H2O2 production on a carbon cathode loaded with a nickel carbonate catalyst and on an oxide photoanode without an external bias

Efficient H2O2 production both on a carbon cathode modified with various metal salts and on an oxide photoanode was investigated. The cathodic current density and faradaic efficiency for H2O2 production (FE(H2O2)) on a carbon cathode in KHCO3 aqueous solution were significantly improved by the loading of an insoluble nickel carbonate basic hydrate catalyst. This electrode was prepared by a precipitation method of nickel nitrate and KHCO3 aqueous solution at ambient temperature. The nickel carbonate basic hydrate electrode was very stable, and the accumulated concentration of H2O2 was reached at 1.0 wt% at a passed charge of 2500C (the average FE(H2O2) was 80%). A simple photoelectrochemical system for H2O2 production from both the cathode and a BiVO4/WO3 photoanode was demonstrated without an external bias or an ion-exchange membrane in a one-compartment reactor under simulated solar light. The apparent FE(H2O2) from both electrodes was calculated to be 168% in total, and the production rate of H2O2 was approximately 0.92 μmol min−1 cm−2. The solar-to-chemical energy conversion efficiency for H2O2 production (STCH2O2) without an external bias was approximately 1.75%.


Introduction
Hydrogen peroxide (H 2 O 2 ) is an environmentally friendly and important chemical oxidant that has been widely applied to pulp bleaching, waste treatment, chemical synthesis of peroxides, and food sterilization. [1][2][3][4][5][6][7][8] The anthraquinone redox process using H 2 and O 2 gases has been used for the industrial largescale production of H 2 O 2 . 2,9 However, there are many problems in this process: a large amount of energy requirement and H 2 consumption in the complicated system producing a huge amount of CO 2 emission, the usage of harmful organic solvents, the requirement of concentration, and transformation from the central process to consumption area. Therefore, some distributed and small-scale electrochemical H 2 O 2 production processes in aqueous solution have been widely investigated. 5,[10][11][12][13][14][15][16][17][18][19][20][21][22] The on-site electrochemical process has many advantages, such as safety without H 2 usage, no organic solvent separation, and controllability on the production amount and concentration for demand. Two kinds of reactions for H 2 O 2 production are present in the electrochemical process: reductive H 2 O 2 production from O 2 on a cathode (eqn (1)), and oxidative H 2 O 2 production from H 2 O on an anode (eqn (2)).
There are many reports on the reductive H 2 O 2 production from O 2 . [23][24][25][26][27][28][29][30] In contrast, the oxidative H 2 O 2 production and accumulation from H 2 O are very difficult. However, we and others have reported that H 2 O 2 can be accumulated when KHCO 3 aqueous solution is used in electrochemical and photoelectrochemical processes. [19][20][21][22][31][32][33][34][35] It is advantageous and highly efficient to produce H 2 O 2 on both electrode sides by combining the cathodic and anodic reactions, as shown in eqn (3), compared to the production on each electrode side under the same electric charge. The apparent faradaic efficiency for the H 2 O 2 production (FE(H 2 O 2 )) could reach 200% in total (100% + 100%) if the production occurred on both sides of the electrode. 36 We have previously demonstrated that the H 2 O 2 production can take place on a BiVO 4 /WO 3 photoanode and an Au cathode at near-neutral pH in KHCO 3 aqueous solution using a simple one-compartment cell without an external bias, as shown in Fig. 1. 19 The apparent faradaic efficiency was 140% in total (FE(H 2 O 2 ) ¼ 90% and 50% on the Au cathode and the BiVO 4 /WO 3 photoanode, respectively). The band gap of BiVO 4 is 2.4 eV, and the theoretical maximum photocurrent is reported to be 7.5 mA cm À2 . 37 Unfortunately, the current density of the Au cathode, a novel metal electrode, was low (À0.18 mA cm À2 at +0.5 V (vs. RHE)). Therefore, improved systems are needed for practical applications.
