Correlation between Photocatalytic Properties of ZnO and Generation of Hydrogen Peroxide — Impact of Composite ZnO/TiO 2 Rutile and Anatase

: The generation of hydrogen peroxide on commercial and synthesized ZnO from different precursors was studied using two model molecules, formic acid (FA) and phenol (Ph), as well as phenolic intermediates, hydroquinone (HQ), benzoquinone (BQ)


Introduction
Water pollution is becoming a universal threat that leads to environmental degradation and poses a risk to humans and the environment worldwide. Advanced Oxidation Processes (AOP) are emerging as an alternative and promising technology for wastewater pollution control [1,2], they are based on the formation of hydroxyl radicals (°OH) exhibiting a higher oxidizing power able to break down organic molecules into degradable molecules or mineral compounds [3,4]. Among different methods, heterogeneous photocatalysis was the most promising process for water problems [5][6][7][8]. Under UV light, the semiconductor generates electron/hole pairs (e − /h + ) leading to highly Reactive Oxygen Species (ROS): the holes (h + ) react with the electron donors adsorbed on the surface to form hydroxyl radicals (°OH), the oxygen is reduced by electrons, producing superoxide (O2°) radicals, and it is in the presence of protons becoming hydro-peroxide (HOO°) radicals [9,10]. Many studies focused on understanding the generation of hydroxyl radicals (°OH) via the oxidation route by trapped holes, assumed to be the most important factor for oxidation processes in photocatalysis [11][12][13][14]. However, hydrogen peroxide (H2O2) has evolved for years as an interesting reservoir for the generation of hydroxyl radicals and a key parameter for understanding photocatalytic mechanisms [15][16][17]. Under UV irradiation, H2O2 can be formed, either directly by the reaction of photo-generated electrons with adsorbed molecular oxygen or by the reduction in the superoxide radicals (O2°−), and the hydro-peroxide (HOO°) already present in the system can be further reduced and transformed into H2O2 [18,19]. Depending on the photocatalyst, H2O2 can be continuously generated, as is the case in the presence of ZnO [20,21], or decomposed and generally cannot be detected in the solution as in the case of TiO2 [22][23][24]. This difference was attributed to the complexation of H2O2 on TiO2 [25], which is not as in the case with zinc oxide. Moreover, the structure of TiO2 also plays a role in the rate of disappearance but also in the nature of reactive oxygen species generated. Sahel et al. [26] showed that hydrogen peroxide was more rapidly decomposed in the rutile phase compared to the anatase phase. Moreover, Hirakawa et al. [27] found that O2°− species were formed during the decomposition of H2O2 in the anatase phase while hydroxyl radicals (HO°) were preferentially generated in the rutile phase. On ZnO, Jang et al. [28] reported that the generation of H2O2 depends on the morphology of the crystal and that the most active ZnO generates the highest amount of H2O2. Furthermore, Domenech et al. [29] reported that the type of electron donors, such as phenol, oxalate, and 2,4-dinitrophenoxyaceticacid have a significant effect on the production of H2O2 in agreement with the role of the hole scavengers necessary for the formation of H2O2 [30]. Mrowetz et al. [31] studied the formation of H2O2 during the degradation of formic acid (FA) and benzoic acid (BA) on TiO2 and ZnO and confirmed the higher formation of H2O2 on ZnO. They suggest that the more important reactivity of TiO2 on the degradation of formic acid, while ZnO is the best on the degradation of benzoic acid, could be linked to the decomposition of H2O2 in the presence of FA, whereas in the presence of BA the formation of phenolic intermediates adsorbed on TiO2 avoids the adsorption of H2O2 and its decomposition. To better understand the role of H2O2, several works studied the impact of the addition of H2O2 on the photocatalytic properties of TiO2 and ZnO [32][33][34][35][36][37][38]. Whereas Khodja et al. [33], Poulios et al. [35], Barakat et al. [37], and Domingues et al. [38] showed a beneficial impact, Pichat et al. [32], Dionysiou et al. [36], and Daneshvar et al. [34] reported that this benefit depended on the ratio between the H2O2 concentration and pollutant concentration. In the presence of a high concentration of H2O2, the °OH radical can react with H2O2 and have a harmful effect. In summary, an understanding of the photocatalytic behavior of hydrogen peroxide is still far from being reached. In most previous studies, researchers considered the formation process of hydrogen peroxide to clarify reaction mechanisms that are occurring, tried to explain why it is detected and why not when using semiconductors, and proposed some possible roads of its generation. However, to our knowledge, the correlation between the disappearance of pollutants and the formation of spontaneous H2O2 on different ZnO as well as the impact of TiO2 composite on its formation has not been established.
