temperature calcination process [108], as shown in Figure 19. NH4TiOF3 mesocrystal nanocubes were used as precursors in our system, and topological transformation from NH4TiOF3 to TiO2 facilitated the release and doping of nitrogen. Oxygen vacancies were also readily produced in the inert heating atmosphere. The significantly improved photodegradation and photoelectrochemical performance under visible light irradiation may be associated with the unique structure of mesocrystals as well as the introduction of foreign and intrinsic defects.

**Figure 19.** Schematic representation of the synthesis of TiOx nanosheets. X-ray powder diffraction (XRD) pattern of (**a**) NH4TiOF3 and (**b**) N/TiO2−x. SEM images of (**c**,**e**) NH4TiOF3 and (**d**,**f**) N/TiO2−<sup>x</sup> [108]. Copyright 2019 The Royal Society of Chemistry.

### **5. Separation of Charges**

Since metals and metal oxides have different working functions, resulting in the formation of a Schottky potential barrier, an effective modification method is to deposit precious metals (Ag, Au, or Pt) on the surface of metal oxides.

Choi et al. presented Ag/TiO2 by a photodeposition method [109]. Due to the different transfer rates of interface charges between electrons and holes to redox species in water, excessive charges can accumulate on photocatalysts [110,111]. By depositing Ag, which can provide a temporary home for excessive electrons, the composite utilized the electron storage capacity to promote the separation of electrons and holes to reduce Cr(VI) in the following dark period. Li et al. prepared a sandwich structure with CdS-Au-TiO2 on a fluorine-doped tin oxide (FTO) substrate [112]. In this composite structure, Au nanoparticles not only acted as an electronic relay between CdS quantum dots (QDs) and TiO2 to increase charge separation occurring on a long-time scale but also served as a plasma photosensitizer that prolonged the photoconversion to improve the absorption range of light. The rate of charge transfer and reverse transfer depends on the relative energy of the hot plasma electrons to the Schottky barrier [112]. The PEC performance is represented in Figure 20.

**Figure 20.** (**a**) Electron relay effect of Au nanoparticles, facilitating the charge transfer from CdS QDs to TiO2 nanorods under the irradiation of incident solar light with a wavelength <525 nm. (**b**) Plasmonic energy transfer from the excited Au nanoparticles to TiO2 through hot electron transfer under the irradiation of incident solar light with a wavelength >525 nm. CB = conduction band, VB = valence band, EF = Fermi energy level, and Φ<sup>b</sup> = Schottky barrier [112]. Copyright 2014 American Chemical Society.

Precious metal deposition can greatly improve the performance of catalysts, but the scarcity of precious metals dramatically limits this modification method and makes it difficult to achieve industrial-scale production. In this case, the search for an inexpensive and efficient doped composite has also attracted much attention. Carbon, abundant on earth, has good electrical conductivity, and its combination with TiO2 can result in excellent photocatalytic performance. Wang et al. demonstrated TiO2–carbon nanoparticles by the sol–gel method and then synthesized core–shell-structured TiO2 and amorphous carbon [113]. This unique morphology and structure result in the modified TiO2 sample exhibiting enhanced responsiveness and excellent photocatalytic activity. Due to the rapid charge transfer in the carbon shell, both the carrier separation efficiency and the photodegradation of pollutants in water is improved. The reduced TiO2 is also more efficient in the production of H2 due to its correct edge position.

### **6. Application of TiO2 Nanomaterials**

Over the past several years, semiconductors, especially titanium dioxide, have been widely used as photocatalysts. It is well known that there are three main steps associated with the photocatalysis process: (1) generation of electrons and holes after the absorption of photons; (2) separation and migration of the charge; and (3) transition of the charge and reaction between the carriers and the reagent. To date, TiO2 has been mainly applied in the areas of environmental conservation, new energy resources, and so on. In this section, we will focus on recent progress in these photocatalytic applications of TiO2.

#### *6.1. Applications in the Environment*

#### 6.1.1. Degradation of Aqueous Pollutants

Industrial development is often accompanied by pollution of the environment, especially water. Photocatalytic water treatment using heterogeneous semiconductors under visible light is considered an eco-friendly technology. Photocatalysis involves the generation of large numbers of electrons and holes on the surface of TiO2 after the absorption of photons; the photogenerated holes have considerable oxidizing capacity and can degrade almost all organic contaminants including carbon dioxide (CO2). However, due to its own deficiencies, such as a wide bandgap and fast recombination of electrons and holes, TiO2 cannot make full use of sunlight to remove the pollutants in water. Wang et al. reported hydrogenation by TiO2 nanosheets with exposed {001} facets maintained by the formation of Ti–H bonds [114]. By annealing the fine-sized pristine hydrothermal product under a high-pressure hydrogen atmosphere, the hydrogenation of F-modified anatase TiO2 nanosheets (with exposed high

percentages of {001} facets) was achieved. Under UV–Vis and visible light irradiation, this material decomposed methylene blue (MB) faster than P25 and pristine TiO2, as shown in Figure 21.

