Molecular Metal Oxide Cluster-Surface Modified Titanium(iv) Dioxide Photocatalysts

TiO₂ has attracted much attention as an ‘eco-catalyst’ for the purification of polluted water and air owing to its UV-light induced powerful oxidation ability, physicochemical stability, and abundance in nature.^[1,2] In general, the photocatalytic activity of TiO₂ strongly depends on the crystal form and the exposed crystal plane.^[3] For the oxidation of organics, anatase usually exhibits a higher UV-light activity than rutile,^[4] which can be improved by coupling with rutile.^[5] The present major challenge in TiO₂ photocatalysis is providing it with visible-light activity at the same time as increasing its UV-light activity. For this purpose, doping of metal ions and anions into TiO₂ has been intensively studied.^[6–12] However, the defects introduced into TiO₂ lead to a decrease in its crystallinity reducing the carrier mobility and increasing the recombination probability. Consequently, if visible-light activity is obtained, the UV-light activity decreases in most cases. On the other hand, bulk- or nano-coupling of appropriate metal oxides with TiO₂ enables the increase of its UV-light activity. For example, the SnO₂-coupling significantly increases the UV-light activity of TiO₂ because of the long-range charge separation assisted by the high electric conductivity of SnO₂.^[13–15] Conversely, Fe₂O₃-^[16] or WO₃-coupling^[17] reduces the UV-light activity, giving no visible-light activity. As an alternative, Kisch and coworkers have devised the photosensitization of TiO₂ by its surface modification with Pt^IV chloride.^[18] This approach is attractive in that a visible-light response can be induced by a simple procedure without the introduction of impurity/vacancy levels. The research groups of Ohno and Hashimoto have reported that the surface modification of rutile TiO₂ with Fe³⁺,^[19,20] Cr,^[21] and Cu^[22] by the impregnation method leads to high visible-light activities for the decomposition of model organic pollutants. However, the effect is small for anatase TiO₂. Libera et al. reported the synthesis of iron(iii)-oxo clusters adsorbed on TiO₂ which showed visible-light absorption and photocatalytic activity for methylene blue degradation.^[23] The interface between the two oxides can result in electron and hole separation upon photoexcitation which is one of the key factors in efficient photocatalysis. Heterostructures of TiO₂ with metals,^[24,25] semiconductors^[26,27] and heterostructures of two different metal oxides^[28–30] have been shown to induce visible light absorption and improved photocatalytic activity when compared with the isolated materials. The formation of a heterostructure of two metal oxides could provide a promising approach for developing materials with improved photocatalytic activity.

Recently, we have shown that the surface modification of anatase and anatase/rutile TiO₂, having the highest level of UV-light activity among the commercial ones, with iron oxide^[31] and nickel oxide clusters^[32] by the chemisorption–calcination cycle (CCC) technique yields a high level of visible-light activity and a concomitant increase in the UV-light activity. In contrast, the surface modification by molecular SnO₂ affords no visible-light activity, while the UV-light activity does increase.^[33] Thus, it is of great importance for the design of this new type of photocatalyst to clarify the action mechanism of the surface modification, which is quite different from those of the usual metal oxide–TiO₂ bulk- and nano-coupling systems.

This short review summarizes our recent experimental and theoretical studies on the iron oxide and tin oxide surface modified anatase/rutile and anatase TiO₂ photocatalysts, respectively. Following the introduction, the experimental results are described, including the metal oxide cluster formation on the TiO₂ surface by the CCC technique (MO/TiO₂), the optical property, photocatalytic activity, and band energy of MO/TiO₂, and the electron transfer from MO/TiO₂ to O₂. Results of simulations using first principles density functional theory (DFT) calculations for FeO_x and SnO₂-modified TiO₂ are then presented. Finally the action mechanisms of the FeO_x and SnO₂ surface modifications on the photocatalytic activity of TiO₂ are discussed.

We developed the CCC technique, where metal complexes such as metal acetylacetonates are adsorbed by chemical bonds, and the organic part is oxidized by post-heating, for preparing metal oxide clusters at a molecular scale (Scheme 1).^[34] For FeO_x-modified TiO₂, Fe(acac)₃ is chemisorbed on the TiO₂ surface by ligand-exchange between the acac-ligand and the surface Ti–OH group (Eqn 1).

