Introduction

Utilization of solar energy as a primary energy source has been strongly demanded to reduce emissions of harmful and/or greenhouse gases produced by burning fossil fuels. However, large fluctuation of solar energy depending on the length of the daytime is a serious problem1,2. To utilize solar energy in the night time, solar energy should be stored in the form of chemical energy and used as a fuel to produce electricity. In this context, H2 has been regarded as the most promising candidate, because H2 can be produced by photocatalytic water splitting and used as a fuel of H2 fuel cells to generate electricity with a high efficiency without emission of harmful chemicals. However, the low solar energy conversion efficiency of H2 production and the storage problem of gaseous H2 have precluded the practical use of H2 as a solar fuel3. In contrast to gaseous H2, H2O2 can be produced as an aqueous solution from water and O2 in the air by the combination of the photocatalytic two-electron reduction of O2 and the catalytic four-electron oxidation of water4,5. H2O2 can be used as a fuel of an H2O2 fuel cell to generate electricity with emission of water and oxygen5,6,7,8,9,10. The energy density of aqueous H2O2 (60%) is 3.0 MJ l−1 (2.1 MJ kg−1), which is comparable to the value (2.8 MJ l−1, 3.5 MJ kg−1) of compressed hydrogen (35 MPa). However, the photocatalytic production of H2O2 from water and O2 has yet to be combined with the consumption of the produced H2O2 in an H2O2 fuel cell because of the insufficient photocatalytic activity4,5. In order to realize the production of H2O2 and its use in an H2O2 fuel cell, a breakthrough is definitely required to improve the photocatalytic efficiency for H2O2 production.

We report herein efficient photocatalytic production of H2O2, which has been made possible by using the most earth-abundant resource, that is, seawater instead of pure water for the photocatalytic oxidation with a semiconductor and the catalytic two-electron reduction of O2 with a cobalt chlorin complex supported on a glassy carbon substrate in a two-compartment photoelectrochemical cell under simulated solar illumination without an external bias potential. The H2O2 produced in seawater was used directly to generate electricity with the open-circuit voltage of 0.78 V and the maximum power density of 1.6 mW cm−2 using an H2O2 fuel cell, and the solar-to-electricity conversion efficiency of the total system is estimated to be ca 0.28%.

Results

Performance of photoanode and cathode

As a semiconductor photocatalyst for the water oxidation, tungsten oxide (WO3), which has a narrow band gap suitable for visible light (<460 nm) absorption, was employed11,12,13,14. In WO3, hole (h+) generated in the valence band (VB) is positive enough to oxidize water with the long lifetime (equation (1))15. The use of seawater instead of pure water has also enabled us to combine the photocatalytic production of H2O2 from seawater and O2 and its use in an H2O2 fuel cell. To perform selective reduction of O2, a cobalt chlorin

complex (CoII(Ch)), which has been proved to function as a catalyst for the efficient and selective two-electron reduction of O2 under homogeneous conditions (equation (2)), has been employed16. The overall photocatalytic reaction is given by equation (3), thus, H2O2 can be produced by the two-electron reduction of O2 by water as an electron donor.

Mesoporous WO3 (m-WO3) prepared by a literature method was deposited on a fluorine-doped tin oxide (FTO) as a photoanode (m-WO3/FTO) and CoII(Ch) was adsorbed on a carbon paper (denoted as CP) as a cathode (CoII(Ch)/CP; see Supplementary Information for details). Photocatalytic production of H2O2 was performed by using a two-compartment photoelectrochemical cell with the m-WO3/FTO photoanode and the CoII(Ch)/CP cathode, which were immersed in Ar-saturated and in O2-saturated aqueous HClO4 solutions (pH 1.3), respectively. These two electrodes were connected to each other by a conducting wire as an external circuit. The cathode and anode cells were separated by a Nafion membrane to prevent the decomposition of H2O2 produced in the cathode cell. Overall schematic diagram is shown in Fig. 1.

Figure 1: Overall scheme of photocatalytic production of H2O2.
figure 1

Photocatalytic production of H2O2 from water and O2 using m-WO3/FTO photoanode and CoII(Ch)/CP cathode in water or seawater under simulated 1 sun (AM 1.5G) illumination.

