Recyclable PhotoFuel Cell for Use of Acidic Water as a Medium

We reported a new PhotoFuel Cell (PFC) comprising two photocatalysts for use of acidic water as a recyclable medium. Nitrogen and oxygen flow was required in the photoanode and photocathode, respectively. In this study, we developed a gas-circulating PFC that needs no gas supply from outside. In the gas-circulating PFC, the reverse reaction of water oxidation at the anode was prevented by the gas flow of photogenerated O2 from the anode to the cathode inside the PFC. The gas-circulating PFC accommodated an organic solvent layer over the aqueous electrolyte for the anode, and also a vent hole in the upper part of the Proton-Conducting Polymer (PCP) film. O2 transferred from the anode electrolyte to the organic solvent due to the solubility difference between the HCl solution and organic solvent. O2 transfer from the gas phase in the anode to that in the cathode was achieved by the vent hole in the PCP film due to the pressure difference due to the progress of the reaction. By the addition of a hexane layer to the anode of the PFC, it was demonstrated to achieve a photocurrent value of 69.7 μA per 1.3 cm2 of photocatalysts. However, in the stability tests for more than 7 h, the small amount of remaining O2 in the electrolyte (2.85 μmol L−1 ) exhibited serious effects on the PFC performance. The ISC , VOC and PMax values of the gas-circulating PFC were 29.2 μA, 1.18 V and 6.10 μW, that were 40%, 74% and 44%, respectively, of those for a N2 and O2 flow-type PFC. Apparently, photocurrents were dramatically suppressed by the reverse reaction at the photoanode in the extended tests for the gas-circulating PFC.


INTRODUCTION
Fossil fuels have been utilized as the essential energy source for industrialization. Industrial CO 2 emissions have led to an increase in the level of atmospheric CO 2 concentration (400 ppm), and the effects of this increase on global warming cannot be underestimated. The development of renewable energy as a replacement for fossil fuels has been slow [1,2].
Among the renewable energies, solar energy has the greatest potential. Although the Silicon Solar Cell (Si SC) has been commercialized, the technology that can convert solar energy to electricity often needs subsidies to spread more widely [3]. Other types of SC, e.g. the dye-sensitized SC and compound SC [4], and Fuel Cells (FC) that use hydrogen/methanol fuel [5][6][7][8], potentially obtained using solar energy [9], have been extensively investigated. However, all the requisites (sustainability, durability and an electromotive force of 1-3 V per cell) have not been fully satisfied.
Recently, a new device was demonstrated: a PFC utilizing two photocatalysts of TiO 2 and silver(0/I)-doped TiO 2 [10] on an electrode film, both immersed in acidic solutions separated by a PCP film [11]. Platinum is not used [5][6][7][8][12][13][14][15][16], and only inexpensive materials were used in the PFC for sustainability. The PFC is similar to photoelectrochemical cells (PEC) consisting of a photoanode and a photocathode to form O 2 and H 2 , respectively, from water [17,18] and also PEC consisting of a photoanode to oxidize water [19] or organic dye fuel [20] and a photocathode to reduce protons. The PFC is different from these PEC in that the medium, water, is recycled in the cell and no fuel is needed except for natural light [21].
Although PFC enable higher electromotive force (1.8 V) [11,21] and use only acidic water as a medium and inexpensive catalysts, PFC need a gas supply to promote surface reactions at both the anode and cathode. At the anode, O 2 gas needs to be evacuated by N 2 flow to prevent a reverse surface reaction: On the other hand, at the cathode reactant O 2 should be supplied for the reaction (1) to proceed. Therefore, PFC need gas cylinders to supply these gases for effective power generation. This situation limits the design of compact enough PFC that can be accommodated in portable electronic devices.
Herein, we developed a PFC in which gas circulates internally. By circulating formed O 2 gas from the anode to the cathode, the reverse reaction at the anode could be prevented. The concept of gas-circulating PFC is shown in Figure 1a, Test 3. The PFC has two photocatalysts (TiO 2 and Ag-TiO 2 ) immersed in HCl solution. The electrolyte at the anode and that at the cathode (Fig. 1b2, 3) are separated by a PCP (Fig. 1b4, 5). As the mechanism of gas circulation, the PFC designed in this study comprises an organic solvent phase on an anode electrolyte (Fig. 1b1) and a vent hole in the upper part of the PCP film (Fig. 1b4, 5). The organic solvent can dissolve a greater amount of O 2 gas than water can. O 2 gas circulated in the PFC as follows. O 2 generated by the anode reaction transferred preferentially into the organic solvent due to the difference in O 2 solubility between the HCl solution and the organic solvent. Then, the dissolved O 2 in the organic solvent moved to the upper gas phase due to the increased O 2 concentration in the solvent in the compartment of the anode, and then to the upper gas phase in the compartment of the cathode through the vent hole in the PCP film due to the pressure difference between the O 2 formed at the anode and O 2 consumed at the cathode (Fig. 1a, 1b1). Finally, the gaseous O 2 that moved to the cathode side reached the photocatalyst on the cathode and was photoreduced.