**Figure 21.** Photocatalytic decomposition of MB (**a**) and •OH generation measurement (**b**) of TiO2 and TiO2–H under UV–Vis light irradiation. Schematic illustration (**c**) of the hydrogenation effect on the structural change in TiO2 and TiO2–H [114]. Copyright 2012 The Royal Society of Chemistry.

Plodinec et al. applied black TiO2 nanotube arrays with Ag nanoparticles, which promoted hydrogenation for the degradation of salicylic acid [115]. The photocatalyst can degrade salicylic acid effectively, and its photocatalytic performance far exceeds that of TiO2 nanotubes and commercial TiO2 P25 (the reference material used for the modeling of photocatalytic processes). Ling et al. prepared TiO2 nanoparticles (with diameters of 10–23 nm) that exhibited photocatalytic activity [116]. The initial degradation rate of phenol by a TiO2 nanocatalyst was 6 times higher than that achieved with H2O2 alone, and the addition of H2O2 to TiO2 can increase the initial concentration of hydroxyl radicals and accelerate the degradation rate. Hao et al. developed a TiO2/WO3/GO nanocomposite (via a hydrothermal synthesis), which presented excellent optical absorbance and displayed excellent photocatalytic activity for the degradation of bisphenol A [117].

In addition to the oxidizing capacity, the photogenerated electrons on TiO2 have strong reducing capacity to remove pollutants, such as Cd(II), Hg(II), As(V), and Cr(VI), from water; these cations can be reduced into less toxic metallic or ion states. Dusadee et al. fabricated a titania-decorated reduced graphene oxide (TiO2·rGO) nanocomposite via a hydrothermal process [110]. Studies on reducing the toxic Cr6+ (hexavalent chromium) ion toxicity using the titanium dioxide x/rGO numerical control have found that photocatalytic reduction of toxic Cr6+ generally increases with the increase in x. In addition, since rGO accelerates electron transport, the combination of photoexcited electrons and holes decreases leads to an increased duration of photocatalytic activity [118]. TiO2 has facilitated many pollutant degradation processes such as the reduction of nitrate, the degradation of acid fuchsin, the decomposition of acetaldehyde, and the dechlorination of CCl4 [119–122]. Due to the continued proliferation of environment pollutants, TiO2 and other nanostructured materials should be vigorously developed in the future to improve the degradation of pollutants by photocatalysis.

### 6.1.2. Degradation of Air Pollutants

Just as industrial and technological developments can result in water pollution, so too can the atmosphere be adversely impacted by toxic pollutants that are emitted from chemical manufacturing plants, power plants, industrial facilities, transportation technologies, etc. Air pollution impacts the health of the global environment and the array of species that live within it, and new techniques are sought to reduce harmful airborne emissions. Highly efficient oxidation and reduction during photocatalysis are considered to be an effective method to degrade inorganic and organic air pollutants to improve air quality [123–125]. Similarly, TiO2 is considered the most promising photocatalyst. Kakeru et al. prepared TiO2 nanoparticles with palladium sub-nanoclusters (<1 nm) using the flame aerosol technique [126]. Under sunlight, these materials can remove NOx at approximately 3 to 7 times the rate of commercial TiO2 (P25, Evonik) (without Pd). Natércia et al. prepared new composite materials of TiO2 (P25) and N-doped carbon quantum dots (P25/NCQD) by a hydrothermal method, which was first used as the photooxidation catalyst of NO under the irradiation of ultraviolet and visible light [127]. The experiment showed that the conversion rate of the P25/NCQD composite material (27.0%) was more than twice that of P25 (10%) without modification, and the selectivity in visible light increased from 37.4% to 49.3%. The photocatalytic performance of the composite material in the UV region was also better than that of P25. Zeng et al. reported a H2 reduction strategy to produce H–TiO2 materials (with enhanced oxygen vacancy concentrations and distributions) that can promote formaldehyde decomposition in the dark [128]. Research of TiO2-based photocatalysts has also been conducted to facilitate removal of tetrachloroethylene [129], acetone [130], benzene [131], phenol [73], etc. from the atmosphere.