The originality of our work was to investigate if a correlation could be established between the photocatalytic activity of the different ZnO and the H2O2 formation independent of the organic compound used. Furthermore, to investigate the impact of the presence of TiO2, rutile, and anatase mechanically mixed with ZnO on the hydrogen peroxide formation and on the initial intermediate formed during the photocatalytic degradation of phenol to check if the known TiO2 rutile generated °OH radicals by H2O2 decomposition was the best compared to known TiO2 anatase generated O2° that is less efficient.

Catalysts Characterization
The impact of the precursor was first determined in relation to the morphology of the ZnO obtained ( Figure 1). As shown, in all the cases after calcination at 400 °C, a spherical morphology is obtained with well-dispersed particles and an average particle size between ~14 ± 2 nm and ~27 ± 2 nm. However, commercial ZnO presents larger particles (37 nm) with non-homogeneous morphologies. The structural properties of the as-prepared ZnO samples were characterized using X-Ray Diffraction. As shown in Figure S1, the crystalline phase of the two precursors indicates the formation of Zn(OH)2 and ZnO2 without any impurities. The XRD spectra of all ZnO samples obtained from different precursors, together with commercial ZnO as reference are presented in Figure 2a. In all the cases and at the same calcination temperature, the samples present the same phase composition with characteristic planes (100), (002), and (101), corresponding to a pure phase Hexagonal Wurtzite of ZnO (JCPDS 36-1451). The crystallite size of the resulting ZnO from zinc acetate, zinc nitrate, and zinc peroxide was calculated using Scherer's equation [39] and values of 19 nm, 23 nm, and 14 nm, respectively, were determined. For commercial ZnO, bigger crystallites (size: 40 nm) are obtained in good agreement with the morphological characteristics shown in Figure  1. The corresponding surface areas and the crystallite size of different samples are presented in Table 1. The change of precursor does not induce a change in the morphology remaining spherical of the resulting samples. However, the use of different types of precursors affects the defect states of ZnO samples, which can be observed using electron paramagnetic resonance (EPR) and Raman (Figure 2b,c).     Figure 2b shows the normalized spectra of commercial ZnO and homemade ZnO, except for ZnO prepared from zinc acetate that exhibits an excessive fluorescence. As shown, the characteristic peaks located at 98 cm −1 , 202 cm −1 , 330 cm −1 , and 437 cm −1 , represent typical vibration modes of ZnO Hexagonal Wurtzite, which correspond to E2 low , 2E2 low , E2 high − E2 low , and E2 high modes, are identified in all the cases. On the other hand, an additional vibration mode at 580 cm −1 was also observed. This peak is recorded to (E1 Lo ) mode and is related to the oxygen vacancies appearing in the structure of ZnO. The difference in defect contribution, represented by the E1 Lo mode, is observed between catalysts. The highest level exhibiting the dominance of oxygen defects in the structure is recorded with the sample prepared from zinc peroxide. To characterize defects and investigate the effects generated by using different types of precursors, Electron Paramagnetic Resonance (EPR) was employed ( Figure 2c).
In all cases and for all samples, there is a first-order EPR signal at g = 1.96. As reported, this resonance was explained by the existence of oxygen vacancies in the structure of ZnO [40,41]. As shown, an additional paramagnetic resonance is detected at g > 2 attributed to zinc vacancies [42][43][44][45], with a different distorted environment as previously observed in the literature [46]. Figure 2d shows the bandgap energy of the ZnO samples, determined using the Kubelka-Munk method by comparing all the samples, a difference in their optical properties is observed. The commercial ZnO exhibits a bandgap value of 3.2 eV, compatible with the value indicated in the literature [47]. However, with the use of different precursors, the bandgap energy of the samples decreases to 3.15 eV (ZnH4T), 3.1 eV (ZnA4T), and 2.9 eV (ZnP4T), respectively. Combined with the Raman characterization, these changes are consistent with the increase of the concentration of oxygen vacancies generated in the ZnO structure, which is more important with zinc peroxide as a precursor and therefore exhibits the lowest energy of bandgap.