Full size image

To evaluate the selectivity to the two-electron reduction of O2 with CoII(Ch)/CP, the number of transferred electrons during the catalytic reaction was estimated by performing the rotating ring disc electrode (RRDE) technique with a glassy carbon disk electrode modified with a CoII(Ch) adsorbed on multi-walled carbon nanotubes (MWCNTs; see Supplementary Information for details). From the ratio of the observed disk current and the ring current at various rotating rates, the average number of transferred electrons was determined to be 2.7 (70% selectivity; Supplementary Fig. 1). CoII(Ch) has previously been reported to catalyse the production of H2O2 with nearly 100% selectivity in PhCN under homogeneous conditions16. The decrease in the selectivity of two-electron reduction of O2 may be attributed to the partial formation of μ-1,2-peroxo CoIII(Ch) dimer, which has been reported to act as a reactive intermediate in the four-electron reduction of O2 (ref. 17).

Photocatalytic production of H2O2

The simulated 1 sun illumination of m-WO3/FTO in the anode cell afforded the efficient photocatalytic production of H2O2 in the cathode cell in the two-compartment photoelectrochemical configuration without an external bias potential. The time courses of photocatalytic H2O2 production are shown in Fig. 2. Very little amount of H2O2 was obtained in the absence of CoII(Ch) on CP electrode, indicating that CoII(Ch) adsorbed on CP efficiently catalyses the two-electron reduction of O2 to produce H2O2 before the charge recombination of photoexcited electron in conduction band and h+ in VB of WO3. The rate of photocatalytic production of H2O2 in seawater was markedly enhanced compared with that in pure water. After the illumination for 24 h, the amount of produced H2O2 in seawater reached ca 48 mM, which is much larger than the value (2 mM) using a one-compartment system reported previously4. No structural change of m-WO3/FTO electrode after the photocatalytic reaction was confirmed by the powder X-ray diffraction measurements (Supplementary Fig. 2). The similar enhancement on photocatalytic activity was observed in an NaCl solution, in which the concentration of Cl− was the same as that of Cl− in seawater. The enhancement effect of Cl− on the photocatalytic activity for water oxidation can be interpreted by the following Cl−-assisted mechanism18,19,20,21. First, the oxidation of Cl− by photogenerated hole to form chlorine (Cl2) occurs before the oxidation of water as given by equation (4)18. Cl2 is in the disproportionation equilibrium with hypochlorous acid (HClO), as given by equation (5), where the population of Cl2 and HClO varies depending on the pH the solution and Cl2 is a major component under an acidic solution below pH 3 (refs 18, 19). Second, HClO is decomposed to O2 and Cl− under solar irradiation, as given by equation (6)20. Thus, the overall reaction of the water oxidation assisted by Cl− is given by equation (1). Indeed, the ultraviolet–visible absorption spectrum

Figure 2: Photocatalytic production of H2O2 in the two-compartment photoelectrochemical cell.
figure 2

Time courses of H2O2 production with m-WO3/FTO photoanode and CoII(Ch)/CP cathode in pH 1.3 water (red circle), in pH 1.3 seawater (blue circle) and in an NaCl aqueous solution (pH 1.3; blue square) under simulated 1 sun (AM 1.5G) illumination. Time course of H2O2 production in the absence of CoII(Ch) on carbon paper under simulated 1 sun (AM 1.5G) illumination in pH 1.3 water is shown as black circle.

Full size image

of the anode cell solution after the photocatalytic reaction for 24 h exhibited the absorption band around at 231 nm, which is identical to the spectrum of a standard HClO/Cl2 solution below pH 3 (Supplementary Fig. 3)19. The amount of O2 evolved in seawater in anode cell after 1 h (12.7 μmol) was more than three times larger than that in water (3.7 μmol) as shown in Supplementary Fig. 4. Thus, the enhancement of photocatalytic production of H2O2 in seawater (Fig. 2) results from the photocatalytic oxidation of Cl− in seawater.

