Production of hydrogen peroxide as a sustainable solar fuel from water and dioxygen

Production of hydrogen peroxide as a sustainable solar fuel from water and dioxygen Satoshi Kato, Jieun Jung, Tomoyoshi Suenobu and Shunichi Fukuzumi* a Department of Material and Life Science, Graduate School of Engineering, Osaka University, ALCA, Japan Science and Technology Agency (JST), Suita, Osaka 565-0871, Japan b Department of Bioinspired Science, Ewha Womans University, Seoul 120-750, Korea


Introduction
Renewable and clean energy resources are urgently required in order to solve global energy and environmental issues. 1,2 Among renewable energy resources, solar energy is by far the largest exploitable resource, 1-6 and thereby it is quite important to obtain sustainable solar fuels such as hydrogen (H 2 ) or others. Hydrogen is a clean energy source to reduce the dependence on fossil fuels and the emissions of greenhouse gases in the long term. [7][8][9][10][11][12][13][14][15][16][17] However, the storage of hydrogen has been very difficult, because hydrogen is a gas having a low volumetric energy density. 18,19 In contrast to hydrogen, hydrogen peroxide (H 2 O 2 ), soluble in water, can be an ideal energy carrier alternative to oil or hydrogen, because it can be used in a one-compartment fuel cell leading to the generation of electricity. [20][21][22][23][24][25][26][27] The output potential of H 2 O 2 fuel cells theoretically achievable is 1.09 V, which is somewhat smaller but comparable to those of a hydrogen fuel cell (1.23 V) and a direct methanol fuel cell (1.21 V). [22][23][24][25][26][27] Thus, a combination of hydrogen peroxide production using solar energy and power generation with a hydrogen peroxide fuel cell provides an ideally sustainable solar fuel. 22 However, photocatalytic production of hydrogen peroxide from water (H 2 O) and dioxygen (O 2 ) using solar energy has remained a great challenge.
H 2 O 2 is currently manufactured in industry by the autoxidation of 2-alkyl anthrahydroquinone by O 2 to the corresponding 2-alkylanthraquinone (the so-called anthraquinone process) using a noble metal such as palladium to regenerate the anthrahydroquinone with H 2 . 28 H 2 used as a reductant is normally produced by steam reforming of natural gases, which emits a signicant amount of CO 2 . 28 H 2 O 2 can also be produced by two-electron photoreduction of O 2 with use of a semiconductor photocatalyst 29 or a homogeneous photocatalyst. [30][31][32][33] In this case, organic reductants such as 2-propanol, acetaldehyde and oxalate are required as sacricial electron sources, resulting in unwanted emission of CO 2 . [29][30][31][32][33] We report herein for the rst time photocatalytic production of H 2 O 2 from H 2 O and O 2 , both of which are earth abundant, without emission of CO 2 by two-electron photoreduction of O 2 by H 2 O that is used as an electron source in acidic aqueous solutions. The high turnover number and quantum yield have been attained by combining an efficient water oxidation catalyst (WOC) with a photosensitiser and a scandium ion that acts as a Lewis  as a compressed gas in a cylinder (0.5 L). Carbon monoxide (CO) gas was purchased from Sumitomo Seika Chemicals Co., Ltd. as a compressed gas in a cylinder (3.4 L). Purication of water (18.2 MU cm) was performed with a Milli-Q system (Millipore, Direct-Q 3 UV).

Synthesis of iridium hydroxide nanoparticles
Iridium hydroxide nanoparticles were synthesised according to the literature. 34 The pH of an aqueous solution of H 2 IrCl 6 was adjusted to $10 by adding 5.0 M NaOH solution with vigorous stirring at 100 C. Aer 1.0 h stirring, precipitates appeared were collected by centrifugation. Then, the precipitates were washed by water three times, dried in vacuo at room temperature and kept at 65 C for 10 h.

Synthesis of [Ru II (Me 2 phen) 3 ]SO 4
The tris(4,7-dimethyl-1,10-phenanthroline)ruthenium(II) sulphate ([Ru II (Me 2 phen) 3 ]SO 4 ) complex was synthesised according to the literature. 35 RuCl 3 was reuxed under N 2 overnight in ethanol-water (v/v 80/20) with 6 equiv. of ligand, Me 2 phen, to form the red-orange [Ru II (Me 2 phen) 3 ]Cl 2 complex. Aer evaporation of the solvent, the product was readily precipitated from acetone with ether. The precipitate, [Ru II -(Me 2 phen) 3 ]Cl 2 , was added to water to be completely dissolved and Ag 2 SO 4 (65 mg) solubilised in water was added to the solution. Aer stirring for 12 h, AgCl was ltered off as a precipitate. An aqueous solution of (NH 4 ) 2 SO 4 was added to the reaction solution to obtain a crystalline product.

