After the first report of ZnO powder as the photosensitizer in 1911
[3], Baur and Perret reported the photoinduced oxidation of water to produce O
2 with ZnO powder in 1924
[4]. In this report, a new concept has been suggested, in which, after the absorption of light energy, the ZnO particle undergoes simultaneous anodic and cathodic processes. In the anodic process, OH
− is oxidized to generate ¼O
2 + ½H
2O. On the other hand, in the cathodic process, the Ag
+ ion is reduced to generate Ag by the formation of intermediate peroxide Ag
2O. However, this photoinduced reaction is associated with the dissolution of ZnO to release Zn
2+, and then the verified reaction scheme was
Therefore, though this reaction takes place only under photoirradiation, it was not a photocatalytic reaction in researcher's time.
Compared with ZnO, TiO
2 is a relatively new material because titanium metal was identified at the end of the 18th century. It was in 1923 that TiO
2 was first manufactured as a white pigment to replace basic lead carbonate which was used in the porcelain industries. Photo-induced effects on the chemical reactions over TiO
2 powder have been investigated before and after World War II in the vicinity of the 1940s. However, as a photosensitizer, in other words, for photocatalytic reaction, the main substance used was ZnO since the photoeffect was first reported as described above. In 1953, Markham and Laidler
[6] reported the reaction mechanism of photo-oxidation on the surface of ZnO, where they suggested that O
2 is formed from two •OH radicals. In the discussion, they cited the report of an XRD study
[7], which described that photo-excitation of TiO
2 caused the change in the XRD pattern to α-Ti
2O
3 crystal. Then, Markham and Laidler described that the absorption of photon energy to TiO
2 may result in the photolysis of the TiO
2 crystal, as represented by
[6].
At that time, this chemical equation had not been confirmed by any researchers, and TiO2 became inexperienced material in the research of solid photosensitizers.
They suggested that when a suitable
p-type semiconductor electrode is coupled with the
n-type TiO
2 semiconductor electrode, efficient electrochemical photolysis of water may occur on the irradiation of both electrodes
[9]. Their report was epoch-making because it was shown that the photon energy could convert to the chemical energy of hydrogen fuel by using a semiconductor electrode. Actually, Yoneyama and co-workers demonstrated the O
2 evolution at the TiO
2 anode and the H
2 evolution at the
p-GaP cathode with the open circuit voltage of 0.58 V
[10]. Though the deterioration in the cell performance was observed due to the unstable
p-GaP cathode, oxidation of water at the TiO
2 anode was confirmed stable. After these observations, the photoelectrochemical systems with metal oxides for water splitting have been investigated widely, which are reviewed by Rajeshwar
[11]. Recent development in the engineering nanostructure interface of photoanode materials toward photoelectrochemical water oxidation can be referred to with a review article
[12].
For the particulate photocatalysts, the decomposition of water vapor to H
2 and O
2 over Fe-doped TiO
2 powder has been reported by Schrauzer and Guth
[13], in which a stoichiometric (2:1) evolution of H
2 and O
2 was confirmed. Photocatalytic water vapor decomposition was enhanced when Pt nanoparticles were photodeposited on a TiO
2 particle, which constructs a small particulate electrochemical cell. Sato and White
[14] reported the generation of O
2 and H
2 with the stoichiometric ratio of 1:2, as shown in
Figure 1. When the irradiation was stopped, the recombination of O
2 and H
2, or reverse reaction of H
2O photodecomposition, occurred at a significant rate. Photocatalytic decomposition of water vapor was recently reviewed by Suguro et al.
[15].
Different from the gaseous system, in the aqueous suspension system, it is difficult to inhibit the reverse reaction of the produced H
2 and O
2. To inhibit the reaction between H
2 and O
2, the surface of the co-catalyst for H
2 production had to be covered with a Cr
2O
3 shell to protect it from the attack of the produced O
2 [18]. From the point of view of a catalyst, TiO
2 is not the best material for OER. The other metal oxides which show a small exchange current for OER have been explored as reviewed by Lewis and coworkers
[19]. Moreover, these materials have been employed as the co-catalyst of semiconductor photocatalysts. For water oxidation, many co-catalysts, such as cobalt oxide, have been used in the newly developed photocatalysts
[20]. The development of photocatalysts for water splitting of the use of such co-catalysts was compiled recently
[21].
3. Mechanism for TiO2 Water Oxidation
3.1. Experimental Approach
3.1.1. ESR and NMR
In the early days of photocatalysis, a spin-trapping ESR technique was used to detect radical species in the reaction. With the UV irradiation on the suspension of Pt-deposited TiO
2 powder, •OH and •O
2H radicals were detected
[22]. Thus, in the O
2 generation by the water oxidation, •OH radical was considered as the reaction intermediate as follows.
The •O
2H radical detected was attributed to being produced by the photocatalytic reduction of the produced O
2 [22].