The effects of Cl− on the catalytic performance of m-WO3/FTO and CoII(Ch)/CP were also investigated by using a photoelectrochemical cell with three electrode configuration. The current–potential (I–V) curves of m-WO3/FTO under simulated 1 sun illumination and dark are shown in Fig. 3a. The onset of photocurrent for the oxidation of water was observed at 0.2 V (versus saturated calomel electrode (SCE)) in water (pH 1.3), which corresponds to the 110 mV of overpotential with respect to the value of flat band potential of WO3 (0.09 V versus SCE at pH 1.3)22. This overpotential is required as a driving force for the migration of the photogenerated electron from conduction band of WO3 to FTO electrode. When the photocatalysis measurements of m-WO3/FTO were performed under the same conditions in seawater instead of pure water, about four times larger photocurrent at 0.3 V (versus SCE) than that in water together with the negative shift of onset potential from 0.2 V (versus SCE) to 0.1 V (versus SCE) was observed (Fig. 3a). In addition, the stability of a photocurrent obtained by applying 0.3 V (versus SCE) was also improved as shown in Supplementary Fig. 5a. These improvements in the stability as well as the photocurrent in seawater can be explained by the efficient quenching of the photogenerated hole in VB by the oxidation of Cl−, as described above. These are consistent with the enhancement of photocatalytic production of H2O2 in seawater (Fig. 2). The Faradaic efficiencies in the photoelectrochemical water oxidation in water and in seawater were determined to be 77% and 93%, respectively, from the simultaneous measurements of O2 evolution (Supplementary Fig. 5b). The effect of Cl− on the electrochemical property of CoII(Ch)/CP was also studied by measuring cyclic voltammograms using the three-electrode electrochemical cell. The addition of tetra-n-butylammonium chloride (0.1 M) to a N2-saturated PhCN solution containing CoII(Ch) (1 mM) and tetra-n-butylammonium hexafluorophosphate (TBAPF6; 0.1 M) resulted in the large negative shift of the redox potential for [CoIII(Ch)]+/CoII(Ch) from 0.37 to 0.01 V (versus SCE; Supplementary Fig. 6a,b). In addition, no catalytic current was obtained for the O2 reduction in the presence of Cl− in an O2-saturated PhCN solution containing HClO4 (10 mM) as a proton source (Supplementary Fig. 6c), suggesting that Cl− inhibits electron transfer from CoII(Ch) to O2 because of the strong axial coordination of Cl− to the reaction centre of CoII(Ch) to form 5- or 6-coordinated inactive species. In contrast, the catalytic current for the O2 reduction with CoII(Ch)/CP measured in seawater (pH 1.3) appeared at ca 0.34 V (versus SCE), which is virtually the same onset potential and catalytic current measured in water (pH 1.3), as shown in Fig. 3b. This result indicates that the coordination of Cl− to CoII(Ch) adsorbed on the electrode surface in an aqueous solution is negligibly weak as compared with that in PhCN23. Hence, the enhancement of photocatalytic production of H2O2 is mainly derived from the acceleration of water oxidation at the photoanode. The predicted operating current of the two-compartment photoelectrochemical cell was defined from the intersection of cyclic voltammograms of CoII(Ch)/CP and I–V curves of the m-WO3/FTO photoanode, giving a value of 0.5 mA at 0.32 V (versus SCE) in water and 1.3 mA at 0.29 V (versus SCE) in seawater, respectively (Fig. 3c).

Figure 3: Photoelectrochemical performance of m-WO3/FTO and electrochemical performance of CoII(Ch)/CP.
figure 3

(a) I–V curves of m-WO3/FTO photoanode in pH 1.3 water (black solid) and in pH 1.3 seawater (red solid) under simulated 1 sun (AM 1.5G) illumination. I–V curves under dark are shown as dashed lines with the same colour definition. Sweep rate: 10 mV s−1. (b) Cyclic voltammograms of CoII(Ch)/CP in a N2-saturated pH 1.3 water (black solid) and an O2-saturated pH 1.3 water (red solid). The dashed lines show the cyclic voltammograms of CoII(Ch)/CP recorded in pH 1.3 seawater. Sweep rate: 20 mV s−1. (c) Cyclic voltammograms of CoII(Ch)/CP in O2-saturated pH 1.3 solutions (black) and I–V curves of m-WO3/FTO photoanode in pH 1.3 solutions (red) under simulated 1 sun (AM 1.5G) illumination.