Anode reaction: Eqn (1) + (2) As for noble-metal-free cathodes, there are several reports on the use of carbon-based materials modied with inexpensive metal compounds mainly at basic pH. [38][39][40][41] However, efficiencies for the H 2 O 2 production with these cathodes were low at near-neutral pH, and an ion-exchange membrane was essential when the pH values of the anodic and cathodic electrolyte solutions were different. 21,22 Recently, we reported that a carbon cathode modied with a biomass-derived W-based electrocatalyst exhibited a relatively high cathodic current for the reductive H 2 O 2 production in a nearneutral KHCO 3 aqueous solution in a two-compartment cell, but the FE(H 2 O 2 ) and the partial current density for H 2 O 2 production were not enough. 20 In this paper, we pursued developing more efficient H 2 O 2 production systems on a photoanode and a noble-metal-free cathode in a KHCO 3 aqueous solution without an ionexchange membrane or an external bias in a onecompartment cell. The system using a single electrolyte in a one-compartment cell without any membrane is very simple. It was found that nickel carbonate on a carbon-based cathode showed excellent electrocatalytic activity on both FE(H 2 O 2 ) and the partial current density for H 2 O 2 production in a KHCO 3 solution. A water-insoluble nickel carbonate electrocatalyst was easily prepared in situ from a nickel salt and KHCO 3 aqueous solution at ambient and mild conditions. The apparent solar-tochemical energy conversion efficiency for H 2 O 2 production (STC H 2 O 2 ) on a combination system of the BiVO 4 /WO 3 photoanode and the cathode loaded with nickel carbonate electrocatalyst without external bias reached 1.75%, which is the highest among all reported values so far.

Fabrication of the carbon cathode loaded with various metals
A water-repellent-treated carbon paper (abbreviated as CP; 1.0 cm 2 , TORAY, TGP-H-090) was used as the substrate plate of the cathode. A conductive carbon powder of Ketjenblack EC600JD (abbreviated as KB; Lion Specialty Chemicals Co., Ltd.) 42 A mixture of 100 mg of KB powder and 0-1000 mg of various metal salts was added to 20 mL of H 2 O. The suspension was thoroughly dispersed by ultrasonication for 30 min and dried overnight in a heating oven at 353 K under an ambient pressure. The ratio of a loaded metal salt of x wt% to the weight of the KB powder was abbreviated as Met x /KB. For example, a loaded metal salt of 10 and 100 wt% vs. the weight of the KB powder were abbreviated as Met 10 /KB and Met 100 /KB, respectively. A mixture of 1.5 mg (standard amount) of the electrocatalyst (KB, Met/KB) and 0.75 mg of 20 wt% Naon solution (Sigma-Aldrich Co., USA) was dispersed in 0.5 mL of ethanol, and loaded on one side of the CP substrate (area 1.0 cm 2 ) at room temperature. This cathode loaded with an electrocatalyst and KB powder on CP was dried at 333 K for 30 min in a heating oven.

HNO 3 treatment for KB powder
HNO 3 -treated KB powder (abbreviated as KB HNO3 ) was prepared by an immersion process. To 38.3 g of concentrated HNO 3 aqueous solution (60%, Fujilm Wako Pure Chemical Corporation), 1000 mg of KB powder was added. The suspension was thoroughly heated at 353 K for 24 h with a string at 300 rpm in a heating oven. They were collected by suction ltration and thoroughly washed with distilled water. Metal-salt loading on KB HNO3 (abbreviated as Met/KB HNO3 ) was prepared and coated on CP as the cathode by the same process.

Fabrication of the BiVO 4 /WO 3 photoelectrode
The BiVO 4 /WO 3 photoelectrode used in this study was synthesized using the method previously reported by Fuku,45 and was prepared on a F-doped SnO 2 (FTO) glass substrate by spincoating method. The precursor of the WO 3 layer was loaded by spin coating (1000 pm, 15 s, 200 mL per 12 cm 2 ) on FTO, and then calcined at 773 K for 30 min. Spin coating of the WO 3 layer was repeated twice using N,N-dimethylformamide (DMF) solutions of tungsten hexachloride (WCl 6 , 4 N; Kojundo Chemical Laboratory, Co., Ltd.) adjusted to 504 and 252 mM for the rst and second coat, respectively. The BiVO 4 layer was also fabricated by spin coating (500 rpm, 15 s, 400 mL per 12 cm 2 ) on the WO 3 layer, and calcined at 773 K for 30 min. The spin coating of

Electrochemical measurements and quantification method
Electrochemical measurement on the cathode The electrochemical properties of the electrocatalyst cathode were evaluated in a 2.0 M KHCO 3 aqueous solution (pH 8.8) with O 2 bubbling (50 mL min À1 ) with an ice bath (273-278 K) by using an electrochemical analyzer (Hokuto Denko, HZ-7000). The current-potential (I-E) characteristics were measured via a three-electrode method using a two-compartment cell. The scan rate was set at 2 mV s À1 . The volume of the electrolyte solutions of the anode and cathode chambers was 35 mL with stirring. The reference electrode used was 3 M Ag/AgCl, and the counter electrode was a Pt wire. Between the anode and cathode chambers, Aciplex (Asahi KASEI) was used as an ion-exchange resin. Aer the reaction was completed, 1.1 mL of the reaction solution was taken out using a syringe. The schematic illustration of a two-compartment cell is shown in Fig. S1(a). † All potentials in this paper were quoted with respect to the reversible hydrogen electrode (RHE), according to the Nernst equation (eqn (4)).