Scheme 1.  Schematic representation of the chemisorption–calcination cycle (CCC) technique exampled by the formation of iron oxide-surface modified TiO₂.

where the subscript s denotes the surface adsorbed species.

On the other hand, [Sn(acac)₂]Cl₂ is chemisorbed onto the TiO₂ surface by ion-exchange between H⁺ and the [Sn(acac)₂]²⁺ ion (Eqn 2).^[35]

X-Ray photoelectron spectroscopy (XPS) of the iron oxide and tin oxide surface modified TiO₂ confirmed that the valence states of Fe and Sn are +3/+2 (mixed valency) and +4, respectively. Post-heating the complex-adsorbed TiO₂ at 773 K in air completely oxidizes the residual acac ligands. The Fe (or Sn) on the TiO₂ surface was dissolved by treatment with 35 % HCl, and the solid was completely dissolved into 96 % H₂SO₄ at 353 K. The Fe (or Sn) amounts in both the solutions were almost equal. Also, no particles were observed on the TiO₂ surfaces of the iron oxide and tin oxide surface modified samples by high resolution-transmission electron microscopy. Thus, FeO_x (or SnO₂) species can be formed on the TiO₂ surface in a highly dispersed state at a molecular level by the CCC technique. The use of hydrolysis-resistant Fe(acac)₃ (or [Sn(acac)₂]Cl₂) as a precursor and the post-heating after its strong chemisorption on the TiO₂ surface are considered to contribute to the formation of the unique FeO_x (or SnO₂) species. The Fe (or Sn) loading amount is expressed by the number of Fe (or Sn) ions per unit of TiO₂ surface area (Γ/ions nm^–2). The sample with a given loading, Γ, is designated as MO(Γ)/TiO₂ below.

The optical properties of MO/TiO₂ structures are of primary importance in determining the photocatalytic activity. Fig. 1A shows UV-vis absorption spectra of FeO_x(Γ)/TiO₂. TiO₂ only shows strong absorption, because of the interband transition, in the UV-light region. As a result of the surface modification of TiO₂ with FeO_x, bandgap narrowing takes place, while the intensity of the d–d transition band is very weak at Γ < 0.36. A similar spectroscopic feature was observed for TiO₂ doped with Cr and N prepared by physical methods such as ion implantation and magnetron sputtering.^[10,35] In contrast, on chemically doping Cr and N ions into TiO₂, fairly strong shoulders by the d–d transition appear in the visible region with the absorption edge almost invariant.^[12]

Fig. 1.  (A) UV-Vis absorption spectra of FeO_x/TiO₂ with varying Γ. (B) UV-Vis absorption spectra of TiO₂ (a) and SnO₂(0.024)/TiO₂ (b).

Fig. 1B shows UV-vis absorption spectra of TiO₂ and SnO₂(0.024)/TiO₂. No change in the absorption spectrum is observed after the surface modification, which is true for the other samples with varying Γ. In this manner, the surface modification of TiO₂ with FeO_x extends light absorption into the visible region, whereas no spectroscopic change is induced by the surface modification with SnO₂.

P-25 (anatase/rutile = 4 : 1, w/w) and ST-01 (anatase), having the highest level of UV-light activities among commercial TiO₂, were used as the standard TiO₂ photocatalyst. To evaluate the relative photocatalytic activities of FeO_x/TiO₂ (k) with respect to that of pristine TiO₂ (k⁰), the rate constants were determined under the same irradiation conditions with the same amount of photocatalysts. As a liquid-phase test reaction, the photocatalytic degradation of 2-naphthol (2-NAP) was carried out under illumination of visible light (λ > 400 nm) and UV light (330 < λ < 400 nm). 2-NAP, which is the starting material of azo-dyes and transparent at λ > 330 nm, was used as a model water pollutant. The 2-NAP degradation apparently follows first-order kinetics. Fig. 2a shows the relative first-order pseudo-rate constants for the FeO_x/P-25 photocatalyzed degradation of 2-NAP under illumination of visible-light (k_vis/k_vis⁰) and UV-light (k_UV/k_UV⁰) as a function of Γ. A high level of visible-light activity is induced by the FeO_x surface modification. However, the k_vis value at Γ ≈ 0.5 (0.7 h^–1) is below half the k_UV value at Γ = 0 (1.6 h^–1). This fact points out the importance of the compatibility of the visible-light with the UV-light activities. A marked increase in the UV-light activity with the surface modification is observed at the same time. In each case, the plot exhibits a volcano-type curve with a maximum at Γ ≈ 0.5 ions nm^–2. Under the optimum conditions, the k_vis and k_UV values increase as compared with those for TiO₂ by factors of 7.9 and 4.2, respectively. As a gas-phase test reaction, the photocatalytic degradation of acetaldehyde (CH₃CHO) was carried out under illumination of visible-light and UV-light. CH₃CHO, which is a volatile organic compound responsible for sick-house syndrome, was used as a model air pollutant. Fig. 2b shows the plots of k_vis/k_vis⁰ and k_UV/k_UV⁰ vs. Γ. Drastic increases in the visible-light and UV-light activities with the FeO_x surface modification are observed. Remarkable enhancing effects by the FeO_x surface modification were also obtained for ST-01.^[31] Evidently, the FeO_x surface modification gives rise to a noticeable visible-light activity and a concomitant significant increase in the UV-light activity.