Full size image

The effect of illumination intensity on the photocatalytic production of H2O2 was examined as shown in Supplementary Fig. 7a, where the produced amount of H2O2 increased in proportion to the intensity of the illumination. The solar energy conversion efficiency for the photocatalytic production of H2O2 in seawater was determined to be 0.55% under simulated 1 sun illumination. The best solar energy conversion efficiency was determined to be 0.94% when illumination intensity was reduced to 0.1 sun (Supplementary Fig. 7b). This efficiency exceeds that of swichgrass (0.2%), which has been considered as a promising crop for biomass fuel24, and also the value in the one-compartment cell (0.25%)4. A much higher solar-to-hydrogen efficiency of 12.3% has recently been achieved by perovskite photovoltaics-based electrolysis2. However, the storage of hydrogen has still been a quite difficult issue, because hydrogen is a gas having a low volumetric energy density. In this contrast, in our system, the produced hydrogen peroxide in seawater can be used directly as a fuel in an H2O2 fuel cell.

H2O2 fuel cell

Finally, the chemical energy of H2O2 produced by the photocatalytic reaction was converted to electrical energy through a H2O2 fuel cell composed of a polynuclear cyanide complexes, FeII3[CoIII(CN)6]2, modified carbon cloth cathode and a nickel mesh anode in a one-compartment cell10. The reaction solution (seawater, pH 1.3) containing ca 48 mM of H2O2 in cathode was transferred to the H2O2 fuel cell. The obtained potential and power density depending on the current density are shown in Fig. 4. The cell has the open-circuit potential and the maximum power density of 0.78 V and 1.6 mW cm−2, respectively. These values agree with those obtained from the H2O2 fuel cell using an aqueous solution (pH 1.0) containing authentic H2O2 (50 mM) and NaCl (1.0 M) as a supporting electrolyte (Supplementary Fig. 8). In addition, the energy conversion efficiency of the H2O2 fuel cell was determined to be ca 50% by the measurement of output energy as electrical energy versus consumed chemical energy H2O2 gas, which is comparable to the efficiency of an H2 fuel cell (Supplementary Figs 9 and 10). Thus, the solar-to-electricity conversion efficiency of the total system is estimated to be ca 0.28% (0.55 × 50%), which is still much lower in contrast to the conventional solar-to-electricity device such as photovoltaic cells. However, there are still many things to do in the one-compartment H2O2 fuel cells including better anode materials to improve the performance, because the one compartment cell without membrane and use of an aqueous solution of H2O2 have significant advantages as compared with H2 fuel cells. The production of chemical energy utilizing solar energy and its conversion to electrical energy based on H2O2 in seawater can provide practical solution to the construction of an ideal energy-sustainable society using seawater, which is the most earth-abundant resource.

Figure 4: Generation of electrical energy in the one-compartment H2O2 fuel cell.
figure 4

I–V (blue) and I–P (red) curves of the one-compartment H2O2 fuel cell with a Ni mesh anode and FeII3[CoIII(CN)6]2/carbon cloth cathode in the reaction solution containing H2O2 (47.9 mM) produced by photocatalytic reaction in seawater as shown in Fig. 2 (blue circle).

Full size image

Methods

Materials

Chemicals were purchased from commercial sources and used without further purification, unless otherwise noted. Benzonitrile (PhCN) used for spectroscopic and electrochemical measurements was distilled over phosphorus pentoxide before use25. Potassium hexacyanoferrate(III) (K3[Fe(CN)6]) and acetylacetone (≥99%) were purchased from Wako Pure Chemical Industries Ltd., Tungsten hexachloride (WCl6, ≥95%) was purchased from Nacalai Tesque. Scandium(III) nitrate tetrahydrate was purchased from Mitsuwa Chemicals. Potassium hexacyanocobaltate (K3[CoIII(CN)6], ≥99.9%) was supplied by Stream Chemicals. Red sea salt was supplied by Red Sea. Oxo[5,10,15,20-tetra(4-pyridyl)porphinato]titanium(IV) ([TiO(tpyp)]) was purchased from Tokyo Chemical Industry Co., Ltd. (TCI). Pluronic P-123, Triton X-100, Nafion perfluorinated ion exchange resin solution, Nafion perfluorinated membrane (Nafion 117) were received from Aldrich Chemical Co. CP electrode (EC-TP1-060T produced by Toray Industry Inc.) was obtained from Toyo Co. Glass slides coated with FTO (transmittance, 83.6%) were supplied by Aldrich Chemicals Co. and cut by Asahi Glass Co., Ltd. Tetra-n-butylammonium hexafluorophosphate (TBAPF6) purchased from Wako Pure Chemical Industries, Ltd. was twice recrystallized from ethanol and dried in vacuo before use. Purified water was provided by a Millipore Milli-Q water purification system (Millipore, Direct-Q 3 UV) with an electronic conductance of 18.2 MΩ cm. Cobalt chlorin complex [CoII(Ch)], mesoporous WO3 (m-WO3) and polynuclear cyanide complex (FeII3[CoIII(CN)6]2) were synthesized according to the literature procedure (vide infra).