An accumulation experiment for H 2 O 2 was conducted at a constant applied bias of +0.5 V vs. RHE.

Simultaneous H 2 O 2 production on both the photoanode and cathode without an external bias or membrane
The simultaneous production of H 2 O 2 from H 2 O oxidation at the photoanode and O 2 reduction at the cathode was performed without applying any external bias using a two-electrode system composed of a BiVO 4 /WO 3 photoanode (irradiation area of 0.2 cm 2 with a white board behind the photoanode) and a Ni 10 / KB HNO 3 cathode (catalyst loading 1.5 mg cm À2 , area 0.2 cm 2 ). An aqueous solution of KHCO 3 (2.0 M) was used as an electrolyte (200 mL), and CO 2 and O 2 gases were each bubbled for 50 mL min À1 into the one-component cell with a pH value of 8.0 in an ice bath (273-278 K). An ice bath was used under CO 2 bubbling to obtain the optimum reaction conditions for the stability of the photoanode. It was conrmed that O 2 gas diffusion is not the rate-limiting factor under our conditions of <10 mA cm À2 . A solar simulator calibrated to AM 1.5G (1 SUN, 0.1 W cm À2 ) was used as the light source. The schematic illustration of our reaction system in a one-compartment cell without membrane is shown in Fig. S1(b). †

Characterization and quantication method
The prepared electrodes were characterized by X-ray uorescence (XRF, Rigaku, Super mini200) measured in a vacuum with a wavelength dispersive spectrometer with Pd-K radiation. The crystal structures of the samples were investigated by X-ray diffraction (XRD, Malvern Panalytical, Empyrean) using Cu Ka radiation at 40 kV and 40 mA. Transmission electron microscopy and energy dispersive X-ray spectroscopy (TEM and EDS, Philips, Tecnai Osiris) measurements were carried out with a eld emission gun operating at 200 kV.
The amount of produced H 2 O 2 was quantied via UV-visible spectroscopy (TECAN, Innite 200 PRO). Then, 1.0 mL of sample was added to 0.9 mL of 3.0 M HCl aqueous solution and 0.1 mL of FeCl 2 in 1.0 M HCl aqueous solution and quantied from Fe 3+ colorimetry (l ¼ 330 nm), as we reported previously. 19,20,32,44 The faradaic efficiency for H 2 O 2 production (FE(H 2 O 2 )) was calculated using eqn (5): We conrmed that the error range of the FE(H 2 O 2 ) value was around AE2%.
The apparent partial current density for H 2 O 2 production at a constant potential (J ap (H 2 O 2 )) was calculated using eqn (6): Here, J(Total) is the total current density at a constant potential and FE(H 2 O 2 ) is the average faradaic efficiency at the passed charge of a constant potential. J ap (H 2 O 2 ) represented the actual H 2 O 2 production rate. The turnover number was calculated using eqn (7). The amount of nickel was measured by XRF.