The XRD of the commercial TiO2 P25, A100, and R100 are given in Figure S2. As shown, pure anatase and rutile samples are, respectively, present in A100 and R100, while for P25 a mixture of the anatase and rutile phase is well observed.
The optical properties of the different samples of TiO2, including the mixture of ZnOanatase and ZnO-rutile together with commercial ZnO, are provided in Figure S3. The bandgap energy of TiO2-based materials is situated between ~3 eV and ~3.2 eV, in the same range with a bandgap value of ZnO (~3.2 eV). As expected, a smaller bandgap is observed with the rutile phase (~3.01 eV). On the other hand, the composites ZnO/TiO2 (anatase and rutile) do not significantly affect the optical properties of the obtained samples. A slight modification is observed by combining with the rutile phase (smaller bandgap ~3.1 eV), which is in agreement with the result of the rutile only, which is close to the visible range.

Characterization of Radicals Intermediates Using EPR
The EPR spectra of ZnO-Com, ZnA4T, ZnH4T, and ZnP4T obtained under UV irradiation for 10 min in aqueous solutions and in the presence of DMPO are presented in Figure 3. For all precursors, a strong signal (marked by @) attributed to °OH is observed (92-100% of the total signal). This species (aN = 14.89 G and aH = 14.89 G) is assigned using a simulation ( Figure S4) [48][49][50][51][52]. Moreover, except for ZnP4T, an additional species is detected at aN = 15.6 G and aH = 19 G (marked by *). This signal is a C-centered (or CO2°) species and could be attributed to the presence of an organic impurity at the surface of ZnO [48,51,52]. After the addition of formic acid (FA) or phenol (Ph) as model pollutants and after UV irradiation, the evolution of the °OH EPR signal is different between FA and Ph ( Figure S5). In the case of formic acid, the °OH EPR signal completely quenches and the organic species generated are the same for all photocatalysts suggesting that the mechanism of the decomposition reaction is similar ( Figure S5a). However, in the case of phenol, a difference is observed between the photocatalysts indicating a difference in the reaction process ( Figure S5b). The EPR signal of organic species is totally absent with ZnP4T, while it exhibits a significant signal of °OH species in the presence of Ph. We also note the absence of the °OH signal in the case of commercial ZnO, while it is always present with a percentage of 30% in the case of ZnA4T ( Figure S6  An estimation of °OH radicals has been performed using the calibration curve provided in Figure S7. The values were provided in Table 1.

Comparison of the Formation of H2O2 from FA and Ph in the Presence of ZnO-Com
In the absence of a pollutant, H2O2 is formed for about 1 h and then it reaches a pseudo plateau, whereas, in presence of a pollutant, the formation of H2O2 is more important ( Figure 4). The formation of H2O2 in water, with no added hole scavengers, is explained considering the reaction of (e − , h + ) pairs formed with oxygen and water following Equations (1)- (6): O2 + e − → O2°− (1) Phenol However, in parallel to the formation of H2O2, the hydroxyl radicals are formed which, in the absence of a pollutant, can react with H2O2 decreasing its formation explaining the pseudo plateau observed in Figure 4a.
H2O2 + °OH → H2O + HO2° In the presence of pollutants, the formation of H2O2 is more important. This behavior has already been observed by some authors [29,30]. As suggested, the photocatalytic generation mechanism of hydrogen peroxide occurs mainly via the reduction in adsorbed oxygen. Accordingly, the increase of H2O2 in the presence of pollutants is explained by a decrease of charged recombination due to the reaction of pollutants with °OH or with h + favoring the reduction of O2 into O2°−. From Figure 4b, it can be noticed that a higher disappearance of phenol than that of HCOOH, led to a higher formation of H2O2. Considering the higher initial concentration of carbon in phenol solution, 1276 µmol/L against 1086 µmol/L for FA, the final amount of H2O2 formed seems to be proportional to the initial concentration of carbon present in the solution. This result agrees with the formation of H2O2 obtained for two concentrations of phenol 20 ppm and 30 ppm corresponding to 766 µmol and 1149 µmol of carbon in the 600 mL of solution, respectively ( Figure 5). Moreover, it can be seen from Figure 6 that decreasing the ZnO concentration from 1 g/L to 0.2 g/L has no effect on the generation of H2O2 and the photocatalytic degradation of phenol. This result shows that 0.2 g/L of catalyst was enough to absorb all the photons and reach about 95% of efficiency after only 180 min of irradiation. In summary, a modification of the ZnO concentration from 0.2 g/L to 1g/L does not lead to any change in the efficiency of the degradation and the generation of H2O2. Moreover, the final concentration of H2O2 formed seems independent of the nature of the pollutant and only depends on the amount of initial carbon present in the solution. Thus, to better understand the mechanism of H2O2 formation in the presence of these two pollutants, the amount of H2O2 formed in presence of commercial ZnO during UV irradiation is reported as a function of the number of carbons removed from HCOOH or phenol (Figure 7).