Sample Preparation of the Photoelectrodes
TiO 2 powder [1.00 g; P25, Degussa; anatase/rutile phases = 7/3; Brunauer-Emmett-Teller (BET) surface area (S BET ) = 60 m 2 g À1 ] was suspended in 3.0 mL of deionized water (<0.06 lS cm À1 ) and then stirred well. The obtained suspension was dried at 373 K for 24 h and heated in air at 673 K for 2 h. The resultant powder was suspended in 75 lL of deionized water and placed on an Indium Tin Oxide (ITO, thickness 1.2-1.6 lm)-coated Pyrex glass plate. The TiO 2 /ITO/Pyrex was dried at 373 K for 18 h and heated in air at 573 K for 30 min. The amount of TiO 2 deposited on the ITO-coated glass plate was 5.0 mg and covered an area of 1.3 cm 2 .
Silver nitrate (159 mg; 99.8%, Wako Pure Chemicals) was dissolved in 10 mL of deionized water. The solution was mixed with 3.33 g of untreated TiO 2 (P25). The mixture was magnetically stirred at a rate of 850 rotations per minute (rpm) and the water was distilled at 353 K. The obtained powder was dried at 373 K for 24 h and heated in air at 673 K for 2 h. The resultant Ag-TiO 2 powders contained 3.0 wt% of Ag. They were suspended in a minimum amount of water and placed on ITO-coated glass (Ag-TiO 2 /ITO/ Pyrex) in a manner similar to that for the TiO 2 /ITO/Pyrex.
The distance between the light exit (/ = 5 mm) and the TiO 2 or Ag-TiO 2 film was 46 mm. The light intensity was 8 mW cm À2 at the center of the photocatalyst film on each electrode.

Test 2: Gas-Closed Conditions
PFC comprising TiO 2 /ITO/Pyrex and Ag-TiO 2 /ITO/Pyrex electrodes were prepared and immersed in HCl solutions (40 mL in each compartment; initial pH 2.0) in a similar procedure to that in Test 1. The two compartments were separated by a 50-lm-thick PCP film. N 2 and O 2 gases purged the compartments of TiO 2 /ITO/Pyrex and Ag-TiO 2 /ITO/Pyrex, but the gas supply was stopped before the start of Test 2. Both the TiO 2 and Ag-TiO 2 photocatalysts were irradiated with UV-visible light in the closed compartments filled with N 2 and O 2 , respectively.

Test 3: Gas-Circulating Conditions inside the PFC
An Ag-TiO 2 /ITO/Pyrex electrode was immersed in 50 mL of HCl solution (initial pH 2.0), whilst TiO 2 /ITO/Pyrex was immersed in 50 mL of the HCl solution phase (initial pH 2.0) and 50 mL of hexane phase was added above the HCl solution phase (Fig. 1). All the area of the TiO 2 film was immersed in the HCl solution phase and not in contact with the hexane phase. N 2 and O 2 gases purged the compartments of TiO 2 /ITO/Pyrex and Ag-TiO 2 /ITO/Pyrex, 3) Figure 1 a) Components of the PFC comprising TiO 2 and Ag-TiO 2 photocatalysts used for Tests 1-5. Oxidation by photogenerated holes at the anode and reduction by photogenerated electrons take place. Remaining electrons at the anode flow to the cathode and combine with the remaining holes. b) Photographs of the PFC for Test 3 (1). Photoanode side (2), photocathode side (3), and the PCP with a vent hole (4, 5).
respectively, but the gas supply was stopped before the start of Test 3. Both the TiO 2 and Ag-TiO 2 photocatalysts were irradiated with UV-visible light, set in the closed compartments filled with N 2 and O 2 , respectively. The two compartments were separated by a 50-lm-thick PCP film; however, in Test 3, gas circulation inside the PFC was enabled through a vent hole (6.0 mm 9 3.0 mm) in the upper part of the PCP film ( Fig. 1a, b4, b5) from the anode compartment in which O 2 was formed to the cathode compartment in which O 2 was consumed. The conditions of UV-visible light irradiation were similar to those for Tests 1 and 2 (Tab. 1).