In the recent study, one- and two-dimensional
1H solid-state NMR techniques were employed to identify the surface hydroxyl groups and the adsorbed water molecules as well as their spatial proximity/interaction in TiO
2 photocatalysts
[23]. Only the bridging OH (i.e., OH
br) is in close spatial proximity to adsorbed H
2O, forming hydrated OH
br. To investigate the role of hydrated OH
br in the hole transfer process, in situ ESR experiments were performed on TiO
2 with variable H
2O loading
[23]. The ESR measurements revealed that the hydrated OH
br groups offer a channel for the transfer of photogenerated holes in the photocatalytic reaction, and the adsorbed H
2O could have a synergistic effect with the neighboring OH
br groups to facilitate the formation and evolution of active paramagnetic intermediates. On the basis of these experimental observations, the detailed photocatalytic mechanism of water splitting on the surface of TiO
2 was proposed, as shown in
Figure 2 [23]. The surface-trapped hole Ti-O• and the adsorbed •O
2− radical were identified by the ESR measurements. In this figure, as the product of the second step, they suggested the side-on coordination of O
22− ions to one Ti ion. However, this structure is probably unreal conformation in an aqueous solution.
Figure 2. OER mechanism proposed from the ESR detection of radicals produced on Pt/TiO
2 (P25) powder. Adapted with permission from
[23]. Copyright 2017, American Chemical Society.
3.1.2. STM
By using the low-temperature STM performed at 80 K, the dissociation of individually adsorbed water molecules has been observed at the five-fold coordinated Ti (Ti
5c) sites of the reduced rutile TiO
2(110)-1 × 1 surface under the irradiation of UV lights with the wavelength shorter than 400 nm
[24]. It was found that two kinds of hydroxyl species are involved in the process of photocatalytic water dissociation. One is always present at the adjacent bridging oxygen sites, OH
br, and the other either occurs as terminal OH
t at Ti
5c sites away from the original ones or even desorbs from the surface
[24]. This observation is in harmony with the suggestion of the solid-state NMR described above. Thus, the initial oxidation step could be expressed by the following equation
[25].
Using a combination of STM and other surface science techniques, the recent progress that provides fundamental insights into TiO
2 photocatalysis were reviewed through direct tracking of the evolution in single molecule photochemistry on TiO
2 single crystal surfaces
[25].
3.1.3. FTIR
Nakato and co-workers investigated the molecular mechanism of water photo-oxidation reaction at atomically flat
n-TiO
2 (rutile) single crystal photoelectrodes of (100) and (110) facet surfaces
[26]. They measured the photoluminescence (PL) of the photoelectrodes in aqueous solutions of various pH values. Combining with the Fourier transfer infrared (FTIR) observations of the intermediates in the reduction of molecular oxygen as well as the oxidation of water with an electron acceptor of Fe
3+ ions, the oxidation process was elucidated, as shown in
Figure 3. Though the nucleophilic attack of water to Ti−O
+−Ti is associated with the [Ti−O• HO−Ti] formation in
Figure 3 [27], an alternative model was proposed in which the bridging oxygen radicals [Ti−OH• Ti] are photogenerated with the intrinsic band-gap surface state
[28]. In both reaction models, the surface-trapped holes become bridging peroxo (Ti-OO-Ti) by combining with the hole generated secondarily in the same crystallite
[29][30]. The bridging peroxo structure was regarded as the intermediate step of water photo-oxidation. Thus, the species generated by two-hole oxidation is equivalent to a chemically adsorbed H
2O
2.
Figure 3. The proposed process of water oxidation and the peak position of the IR–spectra for the intermediates. Adapted with permission from Ref.
[26]. Copyright 2007 American Chemical Society.
When H
218O was used in the FTIR measurements, the observed isotope shift of the peak was small. Then, they suggested that one of the O atoms of the Ti-OOH group originated from O of the surface bridging O of the crystal
[27]. Salvador
[28] reported that the photoinduced holes are trapped at the bridging O of the surface, as stated above. However, in his proposed mechanism, a pair of •OH radicals are produced near the trapped hole, and they form the Ti-OO-Ti surface structure
[28].
