Dynamics of Photogenerated Charge Carriers in TiO 2 / MoO 3 , TiO 2 / WO 3 and TiO 2 / V 2 O 5 Photocatalysts with Mosaic Structure

: Titania is a widely used photocatalytic material possessing such advantages as low cost and high reactivity under the ultraviolet light illumination. However, the fast recombination of photoexcited charge carriers limits its application. Herein, we have synthesized original nanomaterials with mosaic structures that exhibited well-deﬁned heterojunctions and new properties. Using SEM, XRD, EPR spectroscopy, photocatalytic measurements, and photoinduced pathphysiological activity of these photocatalysts, we studied the processes of charge carrier accumulation in TiO 2 / MoO 3 , TiO 2 / WO 3 , and TiO 2 / V 2 O 5 under in situ UV illumination with emphasis on the charge exchange between energy levels of these nanosized semiconductors. It is shown that the accumulation of photoinduced charges occurs in two forms (i) ﬁlled electron traps corresponding to Ti 4 + / Ti 3 + levels and (ii) Mo 5 + centers, both forms contributing to the photoinduced biocide activity of the samples. This work demonstrates that light exposure of heterostructure photocatalysts with mosaic surfaces produces di ﬀ erent types of charge-trapping centers capable of interacting with molecular oxygen yielding peroxo species, which provide long-life light-induced ”self-cleaning” behavior. Such photoaccumulating materials open new opportunities in developing light-driven self-sterilization structures exhibiting a prolonged bactericidal e ﬀ ect up to 10 h after stopping light exposure.


Microscopy
All materials used in the work were purchased from Sigma-Aldrich Chemical Co. and used as supplied. For preparation of solutions, the Milli-Q water was used. The thin-film photocatalysts were prepared using aqueous dispersions of TiO2, MoO3, WO3, and V2O5 contrastingly to hydrolytically, pyrolytically, or mechanochemically derived TiO2/MoO3 (V2O5, V2O5:MoO3) composite photocatalysts investigated in our previous works [14][15][16][17][18]21]. These photocatalysts consisting of the initially synthesized crystallites exhibited well-defined heterojunctions and demonstrated a mosaic structure. The detailed mechanism behind solvothermal synthesis based on the polycondensation of oxo-compounds is discussed elsewhere [14,21]. This synthetic procedure yields perfect crystallites of hexagonal MoO3, hexagonal WO3, and V2O5 xerogel (both crystalline polymorphs possess the lamellar structure) with the medium size of 200-500 nm.
The morphological investigations employing SEM provided evidence about mosaic structure ( Figure 1). The AFM (atomic force microscope) images for TiO2/MoO3, TiO2/WO3, and TiO2/V2O5 heterostructure photocatalysts given in Figure 2 show that the size of building blocks forming the surface of composite film exhibited a considerable decrease when going from TiO2/MoO3 film to TiO2/WO3 film and then to TiO2/V2O5 (the mean roughness evaluated from AFM 10  10 μm plots decreased from 398 nm to 380 and 306 nm, correspondingly). The observed decrease in the composite roughness can be attributed to the well-known fact of a mutual protective action of composite-forming oxide components against their crystallization that hampers rearrangements in the composite film during an annealing and a strong tendency to form aggregates inherent in V2O5 particles. Notwithstanding these variations in the surface roughness that should considerably affect the specific contact angle values, the heterogeneous binary surface of composite films remained hydrophilic (the water contact angle amounted to ~25° even for TiO2/V2O5 The AFM (atomic force microscope) images for TiO 2 /MoO 3 , TiO 2 /WO 3 , and TiO 2 /V 2 O 5 heterostructure photocatalysts given in Figure 2 show that the size of building blocks forming the surface of composite film exhibited a considerable decrease when going from TiO 2 /MoO 3 film to TiO 2 /WO 3 film and then to TiO 2 /V 2 O 5 (the mean roughness evaluated from AFM 10 × 10 µm plots decreased from 398 nm to 380 and 306 nm, correspondingly).

Microscopy
All materials used in the work were purchased from Sigma-Aldrich Chemical Co. and used as supplied. For preparation of solutions, the Milli-Q water was used. The thin-film photocatalysts were prepared using aqueous dispersions of TiO2, MoO3, WO3, and V2O5 contrastingly to hydrolytically, pyrolytically, or mechanochemically derived TiO2/MoO3 (V2O5, V2O5:MoO3) composite photocatalysts investigated in our previous works [14][15][16][17][18]21]. These photocatalysts consisting of the initially synthesized crystallites exhibited well-defined heterojunctions and demonstrated a mosaic structure. The detailed mechanism behind solvothermal synthesis based on the polycondensation of oxo-compounds is discussed elsewhere [14,21]. This synthetic procedure yields perfect crystallites of hexagonal MoO3, hexagonal WO3, and V2O5 xerogel (both crystalline polymorphs possess the lamellar structure) with the medium size of 200-500 nm.
