Modification strategies of TiO2 for potential applications in photocatalysis: a critical review

ABSTRACT TiO2 has received tremendous attention owing to its potential applications in the field of photocatalysis for solar fuel production and environmental remediation. This review mainly describes various modification strategies and potential applications of TiO2 in efficient photocatalysis. In past few years, various strategies have been developed to improve the photocatalytic performance of TiO2, including noble metal deposition, elemental doping, inorganic acids modification, heterojunctions with other semiconductors, dye sensitization and metal ion implantation. The enhanced photocatalytic activities of TiO2-based material for CO2 conversion, water splitting and pollutants degradation are highlighted in this review. GRAPHICAL ABSTRACT


Introduction
Semiconductor photocatalytic technology has attracted tremendous attention for effective utilization of solar energy in water reduction to produce hydrogen, carbon dioxide conversion to chemical fuels and pollutant degradation for environmental remediation (1). In past few years, TiO 2 has been extensively studied as a potential photocatalyst for CO 2 conversion, water splitting and degradation of hazardous contaminants in both aqueous and gaseous system owing to its low cost, high stability, non-toxicity, biologically inertness, redox ability and environmental friendly merits (2)(3)(4)(5)(6). Naturally, TiO 2 exists in three polymorphs such as rutile, anatase and brookite (7). Among these, anatase phase has been proven to be the most efficient photocatalyst, owing to its high photoactivity, cheapness, stability and negative conduction band potential (8). It is widely investigated in photocatalysis and as electrode material (9). Due to the high stability and redox properties of anatase TiO 2 , it has been used in various fields, such as solar cells, water and air purification, cancer therapy and self-cleaning of antibacterial materials (10,11). The presence of high-density surface hydroxyl groups slow down the recombination of photogenerated charge carriers. The highly specific surface area and anatase crystallinity greatly influence the photocatalytic performance of TiO 2 . It is widely accepted that photocatalytic reactions mainly take place on the surface of photocatalysts and the adsorption of pollutants in advance is one of the most important steps in photocatalytic reactions. Thus, the large surface area is generally favorable to enhance photocatalytic activity during photocatalysis (12,13). However, the large intrinsic band gap (E g = 3.2 eV), fast recombination of photogenerated charges and limited utilization of solar-light (ca. 4%) greatly limits the practical applications of anatase TiO 2 . Further, at high temperature (500-600°C) treatment, anatase phase transforms into rutile phase and resulting in the decrease photocatalytic activities under normal conditions (14). To overcome the above-mentioned shortfalls, extensive research is underway to develop new routes to modify TiO 2 by doping metal and nonmetal elements, coupling other semiconductor materials and surface modification with inorganic acids (15)(16)(17). It has been investigated that doping elements such as S, C, N and B could greatly extend the absorption capacity of TiO 2 from UV to visible region. These elements could easily substitute the oxygen atoms from the lattice of TiO 2 and decrease its band gap by contributing their p-orbitals (18)(19)(20)(21). This alters the electronic properties of TiO 2 and results in the enhanced reaction rates for photocatalytic processes. Similarly, the heterojunctions with noble metals and other semiconductors also play a vital role in the enhanced photocatalytic activities of TiO 2 . Besides, modifications with inorganic acids are much meaningful to improve the photocatalytic activities of TiO 2 .
In this review, we have summarized various modification strategies of TiO 2 by doping metals and nonmetals, by constructing heterojunctions with noble metals and semiconductor oxides and modification with inorganic acids. Further, the photocatalytic mechanisms for water splitting, CO 2 reduction and pollutants degradation have been explained elaborately. Although a number of review articles give deep insight into the structure modification and catalytic performances of TiO 2 photocatalyst. This review attempted to provide detail knowledge on the structure modification, photocatalytic mechanisms and applications of TiO 2 in photocatalysis.

TiO 2 photocatalysis
Semiconductor oxide photocatalysts could absorb solarlight, when the energy of the incident photons matches the band gap energy (E g ) of semiconductors. Thus, under solar irradiation, the electrons are excited from the valence band (VB) to its conduction band (CB) and initiate reduction reactions on the catalyst surface. On the other hand, the holes left in the VB take part in oxidation reactions. When the electron-hole pairs are not used in photocatalytic reactions, they recombine very fast and release energy in the form of heat. The basic principles for redox reactions in semiconductor photocatalysts are shown in Figure 1(A). First, photogenerated electron-hole pairs are produced. Second, the electron-hole pairs are separated and diffused to the electrode surface. Third, redox reactions take place on the photocatalyst surfaces (22). Among the semiconductor oxide photocatalysts, TiO 2 an n-type semiconductor shows optical absorption in UV-region (λ < 400 nm) of solar spectrum. Its CB level (−0.3 eV) is more negative and suitable for the redox potential of H + /H 2 (0 eV), while its VB level (2.9 eV) is more positive than the standard redox potential of O 2 /H 2 O (1.23 eV) vs. normal hydrogen electrode (NHE) as depicted in Figure 1 (23). Similarly, TiO 2 is also a potential candidate for both CO 2 conversion to chemical fuels and pollutant oxidation to inorganic minerals. However, the wide band gap and fast recombination of photogenerated charge carrier's (electron-hole pairs) still greatly limits its practical applications. To overcome these shortfalls, various attempts have been made. Among them, doping metal and non-metal elements to extend its optical absorption and enhance charge carriers separation, fabricating heterojunctions with other semiconductors to improve charge transfer and separation, surface modification with inorganic acids, noble metal deposition, metal ion implantation, and dye sensitization are widely investigated (24).