Preparation of CoII(Ch)/CP electrode

CoII(Ch)/CP electrode was prepared by an MeCN solution (1 ml) of CoII(Ch) (0.3 mM), MWCNT (0.63 mg) and 5% Nafion (12 μl). For each experiment, the mixture was sonicated for 20 min and then a 50 μl of the mixture was applied on the both side of surface of a CP with a 3.0 cm2 area by drop-casting and allowed to evaporate to afford a film containing a MWCNT loading of 50 μg cm−2 and a catalyst loading of 30 nmol.

Preparation of m-WO3/FTO electrode

First, FTO glasses were cleaned before use by immersing into an MeOH/HCl (1/1 (v/v)) solution for 30 min and washed by purified water. The resulting FTO glasses were hydroxylated in H2SO4 for 2 h and then boiled in purified water for 30 min with subsequent drying under N2 (ref. 26). m-WO3/FTO electrode was prepared by a solution consisting of 1 ml of water containing 50 mg of m-WO3 and acetyl acetone (30 μl) and 1 drop of Triton X-100. The mixture was sonicated for 5 min and then a 50 μl drop was applied on the surface of FTO electrode with a 2.5 cm2 area and allowed to evaporate to afford a thin film. The resulting electrode was annealed to form crystalline m-WO3 at 400 °C with ramping rate of 2 °C min−1 (held at 400 °C for 2 h) under air to remove surfactant species. The combustion of residual surfactant was confirmed by thermal gravimetric-differential thermal analysis (TG-DTA) of m-WO3 dispersion prepared above (Supplementary Fig. 11). The exothermic current peak along with weight loss observed at around 180 °C was disappeared after the annealing at 400 °C. The morphology of obtained m-WO3/FTO was observed by scanning electron microscope (SEM), as shown in Supplementary Fig. 12. The mesoporous structure of m-WO3 was confirmed by N2 adsorption–desorption isotherm measurements (Supplementary Fig. 13). The measurements performed at 77 K revealed a type IV isotherm27, clearly indicating the presence of mesopores. The Brunauer–Emmett–Teller surface area of m-WO3 was as high as 21 m2 g−1 (Supplementary Fig. 13a). The size of the mesopores was determined to be 8 nm by Barrett–Joyner–Halenda plot (Supplementary Fig. 13b).

Preparation of seawater

The seawater was prepared by dissolving 33.4 g of red sea salt in 1 l of water to form a solution containing ca 550 mM of NaCl.

Characterization of m-WO3/FTO

TG/DTA data were performed on an SII TG/DTA 7,200 instrument. A sample (ca10 mg) was loaded into an Al pan and heated from 25 °C to 600 °C with a ramping rate of 2 °C min−1 under N2. A certain amount of γ-Al2O3 was used as a reference for DTA measurements. Nitrogen-adsorption/desorption measurements were performed at 77 K on a Belsorp-mini (BEL Japan, Inc.) within a relative pressure range from 0.01 to 101.3 kPa. A sample mass of ca 100 mg used for adsorption analysis was pretreated at 150 °C for 2 h under vacuum conditions and kept in N2 atmosphere until N2-adsorption measurements. The resulting sample was exposed to a mixed gas of He and N2 with a programmed ratio and adsorbed amount of N2 was calculated from the change of pressure in a cell after reaching equilibrium (at least 5 min). Powder X-ray diffraction patterns were recorded on a Rigaku MiniFlex 600. Incident X-ray radiation was produced by a Cu X-ray tube, operating at 40 kV and 15 mA with Cu Kα radiation (λ=1.54 Å). The scan rate was 1° min−1 from 2θ=10–70°. SEM images of particles were observed by a FE-SEM (JSM-6320F or JSM-6701F) operating at 10 kV.