Fig. 2.  (a) Plots of the first-order pseudo-rate constants for the FeO_x/P-25 photocatalyzed degradation of 2-naphthol (2-NAP) under illumination of visible light (k_vis/k_vis⁰, solid circle, λ > 400 nm) and UV-light (k_UV/k_UV⁰, solid triangle, 330 < λ < 400 nm) as a function of Γ. (b) Plots of the rate constants for the FeO_x/P-25 photocatalyzed CH₃CHO decomposition under irradiation of visible light (k_vis/k_vis⁰, solid circle) and UV light (k_UV/k_UV⁰, solid triangle) vs. Γ.

Highly active ST-01 was modified by the SnO₂ species by the CCC technique. As predicted from the absorption spectrum, SnO₂/TiO₂ was inactive for the degradations of 2-NAP and CH₃CHO under illumination of visible light. On the other hand, UV-light irradiation led to the degradations of 2-NAP and CH₃CHO apparently with first-order kinetics followed. Fig. 3 shows the first-order pseudo-rate constants for the SnO₂/ST-01 photocatalzyed degradation of 2-NAP (k_L/k_L⁰) and CH₃CHO (k_G/k_G⁰) under UV-light irradiation as a function of Γ. The plots of both the k_L and k_G exhibit parallel volcano-shaped dependency on Γ with a maximum at Γ ≈ 0.0074 ions nm^–2 and the activities further decreased at Γ > 0.04. The SnO₂ surface modification scarcely affected the adsorptivity for 2-NAP in the dark at Γ < 0.04. Certainly, the SnO₂ surface modification has a positive effect on its UV-light activity for both the liquid-phase and gas-phase reactions, although the effects are much smaller than those by the FeO_x surface modification. Also, this optimum Γ value in the SnO₂/TiO₂ system is smaller as compared with that in the FeO_x/TiO₂ system by a factor of more than one-order of magnitude. This result indicates that the optimum loading amount of the metal oxide clusters strongly depends on their type. Interestingly, it has recently been shown that the SnO_x surface modification of ZnGa₂O₄ by an impregnation method remarkably increases not only the UV-light activity but also the visible-light activity.^[36]

Fig. 3.  Plots of the first-order pseudo-constants of SnO₂/ST-01 photocatalyzed degradations of 2-naphthol (2-NAP) (k_L/k_L⁰, solid circle) and CH₃CHO (k_G/k_G⁰, solid triangle) under illumination of UV light (330 < λ < 400 nm) as a function of Γ.

XPS measurements were performed to gain information about the filled energy levels of FeO_x/TiO₂. Fig. 4a shows the valence band (VB)-XPS spectra for FeO_x/TiO₂. The emission from the O2p-VB extends from 3 to 9 eV. Closer inspection of the VB-top (inset) indicates an upwards shift of 0.4 eV with an increase in Γ, which is comparable with the decrease in the bandgap (E_g) with the FeO_x surface modification. The effective mixing between the surface Fe³⁺ levels and O2p as a result of the Ti_s–O–Fe interfacial bond is considered to yield a surface d-sub-band dispersing around the energy level to overlap with the VB(TiO₂).^[37] This interpretation explains the net decrease in the E_g of TiO₂. From the E_g for FeO_x(Γ ≈ 0.5)/TiO₂ (~2.9 eV), the top of the surface d-sub-band is estimated to be situated at +2.4 V vs. a standard hydrogen electrode (SHE).^[38] In addition, Fig. 4b shows VB-XPS spectra for SnO₂(Γ)/TiO₂. In contrast to the FeO_x/TiO₂ system, the VB-top is not changed by the SnO_x surface modification, irrespective of Γ.

