H₂O₂ is not only a high-value industrial chemical, in growing demand on the global market, but also a potentially attractive energy carrier [1, 2]. H₂O₂ is widely used in chemical synthesis, disinfection, wastewater treatment, pulp, etc. [3] Moreover, due to the advantage of a simple membrane-free structure, hydrogen peroxide fuel cells have attracted extensive attention [4, 5]. The traditional method for H₂O₂ production has been the anthraquinone process, which is an indirect batch method requiring sequential hydrogenation, oxidation of a 2-alkylanthraquinone, and extraction of H₂O₂ from organic solvents [6]. However, this multistep method is hazardous, energy-intensive, and difficult for in situ H₂O₂ production. Thus, extensive efforts have been devoted to achieving direct synthesis of H₂O₂ by eco-friendly and low-cost processes [7]. H₂O₂ production by photocatalytic methods, which are powered by clean solar energy, is considered one of the most promising approaches. To date, photocatalysis has been studied in many fields, including hydrogen production, CO₂ reduction, and water treatment [8-24]. However, there is still a great deal of potential for photocatalytic H₂O₂ production. Recently, intense efforts have been made to achieve H₂O₂ production by photocatalytic O₂ reduction [25, 26]. However, research concerning simultaneous H₂ and H₂O₂ production during water splitting is scarce. Compared with the traditional four-electron transfer process, converting water to H₂ and H₂O₂ by a two-electron/two-proton oxidation process is considered as the ideal method, in which all the products are value-added chemicals [27]. Moreover, the system avoids the problem of separation because H₂ and H₂O₂ exist in gas and liquid phases, respectively, which is a significant advantage over the H₂ and O₂ production system. To date, fewer catalysts are known for the selective two-electron oxidation of water to produce hydrogen peroxide, compared with the four-electron oxidation of water to produce oxygen [28, 29]. To the best of our knowledge, two systems, Pt/P25 [30] and Pt/C₃N₄ [31] have been reported for photocatalytic hydrogen and hydrogen peroxide production from pure water. However, the catalytic efficiency is still very low. The H₂ and H₂O₂ evolution rates are 42.75 μmol g^-1 h^-1 (first 30 min) and 34.38 μmol g^-1 h^-1 (average of 2 h), and 10.80 μmol g^-1 h^-1 (average of 25 h) and 13.80 μmol g^-1 h^-1 (average of 25 h), for Pt/P25 and Pt/C₃N₄, respectively.

In this study, we introduce Pt/TiO₂(anatase) as a photocatalyst, which produces hydrogen in the gas phase, and H₂O₂ on the surface of catalyst and in solution. The H₂ and H₂O₂ production efficiencies can reach up to 7410 and 5096 μmol g^-1 h^-1 (first 1 h), respectively, without the assistance of a sacrificial agent. The H₂ production efficiency is 173 and 686 times higher than those of the Pt/P25 and Pt/C₃N₄ systems, respectively. Additionally, the H₂O₂ production efficiency is 148 and 369 times higher than those of the Pt/P25 and Pt/C₃N₄ systems, respectively. Therefore, it is clear that the present system is promising for production of high value-added chemicals by photocatalytic water splitting.

All chemical reagents were analytical grade and used without further purification. Ethylalcohol, methanol, silver nitrate (AgNO₃), ammonia solution and titanium tetrachloride (TiCl₄) were obtained from Beijing Sinopharm Chemical Reagent Co., Ltd. H₂PtCl₆ was obtained from J & K Scientific Ltd.

A solution of 0.2 mol/L TiCl₄ was slowly dropped into a 0.5 mol/L ammonia solution (50 mL) under vigorous stirring at room temperature until the pH value of the solution reached ~8. A white precipitate was formed and then centrifuged and washed with distilled water until chloride ions were not detectable in the washed water (1.0 wt% AgNO₃ solution). A quarter of the precipitation was dispersed in 50 ml of anhydrous ethylalcohol by ultrasonic treatment for 20 min in an autoclave, and was kept at 150 ℃ for 5 h. The products were collected and washed with deionized water three times. After that, the samples were dried and calcined at 400 ℃ for 2 h. The Pt NPs were loaded by an in-situ photo-depostion method using H₂PtCl₆ (Pt 1wt% in the whole sample) as the precusor and MeOH as the electron donor (10%). After 1 h of UV light irradiation, the sample was collected by centrifugation and dried at 80 ℃ overnight to remove the remaining MeOH and H₂O.

The prepared Pt/TiO₂ hybrid (1 mg) was dispersed in 20 mL deionized water in a quartz reactor and sealed with a rubber septum. After purging with argon flow for 30 min to remove dissolved air, the suspensions were irradiated by a 500 W Hg lamp under magnetic stirring at room temperature. The evolved H₂ in the gas phase was examined by a Techcomp gas chromatograph (GC-7900) with a thermal conductivity detector (TCD), 5 Å molecular sieve columns, and Ar carrier.