Electrochemical measurements

Cyclic voltammetry measurements were performed on an ALS 630B electrochemical analyser. Effects of Cl− on CoII(Ch) was investigated in a N2- or O2-saturated PhCN solution containing 0.10 M TBAPF6 as a supporting electrolyte at 298 K using a conventional three-electrode cell with a glassy carbon (GC) working electrode (surface area of 0.3 mm2) and a platinum wire (Pt) as the counter electrode. The GC working electrode was routinely polished with polishing alumina suspension and rinsed with acetone before use. The potentials were measured with respect to the Ag/AgNO3 (1.0 × 10−2 M) reference electrode. All potentials (versus Ag/AgNO3) were converted to values versus SCE by adding 0.29 V (ref. 28). Redox potentials were determined using the relation E1/2=(Epa+Epc)/2.

Electrochemical performance of CoII(Ch) deposited on CP electrode for the catalytic O2 reduction was evaluated in a N2- or O2-saturated aqueous HClO4 (pH 1.3) solution containing NaClO4 (0.1 M) as a supporting electrolyte and in N2- or O2-saturated seawater containing HClO4 (pH 1.3) and NaClO4 (0.1 M) at 298 K using a conventional three-electrode cell consisting of CoII(Ch)/CP as a working electrode and a platinum coil as the counter electrode. All the photoelectrochemical and electrochemical measurements in aqueous solutions were conducted using a reference SCE and all results in this work are presented against the SCE. The conversion of potentials versus SCE to versus normal hydrogen electrode (NHE) was performed according to the following equation (7).

The superior performance of CoII(Ch)/CP in an aqueous solution was confirmed by the comparison with cobalt octaethylporphyrin (CoII(OEP)), which is commonly used as an electrocatalyst for the O2 reduction, modified CP (CoII(OEP)/CP), as shown in Supplementary Fig. 14.

The RRDE measurements were carried out using a BAS RRDE-3A rotator linked to an ALS 730D electrochemical analyser. A three-electrode cell (100 ml) was employed with the RRDE consisting of a platinum ring (Pt) electrode and a GC disk electrode, platinum coil (Pt) as a counter electrode and SCE as a reference electrode. The voltammograms were measured in an O2-saturated aqueous HClO4 solution (pH 1.3) containing NaClO4 (0.1 M) at 5 mV s−1 with various rotating rates (100, 300, 600, 900, 1,200, 1,500, 2,000, 2,500, 3,000, 3,500, 4,000 and 4,500 r.p.m.). A RRDE for the investigation of transferred electrons during catalytic O2 reduction with CoII(Ch)/CP was performed by the modification of GC disk electrode with a thin film of CoII(Ch). The thin film was prepared by a solution consisting of MeCN (1 ml) containing CoII(Ch) (0.3 mM), MWCNT (1.26 mg) and 5% Nafion (12 μl). For each experiment, the mixture was sonicated for 20 min and then a 10-μl drop was applied on the surface of a polished GC disk electrode and allowed to evaporate to afford a thin film containing a MWCNT loading of 100 μg cm−2 and a catalyst loading of 3 nmol.

The number of transferred electrons (n) is determined by following equation n=4ID/(ID+IR/N), where ID is the faradic current at the disk electrode, IR is the faradic current at the ring electrode, and N is the collection efficiency of the RRDE. The N value is measured using an aqueous solution of K3[FeIII(CN)6] (2 mM) as a standard one-electron redox couple ([FeIII(CN)6]3−/[FeII(CN)6]4−) in the presence of KNO3 (0.5 M) and is determined to be N=0.37 when the GC disk electrode of RRDE is loaded with the same amount of MWCNT (100 μg cm−2) as used above (Supplementary Fig. 15).