Doping metal and non-metal elements
It is widely accepted that the lattice structure and surface composition greatly affect the functionalities and efficiency of nanomaterials. Many researchers found that introducing electronically active secondary species into the crystal lattice of nanomaterials greatly alter its optical absorption and suppress the charge carrier's recombination. In 2014, Hou et al. (25) reported the hydrothermal synthesis of N-doped TiO 2 nanotubes, which were immersed in deionized water and various concentration ammonia solution. They investigated that the molar ratio between deionized water and ammonia aqueous solution had a great effect on the morphology of TiO 2 nanotubes. They found that the samples immersed in ammonia solution have higher absorption than that immersed in water. Further, the band gap was effectively reduced from 3.2 to 2.84 eV, after doping nitrogen and facilitated the photogenerated carrier's separation. This resulted in the dramatically enhanced visible-light activity of N-doped TiO 2 nanotubes for methyl orange (MO) degradation. The schematic band gap structure of TiO 2 and N-doped TiO 2 is displayed in Figure 2(A). Recently, McManamon et al. (26) reported the sol-gel synthesis of S-doped TiO 2 nanoparticles. They investigated that S occupy some interstitial sites in the crystal lattice, which results in the generation of surface states between the VB and CB of TiO 2 . It is noticed that the incorporation of S into the crystal lattice dramatically reduced the band gap of TiO 2 from 3.2 to 1.7 eV and facilitated visible-light absorption as shown in Figure 2(B). It was found that after doping S, the visible-light activity of TiO 2 for malachite green degradation was greatly enhanced. Bakar et al. (27) reported the synthesis of S-doped TiO 2 photocatalysts by template-free oxidant peroxide route followed by crystallization through hydrothermal technique. It was investigated that doping S into the crystal lattice of TiO 2 substituted oxygen atoms by sulfur anions and resulted in the narrowed band gap as depicted in Figure 3(A). This greatly extended the optical absorption of TiO 2 from UV to visible region. It was found that the incorporation of electrons deficient S atoms into the crystal lattice of TiO 2 has effectively captured the photoinduced electrons and reduced charge carrier's recombination. Further, S-doping generated surface oxygen vacancies, which acted as trapping center for the photogenerated electrons and reduced the recombination of electron-hole pairs. The synergistic effect of efficiently reduced charge carrier's recombination and extended visible-light absorption resulted in the enhanced visible-light photocatalytic performance of S-doped TiO 2 for methyl orange (MO) degradation. Wu et al. (28) reported the synthesis of C-doped TiO 2 by a facile solvothermal method. It has been demonstrated that the incorporation of carbon atoms into the crystal lattice of TiO 2 substituted oxygen atoms and resulted in the generation of oxygen vacancies. The band gap was considerably reduced from 3.2 to 2.39 eV, and the optical absorption was greatly altered. This resulted in the enhanced visible-light photocatalytic activities of TiO 2 for nitric oxide (NO) and methyl orange (MO) oxidation, which is much superior to P25 and N-TiO 2 photocatalysts. The schematic band gap structure is shown in Figure 3(B).
Sood et al. (29) reported the synthesis of Fe-doped TiO 2 by ultrasonic-assisted hydrothermal method. It was investigated that the incorporation of Fe into the crystal lattice of TiO 2 favored the substitution of Ti, as both have approximately the same ionic size. From UV-visible absorption spectra, it was observed that Fe doping extended the absorption of TiO 2 from UV to visible region. The band gap was effectively reduced from 3.2 to 2.9 eV that resulted in the enhanced visible-light photocatalytic activity for paranitrophenol degradation. The schematic mechanism for pollutant degradation by Fe-doped TiO 2 photocatalyst is depicted in Figure 4    and Ti−O π* molecular orbitals while the CB top has Ti −O σ* molecular orbitals. The energy level of π* orbitals in Cu-doped TiO 2 shifted toward the VB maxima and increased the atomic number of transition metal. As a result, the energy levels of σ*(1) and σ*(2) were shifted up and downward and resulted in the narrow energy gap. The σ*(1) and σ*(2) molecular orbitals of Cu-TiO 2 were found to be located near to the water and peroxides oxidation potentials. Hence, they directly interact with the σ orbitals of the intermediate products thereby reducing the over potential of oxygen evolution reactions.
Tahir et al. (31) reported the synthesis of In-doped TiO 2 nanoparticles by sol-gel method. It was found that the absorption band edge of In-doped TiO 2 was slightly shifted toward lower wavelength direction and resulted in the enlarge band gap. Further, the incorporation of Indium into the crystal lattice of TiO 2 has significantly increased the surface active sites and also suppressed the photogenerated charge carrier's recombination. As a result, the photocatalytic activity for CO 2 reduction under UV-light was obviously increased. The main product during CO 2 reduction over TiO 2 photocatalyst was CO, while Indium-doping remarkably enhanced the CH 4 yield. The photocatalytic processes and charge carriers recombination and separation in TiO 2 and Indoped TiO 2 photocatalyst is depicted in Figure 5(A,B).