The morphological investigations employing SEM provided evidence about mosaic structure ( Figure 1). The AFM (atomic force microscope) images for TiO2/MoO3, TiO2/WO3, and TiO2/V2O5 heterostructure photocatalysts given in Figure 2 show that the size of building blocks forming the surface of composite film exhibited a considerable decrease when going from TiO2/MoO3 film to TiO2/WO3 film and then to TiO2/V2O5 (the mean roughness evaluated from AFM 10  10 μm plots decreased from 398 nm to 380 and 306 nm, correspondingly). The observed decrease in the composite roughness can be attributed to the well-known fact of a mutual protective action of composite-forming oxide components against their crystallization that hampers rearrangements in the composite film during an annealing and a strong tendency to form aggregates inherent in V2O5 particles. Notwithstanding these variations in the surface roughness that should considerably affect the specific contact angle values, the heterogeneous binary surface of composite films remained hydrophilic (the water contact angle amounted to ~25° even for TiO2/V2O5 The observed decrease in the composite roughness can be attributed to the well-known fact of a mutual protective action of composite-forming oxide components against their crystallization that hampers rearrangements in the composite film during an annealing and a strong tendency to form aggregates inherent in V 2 O 5 particles. Notwithstanding these variations in the surface roughness that should considerably affect the specific contact angle values, the heterogeneous binary surface of composite films remained hydrophilic (the water contact angle amounted to~25 • even for TiO 2 /V 2 O 5 Catalysts 2020, 10, 1022 4 of 14 film with the smoothest surface) that should facilitate water adsorption from humid air and thus creates favorable conditions for photocatalyst operation.
Catalysts 2020, 10, x FOR PEER REVIEW 4 of 14 film with the smoothest surface) that should facilitate water adsorption from humid air and thus creates favorable conditions for photocatalyst operation.

XRD
According to XRD analysis, hexagonal MoO3 with an admixture of monoclinic MoO3·H2O, hexagonal WO3 with an admixture of orthorhombic WO3·0.33H2O, and lamellar V2O5 xerogel crystallites were formed ( Figure 3). According to XRD analysis, the resultant titania was anatase (Figure 4) with the medium size of crystallites of ca. 4 nm. The size of crystallites (regions of coherent scattering) was estimated from the broadening of diffraction reflections using the Scherer formula: According to XRD analysis, the resultant titania was anatase (Figure 4) with the medium size of crystallites of ca. 4 nm. The size of crystallites (regions of coherent scattering) was estimated from the broadening of diffraction reflections using the Scherer formula: Catalysts 2020, 10, 1022

of 14
where d XRD is the average size of the coherent scattering region, β is the width of the corresponding diffraction peak at half maximum, λ is the wavelength of the radiation used, θ is the diffraction angle, and k = 0.9.
Catalysts 2020, 10, x FOR PEER REVIEW 5 of 14 where dXRD is the average size of the coherent scattering region, β is the width of the corresponding diffraction peak at half maximum, λ is the wavelength of the radiation used, θ is the diffraction angle, and k = 0.9. The mixtures of TiO2 sol and MoO3 (V2O5) suspensions (2:1 in the case of TiO2/MoO3 and TiO2/V2O5 photocatalysts) were deposited onto the glass substrate by pulverization. The resultant coating was then annealed at 400 °C. The XRD analysis evidenced that the annealing did not result in the recrystallization of the oxide phases forming the photocatalytic film. According to the AFM measurements thus obtained, thin-film photocatalytic coatings had a thickness of ca. 2.5 μm.

Photocatalytic and Pathphysiological Activity of TiO2/MoO3, TiO2/WO3, and TiO2/V2O5 Heterostructure Photocatalysts
The kinetic curves shown in Figure 5a demonstrate degradation of probing dye at the surface of pre-exposed TiO2/MoO3, TiO2/WO3, and TiO2/V2O5 heterostructure photocatalysts; for comparison, the curve for bare titania is given. The efficient storage of photoproduced negative charge observed for all binary photocatalysts resulting in the induced oxidation activity imparted to their surface. This oxidation activity retains for a long time after exposure due to the formation of hydrogen bronzes of variable composition [14,22] which exhibit one-electron oxidation with molecular oxygen yielding peroxo species [16].