Test 4: Gas-Closed Conditions Plus Hexane@Anode
An Ag-TiO 2 /ITO/Pyrex electrode was immersed in 55 mL of HCl solution (initial pH 2.0), whilst TiO 2 /ITO/Pyrex was immersed in 55 mL of the HCl solution phase (initial pH 2.0) and 55 mL of the hexane phase above the HCl solution phase (Fig. 1). All the area of the TiO 2 film was immersed in the HCl solution phase and not in contact with the hexane phase. N 2 and O 2 gases purged the compartments of TiO 2 /ITO/Pyrex and Ag-TiO 2 /ITO/Pyrex, respectively, but the gas supply was stopped before the start of Test 4. Both the TiO 2 and Ag-TiO 2 photocatalysts were irradiated with UV-visible light, set in the closed compartments filled with N 2 and O 2 , respectively. The two compartments were separated by a 50-lm-thick PCP film. The film was complete and did not comprise any vent holes as in Test 3. The conditions of UV-visible light irradiation were similar to those for Tests 1-3 (Tab. 1).

Test 5: Gas-Circulating Conditions inside the PFC Minus Hexane@Anode
An Ag-TiO 2 /ITO/Pyrex electrode was immersed in 40 mL of HCl solution (initial pH 2.0), whilst TiO 2 /ITO/Pyrex was immersed in 40 mL of the HCl solution phase (initial pH 2.0). N 2 and O 2 gases purged the compartments of TiO 2 /ITO/Pyrex and Ag-TiO 2 /ITO/Pyrex, respectively, but the gas supply was stopped before the start of Test 5. Both the TiO 2 and Ag-TiO 2 photocatalysts were irradiated with UV-visible light, set in the closed compartments filled with N 2 and O 2 , respectively. The two compartments were separated by a 50-lm-thick PCP film. The conditions of UVvisible light irradiation were similar to those for Tests 1-4. This PFC had a vent hole (6.0 mm 9 3.0 mm) in the upper part of the PCP film (Fig. 1a, b4, b5) similar to Test 3 (Tab. 1). For Tests 1-5, photocurrent generation (photocurrent tests) was performed by connecting the external parallel circuit for one route via a voltmeter and another route via an ammeter with a resistance of 0.5 X (Fig. 1a). Both catalysts were exposed for 30 min to UV-visible light and then kept in the dark for 30 min, and this process was repeated five-seven times. The current and voltage between the two electrodes were monitored.

Current-Voltage Dependence Tests
The current (i)-voltage (V) characteristics were also measured for several types of PFC using a similar parallel circuit (Fig. 1a). After the photocurrent test for Figure 2 (2.5, 3.5 and 3.5 h of irradiation of UV-visible light for Tests 1, 2 and 4, respectively), the resistance was gradually decreased from 500 kX to 0.3 X over 20 min to plot the i-V dependence. For comparison, a PFC in the conditions of Test 1  The time course of photocurrents of the PFC comprising TiO 2 and Ag-TiO 2 photocatalysts in Tests 1 (gas flow), 2 (gas closed), 3 (gas circulating inside the PFC), 4 (gas-closed conditions plus hexane@anode), and 5 (gas-circulating conditions inside the PFC minus hexane@anode). The pH values of the electrolytes in both the anode and cathode were 2.0.
(Sect. 1.2.1) was also prepared and i-V characteristics were also measured similar to the experiments described above, but in darkness.

Product Analyses during Photocurrent
The HClO formed was evaluated by the N,N-Diethyl-p-PhenyleneDiamine (DPD) method [22]. 20 mg of DPD reagent (0.8 mg of DPD sulfate salt (H 2 SO 4 ) and 19.2 mg of Na 2 SO 4 ) were dissolved in 1.0 mL phosphate buffer solution (pH 6.5). This solution was mixed with 5.0 mL of 0-39.0 lmol L À1 of NaOCl aqueous solutions and used as a standard solution. The mixed solutions (6.0 mL) were used to obtain a linear regression relationship between the absorbance at 515 nm and the concentration of ClO À ions.
Next, 20 mg of DPD reagent were dissolved in 4.0 mL phosphate buffer solution (pH 6.5). This solution was mixed with 2.0 mL of sample solution (HCl aqueous solution) after a photocurrent Test 1 using a PFC (Fig. 1) for 3 h at pH 2.0. The UV-visible spectrum of the mixed solution (6.0 mL) was measured by UV-visible spectrometer (Model V650, JASCO). The concentration of HClO in the sample was determined based on the absorbance at 515 nm and the relationship obtained using NaOCl standard solutions. The reason for the volume difference for the phosphate buffer and sample solutions was to adjust the pH within 6.4 and 6.6 for mixed solutions (6.0 mL).