Modification with inorganic acids
It has been demonstrated that surface modification with inorganic acids can greatly alter the surface carried charge properties of TiO 2 photocatalyst in neutral water. The newly introduced surface groups could greatly promote the separation of photogenerated charge carriers, leading to the enhanced photocatalytic activities. Naturally, it is accepted that the surface modification with inorganic acids greatly promotes the adsorption of O 2 on the photocatalyst surface, which is much crucial for photocatalytic activities (32). He et al. (33) reported the hydrothermal synthesis of sulfuric acid modified TiO 2 nanosheets. They investigated that H 2 SO 4 modified TiO 2 with {0 0 1} and {1 0 1} facets exhibit superior photocatalytic activity for CO 2 conversion to CH 4 and H 2 O oxidation under visible-light irradiation. These enhanced activities were attributed to the promoted surface OH groups, the generated oxygen vacancies and improved visible-light absorption. Further, it was investigated that the surface protonation resulted from acid modification improved the photogenerated charge carrier's lifetime of TiO 2 nanosheets and also prevailed the limitation of unfavorable side length for photogenerated charge carrier's migration. Moreover, the photocatalyst showed high stability during the photocatalytic conversion of CO 2 . The mechanism of CO 2 reduction and water oxidation over bulk anatase TiO 2 and sulfuric acid modified TiO 2 nanosheets is depicted in Figure 6(A,B).   Cao et al. (34) reported the surface modification of TiO 2 nanoparticles by simple post-treatment with monometallic sodium orthophosphate solution. It was found that phosphate modification has greatly promoted the adsorption of O 2 on the surface of TiO 2 photocatalyst. The enhanced adsorption of O 2 was attributed to the substitution of -Ti-OH with -Ti-O-P-OH groups. As a result, the photogenerated electrons were effectively captured by the surface promoted O 2 , and the activities for photocatalytic degradation of pollutants were significantly improved. The surface forms of un-modified and phosphate-modified TiO 2 are depicted in Figure 7(A,B).
Liu et al. (35) reported the cobalt phosphate-modified TiO 2 films, which were prepared by the post-treatment of TiO 2 films with monometallic sodium orthophosphate solution followed by cobalt nitrate solution. They demonstrated that the photoelectrochemical water oxidation of TiO 2 photocatalyst was greatly enhanced after modification with cobalt phosphate. This enhanced activity was attributed to the role of Co (II) ions linked with the surface of TiO 2 through phosphate groups. It was confirmed that the surface linked cobalt (II) ions could effectively capture the photogenerated holes and generate high valence Co ions, which further induce oxidation reaction with H 2 O molecules and return to its original oxidation state. The schematic mechanism for the holes captured by Co (II) and the improved photoelectrochemical (PEC) water oxidation of TiO 2 after modification with Co (II) phosphate is depicted in Figure 8.
Luan et al. (36) reported the synthesis of hydrofluoric acid controlled 001-facet TiO 2 nanoparticles by hydrothermal method. They observed that the 001-facet TiO 2 exhibit exceptional photoactivities for pollutants degradation as compared to the unmodified TiO 2 . The enhanced photoactivities were attributed to the residual HF linked to the TiO 2 surface via coordination bond between Ti 4+ and Fas depicted in Figure 9. The resultant −Ti:F−H bond greatly enhanced the adsorption of O 2 on the surface of TiO 2 . Hence, the photogenerated electrons were effectively captured by the adsorbed O 2 and charge carrier's separation was improved.

Constructing heterojunctions
The utilization of solar energy in a cost-effective manner is a mean of meeting the global energy demand and environmental remediation. Hence, it is highly desirable to convert solar-light into an energy storage medium (37). The photocatalytic water splitting to produce H 2 and O 2 , CO 2 conversion to chemical fuels and pollutant degradation under direct solar irradiation are the main issues of scientific research nowadays. The direct water splitting and CO 2 reduction can be achieved using a single semiconductor. However, the proper redox potentials and effective absorption of visible-light are both the fundamentals of photocatalytic processes (38). In general, the semiconductor photocatalyst should have a wide band gap with proper VB and CB positions to   provide effective separation of photogenerated charges, transfer energetic electrons and high stability toward photocorrosion. Semiconductor TiO 2 fulfill all these requirements, but unfortunately the low quantum yield limits its practical application. This is because; TiO 2 could absorb only UV-light, which accounts for ca. 4% of solar spectrum (39). To overcome this shortfall, constructing heterojunctions with noble metals and other semiconductor oxides with wide and narrow band gaps are widely employed. Constructing heterojunctions could effectively improve the optical absorption, charge carrier's separation and stability of TiO 2 photocatalyst. Therefore, it is highly desirable to choose suitable photocatalysts for coupling, in order to improve the stability and efficiency of TiO 2 photocatalyst. Photocatalysts with large surface area exhibit more active sites, which is much crucial for photocatalytic reactions. Besides, the proper band gap and band edge alignment is also a key point because the relative energy level at interface junction also reveals the charge separation direction and transportation. Based on the above consideration, the heterojunctions are categorized into two classes; noblemetals-TiO 2 heterojunctions and semiconductor oxides-TiO 2 heterojunctions.