The continuous generation of peroxo species during the course of discharging is also responsible for the long-term microbiocide effect as evident by Figure 5b. It is seen from Figure 5b that the survival ratio for heterostructure photocatalysts even exhibited some further decreases when standing in the dark after exposure due to the buildup of the surface concentration of peroxide and then began to decline. From this point of view, the photoaccumulating heterostructure catalysts differed radically from bare TiO2, which lost the light-induced pathophysiological activity almost immediately (within one minute) after UV illumination due to short lifetimes inherent in reactive oxygen species generated at the titania surface under illumination (the lifetime of hydroxyl radicals is below a second, while the lifetime of O2 − does not exceed 50 s [23,24]). The survival ratio versus time dependencies given in Figure 5b permit a conclusion that pre-exposed TiO2/V2O5 photocatalyst demonstrated higher biocide activity as compared to TiO2/MoO3 and TiO2/WO3 photocatalysts. The mixtures of TiO 2 sol and MoO 3 (V 2 O 5 ) suspensions (2:1 in the case of TiO 2 /MoO 3 and TiO 2 /V 2 O 5 photocatalysts) were deposited onto the glass substrate by pulverization. The resultant coating was then annealed at 400 • C. The XRD analysis evidenced that the annealing did not result in the recrystallization of the oxide phases forming the photocatalytic film. According to the AFM measurements thus obtained, thin-film photocatalytic coatings had a thickness of ca. 2.5 µm.

Photocatalytic and Pathphysiological Activity of TiO 2 /MoO 3 , TiO 2 /WO 3 , and TiO 2 /V 2 O 5 Heterostructure Photocatalysts
The kinetic curves shown in Figure 5a demonstrate degradation of probing dye at the surface of pre-exposed TiO 2 /MoO 3 , TiO 2 /WO 3 , and TiO 2 /V 2 O 5 heterostructure photocatalysts; for comparison, the curve for bare titania is given. The efficient storage of photoproduced negative charge observed for all binary photocatalysts resulting in the induced oxidation activity imparted to their surface. This oxidation activity retains for a long time after exposure due to the formation of hydrogen bronzes of variable composition [14,22] which exhibit one-electron oxidation with molecular oxygen yielding peroxo species [16].
The continuous generation of peroxo species during the course of discharging is also responsible for the long-term microbiocide effect as evident by Figure 5b. It is seen from Figure 5b that the survival ratio for heterostructure photocatalysts even exhibited some further decreases when standing in the dark after exposure due to the buildup of the surface concentration of peroxide and then began to decline. From this point of view, the photoaccumulating heterostructure catalysts differed radically from bare TiO 2 , which lost the light-induced pathophysiological activity almost immediately (within one minute) after UV illumination due to short lifetimes inherent in reactive oxygen species generated at the titania surface under illumination (the lifetime of hydroxyl radicals is below a second, while the lifetime of O 2 − does not exceed 50 s [23,24]). The survival ratio versus time dependencies given in Figure 5b permit a conclusion that pre-exposed TiO 2 /V 2 O 5 photocatalyst demonstrated higher biocide activity as compared to TiO 2 /MoO 3 and TiO 2 /WO 3 photocatalysts.

EPR Measurements of Heterostructure Photocatalysts
At the next step of the work, we applied an EPR technique for investigating the nature, concentration of paramagnetic centers (PCs) existing in mixed photocatalysts initially, and their change after UV irradiation. Indeed, all PCs in these systems are kinds of structural "defects" because TiO2, MoO3, WO3, and V2O5 contain transition metal ions, in principle, in the ground diamagnetic states such as Ti 4+ , Mo 6+ , W 6+ , and V 5+ ; hence, one electron that reduced Ti 3+ , Mo 5+ , W 5+ , and V 4+ ions can be assumed as a specific spin probe. Spectra shown in Figure 6 demonstrate the presence of several different PCs in the initial TiO2/MoO3 oxide, e.g., titanium (3+) ions located in the lattice (Ti 3+ lat) or on the surface (Ti 3+ surf), molybdenum Mo 5+ PCs, and nitrogen atoms usually marked as N • for stressing its paramagnetic nature. Spin Hamiltonian parameters, g-tensor and A-tensor characterizing Zeeman and hyperfine (hfi) interactions, determined from EasySpin modeling gave values represented in Table 1, which coincide in the error limits with those reported in [16,17,25,26] and references therein. Earlier, it was shown by double integration of EPR spectra that no more than 1-2% of the total content of nitrogen 14 N atoms are paramagnetic in these systems [26,27], while the others remain diamagnetic. It was proved that the signal from N • -radical definitely appeared after high temperature treatment of oxide-hydroxide precipitate, and its source was NH4 + residues adsorbed on the TiO2 particle surface or incorporated into the TiO2 matrix. Positions of EPR lines of other paramagnetic centers are shown in Figure 6 and agree with the literature very well. We should note that in our case, two types of Mo 5+ centers were observed: (i) a signal Mo 5+ (1) and (ii) a broad single line Mo 5+ (2) with a line width