The hydrogen peroxide measurement was performed in 2.0 mL reaction suspension. A 0.5 mL volume indicator (1% o-tolidine in 0.1 mol/L HCl) was added to the suspension and the mixture was allowed to react for 10 min. The color of the mixture turns to blue. Subsequently, 2.0 mL HCl (1.0 mol/L) was added to the mixture, and the color turns to yellow from blue. Immediately, the mixture was centrifuged and the absorption spectrum of the liquid supernatant was recorded with a UV-vis spectrophotometer. The absorption spectrum of the 2-electron oxidized toluidine has a characteristic maximum at 437 nm.

UV-Vis spectra were measured with a Hitachi U-3900 spectrophotometer. Scanning electron microscopy (SEM) images were taken by using a Zeiss electron microscope. Transmission electron microscopy (TEM) images were obtained by using a 2100F (JEOL) electron microscope at an acceleration voltage of 200 kV. X-ray diffraction (XRD) of the samples was performed with a SmartLab Goniometer with Cu K-beta radiation, operating at 40 kV and 200 mA. X-ray photoelectron spectroscopy (XPS) analysis was carried out on a Thermo Scientific ESCALAB 250Xi spectrometer (all the peaks corrected with reference to the C signal (284.8 eV)). Raman spectrum was recorded on a Renishaw inVia Raman microspectrometer and laser excitation wavelength was 633 nm. The static fluorescence spectral measurements were carried out with an F-4600 (Hitachi) spectrofluorometer. The ESR measurements were conducted by a Bruker ESP-300E spectrometer at 9.8 GHz, X-band, with 100 Hz field modulation.

The valence band of titanium oxide was calculated by the theory of elemental electronegativity. The valence band is calculated by using element electronegativity and material's band gap which comes from the UV-vis diffuse reflectance spectra.

Where χ(A) is the absolute electronegativity of element A, E_g is the band gap, and E₀ is the reduction potential of water (4.5 eV).

The anatase TiO₂ nanoparticles (NPs), with an average diameter of 28.9 ± 3.9 nm, were obtained by an improved hydrothermal strategy, in which TiCl₄ was selected as the Ti precursor [32]. To construct the Pt/TiO₂ hybrid, Pt nanoparticles (NPs) were loaded onto the TiO₂ NPs by in situ photo-deposition [33]. The weight of the Pt NPs was 1.0 wt% compared with the TiO₂ NPs. Both XRD and Raman spectra show a single-phase diffraction pattern assigned to anatase TiO₂ (Fig. 1a) (JCPDS No. 21-1272). Five main diffraction peaks near or at 2θ = 25.3°, 37.8°, 48.0°, 53.9°, and 55.1° are observed, corresponding to the (101), (004), (200), (105) and (211) diffraction planes of anatase TiO₂, respectively [34]. The Raman shifts of 397, 516 and 639 cm^-1 correspond to anatase TiO₂ (Fig. S1) [35]. The morphology of the TiO₂ catalyst was characterized by SEM and TEM (Fig. 1b-f). As shown in Fig. 1b, the TiO₂ NPs display a spherical morphology with an average diameter of 28.9 ± 3.9 nm, from the SEM image statistics. The lattice spacing is 0.35 nm, which is assigned to the (101) crystal plane of anatase TiO₂, from the TEM images (Fig. 1c-d). For the Pt/TiO₂ hybrid (Fig. 1e), the Pt NPs, with a size of 2 nm, are well dispersed onto TiO₂. The d-spacings of 0.35 and 0.23 nm correspond to the (101) and (111) planes of TiO₂ and Pt, respectively [36].

The photocatalytic performance of the catalysts was evaluated under Hg lamp irradiation (λ ≥ 300 nm) at room temperature. As shown in Fig. 2a, our system achieved the highly efficient H₂-production activity of 7410 μmol g^-1 h^-1 for the first 1 h, in deionized water, without an electron sacrificial agent. The produced H₂O₂ was measured using the peroxide indicator o-tolidine, after a period of irradiation, and a typical absorption peak at 437 nm was easily detected using UV-vis spectroscopy [31]. As depicted in Fig. S2, the H₂O₂ production efficiency is 5096 μmol g^-1 h^-1. In contrast, the commercial Pt/anatase TiO₂ hybrid only exhibits an efficiency of 3041 μmol g^-1 h^-1 and 2644 μmol g^-1 h^-1 for H₂ and H₂O₂, respectively (Fig. 2a). It should be noted that the commercial anatase TiO₂ NPs have a similar morphology to our system (Figs. S3-5), and, as shown in Fig. S6, the average pore sizes and Brunauer-Emmett-Teller (BET) surface areas were determined to be 18.2 nm and 46.0 m²/g for the present sample, and 17.8 nm and 53.8 m²/g for the commercial anatase TiO₂ NPs, respectively, by N₂ adsorption. The results indicate that the difference in surface area does not lead to the difference in photocatalytic efficiency. Therefore, compared with the commercial Pt/anatase TiO₂, the present system demonstrates significant advantages not only in H₂ evolution, but also in H₂O₂ production. To demonstrate the potential value in practical applications, a photocatalytic stability experiment was carried out. As depicted in Fig. 2b, the photocatalytic hydrogen evolution activity could be sustained for more than 15 h, and more than 61723 μmol/g H₂ was produced. The amount of H₂O₂ initially increased, and then decreased with an increase in irradiation time. It was proposed that H₂O₂ could decompose to H₂O and O₂ under prolonged irradiation. However, no oxygen was detected during the photocatalytic reaction, and hydrogen peroxide (H₂O₂) was detected as the only oxidative product. The produced O₂ could be easily reduced by the photo-generated electrons, and further mechanistic studies are ongoing. Collectively, these results indicate that the present photocatalytic system is highly efficient and promising for H₂ and H₂O₂ production without separation problems.