Photoelectrochemical measurements

Photoelectrochemical measurements were performed in a home-made quartz cell (light path length=1 cm) composed of the as prepared m-WO3/FTO electrode, a platinum coil counter electrode, and a SCE reference electrode in an Ar-saturated aqueous solution (8 ml) containing HClO4 (pH 1.3) and 0.1 M of NaClO4 at 298 K (Supplementary Fig. 16a). The photoanode was illuminated from the back side of FTO electrode (FTO/electrolyte interface) with a solar simulator (HAL-320, Asahi Spectra Co., Ltd.), where the light intensity was adjusted at 100 mW cm−2 (AM1.5G) at the sample position by a 1SUN checker (CS-20, Asahi Spectra Co., Ltd.). The Faradaic efficiency for O2 evolution was determined by following equation (8), where F denotes Faradaic constant (9.65 × 104 C mol−1).

The Faradaic efficiency for O2 evolution (%)

Photocatalytic production of H2O2

Photocatalytic production of H2O2 was performed in a quartz anode cell (light path length=1 cm) connected with a pyrex cathode cell through a Nafion membrane (Supplementary Fig. 16b). The anode cell consists of the as prepared m-WO3/FTO photoanode for the water oxidation in an Ar-saturated aqueous solution (8 ml) containing HClO4 (pH 1.3) and 0.1 M of NaClO4. The cathode cell is composed of the as prepared CoII(Ch)/CP cathode for the O2 reduction in an O2-saturated aqueous solution (10 ml) containing HClO4 (pH 1.3) and 0.1 M of NaClO4 at 298 K. These two electrodes were connected each other with alligator clips and copper wire as an external circuit. The photoanode was illuminated from the back side of the FTO electrode with the solar simulator (HAL-320, Asahi Spectra Co., Ltd.), where the light intensity was adjusted at 100 mW cm−2 (AM1.5G) at the sample position by the 1SUN checker (CS-20, Asahi Spectra Co., Ltd.). The anode and cathode solution was saturated by continuous bubbling with argon and oxygen gas for 30 min, respectively, before the photocatalytic reaction. The O2 bubbling was continued during the photocatalytic reaction. The cathode cell was kept in dark to prevent the decomposition of produced H2O2 by ultraviolet-light irradiation during photocatalytic reaction. The amount of produced H2O2 was determined by spectroscopic titration with an acidic solution of [TiO(tpypH4)]4+ complex (Ti-TPyP reagent)29. The Ti-TPyP reagent was prepared by dissolving 3.40 mg of the [TiO(tpyp)] complex in 100 ml of hydrochloric acid (50 mM). A small portion of the reaction solution was sampled and diluted with water depending on the concentration of produced H2O2. To 0.25 ml of 4.8 M HClO4 and 0.25 ml of the Ti-TPyP reagent, a diluted sample was added. The mixed solution was then allowed to stand for 5 min at room temperature. This sample solution was diluted to 2.5 ml with water and used for the spectroscopic measurement. The absorbance at λ=434 nm was measured by using a Hewlett Packard 8453 diode array spectrophotometer. A blank solution was prepared in a similar manner by adding distilled water instead of the sample solution to Ti-TPyP reagent in the same volume with its absorbance designated as AB. The difference in absorbance was determined as follows: ΔA434=AB−AS. Based on ΔA434 and the volume of the solution, the amount of H2O2 was determined (Supplementary Fig. 17).

Detection of O2

The concentration of O2 in the anolyte was monitored during both photocatalytic production of H2O2 and photoelecrochemical measurements (vide supra) by using a fluorescence-based oxygen sensor (FOXY Fiber Optic Oxygen Sensor, Ocean Optics). The O2-sensing needle probe was installed in a gas-tight quartz anode cell filled with 8 ml of a solution, which left 7 ml of a headspace, through a rubber septum on the end of the cell. The solution and headspace were purged with argon gas for 30 min before measurements. Two-point calibration of the O2 sensor was performed against solutions (air, 20.9% O2, and Ar, 0% O2) used in each measurement. The amount of O2 leaked in the anode cell during measurements was determined under dark and subtracted from the data obtained under illumination. The amount of dissolved O2 in solutions was recorded as mole %. The total amount O2 evolved in the anode cell was determined using Henry’s Law and converted, using the ideal gas law, into μmol.