EPR Measurements of Heterostructure Photocatalysts
At the next step of the work, we applied an EPR technique for investigating the nature, concentration of paramagnetic centers (PCs) existing in mixed photocatalysts initially, and their change after UV irradiation. Indeed, all PCs in these systems are kinds of structural "defects" because TiO 2 , MoO 3 , WO 3 , and V 2 O 5 contain transition metal ions, in principle, in the ground diamagnetic states such as Ti 4+ , Mo 6+ , W 6+ , and V 5+ ; hence, one electron that reduced Ti 3+ , Mo 5+ , W 5+ , and V 4+ ions can be assumed as a specific spin probe. Spectra shown in Figure 6 demonstrate the presence of several different PCs in the initial TiO 2 /MoO 3 oxide, e.g., titanium (3+) ions located in the lattice (Ti 3+ lat ) or on the surface (Ti 3+ surf ), molybdenum Mo 5+ PCs, and nitrogen atoms usually marked as N • for stressing its paramagnetic nature. Spin Hamiltonian parameters, g-tensor and A-tensor characterizing Zeeman and hyperfine (hfi) interactions, determined from EasySpin modeling gave values represented in Table 1, which coincide in the error limits with those reported in [16,17,25,26] and references therein. Earlier, it was shown by double integration of EPR spectra that no more than 1-2% of the total content of nitrogen 14 N atoms are paramagnetic in these systems [26,27], while the others remain diamagnetic. It was proved that the signal from N • -radical definitely appeared after high temperature treatment of oxide-hydroxide precipitate, and its source was NH 4 + residues adsorbed on the TiO 2 particle surface or incorporated into the TiO 2 matrix. Positions of EPR lines of other paramagnetic centers are shown in Figure 6 and agree with the literature very well. We should note that in our case, two types of Mo 5+ centers were  UV illumination of TiO2/MoO3 sample produced sufficient changes in the EPR spectrum ( Figure  6). Since the intensity of the surface Ti 3+ surf lines and both Mo 5+ PCs increased, that of the bulk Ti 3+ lat centers did not practically change, and of N • atoms decreased ca. fourfold without any changes in the line positions. Evidently, these spectral transformations directly reflect the corresponding changes in PC concentration. Note that in this system, we could also observe the appearance of the O2 − radicals (see Figure 6). Kinetic dependences of these changes under UV irradiation are reproduced graphically in Figure 7. For better comparison, we plotted all these curves in arbitrary units.  UV illumination of TiO 2 /MoO 3 sample produced sufficient changes in the EPR spectrum ( Figure 6). Since the intensity of the surface Ti 3+ surf lines and both Mo 5+ PCs increased, that of the bulk Ti 3+ lat centers did not practically change, and of N • atoms decreased ca. fourfold without any changes in the line positions. Evidently, these spectral transformations directly reflect the corresponding changes in PC concentration. Note that in this system, we could also observe the appearance of the O 2 − radicals (see Figure 6). Kinetic dependences of these changes under UV irradiation are reproduced graphically in Figure 7. For better comparison, we plotted all these curves in arbitrary units.
One can see that UV illumination increased the concentrations of paramagnetic metal ions and decreased that of N • -radical during ca. 50-60 min of treatment, and afterwards they stayed constant reflecting stationary rates of the creation and decay of these PCs. All changes were impressive: from four-to fivefold (Figure 7). It is also obvious that the rate of photogeneration of all these metal PCs at first 60 min of irradiation was practically equal, and this allows the assumption that all observed reduction processes were correlated and not independent.  One can see that UV illumination increased the concentrations of paramagnetic metal ions and decreased that of N • -radical during ca. 50-60 min of treatment, and afterwards they stayed constant reflecting stationary rates of the creation and decay of these PCs. All changes were impressive: from four-to fivefold (Figure 7). It is also obvious that the rate of photogeneration of all these metal PCs at first 60 min of irradiation was practically equal, and this allows the assumption that all observed reduction processes were correlated and not independent.