Synthesis of [Co(Cp*)(bpy)(OH 2 )]SO 4
The [Co(Cp*)(bpy)(OH 2 )]SO 4 complex was synthesised according to the literature. 36 Pentamethylcyclopentadiene (Cp*, 25 mL) and tert-butyllithium ($1.7 M in n-pentane, 90 mL) were combined in an equimolar amount (1 : 1) in n-pentane at 203 K. The solution was stirred and slowly allowed to warm up to room temperature. Aer stirring for further 24 h at room temperature, a white suspension was ltered through an inert gas frit and pentamethylcyclopentadienyllithium (Cp*Li) was ltered off. The anhydrous CoCl 2 (1.32 g) was added to the solution of Cp*Li (1.42 g) in 20 mL of tetrahydrofuran. The mixture was stirred for 3 h at room temperature until the brown solution became green-brown. Aerwards the solution was concentrated to a smaller volume under reduced pressure and extracted with 100 mL of n-pentane. The brown extracts were bubbled by CO gas for 30 min through the solution. Di-m-chloro-bis[chloro-(h 5 -pentamethylcyclopentadienyl)cobalt] ([(m-Cl)(CoCp*Cl)] 2 ) was obtained as a green powder. The [(m-Cl)(CoCp*Cl)] 2 (100 mg) in 20 mL of water was stirred with 1.5 equimolar amount of 2,2 0 -bipyridine (88 mg) for 1 h at room temperature under N 2 . Aer ltering off free 2,2 0 -bipyridine, Ag 2 SO 4 (118 mg) was added to the ltrate. Aer stirring for 12 h, AgCl was ltered off as a precipitate. An aqueous solution of (NH 4 ) 2 SO 4 was added to the reaction solution to obtain a crystalline product.

Quantum yield and quantum efficiency measurements
The quantum yield (QY) of the photocatalytic production of hydrogen peroxide (F) was determined under irradiation of monochromatised light using a Shimadzu spectrouorophotometer (RF-5300PC) through a band-pass lter transmitting l ¼ 450 nm, and estimated as where R (mol s À1 ) represents the H 2 O 2 production rate and I coefficient (Einstein s À1 ) based on the rate of the number of incident photons. In order to produce hydrogen peroxide by two-electron reduction of one molecule of oxygen, two photons are necessary for the electronic transition of the [Ru II (Me 2 phen) 3 ] 2+ photosensitiser. When the total two photons are fully used for production of hydrogen peroxide, QY reaches 100%. Therefore, the coefficient of the right-hand side in eqn (1) is 2 for this photocatalytic system. The total number of incident photons was measured by a standard method using an actinometer (potassium ferrioxalate, K 3 [Fe III (C 2 O 4 )] 3 ) 37 in H 2 O at room temperature under photoirradiation of a Shimadzu spectrouorophotometer (RF-5300PC) through a band-pass lter transmitting l ¼ 450 nm (slit width of 5.0 mm) at room temperature. For the same quartz cuvette (light path length ¼ 1 cm) with 3.0 mL solution as used in the production of hydrogen peroxide experiments, the rate of photon ux of the incident light (I) was determined to be 1.11 Â 10 À9 Einstein s À1 . Ir(OH) 3  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 l ¼ 434 nm was measured using a Hewlett Packard 8453 diode array spectrophotometer (A S ). A blank solution was prepared in a similar manner by adding distilled water instead of the sample solution in the same volume with its absorbance designated as A B . The difference in absorbance was determined as follows:  3 ]SO 4 (100 mM) and Sc(NO 3 ) 3 (100 mM) in a quartz cuvette (light path length ¼ 1 cm). The solution was saturated by bubbling with oxygen gas for $30 min. The photocatalyst was irradiated with a solar simulator (HAL-320, Asahi Spectra Co., Ltd.). The light intensity was adjusted to be 10 mJ cm À2 s À1 (Air Mass 1.5 (AM1.5)) at the sample position for the whole irradiation area (1.0 Â 3.0 cm 2 ) using a 1 SUN checker (CS-20, Asahi Spectra Co., Ltd.) at room temperature. The amount of produced hydrogen peroxide was determined by the titration with the Ti-TPyP reagent (vide supra).