Noble-metals-TiO 2 heterojunctions
Recently, it has been investigated that incorporation of noble metals into the crystal lattice of wide band gap semiconductors such as TiO 2 have significantly improved the photocatalytic activity of TiO 2 in visible-light region due to surface plasmon resonance (SPR) effect of noble metals. It has been demonstrated that the surface plasmon resonance effect mainly arises from the collective oscillations of electrons on surfaces of noble metal, and these energetic electrons transfer into the conduction band of the coupled semiconductor due to the Schottky barrier. Schottky barrier between noble metals and semiconductor oxides arises via Fermi levels equilibrium, as a result built-in electric field is formed at the interface which favors separation of photogenerated charges as depicted in Figure 10   ( Figure 11(B)), showed that the Pt nanoparticles with average diameter of ∼15 nm are in close contact with TiO 2 nanoparticles. HRTEM image of Pt/TiO 2 nanocomposite is shown in Figure 11(C), while the dark-field TEM image of mesoporous Pt/TiO 2 nanocomposite and Pt/ TiO 2 nanocomposite is shown in Figure 11(D and E). The photocatalytic activity of both type nanocomposites was evaluated for degradation of dichloroacetic acid (DCA). The result showed that the photocatalytic activity of TiO 2 for DCA degradation was significantly enhanced by the addition of Pt islands and directly associated with the particle size of Pt. The activity of the nanocomposites prepared via photodeposition method was much higher than that of nanocomposites prepared by the insitu method, even higher than the commercially available TiO 2 (P25). This enhanced activity was attributed to the well-dispersed small-sized (3 nm) Pt nanoparticles on the surface of TiO 2 . They demonstrated that when Pt/ TiO 2 nanocomposites are irradiated with photons having energy higher than the band gap of TiO 2 , photogenerated electrons from the valance band are excited to the conduction band of TiO 2 leaving the holes in the valance band. The closely contact Pt nanoparticles act as electrons sink to promote the reduction of oxygen on their surfaces and the holes in the valance band migrate to the surface of photocatalysts for oxidation of organic pollutants as depicted in Figure 11(F).
Jun Fang et al. (43) reported the synthesis of Au-TiO 2 nanocomposites by a co-polymer assisted sol-gel method. From TEM micrograph (Figure 12(A)), it was confirmed that Au and TiO 2 nanoparticles are in close contact with each other. HRTEM image (Figure 12(B)), showed well resolved (101) direction lattice fringes with d-spacing (d = 0.35 nm) and (200) direction lattice fringes with d-spacing (d = 0.23 nm) for TiO 2 and Au, respectively. The activity of the as-prepared nanocomposites was evaluated for photocatalytic water splitting in the presence of electron donor (ascorbic acid), under visible light as shown in Figure 12(C). All the samples showed excellent photocatalytic activity for H 2 evolution, which were attributed to the surface Plasmon resonance effect of Au nanoparticles. To study the SPR effect of Au nanoparticles, they performed controlled wavelengths irradiation (500 ± 20 nm) experiments. It was confirmed that the surface plasmon excited hot electrons transfers from Au nanoparticles to the CB of TiO 2 ; as a result, the visible-light photocatalytic activity of TiO 2 for water reduction was significantly enhanced. The schematic mechanism for hydrogen evolution over Au-TiO 2 nanocomposite is depicted in Figure 12(D). Dong Yang et al. (44) reported the synthesis of Ag/ TiO 2 nanocomposites by two different methods. Firstly, they prepared Ag/TiO 2 heterojunctions by simple ultrasonic treatment and stirring of AgNO 3 and TiO 2 nanotubes (TNT) with the aid of 3-(3, 4 Dihydroxyphenyl) propionic acid (DiHPP), which acts both as a linker and reducer. Secondly, they followed photochemical reduction method for the synthesis of Ag/TiO 2 heterojunctions without adding DiHPP. In the prepared Ag/ TNT composite, about 3.8 nm in diameter of Ag nanocrystals were uniformly dispersed over the TiO 2 nanotubes (TNT) surface and formed the heterojunction structure with TiO 2 nanotubes (TNT). They demonstrated that the DiHPP first links to the TiO 2 nanotubes (TNT) surface through the bidentate chelation of catechol group with Ti 4+ and then acts as both an anchor and a

Semiconductors-TiO 2 heterojunctions
In semiconductors-TiO 2 heterojunctions, both the components can be excited by photons to generate charge carriers (electron-hole pairs). The charge transfer direction would mainly depend on the relative VB and CB position of both semiconductors. Primarily, the heterojunctions between TiO 2 and other semiconductors are categorized into three types as shown in Figure 14. In Type-I heterojunction, the two semiconductors should be either n-type or p-type. Suppose the two semiconductors are represented by A and B. The CB of B should be higher and its valance band should be lower than that of A. Therefore, both the electrons and holes would transfer to semiconductor A. In type-II heterojunctions, the photogenerated electrons would transfer from B to A, while the photogenerated holes would transfer from A to B. Such type of heterojunction is much beneficial, because of the effectively separation of electron-hole pairs. Type-III heterojunction is similar to type-II except for the more prominent difference in VB and CB positions, which requires a huge driving force for photogenerated charge transfer.