In the dark conditions after switching off illumination, created paramagnetic centers were re-oxidized by air dioxygen and by electron transfer reactions between reduced PCs according to their red-ox potentials. These processes were much slower (over 30-40 h, see Figure 8) due to complex character of the electron transfer in such nanosized heterojunction systems [14][15][16][17][18]25]. Note that (i) at the first 10 h in the dark, the decay was exponential, and later it became practically linear fitted; (ii) the rate of the decay processes varied on the metal nature: Ti 3+ surf concentration changed faster than that of V 4+ ; (iii) concentration of N • radicals increased linearly during all 40 h (Figure 8).  UV irradiation of TiO2/WO3 also produced noticeable changes ( Figure 9): initial EPR signal recorded in dark conditions (g  1.63 and H  23.0 mT) transformed to a new spectrum with parameters listed in Table 1, which correlate with g-values published in [28][29][30][31]. The observed three-axes anisotropy is usual for W 5+ PCs. Unfortunately, we could not quantitatively characterize In the dark conditions after switching off illumination, created paramagnetic centers were re-oxidized by air dioxygen and by electron transfer reactions between reduced PCs according to their red-ox potentials. These processes were much slower (over 30-40 h, see Figure 8) due to complex character of the electron transfer in such nanosized heterojunction systems [14][15][16][17][18]25]. Note that (i) at the first 10 h in the dark, the decay was exponential, and later it became practically linear fitted; (ii) the rate of the decay processes varied on the metal nature: Ti 3+ surf concentration changed faster than that of V 4+ ; (iii) concentration of N • radicals increased linearly during all 40 h (Figure 8). One can see that UV illumination increased the concentrations of paramagnetic metal ions and decreased that of N • -radical during ca. 50-60 min of treatment, and afterwards they stayed constant reflecting stationary rates of the creation and decay of these PCs. All changes were impressive: from four-to fivefold (Figure 7). It is also obvious that the rate of photogeneration of all these metal PCs at first 60 min of irradiation was practically equal, and this allows the assumption that all observed reduction processes were correlated and not independent.
In the dark conditions after switching off illumination, created paramagnetic centers were re-oxidized by air dioxygen and by electron transfer reactions between reduced PCs according to their red-ox potentials. These processes were much slower (over 30-40 h, see Figure 8) due to complex character of the electron transfer in such nanosized heterojunction systems [14][15][16][17][18]25]. Note that (i) at the first 10 h in the dark, the decay was exponential, and later it became practically linear fitted; (ii) the rate of the decay processes varied on the metal nature: Ti 3+ surf concentration changed faster than that of V 4+ ; (iii) concentration of N • radicals increased linearly during all 40 h (Figure 8).  UV irradiation of TiO2/WO3 also produced noticeable changes ( Figure 9): initial EPR signal recorded in dark conditions (g  1.63 and H  23.0 mT) transformed to a new spectrum with parameters listed in Table 1, which correlate with g-values published in [28][29][30][31]. The observed three-axes anisotropy is usual for W 5+ PCs. Unfortunately, we could not quantitatively characterize UV irradiation of TiO 2 /WO 3 also produced noticeable changes ( Figure 9): initial EPR signal recorded in dark conditions (g ≈ 1.63 and ∆H ≈ 23.0 mT) transformed to a new spectrum with parameters listed in Table 1, which correlate with g-values published in [28][29][30][31]. The observed three-axes anisotropy is usual for W 5+ PCs. Unfortunately, we could not quantitatively characterize kinetics under illumination and in the dark afterwards due to very noisy EPR spectra recording (Figure 9).
Catalysts 2020, 10, x FOR PEER REVIEW 9 of 14 kinetics under illumination and in the dark afterwards due to very noisy EPR spectra recording ( Figure 9).  Figure 9. EPR spectra of TiO2/WO3 photocatalyst before (a) and after 10 min (b) of UV light illumination. Figure 10 represents changes in EPR spectra after 10 min of UV illumination of TiO2/V2O5 photocatalyst, which led to an approximately twofold increase in the spectrum intensity. The kinetics of V 4+ signal grove and its decay in the dark conditions are shown in Figures 7 and 8. We would stress that the EPR signal intensity of V 4+ PCs in mixed oxides containing vanadium pentoxide as one of the components is so high that spectra of Ti 3+ or Mo 5+ paramagnetic species cannot be observed due to their much lower amplitude as well as the high oxidative properties of V 5+ ions in V2O5. It has been shown recently that the position of energy levels corresponding to the charge accepting states in the bandgap of TiO2, MoO3, and V2O5 (Mo 6+ /Mo 5+ , V 5+ /V 4+ , Ti 4+ /Ti 3+ ) can be determined from EPR measurements under in situ illumination [16]. The energies of these levels measured against the valence band of corresponding semiconductors are collected in Table 2. It should be noted that Ti 4+ /Ti 3+ states involved into the charge exchange correspond to the surface states, whereas the population of lattice Ti 4+ /Ti 3+ states is not modulated under illumination [16].