EPR measurements
The EPR spectrum was taken on a JEOL X-band spectrometer (JES-RE1XE) under nonsaturating microwave power conditions (1.0 mW) operating at 9.2 GHz. Distilled water (1.0 mL) containing [Ru II (Me 2 phen) 3 ]SO 4 (20 mM) and Sc(NO 3 ) 3 (100 mM) in the EPR tube was saturated by bubbling with pure O 2 for $30 min. The solution was frozen at 77 K aer visible light irradiation (l > 420 nm). The magnitude of the modulation was chosen to optimise the resolution and the signal to noise ratio (S/N) of the observed spectrum (modulation width, 2.0 G; modulation frequency, 100 kHz). The g values were calibrated using an Mn 2+ marker.

Characterisation of Ir(OH) 3 nanoparticles
X-ray photoelectron spectra (XPS) were recorded using a Kratos Axis 165 with a 165 mm hemispherical electron energy analyser. The incident radiation was Mg Ka X-ray (1253.6 eV) at 200 W and a charge neutraliser was turned on for acquisition. Each sample was attached on a stainless stage with a double-sided carbon scotch tape. The binding energy of each element was corrected by the C 1s peak (284.6 eV) from residual carbon. TG/DTA data were recorded on an SII TG/DTA 7200 instrument. Each sample ($3.0 mg) was heated from 298 K to 373 K (held at 373 K for 10 min) and from 373 K to 873 K with a ramp rate of 2 K min À1 . A certain amount of g-Al 2 O 3 was used as a reference for DTA measurements. Nitrogen adsorption-desorption at 77 K was performed with a Belsorp-mini (BEL Japan, Inc.) within a relative pressure range from 0.01 to 101.3 kPa. A sample mass used for adsorption analysis was pretreated at 333 K for 30 min under vacuum conditions and kept in a N 2 atmosphere until N 2adsorption measurements. The sample was exposed to a mixed gas of He and N 2 with a programmed ratio and the adsorbed amount of N 2 was calculated from the change of pressure in a cell aer reaching equilibrium (at least 5 min). The transmission electron microscopy (TEM) image of iridium hydroxide, which was mounted on a copper microgrid coated with elastic carbon, was observed using a JEOL JEM-2100 operating at 200 keV.

Characterisation of Ir(OH) 3 nanoparticles
The TG curve of Ir(OH) 3 nanoparticles is shown in Fig. 1a, which exhibits two consecutive steps. The rst step of weight loss with an endothermic peak at 393 K corresponds to the removal of physisorbed water. The weight loss at the second step of Ir(OH) 3 starting from 583 K was attributed to dehydration of Ir(OH) 3 .
The TG curve of commercially available IrO 2 in Fig. 1b showed no such dehydration step.
To determine surface conditions of Ir(OH) 3 , X-ray photoelectron spectroscopy (XPS) measurements were carried out for the energy regions of Ir 4f, O 1s and C 1s with reference to commercially available IrO 2 . As reported previously, the binding energy of Ir 4f 5/2 reects the valence of Ir ions sensitively where the binding energies of Ir 4f 5/2 for Ir 0 , Ir III and Ir IV are reported to be 61.0 eV, 62.0 eV and 63.7 eV, respectively. [40][41][42] The XPS spectra of Ir 4f and O 1s for the iridium hydroxide and IrO 2 are shown in Fig. 2. The binding energy of Ir 4f 5/2 of both the iridium hydroxide and IrO 2 was 62.2 eV, which is close to the reported binding energy of 62.0 eV for Ir(III) species. The O 1s peaks of the iridium hydroxide and IrO 2 appeared at 531.5 eV and 530.5 eV, respectively. The higher binding energy of the O 1s peak of the iridium hydroxide results from the formation of hydroxide species as oen reported previously. 43,44 Iridium hydroxide was investigated by transmission electron microscopy (TEM). The TEM image is displayed in Fig. 3, which indicates that the size is in the range of 50-100 nm with an undened shape. BET surface areas of Ir(OH) 3 and IrO 2 (commercially available) are shown in Table S1 in the ESI. † The BET surface area of Ir(OH) 3 was 28 times higher than that of IrO 2 .