. remaining in the solution during phenol degradation was reported in Figure 8a. The intermediate products formed during the degradation of phenol are initially aromatic compounds, hydroquinone (HQ), benzoquinone (BQ), and catechol (CAT) [53][54][55]. Then, some carboxylic acids are formed similar to those observed in the case of TiO2 [56] but not quantified in this study. Figure 8b represents the number of H2O2 formed as a function of Total Organic Carbon (TOC) removed from the HCOOH or phenol. The results indicate that the number of moles of H2O2 formed per number of carbon atoms removed is similar to the degradation of these two molecules (formic acid and phenol). After removing the amount of H2O2 formed in the absence of a pollutant, we found that for 1 mole of carbon removed about 1 mole of H2O2 is formed along the reaction for these two pollutants.

Influence of the Nature of ZnO Samples on H2O2 Formation
After investigating the formation of H2O2 in the presence of ZnO-Com and showing that the formation of H2O2 seems to be correlated with the disappearance of the number of carbon during the degradation of FA and phenol with a factor of about 1 along the degradation, we investigated the behavior of ZnA4T, ZnH4T, and ZnP4T concerning the formation of H2O2 and the disappearance of carbon during the degradation of FA and Ph. The formation of H2O2 and the disappearance of FA and Ph in the presence of different ZnO samples are reported in Figure 9. With whichever ZnO samples, H2O2 is detected during the degradation of FA and Ph. In the absence of a pollutant, the amount of H2O2 initially increases and reaches a plateau of around one hundred µmol. However, it is difficult, in the absence of a pollutant, to determine the difference in H2O2 formation between the different catalysts although it seems that less H2O2 was formed with the sample ZnP4T. We noticed that the higher disappearance of FA and Ph, the greater the formation of H2O2 as previously observed with ZnO-Com. So, it means that the formation of H2O2 depends on the same physicochemical parameter as those influencing the photocatalytic properties. The results are in agreement with the EPR results of ZnO-com in the presence of phenol (Figure S6a), which explains its much higher activity. In the literature, Jang et al. [28] found that the highest activity for H2O2 generation is obtained with an exposed (001) polar face and depends on the morphology of the crystal. However, in our case, the different ZnO samples have similar morphologies and the same exposed face and cannot explain the difference observed. In agreement with our previous publication [57], the differences observed can be explained by considering the harmful effect of defects for photocatalysis. The defects decrease from ZnP4T to ZnH4T and ZnO-com. In the case of ZnA4T, the higher amount of H2O2 observed can be explained by considering its large surface area. Moreover, whatever the ZnO sample, the ratio between the number of H2O2 formed and the number of carbon atoms removed is around 1, in the presence of FA and Ph (Figure 10), as previously obtained for ZnO-Com. It is also interesting to note that the highest photocatalytic performance and the highest H2O2 generation is obtained with the ZnO samples with less oxygen vacancies, which is in agreement with our previous results [57].
(a) (b) Figure 10. Moles of H2O2 formed as a function of moles of HCOOH (a) and phenol (b) degraded.