Due to the low signal-to-noise ratio exhibited by TiO2/WO3 that did not allow the determination of reliable energy value of W 6+ /W 5+ states from EPR spectra, this energy was estimated from the X-ray photoelectron spectra for tungsten oxide bronzes [32]. Using the photocurrent onset potential Figure 9. EPR spectra of TiO 2 /WO 3 photocatalyst before (a) and after 10 min (b) of UV light illumination. Figure 10 represents changes in EPR spectra after 10 min of UV illumination of TiO 2 /V 2 O 5 photocatalyst, which led to an approximately twofold increase in the spectrum intensity. The kinetics of V 4+ signal grove and its decay in the dark conditions are shown in Figures 7 and 8. We would stress that the EPR signal intensity of V 4+ PCs in mixed oxides containing vanadium pentoxide as one of the components is so high that spectra of Ti 3+ or Mo 5+ paramagnetic species cannot be observed due to their much lower amplitude as well as the high oxidative properties of V 5+ ions in V 2 O 5 .
( Figure 9).  Figure 9. EPR spectra of TiO2/WO3 photocatalyst before (a) and after 10 min (b) of UV light illumination. Figure 10 represents changes in EPR spectra after 10 min of UV illumination of TiO2/V2O5 photocatalyst, which led to an approximately twofold increase in the spectrum intensity. The kinetics of V 4+ signal grove and its decay in the dark conditions are shown in Figures 7 and 8. We would stress that the EPR signal intensity of V 4+ PCs in mixed oxides containing vanadium pentoxide as one of the components is so high that spectra of Ti 3+ or Mo 5+ paramagnetic species cannot be observed due to their much lower amplitude as well as the high oxidative properties of V 5+ ions in V2O5. It has been shown recently that the position of energy levels corresponding to the charge accepting states in the bandgap of TiO2, MoO3, and V2O5 (Mo 6+ /Mo 5+ , V 5+ /V 4+ , Ti 4+ /Ti 3+ ) can be determined from EPR measurements under in situ illumination [16]. The energies of these levels measured against the valence band of corresponding semiconductors are collected in Table 2. It should be noted that Ti 4+ /Ti 3+ states involved into the charge exchange correspond to the surface states, whereas the population of lattice Ti 4+ /Ti 3+ states is not modulated under illumination [16].
Due to the low signal-to-noise ratio exhibited by TiO2/WO3 that did not allow the determination of reliable energy value of W 6+ /W 5+ states from EPR spectra, this energy was estimated from the X-ray photoelectron spectra for tungsten oxide bronzes [32]. Using the photocurrent onset potential It has been shown recently that the position of energy levels corresponding to the charge accepting states in the bandgap of TiO 2 , MoO 3 , and V 2 O 5 (Mo 6+ /Mo 5+ , V 5+ /V 4+ , Ti 4+ /Ti 3+ ) can be determined from EPR measurements under in situ illumination [16]. The energies of these levels measured against the valence band of corresponding semiconductors are collected in Table 2. It should be noted that Ti 4+ /Ti 3+ states involved into the charge exchange correspond to the surface states, whereas the population of lattice Ti 4+ /Ti 3+ states is not modulated under illumination [16]. Table 2. Energy position of levels involved in the charge storage in TiO 2 /MoO 3 , TiO 2 /WO 3 , and TiO 2 /V 2 O 5 photocatalysts (energy is measured against the valence band of semiconductor).
Due to the low signal-to-noise ratio exhibited by TiO 2 /WO 3 that did not allow the determination of reliable energy value of W 6+ /W 5+ states from EPR spectra, this energy was estimated from the X-ray photoelectron spectra for tungsten oxide bronzes [32]. Using the photocurrent onset potential values, E on , given in Table 3 as the rough estimation of the energy position of the bottom of the conduction band of highly doped TiO 2 , MoO 3 , WO 3 , and V 2 O 5 , one can obtain the detailed energy diagram of the energy storage photocatalysts under consideration (Figure 11). Table 3. The band gap energies (E g ) and the photocurrent onset potentials (E on ) for oxides used for preparation of heterostructured photocatalysts.

Oxide
E g *, eV E on , V values, Eon, given in Table 3 as the rough estimation of the energy position of the bottom of the conduction band of highly doped TiO2, MoO3, WO3, and V2O5, one can obtain the detailed energy diagram of the energy storage photocatalysts under consideration ( Figure 11).  [16]; ** the value was obtained from the X-ray photoelectron spectra [32].  The EPR measurements strongly show that in the case of TiO2/MoO3 photocatalyst, there were two mechanisms of accumulation of negative charge photoproduced in TiO2: (i) injection of electrons into MoO3 followed by redox transformations of oxide accompanied by the formation of Mo(V) centers readily detectable in EPR spectra; (ii) "indirect" filling of Ti 4+ /Ti 3+ surface states via electron transfer from Mo 6+ /Mo 5+ states which lie at higher energies. The latter process in the case of bare TiO2 appeared to be completely hindered since the rate constant of electron trapping at these states was much lower as compared to the rate constant of surface recombination. In the TiO2/MoO3 heterostructure photocatalyst, Ti 4+ /Ti 3+ states were in equilibrium with the Mo 6+ /Mo 5+ states, and their filling was not compete with recombination. By contrast, it is seen from the energy diagram given in Figure 11 that both W 6+ /W 5+ and V 5+ /V 4+ states lay below Ti 4+ /Ti 3+ states and thus did not contribute to their filling. As a result, in the case of TiO2/WO3 and TiO2/V2O5 photocatalysts, the redox transformations in WO3 and V2O5 appeared to be the only mechanism responsible for charge storage.