The apparent solar-to-chemical energy conversion efficiency (STC) for the H 2 O 2 and O 2 production on the BiVO 4 /WO 3 photoanode (irradiation area of 0.2 cm 2 with a white board behind the photoanode) and on the Ni 10 /KB HNO 3 cathode (catalyst loading 1.5 mg cm À2 , area 0.2 cm 2 ) was calculated using eqn (8) V H 2 O 2 is the H 2 O 2 production rate (in moles per second per square centimeter), DG is the Gibbs free energy from eqn (3) (DG ¼ +116.8 kJ mol À1 ), 46 and Int is the intensity of the incident simulated solar light (0.1 W s À1 cm À2 ). The J(Total) of the CP substrate was negligibly small. On the other hand, the J(Total) largely increased when KB powder was loaded on the CP substrate. KB powder has a high surface area (1270 m 2 g À1 ) and high conductivity. 47 It was surmised that the electron could be transferred through the conductive KB powder network, and that the reduction for H 2 O 2 production could take place on the KB surface. The optimum loading amount of KB powder on the CP substrate was around 1.5 mg cm À2 , and the current did not increase by further loading (Fig. S2 †). The exfoliation of the powder from the CP substrate was observed by overloading ( Fig. S2d †). Therefore, the loading amount of Met 10

Results and discussion
The theoretical reduction potentials of the four-electron reduction of O 2 (+1.23 V, eqn (9)) and successive reduction of H 2 O 2 (+1.77 V, eqn (10)) were signicantly positive compared to that of H 2 O 2 production via two-electron reduction of O 2 (+0.68 V, eqn (1)). In the case of the J(Total) improvement with decreasing FE(H 2 O 2 ), the J ap (H 2 O 2 ) was not signicantly improved. It was surmised that these undesirable reactions might be accelerated by the modication of these metal nitrates or ammonium salt of Cu, Cr, Mn, Co, Ga, W, and Mo. On the other hand, the J(Total) and FE(H 2 O 2 ) were both increased when Ni or Zn nitrate was modied. In particular, the Ni 10 /KB cathode showed the highest performance of J ap (H 2 O 2 ) on the H 2 O 2 production (À6.9 mA cm À2 ) among all metal salts in Table 1. The J ap (H 2 O 2 ) was also improved by modication of all various Ni salts (nitrate, sulfate, acetate, chloride, Ni oxide, Ni hydroxide, and nickel carbonate (Table S1, † 10 wt% of Ni salt)), suggesting that the positive effect of the current density and the FE(H 2 O 2 ) was caused mainly by the presence of Ni salt itself, rather than by the effect of anions. Ni nitrate showed the highest activity among them. Then, Ni nitrate was mainly used as a precursor for loading to a carbon electrode for subsequent experiments. An improved effect of the NiO-loaded KB cathode was not obvious compared to those with other nickel salts. NiO particles are hardly soluble in aqueous solution. In the case of soluble nickel-salts loaded KB cathodes, their properties were positively changed through the process of dissolution and precipitation. In the case of NiO, the bulk of NiO may be not changed in KHCO 3 solution. Fig. 2 shows the dependence of the loading amount of Ni nitrate over the KB cathode on FE(H 2 O 2 ), J(Total), and J ap (H 2 O 2 ). All values of the FE(H 2 O 2 ), J(Total), and J ap (H 2 O 2 ) showed volcano shape proles, depending on the amount of Ni nitrate modication, and had the best values at 10 wt% of Ni nitrate on the KB cathode (Ni 10 /KB). As for the Ni-nitrate-loaded cathode without KB powder (the rightmost data in Fig. 2), the J(Total) and the J ap (H 2 O 2 ) were very small, while the FE(H 2 O 2 ) was equivalent to pristine KB powder. The current-potential dependences of some typical cathodes with and without Ni nitrate are shown in Fig. 3. It was surmised that the loaded nickel compound itself on the CP substrate might be hardly conductive, and the excess amount of Ni compound hindered the electron transfer though the network of the conductive KB.  The TEM and EDX images of Ni 10 /KB are shown in Fig. 4. The Ni 10 /KB sample was prepared by immersion in KHCO 3 electrolyte. Although it was difficult to recognize the clear particle shapes of the Ni compounds based on a TEM image only, the presence of a small aggregation of Ni element was conrmed by TEM-EDX image analysis. The aggregation size of the Ni compounds was approximately 10-40 nm in the EDX mapping images. On the other hand, when a larger amount of Ni was loaded (Ni 100 /KB), a large aggregation (>0.5 mm) of the Ni element was observed by SEM-EDX measurements (Fig. S3 †). It was speculated that the large aggregation of the Ni compounds might decrease the current density at a higher loading amount of Ni(NO 3 ) 2 S 50 wt% in Fig. 2.