### *6.2. Applications in Energy*

#### 6.2.1. Photocatalytic Hydrogen Generation

With the extensive use of nonrenewable fossil fuels, mankind is facing an unprecedented energy crisis. The photogenerated electrons on TiO2 have strong reducing capacity, enabling hydrogen production from the photocatalytic splitting of water. Moreover, hydrogen combustion produces only water and no harmful emissions, and therefore its potential as a truly clean energy source has received considerable attention since it was discovered. Zou et al. reported a self-modified TiO2 material with paramagnetic oxygen vacancies [132]. For the synthesis of Vo-TiO2 (Vo: denotes a paramagnetic oxygen vacancy), they chose a porous amorphous TiO2 material as a precursor that possessed a high surface area of 543 m2 g<sup>−</sup>1. The precursor was calcined in the presence of imidazole and hydrochloric acid at an elevated temperature in air to obtain the Vo-TiO2 material [132]. The Vo-TiO2 sample (for H2 evolution from water) used methanol as a sacrificial reagent under visible light (≥400 nm) at room temperature, and the H2 production rate was approximately 115 μmol h−<sup>1</sup> g<sup>−</sup>1, which is substantially higher than that achieved with Vo-Ti3+-TiO2 (32 μmol h−<sup>1</sup> g−1). Zhou et al. introduced an ordered mesoporous black TiO2 material that utilized a thermally stable and high surface area mesoporous TiO2 as the hydrogenation precursor for treatment at 500 ◦C [133]. The samples possessed a relatively high surface area of 124 m<sup>2</sup> g−<sup>1</sup> and exhibited a photo response that extended from ultraviolet to visible light. As shown in Figure 22, the ordered mesoporous black TiO2 material exhibits a high solar-driven hydrogen production rate (136.2 μmol h−1), which is almost twice as high as that of pristine mesoporous TiO2 (76.6 μmol h−1). Zhong et al. constructed a covalently bonded oxidized graphitic C3N4/TiO2 heterostructure that markedly increased the visible light photocatalytic activity for H2 evolution by nearly a factor of approximately 6.1 compared to a simple physical mixture of TiO2 nanosheets and O-g-C3N4 [134].

### 6.2.2. Photocatalytic CO2 Reduction into Energy Fuels

In addition to reducing water to hydrogen, the photogenerated electrons on TiO2 are capable of generating valuable solar energy fuels, such as CH4, HCO2H, CH2O, CH3OH, and CO2, which are considered highly viable energy sources that can alleviate the problems associated with the production of greenhouse gases from the combustion of fossil fuels. Slamet et al. prepared Cu-doped TiO2 through an improved impregnation method for photocatalytic CO2 reduction [135]. Both the distribution of copper on the catalyst surface and the grain size of copper–titania catalysts (crystallite size of approximately 23 nm) were uniform, and it was determined that Cu doping can greatly enhance the photocatalytic performance of TiO2 with respect to CO2 reduction. Liu et al. found that

copper-loaded titania photocatalysts, prepared via a one-pot, sol–gel synthesis method, comprised highly dispersed copper and that CO2 photoreduction exhibited a strong volcano dependence on Cu loading, which reflected the transition from 2-dimensional CuOx nanostructures to 3-dimensional crystallites; optimum CH4 production was observed for 0.03 wt.% Cu/TiO2 [136].

**Figure 22.** Photocatalytic hydrogen evolution of ordered mesoporous black TiO2 (a) and pristine ordered mesoporous TiO2 materials (b). (**A**) Cycling tests of photocatalytic hydrogen generation under AM 1.5 and visible light irradiation. (**B**) The photocatalytic hydrogen evolution rates under single-wavelength light and the corresponding QE. The inset enlarges the QE of single-wavelength light at 420 and 520 nm [133]. Copyright 2014 American Chemical Society.