Fig. 4.  Valence-band X-Ray photoelectron spectroscopy (XPS) spectra for FeO_x(Γ)/TiO₂ (a) and SnO₂(Γ)/TiO₂ (b).

Information about empty levels can be obtained by photoluminescence (PL) spectroscopy. Fig. 5 shows PL spectra of FeO_x(Γ)/TiO₂ at 77 K: excitation wavelength = 320 nm. TiO₂ (ST-01) has a broad emission band centered at 538 nm (E₁). Heating it at 773 K for 1 h in air remarkably weakens the E₁ signal intensity. This PL band is assignable to the emission from the surface oxygen vacancy levels of anatase TiO₂.^[39] On modifying TiO₂ with the FeO_x species, the intensity further decreases to disappear at Γ > 0.044 ions nm^–2, while two new emissions appear at 423 (E₂) and 468 nm (E₃). The E₂ and E₃ signals can be attributed to the emissions from extrinsic levels resulting from the surface FeO_x species. These findings suggest that the excited electrons in the conduction band (CB) of TiO₂ are transferred to the empty surface FeO_x levels in preference to the surface oxygen vacancy levels. Additional surface oxygen vacancy is not induced by the surface modification in contrast to the Fe-doping, which is probably because the local charge balance is conserved on the surface.

Fig. 5.  Photoluminesence spectra of FeO_x(Γ)/TiO₂ at 77 K: excitation wavelength = 320 nm.

In the oxidative decomposition of organic pollutants, the key to increasing the photocatalytic activity of TiO₂ is to enhance the transfer of the excited electrons to O₂.^[40,41] Current (I)–potential (E) curves were measured for the mesopoporous TiO₂ nanocrystalline film-coated fluorine-tin oxide (FTO) (mp-TiO₂/FTO) electrodes in an aerated 0.1 m NaClO₄ aqueous solution in the dark. Fig. 6 shows the I–E curves of mp-TiO₂/FTO, FeO_x/mp-TiO₂/FTO, and SnO₂/mp-TiO₂/FTO. In the presence of O₂, cathodic currents attributable to the O₂ reduction are observed at E < –0.2 V, whereas the current hardly flowed at –0.8 < E < –0.2 V without O₂. With the FeO_x surface modification, the reduction current at E < –0.55 V greatly increases, i.e., the surface FeO_x species mediate the electron transfer from TiO₂ to O₂. Also, a smaller enhancing effect is observed with the SnO₂ surface modification.

Fig. 6.  Current–potential curves of mp-TiO₂/FTO (a), FeO_x/mp-TiO₂/FTO (b), and SnO₂/mp-TiO₂/FTO (c) in an aerated 0.1 m NaClO₄ aqueous solution in the dark.

For the calculations of surface-modified TiO₂ we use the DFT approach with corrections for on-site Coulomb interactions, DFT+U, to describe Fe and Ti oxidation states consistently; no such correction is applied to SnO₂, since DFT adequately describes this system.

For modelling TiO₂ rutile (110) and anatase (001) surfaces, we use a three dimensional periodic slab model within the VASP code.^[42] The valence electrons were described by a plane wave basis set and the cut-off for the kinetic energy is 396 eV. There are four valence electrons for Ti, eight for Fe, four for Sn, and six for O. The exchange-correlation functional was approximated by the Perdew–Wang 91^[43] functional. The Monkhorst–Pack scheme was used for K-point sampling with a 2 × 1 × 1 sampling grid.

To describe Ti 3d states the DFT+U approach was used where U = 4.5 eV. The need to introduce the U parameter in order to describe the electronic states of d shells properly is well known.^[44,45] Fe 3d states were described with U = 6.5eV and J = 1 eV, which are typical values from the literature.^[46] For Sn, the electronic states are consistently described by DFT so no U correction was applied. The DFT+U approach gives a relatively correct d state description but still gives an underestimation of the bandgap and this depends on the precise DFT+U set up. We are aware of this important issue but are primarily concerned with qualitative changes in the bandgap upon surface modification. With this in mind, the simulation results are important for understanding the experimental results.