The Ni amount with KB powder (Ni 10 /KB and Ni 100 /KB) measured by XRF before and aer washing with distilled water or KHCO 3 aqueous solutions is shown in Table 2. The Ni signal by XRF before the washing process completely disappeared aer washing with distilled water. On the other hand, aer washing with KHCO 3 aqueous solution, the Ni signal was clearly detected, and more than 80% of Ni remained on the KB cathode. Ni(NO 3 ) 2 is easily dissolved in water, while nickel carbonates have a very poor water solubility (solubility product (pK sp ) ¼ 11.2). 48 Actually, the green transparent aqueous solution of Ni(NO 3 ) 2 changed to colorless by the addition of KHCO 3 , and insoluble green powders were precipitated (Fig. S4 †). The positive effect of Ni loading on the I-E curve disappeared by washing with distilled water, not by washing with a bicarbonate solution (Fig. S5 †). Moreover, the presence of Ni species on the surfaces of the cathodes was also investigated by XPS (Fig. S6 †). The peaks of the spectrum of Ni2p 3/2 and Ni2p 5/2 were observed on Ni 10 /KB (a) before and (b) aer washing with KHCO 3 aqueous solution. However, they were not observed (c) aer washing with distilled water. While the apparent surface coverage of Ni compounds to carbon before washing with KHCO 3 aqueous solution as calculated by XPS intensity was not large (less than 1%), the exact Ni coverage was difficult to estimate due to the low sensitivity and presence of carbon impurities.
We compared our electrocatalyst powder sample with reference reagents of nickel(II) carbonate basic hydrate (NiCO 3 -$2Ni(OH) 2 $4H 2 O; Fujilm Wako Pure Chemical Corporation, 44% as Ni), nickel(II) hydroxide (Ni(OH) 2 ; Fujilm Wako Pure Chemical Corporation, 95%) and nickel(II) nitrate hexahydrate (Ni(NO) 2 $6H 2 O; Fujilm Wako Pure Chemical Corporation, 99.9%) using XRD and thermogravimetric-differential thermal analysis (TG-DTA). Our synthetic powder sample was prepared by immersing Ni(NO 3 ) 2 in a KHCO 3 aqueous solution. As shown in the XRD patterns (Fig. 5), the shape and the broad peak position at around 17 and 35 of our powder sample were similar to those of the reference nickel carbonate rather than Ni(OH) 2 . In the TG-DTA results (Fig. S7 †), the curves of TG and DTA of our electrocatalyst sample (a) were very similar to those of the reference reagent of NiCO 3 $2Ni(OH) 2 $4H 2 O (b), but different from those of Ni(OH) 2 (c) and Ni(NO) 2 $6H 2 O (d). Therefore, it was concluded that the loaded Ni(NO 3 ) 2 on the KB cathode was immediately changed to insoluble nickel(II) carbonate basic hydrate when the cathode electrode was soaked in a KHCO 3 solution, and that the small particles (10-40 nm) of Ni carbonate formed by this process can function as an excellent catalyst for H 2 O 2 production on the KB cathode. This simple preparation of the insoluble electrocatalyst at ambient conditions is a signicant advantage for practical applications.
We conrmed that the H 2 O 2 production properties on FE(H 2 O 2 ) and J ap (H 2 O 2 ) over Ni 10 /KB were almost the same as those over pristine KB in KOH and potassium hydrogen phosphate aqueous solution, suggesting that a positive nickel salt effect was not observed in KOH and potassium hydrogen phosphate aqueous solution, in contrast with the KHCO 3 aqueous solution. The particle of nickel carbonate basic hydrate was stable and insoluble in KOH (pH > 14) and potassium hydrogen phosphate (pH > 8.5, which is the same as KHCO 3 ) aqueous solution. The surface condition is very important for electrocatalytic reactions. The loaded Ni(OH) 2 on KB could show a positive effect in KHCO 3 aqueous solution, as shown in Table 1. It was conrmed using XPS measurement that the surface carbonate was reduced aer the immersion in KOH or potassium hydrogen phosphate aqueous solution. From all results, it is speculated that the nickel adsorbed with carbonate, rather than nickel element itself, on the outermost surface and at the interface of the loaded electrocatalyst with KB possibly being the active site for H 2 O 2 production.