Photocurrent Tests
The time courses of the photocurrent tests are summarized in Figure 2. For Test 1, in response to the UV-visible irradiation, photocurrents gradually increased from 43.7 to 60.5 lA in five cycles at pH 2.0. In contrast, the photocurrents increased and converged to a constant value (26.7 lA) in the test at pH 3.0 (not shown) [11]. Thus, the active metallic Ag 0 species for photoreduction of O 2 at the cathode should gradually increase, transformed from Ag 2 O by the reaction with protons, in the five cycles at pH 2.0. The transformation of the active Ag species (Ag 0 , Ag 2 O and AgCl) was reported based on an extended X-ray absorption fine-structure study [11].
At pH 4.0, a PFC current of 3 lA was obtained (not shown) for 2.5 h irradiated by UV-visible light (0.28 lmol-e À generated in total) [11]. The pH values starting from 4.0 for the HCl solution (40 mL, 4.0 lmol-H + ) remained constant after the PFC test at both the anode and cathode. The pH change due to each half reaction was calculated as ±0.03. As this value is nearly the accuracy of pH monitoring under these conditions, we cannot judge that the photooxidation of water (at the anode) and photoreduction of O 2 (at the cathode) exclusively took place. The possibility of a by-reaction, e.g. the consumption of acid, remained. For the test at pH 2.0, the initial proton concentration was 400 lmol in 40 mL of electrolyte solution at both the anode and cathode, while the PFC current for 2.5 h was 4.7 lmol-e À in total. The pH change due to each half reaction is calculated as ±0.005. It was even more difficult to judge the possibility of photooxidation of water and/or anions of acid under the conditions. In Test 2, the photocurrent in the first cycle (34.7 lA) was similar to that in Test 1 (43.7 lA). In response to the UV-visible irradiation, photocurrents gradually increased to 35.8 lA in three cycles. In contrast to the gradual increase in photocurrents throughout 5 h of the study in Test 1, the photocurrents peaked at 35.8 lA in the third cycle and gradually decreased to 32.2 lA in the seventh cycle in Test 2 (Fig. 2). This decrease by 3.6 lA suggested partial deactivation of the TiO 2 photocatalyst at the anode. One of the major reasons is the reverse reaction of produced O 2 which remained dissolved in the HCl solution to water (Eq. 1). In contrast to the gas-flowing Test 1, the anode compartment of TiO 2 /ITO/Pyrex was purged with N 2 gas and produced O 2 remained in the compartment for Test 2 (Tab. 1).
The photocurrent in the first cycle (54.8-40.3 lA) was clearly higher in Test 3 than that in Tests 1 and 2. The photocurrent decreased to 36.5 lA at the end of the second cycle (Fig. 2) and continued to decrease gradually to 31.0 lA at the end of the seventh cycle. The photocurrents in the first cycle were probably higher because N 2 was purged in the anode compartment more completely than in Tests 1 and 2 by the effect of hexane to preferably dissolve a contamination level of O 2 (~0.04 ppb). This evaluation was based on the impurity of O 2 (< 1 ppm) in the N 2 gas used and the equilibrium of dissolution. Conversely, the decrease in the photocurrent in seven cycles suggested an even smaller amount of formed O 2 remained in the anode and inhibited the PFC performance considerably, and also unexpected reverse diffusion of O 2 from the cathode into the anode was not excluded through the vent hole.
In Test 4, the photocurrent in the first cycle (69.7 lA) was clearly higher than that in Tests 1-3. However, in response to the UV-visible irradiation, photocurrents gradually decreased to 51.9 lA in seven cycles (Fig. 2). The photocurrent in Test 4 was always higher than that in Tests 2 and 3 throughout the photocurrent test. Thus, the positive effects of hexane on PFC were confirmed (Tab. 1), but unexpectedly, the vent hole in Test 3 contributed negatively to the photocurrent generation. Furthermore, the linear decrease of the photocurrent as a function of time that elapsed during Test 4 was more gradual than the exponential-like decrease as a function of time in Test 3.
In Test 5, the photocurrent in the first cycle (41.0-33.0 lA) was similar to that in Test 2 (Fig. 2). Then, it decreased to 29.2 lA in the second cycle. Later than the third cycle, the photocurrents became stable at 28.4 lA. This stabilized current was lowest among the Tests 1-5, as the result of produced O 2 which remained dissolved in the HCl solution to water, and the unexpected reverse diffusion of O 2 in the cathode into the anode through the vent hole, as in Test 3 (Tab. 1).
In Test 2, 1.11 lmol of O 2 were generated during the 7-h photocurrent test based on the total photocurrent for 3.5 h and assuming the next reaction (Eq. 1 is the reverse reaction of Eq. 4): Thus, 1.85 and 0.157 lmol of O 2 should be distributed to the hexane and HCl solutions, respectively, in equilibrium. Thus, the O 2 concentration in the anode electrolyte (HCl solution) was calculated as 2.85 lmol L À1 . Hence, the O 2 concentration at the end of Test 4 was only 10% of that at the end of Test 2. This difference accounted for the higher current density in Test 4 by a factor of two as compared with Test 2 (Fig. 2).