Heterojunctions with wide band gap semiconductors.
Since the first report on solar-driven photoelectrochemical (PEC) energy conversion by TiO 2 semiconductor photoelectrode in 1972, TiO 2 has been widely investigated and regarded as one of the most promising candidates for water splitting, owing to its low cost, high stability, strong optical absorption and suitable band edge positions (45). However, TiO 2 normally suffers from the intrinsic drawback of photocorrosion, which reduces its photoactivity and photostability. To solve this issue, research has been proliferated to improve the photostability and photocatalytic properties of nanosized TiO 2 by coupling with ZnO. It has been wellestablished to remarkably enhance the photogenerated charge carrier's separation efficiency because of the formation of heterojunction between them (46). Recently, some researchers (47)(48)(49) found that the photoelectrochemical performance of TiO 2 photoanodes could be obviously improved by constructing heterojunction with ZnO. Zha et al. (50) reported TiO 2 /ZnO nanocomposites prepared via solvothermal method. They demonstrated that the as-prepared nanocomposites exhibit exceptional photocatalytic activity (97% in 30 min) for methyl orange (MO) degradation under UV light; which is attributed to the hierarchical nanostructure with a large surface area. The SEM images (Figure 15(A,B)) and TEM image (Figure 15(C)) of the heterojunctions show multi-layers of fan blades like morphology with an average length of 2.5-3 mm. The HRTEM image ( Figure  15(D)) reveals that the lattice fringes with d-spacing 0.28 nm in (101) planes correspond to anatase TiO 2 , while the later at (100) planes are attributed to hexagonal wurtzite ZnO. The photocatalytic mechanism over TiO 2 /ZnO heterojunctions is shown in Figure 15(E). As the energy band gap of ZnO is 3.37 eV, hence, its conduction band (CB) and valence band (VB) potential is a little bit more negative than that of TiO 2 . When TiO 2 / ZnO nanocomposite is irradiated by photon energy to surpass its band gap energy, the photogenerated electrons from its VB are excited to the CB, which then transfer to the CB of TiO 2 due to the potential difference between them. Simultaneously, the holes from the VB of TiO 2 transfer thermodynamically to the VB of ZnO. This results in the enhanced photoactivities of TiO 2 .
Recent reports (51)(52)(53) show that SrTiO 3 (3.4 eV) could effectively improve the charge separation and photoelectrochemical performance of TiO 2 by shifting its Fermi level to more negative potential in the formed SrTiO 3 /TiO 2 heterojunction. Guo et al. (54) reported SrTiO 3 /TiO 2 nanosheet heterojunctions prepared via hydrothermal method. They demonstrated that the asprepared SrTiO 3 /TiO 2 heterojunction exhibit porous structure with a high surface area. It was found that the material exhibit enhanced charge carriers separation, which is significant for dye sensitized solar cells (DSSCs). The photoelectric conversion efficiency and short-circuit current density of DSSCs based on SrTiO 3 /TiO 2 heterojunctions were obviously enhanced as compared to the pure TiO 2 nanosheets. The short-circuit current density of 12.55 mA cm −2 and photoelectric conversion efficiency of 7.42% was observed for DSSCs based on the SrTiO 3 /TiO 2 heterojunctions under solar illumination. The photogenerated charge transfer and separation in SrTiO 3 /TiO 2 heterojunction under UV-light irradiation is shown in Figure 16.
It is widely accepted that TiO 2 exhibit fast recombination of photogenerated charge carrier's which leads to the low photocatalytic efficiency. In order to overcome this shortfall, recent studies (55-58) reveal that photogenerated charge separation and catalytic activities of TiO 2 can be obviously improved after coupling nanocrystalline SnO 2 . Wang et al. (59) reported hierarchical SnO 2 / TiO 2 heterojunctions. They demonstrated that under UV light irradiation, the nanocomposites exhibit enhanced photocatalytic activities for Rhodamine B (RhB) degradation as compared to the bare TiO 2 nanofibres. This enhanced photoactivity was attributed to the UV-light excited electrons transfer from the CB of TiO 2 to the CB of SnO 2 . To well understand the charge transfer mechanism and the enhanced photoactivities, they proposed a schematic mechanism as shown in Figure 17(A). The band gap and work function of anatase TiO 2 is 3.2 and 4.2 eV, respectively. Its electron affinity is approximately 4.2 eV. On the other hand, the band gap and work function of rutile SnO 2 are 3.5 and 4.4 eV, respectively. Its electron affinity is 0.5 eV higher than that of TiO 2 . The Fermi energy level of TiO 2 is higher than that of SnO 2 , hence the photogenerated electrons of TiO 2 easily transfer to SnO 2 and the photogenerated holes of SnO 2 transfer to the VB of TiO 2 . Therefore, the efficient charge separation will increase the lifetime of charge carriers and then enhance photocatalytic activity. On the other hand, in a single component system, most of the charges quickly recombine without doing any photochemical process and only a few charges take place in photocatalytic reactions. This results in the low photocatalytic activity. As can be seen from Figure 17(B), the PL emission spectrum of TiO 2 (TS0) is much intense compared to that of SnO 2 /TiO 2 nanocomposite (TS1). This indicates that the charge recombination is greatly inhibited in the SnO 2 /TiO 2 nanocomposite. This resulted in the enhanced photoactivity for RhB degradation.