It has been reported that the activity of a carbon-based cathode for H 2 O 2 production from O 2 could be improved by nitric acid treatment, where the carbon surface was changed in concentrated nitric acid for a long time at high temperature. 49,50 Therefore, we tried to combine nitric acid treatment and Ni carbonate effect on the KB cathode (abbreviated as Ni/KB HNO 3 ).   Fig. 6 shows the dependence of the FE(H 2 O 2 ) on the applied potentials for Ni 10 /KB HNO 3 . The FE(H 2 O 2 ) increased with the applied potential positively, and this behavior was similar to those in previous reports of carbon-based cathodes in KOH. 51 The highest FE(H 2 O 2 ) at 5C was 82-84% at around +0.5 to +0.6 V (vs. RHE). We also measured faradaic efficiency by the rotating ring disk electrode (RRDE) method (Fig. S8 †), and found that the value of the FE(H 2 O 2 ) by the RRDE method was 85% at +0.5 V (vs. RHE), which was almost consistent with the results of the accumulated H 2 O 2 aer passing 5C electrons of using a Fe 3+ colorimetry method. The production rates of H 2 O 2 on Ni 10 / KB HNO 3 at +0.5 and +0.2 V (vs. RHE) were 121.9 and 330.3 mmol L À1 h À1 g À1 cm À2 , respectively. To the best of our knowledge, these rates for reductive H 2 O 2 production were the highest among all reports on noble-metal-free cathodes in aqueous solutions having near-neutral pH (Table S2 †). Fig. 7 shows the long-term evaluation result of H 2 O 2 production on the Ni 10 / KB HNO 3 cathode with large amounts of passed charge. H 2 O 2 production was increased linearly at a constant applied bias at +0.5 V (vs. RHE), and the FE(H 2 O 2 ) was around 80% passing charges up to 2500C, suggesting that the produced H 2 O 2 was not further reduced to H 2 O sequentially (eqn (10)) for a long term. From the linear production and the agreement of the FE(H 2 O 2 ) by colorimetry and RRDE methods for long-and short-term quantitative measurements, respectively, we concluded that the total H 2 O 2 selectivity was determined by the initial H 2 O 2 selectivity on the cathode with the electrocatalyst. The amount of the produced H 2 O 2 was 10.4 mmol and the amount of nickel was 0.037 mmol cm À2 (calculated from XRF) using Ni 10 /KB HNO 3 . Therefore, the turnover number was calculated as more than 14 000 from eqn (7). The H 2 O 2 accumulation concentration reached 1.0 wt% in 35 mL of 2.0 M KHCO 3 aqueous solution. In addition, the I-E curves of the cathode hardly changed even aer repeated 150 cycles of I-E measurements during 36 h (Fig. S9 †), and the cathodic current remained around À11 mA cm À2 , suggesting that the modication effect of the Ni 10 /KB HNO 3 cathode was very stable under the H 2 O 2 production conditions.
Nitric acid treatment and Ni carbonate catalyst had different effects. The I-E curve of the Ni 10 /KB aer washing with distilled water cathode (Fig. S5(c) †) was overlapped to that of the pristine KB cathode (Fig. S5(a) †), and the FE(H 2 O 2 ) of Ni 10 /KB aer washing with water (62%) was almost similar as that of the pristine KB (59%). It is suggested that the Ni 10 /KB cathode returned to the original KB just by washing with pure water and reversibly due to the removal of the loaded Ni compound. On the other hand, the mechanism of the nitric acid treatment effect was explained by the change of the carbon surface structure, especially by the ratio of sp 2 and sp 3 bonded carbons. 51,52 The former originated mainly from graphitic and/ or aromatic bonded carbon (C]C), and the latter was from aliphatic and/or diamond bonded carbon (C-C). By XPS measurement, the sp 2 /sp 3 ratio of C 1s became higher, and the activity was higher. 49,53 It was also conrmed that the sp 2 /sp 3 ratio and the peak of the COO functional group at around 289.3 eV in our samples increased aer nitric acid treatment ( Fig. S10 †). The increase of sp 2 /sp 3 may indicate the increase of the graphitic carbon structure with high conductivity. The C 1s XPS spectrum of KB HNO3 did not change even when the sample was washed with distilled water thoroughly (Fig. S11 †), suggesting that the carbon surface structure of the KB powder was irreversibly oxidized or dehydrated by concentrated nitric acid treatment at 353 K. The positive effect of nitric acid treatment was not observed when the immersed time was short (<1 h), and/or the temperature was low at around room temperature. In the case of KB HNO3 and Ni 10 /KB HNO3 , the XPS spectra of C 1s were hardly changed by Ni carbonate loading (Fig. S10(c and d) †). From the results, it was concluded that the mechanism of effect by Ni carbonate catalyst was different from that by nitric acid treatment. It was surmised that the presence of small particles of insoluble Ni carbonate catalyst (mainly NiCO 3 $2Ni(OH) 2 $4H 2 O) on the carbon surface of both KB and KB HNO3 could accelerate the two-electron process from O 2 to H 2 O 2 effectively. The detailed mechanism of the nickel carbonate basic hydrate effect is under investigations.