i-V Characteristic Tests
The obtained i-V characteristics are shown in Figure 3. Furthermore, the properties of PFC calculated based on i-V data in Tests 1, 2 and 4 are summarized in Table 2. In Test 1, the i value gradually increased as the cell voltage decreased, starting from open circuit voltage (V OC ) at 1.59 V, which is a similar trend to the i-V dependence of the SC [24] or the Ohmic i-V characteristics of the PFC comprising TiO 2 and Ag-TiO 2 photocatalysts in Test 1 (gas flow), Test 2 (gas closed), Test 4 (gas-closed conditions plus hexane@anode), and control Test 1' in darkness (gas flow).
loss region of a Polymer Electrolyte Fuel Cell (PEFC). When the voltage became less than 0.45 V, the current increased linearly from 20 to 74 lA (short-circuit current, I SC ). This increasing trend was different from that in a SC or the transportation limit region of a PEFC [7,8].
In Test 2, the shape of the i-V characteristic curve was similar to that for Test 1. However, the increase in currents in the lower voltage region (! 0.45 V) was less steep than the slope in Test 1. The values of I SC (29.8 lA), V OC (1.23 V) and P Max (5.51 lW) were 40, 77 and 39% of the corresponding values obtained in Test 1. The lower photocurrents in Test 2 suggested that the reverse reaction 1 that occurred in the anode seriously affected the performance of the PFC.
In Test 4, the curve of i-V characteristics was similar to that for Test 2; the increase in photocurrents in the lowvoltage region (! 0.45 V) was less steep than the slope in Test 1. The values of I SC (29.2 lA) and V OC (1.18 V) were slightly lower than the corresponding values in Test 2, whilst the value of P Max (6.10 lW) was clearly higher than that in Test 2 (5.51 lW). This low performance (Fig. 3) was inconsistent with the better performance in photocurrent Test 4 than that in Tests 1 and 2 (Fig. 2). This trend cannot be rationalized based on the remaining amount of produced O 2 dissolved in the anode. One of the possibilities is the polarization of photoelectrodes during the i-V measurements at high voltage to modify the band bending (charge balance at the surface) at the surface of photocatalysts [25][26][27][28][29].
In a control Test 1' in darkness, the value of the current was always less than the detection limit by changing the resistance from 500 kX to 0.3 X. The corresponding voltage was less than 0.002 V. Therefore, the values of I SC , V OC and P Max were less than 0.1 lA, 0.002 V and 0.0002 lW, respectively. This indicated electricity of the PFC was generated exclusively by irradiation with UV-visible light.   The time course of dissolved O 2 concentrations during the photoirradiation test on TiO 2 . UV-visible light and FeCl 3 Á6H 2 O (2.31 mmol L À1 ) were used as excitation light and a sacrificial reagent, respectively. The initial pH value was 2.0. the gas phase. The increase in dissolved O 2 in the first cycle was the quickest among all the cycles. The increase in dissolved O 2 was 63.1 lmol L À1 per 0.25 h. The slope of the dissolved O 2 gradually became less steep as the cycles were repeated. As a result, in the fifth cycle, it was 36.2 lmol L À1 per 0.25 h. This trend is contradicted by the gradual increase in the photocurrent in Figure 2, Test 1 as the cycles repeated. We suspect that the major reason was that the transportation of dissolved O 2 to the gas phase became faster as the concentration of dissolved O 2 increased, as the cycles were repeated (Fig. 4).