The charged Ti 4+ /Ti 3+ states at the surface of titania nanoparticles can be readily oxidized with oxygen from the air yielding peroxo species, which resulted in a high initial rate of probing dye The EPR measurements strongly show that in the case of TiO 2 /MoO 3 photocatalyst, there were two mechanisms of accumulation of negative charge photoproduced in TiO 2 : (i) injection of electrons into MoO 3 followed by redox transformations of oxide accompanied by the formation of Mo(V) centers readily detectable in EPR spectra; (ii) "indirect" filling of Ti 4+ /Ti 3+ surface states via electron transfer from Mo 6+ /Mo 5+ states which lie at higher energies. The latter process in the case of bare TiO 2 appeared to be completely hindered since the rate constant of electron trapping at these states was much lower as compared to the rate constant of surface recombination. In the TiO 2 /MoO 3 heterostructure photocatalyst, Ti 4+ /Ti 3+ states were in equilibrium with the Mo 6+ /Mo 5+ states, and their filling was not compete with recombination. By contrast, it is seen from the energy diagram given in Figure 11 that both W 6+ /W 5+ and V 5+ /V 4+ states lay below Ti 4+ /Ti 3+ states and thus did not contribute to their filling. As a result, in the case of TiO 2 /WO 3 and TiO 2 /V 2 O 5 photocatalysts, the redox transformations in WO 3 and V 2 O 5 appeared to be the only mechanism responsible for charge storage.
The charged Ti 4+ /Ti 3+ states at the surface of titania nanoparticles can be readily oxidized with oxygen from the air yielding peroxo species, which resulted in a high initial rate of probing dye degradation at the pre-exposed photocatalyst ( Figure 1). However, due to rapid consumption of the charges accumulated at Ti 4+ /Ti 3+ states, their contribution to the long-term pathophysiological activity was marginal. As a result, the photoinduced biocide activity of TiO 2 /MoO 3 photocatalyst (as evident by Figure 2) appeared to be less than the photoinduced biocide activity of TiO 2 /WO 3 and TiO 2 /V 2 O 5 photocatalysts accumulating photoinduced charges solely in the form of hydrogen bronzes (and, therefore, demonstrating a lower initial rate of probing dye degradation- Figure 1). The oxidation of the hydrogen bronzes occurred slowly being governed by the diffusion of redox centers from the bulk of nanocrystals of redox-active oxides to their surface that ensure continuous production of peroxide (and guarantees high antimicrobial activity level) during a long period of time. The energy diagram given in Figure 11 also shows that combining nanoparticles of different oxides in the photocayalyst film, it is possible to realize cascade effects that ensure efficient separation and accumulation of photoproduced charges.
The EPR measurements thus provide evidence that in the case of heterostructure photocatalysts with mosaic surfaces (i.e., comprising photogenerating and redox-active components), the exposure to the actinic light results in the generation of charge-trapping reaction centers of several types exhibiting different oxidation activity and long-term stability. As a result, these centers make different contributions in the induced pathophysiological behavior of energy storage photocatalysts.

Materials and Methods
The thin-film photocatalysts were prepared using aqueous dispersions of TiO 2 , MoO 3 , WO 3 , and V 2 O 5 (Sigma-Aldrich, St. Louis, MO, USA). The aqueous sol of titania was prepared by adding 12.5% NH 4 OH dropwise to 2.5 M TiCl 4 + 0.65 M HCl aqueous solution (Sigma-Aldrich, St. Louis, MO, USA) cooled to 0 • C under vigorous stirring until pH 5 was reached. The obtained precipitate after thorough washing with distilled water was dispersed by ultrasonic treatment. The size of the obtained TiO 2 particles (anatase) was of ca. 4 nm. The aqueous suspensions of MoO 3 and V 2 O 5 were synthesized via the thermally-induced polycondensation of corresponding oxo-compounds in aqueous medium. The oxo-acids used as precursors were prepared by acidification of aqueous solutions of sodium molybdate and sodium vanadate on a resin. The two-step synthetic route yielding oxide particles of submicron size [21] was used. At the first step, 0.5 M solutions of molybdic, vanadic, and tungstic acids were heated at 100 • C for 4 min; the resultant solution was diluted 1:5 to stop further nucleation as well as to provide dispersion of grown oxide particles and then incubated for 4 h at 100 • C (the solution volume was maintained constant).