The property of H 2 O 2 production on the BiVO 4 /WO 3 photoanode is known to be highly excellent, and we also conrmed that the FE(H 2 O 2 ) was around 90% initially on our BiVO 4 /WO 3 photoanode in KHCO 3 aqueous solution. 19,29,32 Finally, the combination of the Ni 10 /KB HNO 3 cathode with the BiVO 4 /WO 3 photoanode was investigated in the one-compartment cell without any membrane between electrodes in 2.0 M KHCO 3 aqueous solution under the solar simulator AM 1.5G (1 SUN). Fig. 8 shows the current-time dependence under simulated solar light irradiation in the two-electrode system without an applied bias potential. The current was not observed between electrodes in the dark condition. When the simulated solar light was irradiated to the photoanode, a photocurrent of >1.75 mA cm À2 was observed initially, and the photocurrent was maintained at around 1.5 mA cm À2 . This photocurrent value was in agreement with that in the I-E curve at 0 V of the potential (Fig. S12 †). The average FE(H 2 O 2 ) from both electrodes was calculated to be 168% in total at passed charge aer 0.5C using eqn (5), and the production rate of H 2 O 2 was estimated to be 0.92 mmol min À1 cm À2 . The values of the apparent solar-to-chemical energy conversion efficiency for the H 2 O 2 production (STC H 2 O 2 ) were estimated to be 1.75% without applying an external bias, using eqn (8). This value is, to the best of our knowledge, the highest among all the reported values for H 2 O 2 production systems using simulated solar light (Table S3 †). We successfully achieved highly efficient H 2 O 2 production by only using electrolytes, oxygen, and simulated solar light, without using external electricity, a membrane, and a noble metal.

Conclusion
We have successfully prepared a highly active non-noble-metal electrocatalyst for H 2 O 2 production from O 2 in 2.0 M KHCO 3 aqueous solution using an in situ preparation process. The small particles (10-40 nm) of Ni carbonate prepared by immersion of Ni nitrate in a KHCO 3 solution are remarkably effective on the HNO 3 pretreated KB electrode. The highest values of the FE(H 2 O 2 ) and J ap (H 2 O 2 ) of 82% and À9.8 mA cm À2 , respectively, were obtained using a Ni 10 /KB HNO 3 cathode at +0.5 V vs. RHE. The cathode is very stable, and the H 2 O 2 accumulation concentration can reach 1.0 wt%. Finally, the H 2 O 2 production on a BiVO 4 /WO 3 photoanode and a Ni 10 /KB HNO 3 cathode in a one-compartment photochemical cell was demonstrated without applying an external bias, and the apparent FE(H 2 O 2 ) was 168% in total. The production rate of H 2 O 2 was 0.92 mmol min À1 cm À2 . The value of STC H 2 O 2 was estimated to be 1.75% without applying an external bias. This H 2 O 2 production system on both electrodes without external bias has great economic advantages compared to a system using a one-side electrode with bias. This one-compartment cell system without any membrane is very simple, and it will be developed using photocatalyst sheets as both cathode and anode electrodes for a large area application. 19 In order to accumulate the produced H 2 O 2 further in this system, the development of the suppression method about the H 2 O 2 successive decomposition on the photoanode is very important. 20 Moreover, the elucidation on the mechanism of the nickel carbonate basic hydrate at the atomic level by computational chemistry is also needed. They are under investigations.

Conflicts of interest
There are no conicts of interest to declare.