Heterojunctions with narrow band gap semiconductors.
Hematite (a-Fe 2 O 3 ) a narrow band gap (∼2.1 eV) semiconductor oxide has attracted tremendous attention owing to its potential applications as a low-cost, high stability, natural abundance, non-toxicity, visible-   light responsive and electrode material for photoelectrochemical (PEC) water splitting (60,61). The theoretical solar-to-chemical energy conversion efficiency for Fe 2 O 3 is reported to be 16% (62). Coupling a-Fe 2 O 3 to semiconductor TiO 2 has seldom been reported, however, recently Li et al. (63) reported the branch-shaped a-Fe 2 O 3 /TiO 2 nanocomposites, which were prepared by a combine electrospinning and hydrothermal approach. The nanocomposites consist of 3D porous TiO 2 nano-fibers (diameter ∼ 70 nm) and the coupled a-Fe 2 O 3 nanorods (length ∼ 100-200 nm). The nanocomposites exhibit large surface area (∼ 42.8 m 2 g −1 ). They investigated that the resultant nanocomposites exhibit excellent visible-light photocatalytic activities for degradation of organic dyes, methylene blue (MB), including Congo red (CR), methyl orange (MO) and eosin red (ER). These enhanced activities were attributed to the large surface area, extended visible-light absorption and improved photogenerated charge separation. They proposed a charge transfer and separation mechanism, which is shown in Figure 18. They demonstrated that under visible-light (400 cutoff) irradiation, the excited highlevel energy electrons from the valance band (VB) of Fe 2 O 3 would transfer thermodynamically to the conduction band (CB) of TiO 2 , and react with molecular oxygen (O 2 ) to produce the superoxide anion (O ·− 2 ) and hydrogen peroxide (H 2 O 2 ). While the photogenerated holes in the VB of Fe 2 O 3 will react with hydroxyl species on the catalyst surface and produce the highly reactive hydroxyl radicals ( . OH), which are responsible for the oxidation of organic dye pollutants.
Mu et al. (64) reported the synthesis of In 2 O 3 -TiO 2 nanocomposites via electrospinning and solvothermal techniques. They investigated that the In 2 O 3 nanoparticles with an average particle size of about 30 nm were successfully grown onto the surface of TiO 2 as confirmed by means of TEM analysis (Figure 19(A)). Further, it was observed that TiO 2 nanoparticles exhibit an average particle size of about 130 nm. Moreover, the In 2 O 3 -TiO 2 heterojunction was confirmed by HRTEM image (Figure 19 (B)). The lattice fringes at (101) plane with d-spacing 0.31 nm were attributed to anatase TiO 2 , while the fringes at (222) plane with d-spacing 0.29 nm were attributed to In 2 O 3 . From selected area electron diffraction (SAED) patterns (Figure 19(C)), it was confirmed that the nanoparticles exhibit polycrystalline structure. The visible-light photocatalytic activity of the fabricated nanocomposites was evaluated for RhB degradation. It was investigated that the nanocomposites exhibit superior photocatalytic activity for RhB degradation as compared to the TiO 2 and In 2 O 3 alone. From photocatalytic stability recycle test, it was confirmed that the photocatalyst exhibit high stability. The photocatalytic mechanism for In 2 O 3 -TiO 2 nanocomposites was designed as shown in Figure 19   CdS is a visible-light responsive n-type semiconductor, which has been widely investigated among narrow band gap semiconductors. It has been regarded as an attractive material owing to its high efficiency for water reduction to evolve hydrogen. The narrow band gap (∼2.4 eV) with sufficient negative conduction band position for the reduction potential of protons made this material more valuable for practical applications (65). However, the poor stability, particle agglomeration and high recombination rates of photogenerated charges still greatly impair its practical use for large-scale water splitting (66,67). Therefore, the fabrication of CdS-TiO 2 nanocomposites has been attempted to diminish recombination, improve its surface area and to stabilize the material. Liu et al. (68) reported One-dimensional (1D) CdS@TiO 2 core −shell nanocomposites (CSNs) prepared via two-step solvothermal method. The visible-light photocatalytic activities of the as-prepared 1D CdS@TiO 2 CSNs nanocomposites were evaluated for selective oxidation of alcohols to aldehydes. It was investigated that the nanocomposites exhibit obviously enhanced visiblelight activity for both conversion and yield as compared to bare CdS NWs, which is attributed to the prolonged lifetime of photogenerated charge carriers after coupling TiO 2 . Moreover, it is noticed that the photogenerated holes from CdS core can be stuck by the TiO 2 shell, as confirmed by means of controlled radical trapping experiments and the selective reduction of heavy-metal ions, Cr(VI), over 1D CdS@TiO 2 CSNs. Hence, it is concluded that the reaction mechanism of visible-light catalytic oxidation of alcohols over 1D CdS@TiO 2 CSNs is actually different from that over 1D CdS NWs. It is demonstrated that under visible-light irradiation (λ = 520 ± 15 nm), the excited electrons from the valence band (VB) of CdS NWs in the 1D CdS@TiO 2 CSNs would transfer thermodynamically to the TiO 2 shell. Simultaneously, the holes in the valance band of CdS NWs would stuck by the TiO 2 shell. As a result, the lifetime of charge carriers will prolong. The photogenerated electrons react with molecular oxygen and produce superoxide radicals (O ·− 2 ), which oxidize the alcohols to the corresponding aldehydes ( Figure 20).