Solar-to-H2O2 energy conversion efficiency

Measurement of solar energy conversion efficiency of the photocatalytic production of H2O2 was carried out in a quartz anode cell (light path length=1 cm) connected with a pyrex cathode cell through a Nafion membrane as used in photocatalytic production of H2O2 as described above. The photoanodes were illuminated from back side of the FTO electrode with the solar simulator (HAL-320, Asahi Spectra Co., Ltd.), where the light intensity was adjusted at 10–100 mW cm−2 (AM1.5G) at the sample position by the 1SUN checker (CS-20, Asahi Spectra Co., Ltd.). The amount of produced H2O2 was determined by the titration with the Ti-TPyP reagent (vide supra). The solar energy conversion efficiency was determined by following equations (3 and 9), where output energy as H2O2 was calculated by

Solar Energy Conversion Efficiency (%)

the multiplication of enthalpy change (ΔH=98.3 kJ mol−1) and the produced amount of H2O2 (the concentration of H2O2 × volume of cathode solution (10 ml)). Energy density of incident solar light was adjusted at 10–100 mW cm−2 (0.1–1 SUN, Air Mass 1.5 (AM1.5)) at the sample position for whole irradiation area (2.5 cm2) by the 1 SUN checker (CS-20, Asahi Spectra Co., Ltd.) at room temperature.

Spectroscopic measurements

Ultraviolet–visible spectroscopy was carried out on a Hewlett Packard 8453 diode array spectrophotometer at room temperature using quartz cell (light path length=1.0 cm).

H2O2 fuel cell

FeII3[CoIII(CN)6]2 was mounted onto a carbon cloth by drop-casting or by spraying a dispersion of FeII3[CoIII(CN)6]2 in isopropanol with an airbrush (TAMIYA Spray-work HG). An aqueous solution of Nafion (0.2 wt.%) was used to protect the film of FeII3[CoIII(CN)6]2 on a carbon cloth. A Ni mesh (150 mesh) and FeII3[CoIII(CN)6]2 that was mounted onto a carbon cloth were immersed in the solution of H2O2. The performance tests were conducted in a one-compartment cell with the reaction solution containing H2O2 produced by the photocatalytic reaction transferred from the cathode cell of the two-compartment cell system. The current and power values normalized by the geometric surface area of an electrode were recorded on an ALS 630B electrochemical analyser and KFM 2005 FC impedance meter at 25 °C. The performance tests in solutions containing various concentrations of standard H2O2, HClO4 (pH 1) and NaCl (1.0 M) were performed for the control experiment (Supplementary Fig. 8).

Energy conversion efficiency of H2O2 fuel cell

FeII3[CoIII(CN)6]2/carbon cloth and Ni mesh electrodes were prepared as noted above. Each electrode was connected with Pt wire and protected by PP (polypropylene) sheet to avoid electrical short circuit. The performance tests were conducted in a well-sealed one-compartment cell with a rubber septum (Supplementary Fig. 9). The reaction solution containing H2O2 (0.3 M), NaCl (1.0 M) and Sc(NO3)3·4H2O (0.1 M)10 and the headspace (6.5 ml) of the one-compartment cell were purged separately with argon gas for 30 min before measurements. After the argon-saturated reaction solution was transferred to the one-compartment cell using gas-tight syringe, cell voltage, applying constant current of 3.3 mA, was recorded on a KFM 2005 FC impedance meter at 25 °C. The amount of evolved O2 gas in the headspace of one-compartment cell was quantified by a Shimadzu GC-17A gas chromatograph (Ar carrier, a capillary column with molecular sieves (Agilent Technologies, 19095PMS0, 30 m × 0.53 mm) at 313 K) equipped with a thermal conductivity detector. The energy conversion efficiency of H2O2 fuel cell was determined by following equations (3) and (10), where consumed chemical energy as H2O2 was calculated by the multiplication of enthalpy change (ΔH=−98.3 kJ mol−1) and twice of the produced amount of O2 (Supplementary Fig. 10).

Energy Conversion Efficiency of H2O2 Fuel Cell (%)

Additional information

How to cite this article: Mase, K. et al. Seawater usable for production and consumption of hydrogen peroxide as a solar fuel. Nat. Commun. 7:11470 doi: 10.1038/ncomms11470 (2016).