### 6.2.3. Solar Batteries

Since semiconductors absorb photons to produce photonic carriers and the photonic carriers move and separate at the same time, electric energy can be obtained through charge transport. TiO2 can also be applied to dye-sensitized solar cells, Li-ion batteries, Na-ion batteries, and supercapacitors. Liu et al. synthesized a spring-like Ti@TiO2 nanowire array wire that could be used as a photoanode in dye-sensitized solar cells; this configuration exhibited a conversion efficiency maintenance rate of more than 95.95% [137]. Another study reported the use of anatase TiO2 nanotubes on rutile TiO2 nanorod arrays as photoanodes in quantum dot-sensitized solar cells, which have a small thickness of 1 μm and an excellent solar energy conversion efficiency of approximately 1.04%; this is almost 2.7 times higher than the conversion efficiencies measured for solar cells using the original TiO2 nanorod array photoanodes, as shown in Figure 23 [138]. Chen et al. implemented a C@TiO2 nanocomposite as the anode material for lithium-ion batteries, which utilize the esterification of ethylene glycol with acetic acid in the presence of potassium chloride. Li-ion batteries utilizing the C@TiO2 nanocomposite anode exhibited excellent rate performance and specific capacity (237 mA h−<sup>1</sup> g−1), and a coulomb efficiency (CE) of approximately 100% after 100 cycles [139]. Su et al. synthesized anatase TiO2 via a template approach for use as the anode in Na-ion batteries; use of the template-synthesized TiO2 resulted in better battery performance in comparison to that achieved when amorphous and rutile TiO2 was used as the anode material. Compared to other crystalline phases of titanium dioxide, anatase titanium dioxide produced the highest capacity, 295 mA h−<sup>1</sup> g−1, in the second cycle, tested at a current density of 20 mA g−<sup>1</sup> [140]. Kim et al. developed a black-colored TiO2 nanotube array synthesized by electrochemical self-doping of an amorphous TiO2 nanotube array and N2 annealing; the material exhibited good stability, high capacitance, and electrocatalytic performance, and is an excellent material for supercapacitors and oxide anodes [141].

#### 6.2.4. Supercapacitors

Yang et al. developed a hybrid material, covalently coupled ultrafine H–TiO2 nanocrystals/ nitrogen-doped graphene, via the hydrothermal route [142]. Due to the strong interaction between H–TiO2 nanocrystals and NG plates, the high structural stability of the H–TiO2 nanocrystal aggregation is inhibited. At the same time, the NG matrix plays the role of electron conductor and mechanical skeleton, imparting good stability and electrochemical activity on most of the well-dispersed ultrafine

H-TiO2 nanocrystals [142]. The material exhibited a high reversible specific capacity of 385.2 F g−<sup>1</sup> at 1 A g−<sup>1</sup> and excellent cycling stability with 98.8% capacity retention. Parthiban et al. reported a blue titanium oxide (B-TiO2) nanostructure that was applied via a one-pot hydrothermal route and hydrothermal oxidation [143]. The B–TiO2 nanostructure indicated excellent cycling stability with approximately 90.2% capacitance retention after 10,000 charge–discharge cycles.

**Figure 23.** (**a**) Electron lifetime as a function of Voc for TiO2 NRA and H–TiO2 NRA electrodes with various reaction times. (**b**) Recombination resistance (Rrec) of the QDSCs made from TiO2 NRAs and H-TiO2 NRAs at various forward biases in the dark. (**c**) Transient photovoltage responses of CdS–TiO2 NRAs and CdS–H-TiO2 NRAs. The wavelength of the laser pulse was 532 nm. Inset: schematic setup of TPV measurements. (**d**) Schematic configuration for our device showing the interfacial charge transfer and recombination processes [138]. Copyright 2015 The Royal Society of Chemistry.

#### *6.3. Other Applications*

#### 6.3.1. Antibacterial and Wound Healing

It is generally believed that electron–hole pairs formed under light illumination, such as •O2<sup>−</sup> and •OH, not only destroy all chemical contaminants but also kill microorganisms. Liu et al. proposed a TiO2/Ag2O heterostructure (produced by a facile in situ precipitation route) to enhance antibacterial activities [144]. Yu et al. synthesized a TiO2/BTO/Au heterostructured nanorod arrays (exhibiting piezophototronic and plasmonic effects) by using a simple process that combined hydrothermal and PVD methods. This material can be used as an antibacterial coating for efficient light driven in vitro/in vivo sterilization and wound healing [145].

### 6.3.2. Drug Delivery Carriers

TiO2 has the advantages of nontoxicity, stability, biocompatibility, and natural abundance. The preparation of TiO2 with a high specific surface area can be advantageous in drug delivery carrier applications. Johan et al. controlled the kinetics of drug delivery from mesoporous titania thin films via surface energy and pore size control [146]. Different pore sizes ranging from 3.4 nm

to 7.2 nm were achieved by the use of different structural guiding templates and expansive agents. In addition, by attaching dimethyl silane to the pore wall, the surface energy of the pore wall could be altered. The results indicated that the pore size and surface energy had significant effects on the adsorption and release kinetics of alendronate [146]. Biki et al. designed silica-supported mesoporous titania nanoparticles (MTN) coated with hyaluronic acid to cure breast cancer by effectively delivering doxorubicin (DOX) to the cancer cells [147]. Guo et al. deposited (onto the surface of MTN) hyaluronic acid and cyclic pentapeptide (ADH-1), which target CD44-overexpressing tumor cells and selectively inhibit the function of N-cadherin, respectively, to overcome the drug resistance of tumors [148].