The rutile (110) surface is terminated by two coordinated bridging oxygens and the surface contains five-fold and six-fold coordinated Ti atoms. The anatase (001) surface is terminated by two coordinated oxygen atoms while the oxygen atoms in the surface are three coordinated. The Ti atoms in the surface are five-fold coordinated. All surfaces have a 12 Å vacuum gap. We used a 4 × 2 surface supercell for both surfaces. For consistency in the calculation we applied the same supercell for the bare TiO₂ surface and free clusters. While the anatase (001) surface is not the most stable surface of anatase (which is (101)), the consensus from experimental work is that this surface is the most active for photocatalysis^[47,48] and hence we consider this surface in our simulations.

The clusters are positioned on the TiO₂ surfaces and the adsorption energy is computed from:

where E((MO_x)_n – TiO₂) is the total energy of the MO_x cluster supported on the TiO₂ surface and E(MO_x) and E(TiO₂) are the total energies of the free MO_x cluster and the bare surface, respectively.

A negative adsorption energy signifies that the absorption of the metal oxide cluster is favourable. For both oxide–TiO₂ systems, several FeO_x and SnO₂ adsorption configurations were tested and those presented in Fig. 7 are the most stable we have found.

Fig. 7.  Atomic structures of (a) (FeO)₂ on rutile 110, (b) Sn₂O₄ on anatase 001. (Grey spheres are Ti atoms, red spheres are oxygen atoms from the cluster and the surface, blue spheres are (a) Fe atoms and (b) Sn atoms.)

We performed calculations for the following oxide clusters: FeO, (FeO)₂, Fe₂O₃, SnO₂, and Sn₂O₄ and for the purpose of this summary we present only the most stable structures that we have found for (FeO)₂ on rutile (110) and Sn₂O₄ on anatase (001), with further details presented in the literature.^[33,49] We adsorbed the oxide nanoclusters on the TiO₂ surfaces and ran a full ionic relaxation.

Fig. 7 presents the atomic structure of (FeO)₂ on the rutile (110) surface, and Sn₂O₄ on the anatase (001) surface, with the computed adsorption energies shown. In both composite materials the computed adsorption energies are negative, being –1.84 eV for (FeO)₂/TiO₂ and –5.29 eV for Sn₂O₄/TiO₂.

These negative adsorption energies indicate the stability of the adsorbed iron and tin oxide clusters when supported at the TiO₂ surface. While examining the atomic structures, both materials show an interaction between the surface and the cluster with the formation of new bonds. In the case of (FeO)₂ the Fe atoms are bonded to two bridging oxygen atoms with distances of 2.11 and 2.29 Å. Two atoms from the cluster bind to Ti atoms from the surface with a bond length of 1.89 Å. A Sn₂O₄ cluster creates new bonds with the anatase (001) surface, where two O atoms from the surface have migrated from the surface layer to bind to Sn atoms of the cluster, undergoing a displacement of 1.11 Å. The two O atoms from the cluster bond to Ti from the surface. Thus the growing number of new metal–oxygen bonds results in the stability of the presented structures.

The electronic structure of the composite material can be used to understand its photocatalytic properties. We plot in Fig. 8 the projected electronic density of states (PEDOS) for bare rutile (110) and anatase (001), as well as the Fe 3d states from the (FeO)₂ cluster and Ti 3d states from the rutile (110) surface and the Sn 5s states from the Sn₂O₄ cluster and Ti 3d states from the anatase (001) surface.

Fig. 8.  Spin polarized electronic density of states (DOS): (a) projected on Ti 3d surface states of rutile (110), (b) projected on Ti 3d surface states of anatase (001), (c) projected on Ti 3d surface states and Fe 3d cluster states on Fe₂O₂-TiO₂ rutile (110), and (d) projected on Ti 3d surface states and Sn 5s cluster states on Sn₂O₄-TiO₂ anatase (001). The zero of energy is the Fermi level. Cluster states are enhanced by a factor of 10 to make them more visible relative to the surface.