Photocatalytic production of
The quantum yield (F) of generation of [Ru III (Me 2 phen) 3 ] 3+ (21%) at the initial stage (0-1 min) was signicantly larger than that of [Ru III (bpy) 3 ] 3+ (1.6%) in the presence of 2.0 M H 2 SO 4 (Fig. S1 in the ESI †) because the one-electron oxidation potential of [Ru(Me 2 phen) 3 46 The F value at the initial stage (0-1 min) increased with increasing the concentration of H 2 SO 4 to reach 72% in the presence of 4.0 M H 2 SO 4 (  (Fig. S2 in the ESI †). At H 2 SO 4 concentrations higher than 2.0 M, the catalytic reactivity of Ir(OH) 3 nanoparticles decreased because Ir(OH) 3 nanoparticles became partially dissolved under strongly acidic conditions. Ir(OH) 3 nanoparticles were clearly dissolved in water with 3.0 M H 2 SO 4 . The photocatalytic reactivity increased with increasing the amount of Ir(OH) 3 nanoparticles, but it decreased through a maximum value with further increase in the amount of Ir(OH) 3 (Fig. 5b) because of the competition of the visible light absorption of [Ru II (Me 2 phen) 3 ] 2+ with that of Ir(OH) 3 nanoparticles. The F value of the photocatalytic H 2 O 2 production at l ¼ 450 nm was determined using a ferrioxalate actinometer to be 20% (0-30 min, Fig. S3 in the ESI †), which agrees with the F value of generation of [Ru III (Me 2 phen) 3 ] 3+ without WOC (vide supra).
Isotope-labelling experiments using 18     photocatalytic H 2 O 2 production, in which the produced H 2 O 2 comes from O 2 in the gas phase. Aer the reaction, the solution was carefully deaerated by bubbling with He gas to remove 18 O 18 O and an excess of MnO 2 , which catalysed the decomposition of H 2 O 2 to O 2 , was added to the solution. The evolved oxygen in the headspace of a reaction tube was separated using a gas chromatograph equipped with a molecular sieve column and analysed using a mass spectrometer (Fig. 6) (Fig. 8a). In the presence of Sc 3+ , however, the recovery of absorption becomes signicantly slower because of the binding of Sc 3+ to O 2 _ À , which prohibits the back electron transfer (Fig. 8b). The slower recovery was also observed for other metal ions (Fig. 9).
Thus, the acceleration effect of Sc(NO 3 ) 3 on the photocatalytic production of H 2 O 2 from H 2 O and O 2 results from the inhibition of back electron transfer by binding of Sc 3+ to O 2 _ À , which leads to more efficient generation of [Ru III (Me 2 phen) 3 ] 3+ . In addition, the scandium ion inhibited disproportionation of H 2 O 2 by Ir(OH) 3 under the same pH conditions as shown in Fig. 10.    was irradiated with visible light (l > 420 nm), the O 2 _ À -Sc 3+ complex was detected by the EPR spectrum in frozen H 2 O at 77 K as shown in Fig. 11, where the superhyperne structure due to the binding of Sc 3+ (I ¼ 7/2) was observed at g zz . The g zz value of O 2 _ À -Sc 3+ (2.036) is larger than the value reported in frozen acetonitrile at 143 K (2.030), 47     [Co III (Cp*)(bpy)(H 2 O)] 2+ (Fig. S7 in 2+ act as an efficient homogeneous photocatalyst and water oxidation catalyst, respectively. The F value of the photocatalytic H 2 O 2 production under visible light irradiation at l ¼ 450 nm was determined using a ferrioxalate actinometer to be 37% (0-30 min, Fig. S8a in the ESI †). The value of conversion efficiency from solar energy to chemical energy was also determined to be 0.25% (0-10 min, Fig. S8b in the ESI †). This value has reached the solar energy conversion efficiency of switchgrass (0.2%), 53 a promising crop for biomass fuel.
In conclusion, efficient photocatalytic production of H 2 O 2 from H 2 O and O 2 has been achieved using [Ru II (Me 2 phen) 3 ] 2+ as a photocatalyst and Ir(OH) 3 nanoparticles or [Co III (Cp*)-(bpy)(H 2 O)] 2+ as a WOC in water containing H 2 SO 4 or Sc(NO 3 ) 3 as shown in Scheme 1. The photocatalytic production of H 2 O 2 from H 2 O and O 2 using solar energy reported in this paper provides the most convenient and sustainable solar fuel that can be converted to electricity using an H 2 O 2 fuel cell. Further improvement of the catalytic reactivity and the more detailed elucidation of the catalytic mechanism are now in progress.