The X-ray diffraction analysis of the samples was performed with the use of a PANalytical Empyrean difractometer with Cu K α radiation. The morphological investigations emploed SEM (LEO 1420, KARL ZEISS, Germany) and AFM microscope (NT-206, MICROTESTMACHINES, Gomel, Belaru).
As the UV light source in the photocatalytic measurements, the high-pressure mercury lamp Philips HPK 125 W was used. The intensity of the incident light was of ca. 10 mW/cm 2 .
The photoinduced oxidation activity of TiO 2 /MoO 3 and TiO 2 /V 2 O 5 systems was evaluated using the reaction of degradation of probing dye (Rhodamine 6G) deposited onto the photocatalyst surface which was preliminary exposed to UV light for 10 min. Rhodamine 6G was applied onto the photocatalyst surface at different times after exposure in an amount of ca. 2 × 10 −8 mol/cm 2 , and a diffuse reflectance, R, at 530 nm that corresponded to the absorption of the dye in the adsorbed state was measured. The value proportional to the surface dye concentration (Γ) was calculated from the reflectance data using the Kubelka-Munk function [33]: The photocurrent onset potentials for the TiO 2 , MoO 3 , WO 3 , and V 2 O 5 thin films deposited onto ITO were obtained from photocurrent versus potential dependencies measured in 0.25 Na 2 SO 4 + 0.1 CH 3 COONa electrolyte using the Autolab PGStat 204 potentiostat (Metrohm Autolab, Utrecht, The Netherlands). All potentials are given against saturated Ag/AgCl reference electrode.
The photoinduced pathphysiological activity of photoaccumulating catalyst was investigated using Escherichia coli ATCC 8739 bacteria. For this purpose, the samples of thin-film photocatalysts at glass substrates (4 × 4 cm) pasted with the agar at the bottom of Petri dish were exposed to UV light for 10 min and left in the dark for different times to evaluate the deactivation kinetics. Then, 5 mL of suspension of microorganisms in 0.5% LB agar medium was poured into the dish. Plates were then incubated for 48 h at 37 • C, and the number of arising colonies, N, was calculated. In the control experiments, the photocatalyst-free glass substrates were used. To evaluate the biocide activity of the pre-illuminated photocataylys, the survived ratio, S = (N/N 0 )·100%, was used (N 0 is the number of colony-forming units in the case of the nonexposed photocatalyst used as the control sample). The respective data were the average values obtained from the triplicate runs. The standard deviations of these replicate experiments were within 14%.
The EPR spectra were recorded with a EPR spectrometer (ELEXSYS-E500, X-band, the sensitivity up to 10 10 spin/G, Bruker, Karlsruhe, Germany). For investigation of photogenerated paramagnetic centers (PCs), the samples were illuminated directly in the cavity of the spectrometer with the use of a 50 W high pressure mercury lamp. The concentration of PCs was evaluated using CuCl 2 ·2H 2 O monocrystal with the known number of spins as the standard. The EPR spectra simulation permitting the determination of g-factor values of paramagnetic centers was carried out with the use of EasySpin MATLAB toolbox (Comprehensive Software Package for Spectral Simulation and Analysis in EPR, Stoll, S.; Schweiger, A., Physical Chemistry Laboratory, ETH Hönggerberg, Zürich, Switzerland, 2006) [34].

Conclusions
The obtained results provide evidence that the structure of energy levels involved in the separation of photoproduced charges in the case of TiO 2 /MoO 3 photocatalyst creates favorable conditions for filling the energy states of TiO 2 (the latter process is not observed for bare TiO 2 ). As a result, the accumulation of photoinduced charges occurs in two forms: (i) filled electron traps corresponding to Ti 4+ /Ti 3+ levels and (ii) Mo 5+ centers produced as the result of electron trapping by MoO 3 , both forms contributing to the photoinduced biocide activity of the heterostructure photocatalyst. This effect of "indirect" filling of Ti 4+ /Ti 3+ surface states resulted from the cascade effect is not observed for TiO 2 /WO 3 and TiO 2 /V 2 O 5 heterostructures due to deeper lying W +6 /W +5 and V 5+ /V 4+ levels involved in the electron trapping. In the case of TiO 2 /WO 3 and TiO 2 /V 2 O 5 photocatalysts, the electron trapping is irreversible and occurs at the bulk W +6 /W +5 and V 5+ /V 4+ states. This permits the conclusion that the exposure of energy storage heterostructure photocatalysts with mosaic surfaces (i.e., comprising photogenerating and redox-active components) to actinic light can result in the generation of different types of charge-trapping centers capable of interacting with molecular oxygen yielding peroxo species; these centers differ as to their long-term stability and contribution to the induced pathophysiological behavior.