For (FeO)₂-modified TiO₂, the DOS shows that the Fe 3d states of adsorbed FeO_x lies above the valence band of TiO₂ while the conduction band derives from Ti 3d from the TiO₂ surface. This result explains the photocatalytic properties found for FeO_x–TiO₂, because the state from FeO_x that lies above the TiO₂ valence band gives the bandgap narrowing compared with bare TiO₂ surface pushing the light absorption into the visible region. Moreover we can expect improved charge separation because upon visible-light excitation the excited electrons will be found on TiO₂ and the holes on the iron oxides.^[37]

The behaviour of Sn₂O₄ on TiO₂ anatase (001) is, however, different; the resulting bandgap is unchanged in comparison to the bare TiO₂ surface. Examining the PEDOS plot we found that empty tin oxide states lie above the conduction band of the TiO₂ surface and there are no tin oxide states found in the VB–CB energy gap of TiO₂. Therefore, SnO₂ modification of anatase will give no bandgap shift to the visible-light region. The fact that the Sn 5s states are dispersed above the CB of TiO₂ has the consequence that these states will be acceptors of electrons during UV-light excitation.

The presented results for both oxides explain the experimental results that show FeO_x-modified TiO₂ shifts the photocatalytic activity to the visible region, and Sn₂O₄-supported material enhances only the UV-light activity.

The energy band diagram scheme of oxide-modified TiO₂ derived on the basis of the results above is shown in Scheme 2a for FeO_x. As indicated by the VB-XPS measurements and DFT simulation, the loading of the FeO_x species mainly modifies the electronic levels of TiO₂ around the VB edge. Visible-light absorption triggers electronic excitation from the surface d sub-band, derived from iron oxide, to the CB (derived from TiO₂) (p₁, Scheme 2). The charge separation is achieved by this visible-light induced surface-to-bulk interfacial electron transfer (IET). On illumination with UV light (p₂), the electrons in the VB(TiO₂) are excited to the CB(TiO₂). In both the cases, without the FeO_x surface modification, the excited electrons in the CB(TiO₂) are rapidly trapped at the surface oxygen vacancy levels to undergo recombination with the holes. The FeO_x surface modification permits the preferential electron transfer from the CB(TiO₂) to the surface FeO_x levels (p₃). The electrons effectively reduce adsorbed O₂ by the action of the surface FeO_x species as an excellent mediator (p₄). As a result of the effective suppression of the recombination by the surface oxygen vacancy levels, the photocatalytic activities drastically increase under illumination of visible and UV light. A feature of the surface modification in the anodic process should also be stressed in that the holes generated in the surface d sub-band take part in the oxidation process without diffusion (p₅).^[21]

Scheme 2.  Energy diagram schemes of Fe₂O₃-TiO₂ bulk-coupling system (a) and FeO_x surface-modified TiO₂ (b).

On the other hand, the loading of the SnO₂ species modifies the unfilled electronic levels, while the E_g is invariant.^[37,49] Scheme 2b shows the energy band diagram scheme of SnO₂/TiO₂ derived from experimental and simulation results. On UV-light irradiation, the electrons in the VB(TiO₂) can be promoted to either the CB(TiO₂) or the SnO₂ states around the CB edge of TiO₂. The latter process, i.e., the UV-light induced bulk-to-surface IET, leads to the charge separation. Subsequently, the electrons in the Sn states are transferred to O₂ with an aid of the function as a mediator.

This short review summarizes our work on FeO_x and SnO₂ surface-modified TiO₂ photocatalysts. Extremely small metal oxide clusters are formed on the TiO₂ surface by the CCC technique with the loading amount strictly controlled. The compatibility of high visible-light and UV-light activities is achieved by the FeO_x surface modification, while only UV-light activity somewhat increases with the SnO₂ surface modification. Both the effects can be rationalized in terms of the experimental characterization and the DFT calculations. For FeO_x supported on TiO₂ rutile (110) we: (a) obtained a narrowed bandgap, giving visible-light absorption, when compared with pure TiO₂, (b) determined that the top of the VB comes from the Fe 3d states of the cluster and the CB arises from Ti 3d states of the surface, and (c) subsequently observed an expected charge separation and consequent reduction of electron–hole recombination. SnO₂ species supported on TiO₂ anatase (001) presents the following features: (a) no change in the bandgap, so no visible-light activity as confirmed by the experiment, (b) Sn 5s states dispersed and centred higher than the CB edge, and (c) under UV light excitation these states can be acceptor levels of electrons from the VB (TiO₂). Based on these promising results, systematic experimental and simulation studies on the surface modification of TiO₂ with various metal oxides are currently in progress.