Photocatalytic Decomposition of H2O2
The production of hydrogen peroxide shown in Section 2.3 has been studied in the presence of various pollutants (FA and Ph) and the impact of using different ZnO was also investigated. For this section, only phenol is considered. Taking into account the different behaviors of ZnO and TiO2 anatase and TiO2 rutile toward the formation and the decomposition of H2O2 [27], we have mixed 50% of ZnO Com with 50% of TiO2 anatase (and rutile) to determine the impact of both of these TiO2 phases on the formation of H2O2 generated in the presence of ZnO and on the formation of the first intermediate products, hydroquinone, catechol, and benzoquinone. The formation of H2O2 and the first intermediate products (hydroquinone (HQ), benzoquinone (BQ), and catechol (CAT)) formed during the degradation of phenol in the presence of ZnO alone or mixed with TiO2 anatase and TiO2 rutile are presented in Figure 11a  On the composites ZnO-TiO2 (anatase) and ZnO-TiO2 (rutile), a decrease in the amount of H2O2 generated compared to ZnO was noticed, which is slightly more important in the presence of ZnO-TiO2 (rutile). The smaller formation of H2O2 in the presence of composites is due to its decomposition on TiO2 and the difference observed in the concentration of H2O2 generated with both of the composites is in good agreement with our previous work indicating the more important decomposition of H2O2 in the presence of the rutile phase [26]. Moreover, in Figure 11b it can be observed that at the same conversion of phenol, the yield of phenolic intermediates in the presence of the two composites was lower compared to this one obtained in the absence of TiO2. These results confirm the interaction of H2O2 generated with the accessible sites of TiO2 to form additional radicals that can participate in the photocatalytic mechanism and enhance the oxidation of intermediate compounds. On the other hand, the yield of the phenolic intermediate is slightly lower for the ZnO-TiO2 rutile compared to the composite ZnO-TiO2 anatase. This difference could be attributed to the nature of the reactive oxygen species (ROS) derived from the decomposition of H2O2 under UV irradiation on TiO2 anatase and rutile. Some publications reported that °OH radicals are preferentially formed in the decomposition of H2O2 on rutile, whereas the production of O2°− is favored in the presence of TiO2 anatase [17,25,27].
The preparation of these three ZnO samples are reported below. The elaboration of ZnA4T:Zn(CH3COO)2·2H2O was precipitated u KOH in methanol with a ratio of 4. During synthesis, magnetic stirring was maintained and the temperature was fixed at 60 °C for 3 h under reflux. The resulting product was separated from a solution by centrifugation, washed with ethanol and water, dried at 60 °C, and finally calcined at 400 °C for 2 h under air flux with a heating rate of 5 °C·min −1 .
The elaboration of ZnH4T was formed by calcination at 400 °C for the Zn(OH)2 previously obtained. The ZnP4T was elaborated by mixing an aqueous solution of Zn(OH)2 and H2O2 (1M), which is kept under stirring at 70 °C for 2 h. Then, the precipitate was separated and washed several times by centrifugation, then heated at 60 °C overnight. The obtained ZnO2 was further calcinated at 400 °C for 2 h to obtain ZnP4T.

Characterization
The as-prepared photocatalysts were characterized through X-ray Diffraction (XRD) using an A25 Bruker D8 Advance diffractometer (Billerica, MA, USA) with Cu-Kα radiation (λ = 0.15406 nm) at 40 KV, and a scanning range between 2θ = 4-80° with a scan rate of 0.02°/s. The UV-visible diffuse reflectance considering the Kubelka-Munk method provided the direct bandgap transition for ZnO: (αhѵ) 2 versus (hѵ) where α is the absorption coefficient and hѵ is the photon energy. The BET surface areas were measured on an ASAP 2020 instrument using nitrogen physisorption at 77 K. The catalysts were degassed at 200 °C for 3 h under vacuum. The Raman experiments were performed with a spectral resolution of 4 cm −1 and an Ar+ ion laser at 514 nm (Horiba Jobin Yvon LabRAM-HR equipment, Palaiseau, France). A CCD detector cooled at −75 °C was used. The morphology of the samples was carried out using Transmission Electron Microscopy (TEM) using a JEOL 2010 microscope (Tokyo, Japan) with 200 KV and the particle size distribution was determined using Image J (Fiji 2.0.0-rc-68/1.53t). The EPR assays were all carried out at room temperature using a Bruker E500 spectrometer operating at X-band (9.34 GHz), sensitive cavity, and with 100 kHz modulation frequency. The instrument settings were as follows: microwave power; 22 mW; modulation amplitude; 1 G. The hyperfine coupling constants (a and g values) were obtained with a simulation of experimental spectra using easyspin (Matlab 2016b (Massachusetts 01760 USA), Easyspin (easyspin-5.2.33)). The aqueous solutions of a spin trapping DMPO (5,5-Dimethyl-1-Pyrroline-N-Oxide, TCI chemicals) were prepared in capillary tubes. The irradiation (0.5-20 mn), Thorlab LED365 nm (Newton, New Jersey, USA), was directly performed in the EPR cavity while the spectrum was recording.