For the water splitting reaction (2H₂O → H₂ + H₂O₂), it is widely accepted that H₂ is generated on Pt [37]. For the mechanism of H₂O₂ formation, it has been proposed that H₂O₂ is mainly derived from the recombination of hydroxyl radicals (·OH). During the photocatalytic reaction, ·OH is confirmed as the crucial reactive intermediate. To gain insight into the process, XPS measurements were first used to investigate the surface properties of the TiO₂ NPs (Figs. S7-10). Before irradiation, TiO₂ and Pt/TiO₂ samples show two main peaks, corresponding to surface lattice oxygen (529.7 eV, 529.8 eV) and surface hydroxyls (531.6 eV, 531.9 eV). At the same time, for the Pt/TiO₂ sample, a weak peak near 533.3 eV, associated with physisorbed water is observed. After 1 h of irradiation, there is a distinct change in the spectrum. The peak at 533.2 eV, associated with physisorbed water, is greatly enhanced, and the hydroxyl peak is significantly increased, compared with the other samples (Fig. 3a) [38, 39]. In contrast, Ti atoms demonstrate a similar bonding environment before and after irradiation (Fig. 3b). The binding energies of Ti 2p_3/2 and 2p_1/2, 458.6 and 464.4 eV, respectively, are the characteristic peaks of Ti⁴⁺ in TiO₂ [40]. Therefore, it is proposed that the hydroxyls and adsorbed water have a significant impact on the formation of H₂O₂. In order to capture the reactive intermediate during the irradiation reaction, electron spin resonance (ESR) and fluorescence labeling measurements were performed. As shown in Fig. 3c, a characteristic quartet of peaks (1:2:2:1, a_N = a_H = 15.4 G) associated with the 5, 5-dimethyl-1-pyrroline-N-oxide-OH (DMPO-OH) radical adduct is immediately detected after 60 s of UV light irradiation [41]. Additionally, when terephthalate (TANa) was added to the system, an obvious fluorescence signal at 422 nm, assigned to 2-hydroxyterephthalic acid (TAOH), was detected (Fig. 3d) [42]. The ESR and fluorescence results imply that the generated ·OH radicals play a significant role in the formation of H₂O₂ [30]. The bandgap and valence band for the present anatase TiO₂ are 3.21 and 2.97 eV, respectively, confirmed by diffuse reflectance UV-vis spectroscopy (Fig. 3e). The oxidation potentials E_ox(·OH/OH^-) and E_ox(·OH_ads/OH_ads^-) are 1.89 and 1.5 V versus the NHE, respectively [43, 44]. Therefore, the H⁺ photogenerated by the anatase TiO₂ has enough oxidation potential to convert ·OH to OH^- and OH_ads^- to ·OH_ads. Furthermore, compared with the commercial anatase TiO₂ NP sample, the present anatase NPs possess more hydroxyls and adsorbed water (Fig. S11). Therefore, it can be concluded that the surface properties, and the quantities of hydroxyls and physisorbed water are related to the catalytic performance. Based on the above analysis, the reaction mechanism is depicted in Fig. 3f. To date, different oxidation products have been reported for photocatalytic water splitting using anatase. No oxygen was detected, or the final ratios of H₂/O₂ were below 2 [28, 29, 45]. Therefore, water oxidation products other than oxygen play a significant role. In the present Pt/TiO₂ (anatase) system, we detected H₂O₂ as the water oxidation product, and further mechanistic investigation is underway.

We have developed a much more valuable and efficient photocatalytic water splitting system by converting water to H₂ and H₂O₂ simultaneously. The stable and easily-prepared Pt/TiO₂(anatase) hybrid is introduced as the catalyst. The H₂ and H₂O₂ production rates can reach up to 7410 and 5096 μmol g^-1 h^-1. More importantly, H₂ and H₂O₂ are separated automatically, avoiding additional separation costs. The excellent activity is attributed to the more favored two-electron oxidation of water to H₂O₂, compared with the four-electron oxidation of water to O₂. Therefore, photocatalytic oxidation of water to H₂O₂ is a promising method to achieve large-scale H₂ and H₂O₂ production without a hazardous and energy-intensive process.