Dye-sensitized TiO 2
Sensitization of TiO 2 with organic dyes is a promising and extremely effective technique used for renewable solar H 2 production, electricity generation, and various organic pollutants oxidation. In this technique, the dye molecules anchored on TiO 2 are excited with solar photons to generate highly excited electrons which then transfer to the conduction band of TiO 2 . Subsequently, the electrons in the conduction band of TiO 2 are transported to the interfacial electron acceptors (e.g. H 2 O, I 3 − , H + , O 2 , etc.) and induce various redox reactions as depicted in Figure 21. In this way, the effective solar energy utilization and interfacial charge transfer from the excited dye to the conduction band of TiO 2 should be the key parameters to accomplish a high level of photo-conversion efficiency (69).
Choi et al. (69) reported the synthesis of organic dyesensitized TiO 2 particles for H 2 evolution from H 2 O and 4chlorophenol and Cr(VI)) pollutants oxidation under visible-light (λ > 420 nm). They investigated the effect of various anchoring groups. For this, they prepared Ru-free organic dyes of donor-acceptor configuration with different numbers (n = 1, 2, and 3) of carboxylate-  anchoring groups which were referred as D1, D2 and D3, respectively. They found that all the three kinds of dyes are effective for H 2 production in the presence of electron donors such as tri-ethanol-amine (TEOA) and ethylene-diamine-tetra-acetic acid (EDTA) with the orders: D3-D2 > D1 (TEOA) and D3 > D2-D1 (EDTA). From FTIR analysis, it was confirmed that D1 and D3 are probably anchored on the surface of TiO 2 mainly by means of bidentate modes with a single and double carboxylates, respectively. Further, they investigated that D2 is anchored with single and double carboxylates, depending on the competing electron donors. They demonstrated that the carboxylates number is less important in the sensitized pollutant oxidation because of the different photochemical conditions and reaction pathways. The tri-branched organic dye molecules with mono-, di-, and tri-carboxylate anchoring groups (sensitizers D1, D2 and D3, respectively are depicted in Figure 22. They investigated that the extended visible-light absorption and the enhanced interfacial charge transfer from excited dye molecules to the conduction band of TiO 2 are much significant in DN-F05-TiO 2 . They found that the DN-F05 dye-sensitized TiO 2 -Pt photocatalyst display superior turnover number (1864) and noticeable quantum efficiency (AQE ∼44%) compared to their analogous simple architecture dye molecules. Further, they noticed that the additional donating-groups such as dibutoxyphenyl and dihexylcyclopentyl p-conjugated-bridge units provide exceptional surface protection through steric hindrance. This leads to the high photocatalytic activity. The mechanism of H 2 evolution over the dye-sensitized TiO 2 -Pt photocatalyst is shown in Figure 23. The HOMO and LUMO positions of the dye attribute to the formal redox potentials and the formal excited state oxidation potential, respectively. In H 2 evolution process, first, the dye molecules are excited from the ground state to the excited state by the absorption of incidentphoton flux, due to the intra-molecular p-p* transition. Second, the excited electrons immediately transfer to the conduction band of TiO 2 before quenching by emission and subsequent formation of the oxidized state dye. Third, the transferred electrons to the conduction band of TiO 2 are transported toward the co-catalyst Pt site. Fourth, these electrons reduce H 2 O molecules at Pt site. Finally, the oxidized dyes reduce to its ground state by accepting electrons from the oxidized sacrificial electron donors. Hence, the reduction of the oxidized dye and the regenerated by sacrificial electron donors resolve the efficiency of H 2 evolution. Lee et al. (71) reported the fabrication of phenothiazine-based organic dyes/TiO 2 -Pt composites with systematic alkyl-chains and two-anchoring groups. They demonstrated that the composite showed excellent visible-light photoactivities for H 2 O splitting. Further, they investigated that the phenothiazine dyes with longer alkyl-chains exhibit high stability during photocatalytic processes. Their stability was attributed to the cationic radical produced by one electron oxidation. They  demonstrated that the two-anchoring groups can be introduced at position 3 and 7 of the phenothiazine dyes. However, the multi-anchoring groups can greatly improve the photostability of the photocatalysts during photocatalytic processes. They demonstrated that the phenothiazine-based dyes with a directly connected anchoring group exhibit enhance visible-light harvesting efficiencies compared to the common donor-acceptor sort of dyes with additional p-conjugated bridges. The alkyl groups on nitrogen can induce orientation of dyes on TiO 2 , which may possibly result in the efficient electron transfer from the excited dye molecules to the conduction band of TiO 2 . This results in the improved photoactivities of the composites for H 2 O splitting. The schematic of phenothiazine-based dyes (P1-P5) with two anchoring groups is depicted in Figure 24.