Photocatalytic Experiments and Analytical Procedure
For photocatalytic degradation tests, a 1 L Pyrex photo-reactor thermostated by water external circulation was employed. The formic acid (99% pure) and phenol (99% pure) were supplied, respectively, from Across Organics. For all the experiments, the concentration of the photocatalyst was set at 1 g·L −1 ; 600 mg of catalyst were added to 600 mL of a solution of formic acid (50 ppm) or phenol (20 ppm) with bottom magnetic stirring for 30 min in the dark to reach equilibrium. A pre-heated UV-A T8 8W diving lamp (Vilber Lourmat, Collégien, France) was positioned in a quartz tube in the middle of the reactor, providing a 2.4 mW/cm² irradiation; 1 mL of the solution was sampled and filtrated on an MILLEX HVLP 0.45 µm hydrophilic filter (Millipore, Burlington, MA, USA) for the HPLC analysis and H2O2 measurements.
For analytical characterization, high-performance liquid chromatography (HPLC) VARIAN PROSTAR (Agilent Technologies, Santa Clara, CA, USA) with an automated sampler was employed for formic acid analysis. An H2SO4 (5 × 10 −3 mol·L −1 ) mobile phase with a flow rate of 0.7 mL min −1 was used, equipped with a Coregel-87H3 column (300 mm × 7.8 mm-Concise Separations). For detection, a Prostar325 UV-Vis module was set at 210 nm. To follow the phenol disappearance and the aromatic intermediates, a 1290 Infinity HPLC system (Agilent Technologies, Santa Clara, CA, USA) equipped with a Nucleosil 250 × 4.6 mm 5 µm C18 column (Macherey-Nagel, Düren, Germany) was employed. The mobile phase was composed of 20% methanol and 80% of aqueous H3PO4 (1 mM) solution. A PDA detector was used at 210 nm wavelength. All of the HPLC analysis was performed at 40 °C.
The Total Organic Carbon (TOC) of phenol was quantified during photocatalytic tests at different times using a Shimadzu model TOC-VSCH with total organic carbon and equipped with an autosampler. The detection limit of the TOC analyzer is 0.5 mg·L −1 and the quantification limit is 1 mg·L −1 .
The hydrogen peroxide measurements were performed on a 6850 UV-Visible spectrometer from Jenway (Cole Parmer, Vernon Hills, IL, USA). For analysis, 500 µL of the solution was sampled and filtrated then mixed with 200 µL of ammonium molybdate ((NH4)2MoO4, 0.01 M), 300 µL of H2SO4 (1M), 2 mL of potassium iodide (KI, 0.1 M), and 7 mL of water. After being shaken, the solution rests 10 min before being measured at 361 nm.

Conclusions
The elaboration of ZnO samples from different precursors, zinc acetate, zinc hydroxide, and zinc peroxide were carried out, characterized by RAMAN and EPR and their photocatalytic performance were evaluated using two model molecules, formic acid (FA) and phenol (Ph).
We showed that the most efficient photocatalyst in the degradation of FA or Ph generates the most important amount of H2O2 correlated to the less important oxygen vacancies. The highest amount of H2O2 was recorded with commercial ZnO presenting the most important activity for the degradation of FA and Ph, while the sample from ZnO2 (ZnP4T) generated the lowest amount compared to ZnA4T and ZnH4T. Furthermore, our results show that the number of moles of H2O2 formed per number of carbon atoms removed during the degradation of FA and Ph was similar. In all cases and regardless of the precursor or the pollutant, a correlation was found between the amount of H2O2 generated and the amount of carbon removed by a factor of the order of 1.
We also proved that the presence of TiO2 decreases the amount of H2O2 generated with ZnO. This reduction is due to the decomposition of H2O2 on TiO2, which is accompanied by an improvement of the degradation of the first phenolic intermediates, more significant in the presence of rutile TiO2. These results are in good agreement with the formation of °OH and O2°− radicals in the presence of rutile TiO2 and anatase, respectively, and highlight the important role of °OH in the photocatalytic process.