Product Analyses during Photocurrent
The increase rate of dissolved O 2 in the first cycle was: 63:1 lmol L À1 Â 0:090 L=0:25 h=0:151 g cat In comparison, the total photocurrent in Test 2 was average current: 34:0 lA Â 3 600 s h À1 =9:65 Â 10 4 A s mol-e À1 =0:0050 g cat The electron flow rate corresponded to the O 2 generation rate of 63 lmol-O 2 h À1 g cat À1 . The net O 2 generation rate in Figure 4 should be even higher because the O 2 transported to the gas phase was not included for the evaluation in Figure 4. This difference in the amount based on the PFC current was not inconsistent. In the presence of a sacrificial oxidation agent, Fe 3+ , photoexcited electrons to the conduction band of TiO 2 easily diffuse with sufficient Fe 3+ and react in contrast so that the electrons need to travel to the cathode to react with holes at the valence band of [Ag-]TiO 2 in the PFC. The concentration of holes was determined by the balance of charge separation by light and the cathode reaction rate of O 2 reduction [11,21].

Analysis of Hypochlorous Acid Formed at the Anode from Cl À Ions
The absorbance at 515 nm for standard mixed solutions (6.0 mL) was plotted as a function of NaOCl concentrations in the solution (6.0 mL). Then, a standard curve was calculated as follows (Fig. 5): 20 mg of DPD reagent in 4.0 mL phosphate buffer solution (pH 6.5) were mixed with 2.0 mL of sampled solution from the anode solution (40 mL) of the PFC after Test 1 for 3 h. The absorbance of this mixed solution was 0.018 ± 0.008. Therefore, the amount of HClO in the mixed solution (6.0 mL) was 3.98 nmol based in Equation (8).
Thus, the amount of HClO in the anode solution was calculated as: 3:98 nmol Â 40 mL=2:0 mL ¼ 0:080 AE 0:056 lmol ð9Þ For 3 h of Test 1 (pH 2.0), total electron flow was 3.92 lmol based on the integration of photocurrents. Two holes reacted with Cl À ions (Eq. 2) correspond to one HClO formed from Cl 2 (Eq. 3). As the concentration of HClO was the impurity level for the anode solution in Test 1, the amount of HClO was nearly the detection limit. Therefore, the upper limit of selectivity to proceed with Equation (2) for photogenerated holes was: 2 Â 0:080 lmol-HClO=3:92 lmol-h þ ¼ 0:041 AE 0:031 ð10Þ The participation of chlorine ions in the anode reaction cannot be denied; however, the ratio was as low as 4.1% at most.

CONCLUSIONS
First, the anode reaction in the PFC was demonstrated exclusively to be the photooxidation of water over TiO 2 based on the O 2 evolution test, and negligible (or no) photooxidation of Cloccurred based on the detected amount of HClO. The selectivity of photooxidation of Clwas 4.1% at most as compared with the photooxidation of water. Next, a product gas-trapping PFC was demonstrated to achieve a  Absorbance at 515 nm for standard mixed solutions as a function of the concentration of HClO standard solutions. photocurrent value of 69.7 lA per 1.3 cm 2 of photocatalysts (53.6 lA cm À2 ). Gas trapping was enabled by accommodating an organic solvent layer over the aqueous electrolyte solution phase in the anode and utilizing the solubility difference of produced O 2 . However, gas circulation by a vent hole in the upper part of the PCP film due to pressure differences was not well achieved. To transfer generated O 2 to an anode, a greater pressure difference was needed. The performance of the gas-closed PFC equipped with a hexane layer on the anode (Test 4) was the best at the initial stage of the photocurrent test due to the O 2 solubility effect of hexane rather than the other conditions. However, the photocurrents gradually decreased during the measurements for 7 h. The small amount of remaining O 2 in the anode electrolyte (2.85 lmol L À1 ) reduced the PFC performance considerably in Test 4. The I SC , V OC and P Max values of the gascirculating PFC were 29.2 lA (22.5 lA cm À2 ), 1.18 V and 6.10 lW (4.69 lW cm À2 ), respectively. These values were similar to corresponding values for the gas-closed PFC.
The feasibility of a gas-trapping PFC was successfully demonstrated in this study. Further, the improvement of power efficiency is required by the optimization of photocatalysts and the thickness and density of photocatalyst layers in order to be accommodated in portable electronic devices, remote-control light gliders, and environmental cameras/ sensors.