Recently, Nakayama et al. found that H2O2-treated TiO2 can enhance the ability to produce reactive oxygen species (ROS) in response to X-ray irradiation [149]. As shown in Figure 24, the atomic packing factor (APF) intensity indicated that hydroxyl radical production in the TiOx (H2O2-treated TiO2) nanoparticles increased in a radiation dose-dependent manner in comparison to that of the non-H2O2-treated TiO2 nanoparticles. This behavior allows H2O2-treated TiO2 nanoparticles to act as potential agents for enhancing the effects of radiation in the treatment of pancreatic cancer. Dai et al. designed and synthesized a novel nanodrug delivery system for the synergistic treatment of lung cancer [150]. They loaded DOX onto H2O2-treated TiO2 nanosheets. In this way, chemotherapy and radiotherapy were combined effectively for the synergistic therapy of cancers.

**Figure 24.** ROS production by the TiOxNPs, PAA-TiOxNPs, and TiO2 NPs under X-ray irradiation. (**A**) Atomic packing factor (APF) intensity indicating that hydroxyl radical production in the TiOxNPs and the PAA-TiOxNPs increased in a radiation dose-dependent manner, but that of the TiO2 NPs did not. Irradiated radiation doses were 0, 5, 10, and 30 Gy. Data are shown as the mean ± SD from 5 independent experiments. (**B**) Production and scavenging of ROS by 1 mM vitamin C (Vit. C) or 1 mM glutathione (GSH). Histograms show the mean ± SD calculated from 5 independent experiments. (**C**) Hydrogen peroxide production from the TiOxNPs under X-ray irradiation [149]. Copyright 2016 Springer Nature Switzerland AG.

#### **7. Conclusions**

As discussed in this review article, TiO2-based nanomaterials with wide band gaps have advantages associated with natural geologic abundance, nontoxicity and stability but they also exhibit inherent deficiencies and limitations related to ineffective visible light responses and other photocatalytic properties. The present review aimed to summarize key studies related to the marked enhancement of the photocatalytic performance of TiO2 by analyzing its electrical structure and photocatalytic reaction process. We have highlighted TiO2 photocatalysts with well-defined electrical and structure design, as well as tailored facets, dimensions, and remarkable morphologies, which are promising with respect to enhancing the photocatalytic properties of TiO2. All works presented in this review has enabled the authors to obtain an in-depth understanding of the TiO2 photocatalytic process, and the critical design of TiO2 nanostructures with enhanced light absorption, high surface area, desired photostability, and charge carrier dynamics. We hope that this review will guide the future development of more robust TiO2-based photocatalysts for large-scale applications.

Finally, photocatalysis technology is one of the most active research fields in the world in recent years. However, photocatalysis technologies based on TiO2 semiconductor still suffer from several key scientific and technological problems, such as low solar energy utilization rate, inferior quantum yield, and difficult recovery, which greatly restricts its wide application in industry. The fundamental solution to improve solar energy absorption is energy band engineering, designing and regulating the bandgap to optimize the harvesting of incident photons. Narrow bandgap and direct semiconductor are more likely to make use of low energy light, but they are restricted by very high electron and hole recombination rate and the incompatible band-edge position. High quantum yield is inevitable for an idea photocatalysis in practical solar engineering, but it cannot be achieved simply doping or inducing intrinsic defects. More works are needed to do to search high quantum yield. All of the above problems depend on the deepening of basic research. Although at present, photocatalysis technology is still a long way from large-scale production and application, its huge potential excellent performance provides a good way for our development. In the near future, with the breakthrough of these key issues, the practical application of nano-photocatalytic materials will certainly be realized to improve our environment, provide cleaner energy, and bring more convenience to our daily life.

**Author Contributions:** X.K. and S.L. collected references, prepared figures, and wrote the original draft of the manuscript, they contributed equally to this work; Z.D. and Y.H. collected references and analyzed the data; X.S. gave valuable advice; Z.T. acted as a project director and contributed to subsequent revisions. All authors agreed to the final version of the paper.

**Funding:** This research was funded by the National Natural Science Foundation of China grant number 21571028, 21601027, the Fundamental Research Funds for the Central Universities grant number DUT16TD19, DUT17LK33, DUT18LK28 and the Education Department of the Liaoning Province of China grant number LT2015007.

**Conflicts of Interest:** The authors declare no conflicts of interest.

### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