Bae et al. (72) reported the synthesis of a variety of Rubipyridyl with di-, tetra-, and hexa-carboxylate functionalities (C2, C4 and C6) and with di-, tetra-, and hexa-phosphonate anchoring groups (P2, P4 and P6) sensitized TiO 2 samples for visible-light catalytic H 2 evolution in aqueous suspension containing an electron donating agent (EDTA). They investigated that the efficiencies of C and P sensitizers depend on the number and type of anchoring groups. The adsorption of P sensitizers on TiO 2 surface was strong enough not to be hindered by EDTA, whereas the adsorption of C sensitizers was significantly inhibited. Thus, the P-TiO 2 exhibited exceptional visible-light activity for H 2 evolution compared to that of the C-TiO 2 . Among the six different kinds of sensitizers, P2 was the most active for H 2 evolution.
When the Ru-bipyridyl-TiO 2 samples were irradiated under visible-light, the excited electrons of dye molecules were transferred to the conduction band (CB) of TiO 2 which then initiated interfacial chemical reaction. The schematic of visible-light induced H 2 evolution over dye-sensitized TiO 2 is depicted in Figure 25.

Metal ion-implanted TiO 2
The metal ion implantation is another strategy used to modify the electronic structure of semiconductors. In this technique, the metal ions are accelerated in the electronic field and injected to the target (sample) as the ion beam. These metal ions interact with the sample surfaces in different ways depending on their kinetic energy. Further, the metal ions are highly accelerated to have enough kinetic energy (50-200 keV) and then implanted into the semiconductors bulk (73). Yamashita et al. (73) reported the synthesis of (V + , Mn + , Fe + ) metal ionimplanted TiO 2 photocatalysts. They investigated that at high energy acceleration and subsequent annealing of these metal ion-implanted TiO 2 photocatalysts in O 2 at 725 K, the absorption spectra was remarkably shifted toward visible-light region. The resultant photocatalysts showed exceptional photocatalytic activity for 2-propanol (diluted in H 2 O) degradation under visible-light irradiation. They confirmed by means of XAFS analysis that the implanted metal ions are sited at Ti 4+ lattice positions in TiO 2 after annealing. They demonstrated that the substitution of Ti ions in TiO 2 matrix with the implanted metal ions is much important to modify TiO 2 for visible-light photocatalytic activities. Takeuchi et al. (74) reported the synthesis of transition metal ions (Cr and V) implanted TiO 2 thin film photocatalysts by an advanced metal ion-implantation technique. They observed that the absorption spectra of the resultant thin film photocatalysts were remarkably shifted toward visible-light region. As a result, the photocatalysts  showed significant visible-light catalytic decomposition of NO x into N 2 and O 2 environment at 275 K. Zhou et al. (75) reported the fabrication of V + ion-implanted P25 TiO 2 photocatalysts by an ion implantation method. They investigated that the materials exhibited remarkably enhanced visible-light catalytic activity for the degradation of formic acid. From HRTEM micrographs, they have confirmed that the V ions exist in the form of VO 2 (T) in the P25 TiO 2 matrix. Based on the photoluminescence spectra, they suggested that the enhanced visiblelight activity of the V + ion-implanted P25 TiO 2 photocatalyst for formic acid degradation is also attributed to improved charge separation. Lam et al. (76) reported the synthesis of chromium (Cr) ion-implanted TiO 2 thin films. They investigated that the visible-light catalytic activity of the resultant Cr/TiO 2 photocatalyst for degradation of gaseous formaldehyde was considerably enhanced. Wang et al. (77) reported the fabrication of Sn ion-implanted TiO 2 thin films by an ion implantation technique. They used different doses of Sn ion implantation for surface modification of TiO 2 . They demonstrated that the Sn 4+ ions substitute Ti lattice site in TiO 2 matrix. They investigated that TiO 2 with an optimum dose of implanted Sn ions exhibited enhanced photoactivity for Rhodamine B (RhB) degradation. The enhanced photoactivity was attributed to the extended light absorption from ultraviolet to visible region and enhanced charge separation as a result of Sn 4+ introduced electron and holes trapping states near the conduction band bottom and valance band top of TiO 2 .

Conclusions
In this review, we have described various modification strategies of TiO 2 and their applications in photocatalysis. We have discussed basic concepts related to surface modification including metal nanoparticles deposition, inorganic acids modification, elemental doping, heterojunctions with other semiconductors, dye sensitization and metal ion implantation. Further, this review emphasizes the basic mechanisms for photophysical and photochemical processes and the physical concepts suitable for material properties interpretation and structural property correlations. This review would help us to synthesize low-cost TiO 2 -based material and also help us to deeply understand the photophysical and photochemical processes involved during the photocatalytic CO 2 conversion, water splitting, and pollutants degradation over TiO 2 -based materials.

Disclosure statement
No potential conflict of interest was reported by the authors.