As shown in
Figure 3, Nakamura et al. assigned the 838 cm
−1 peak in the FTIR spectra to the OO vibration of the Ti-OOH group and the peak at 812 cm
−1 to the Ti-OO-Ti group. Researchers also measured the FTIR spectra of the adsorbed H
2O
2 on rutile TiO
2 powders and the reaction intermediates under the UV irradiation
[31]. However, no absorption peaks at around 800 cm
−1 could be observed, but the peaks around 1000 cm
−1 and 934 cm
−1, which could be assigned to Ti-OOH and Ti-OO-Ti, respectively, were observed
[31]. This assignment is supported by the result of the DFT calculation described later, in which the O-O stretching mode of Ti-OOH was calculated to be 1002 cm
−1 with a large intensity. Since researchers did not use an electron scavenger in the FTIR experiments, O
2 in air was reduced, and the signals of superoxo Ti-OO
− were increased. On the other hand, Nakamura et al. used Fe
3+ ions as the sacrificial electron scavenger. Since the FTIR signal of Fe(IV)=O appears at 830–840 cm
−1 as reported in the literature
[32], the signal at 838 cm
−1 probably originates from an Fe ion compound and should not be attributable to Ti-OOH. Due to the uncertainty of the assignments of FTIR spectra, it is not clearly concluded that the bridging O atom is involved in the Ti-OO-Ti structure.
3.1.4. Experimentally Suggested Mechanism for Rutile and Anatase
Researchers have experimentally investigated the reactive oxygen species, such as •OH, H
2O
2, and •O
2−, which are generated from H
2O by TiO
2 photocatalysis and photoelectrodes
[33]. Nakabayashi and Nosaka examined the facet dependence of water oxidation at anodically polarized TiO
2 single-crystal photoelectrodes
[34]. Though the Faraday efficiencies of the oxygen evolution were almost 100%, the intrinsic photocurrent was increased in the order of (100) < (110) < (001). On the other hand, the formation of •OH radicals simultaneously measured was reverse order; that is, the Faraday efficiencies were 0.59, 0.23, and 0.13%. Only for the crystal of (100) facet does the photocurrent decrease with the irradiation time, but it could be recovered by depolarization of the electrode. This observation showed that the surface structure changed to inhibit the O
2 production at the surface of (100) single crystals. Furthermore, by the addition of H
2O
2, the formation of •OH radical was increased for the (100) and (110) crystals
[35], indicating that H
2O
2 was not formed from •OH radical but H
2O
2 produced •OH radical at the TiO
2 photo-anodes. Namely, the Ti-OO-Ti structure promotes the •OH radical generation. Thus, •OH radical could be formed in the process of the oxidation of Ti-OO-Ti species, but •OH radical is not the precursor of the Ti-OO-Ti formation. This conclusion contradicts the reaction mechanism suggested by Salvador
[28] in which a pair of •OH radicals form the Ti-OO-Ti structure.
When the rutile surface is the ideal crystal structure without reorganization, the Ti-OO-Ti structure of each crystal surface can be illustrated as in
Figure 4. The difference in the O
2 production efficiency could be explained by the Ti-OO-Ti configurations for three kinds of crystal surfaces
[34]. When the two Ti ions forming bridged O could not alter their position, the formation of these structures seems difficult. Furthermore, since the dihedral angle of H
2O
2 is 93.4°, the Ti-OO-Ti structure cannot lay on a plane, but the Ti-O-Ti can. Thus, the formation of Ti-OO-Ti at the rutile surfaces of (100), (110), and (001) seems to have some difficulty in configuring the structure.
Figure 4. Local configuration of peroxo, Ti-OO-Ti, at distinct rutile TiO
2 surfaces. Adapted with permission from Ref.
[34]. Copyright 2013 American Chemical Society.
In order to discuss a rational reaction model for the oxidation of water at the surface of TiO
2 crystal, researchers simply suggested that the oxidation and reduction sites are assigned to bridging OH and terminal Ti-OH, respectively
[33], based on the fact that the surface Ti is positively charged while the surface O
br is negatively charged, as expected from the following equations of ionization equilibrium in aqueous solution.
Figure 5 shows the plausible water oxidation processes at the bridging O site of the anatase and rutile TiO
2 surfaces
[31]. A photoinduced hole attacks the bridging O atom at first, followed by the attack of water resulting in the formation of a pair of Ti-O• and Ti-OH
[29]. As shown in
Figure 5, at the anatase surface, since the Ti-Ti distance is too large to form peroxo Ti-OO-Ti structure, a surface-trapped hole should be isolated, and it is useful to oxidize organic molecules RH. A certain fraction of the trapped holes desorbs as •OH radicals into the solution
[36][37]. On the other hand, at the surface of rutile, when a second hole is generated in the particle, it migrates to combine with the existing hole to form bridging peroxo species at the surface
[29][38]. The distance between Ti atoms at the rutile surface (2.96 Å) could be shorter than that of anatase (3.79 Å) and be favorable to forming the Ti-OO-Ti structure
[31][39]. Thus, rutile crystal is relatively more active for the water oxidation to O
2 evolution against anatase crystal
[33][36]. This explanation by the surface Ti-Ti distance for the difference in the water oxidation between rutile and anatase was supported by a theoretical calculation in a recent report
[40].
Figure 5. Plausible photocatalytic oxidation paths at the surface of TiO
2 of anatase and rutile polymorphs. Reproduced from Ref.
[31] with permission from the PCCP Owner Societies.