Transition metal (Ni, Fe, and Cu) hydroxides enhanced α-Fe₂O₃ photoanode-based photofuel cell

Abstract

Photofuel cells have been demonstrated to be a promising strategy for generating electricity using biomass. Here, we present a photofuel cell with a visible light α-Fe₂O₃ based photoanode that can be directly powered by a variety of biomasses such as methanol, glycerol, glucose, cellubiose and starch. The photocurrent density and power density of the photofuel cell are significantly enhanced by loading cocatalysts (metal hydroxides, e.g. Ni(OH)₂) on the α-Fe₂O₃ photoanode. The power density of the photofuel cell powered by glucose is enhanced over two times from 0.082 mA cm⁻² for α-Fe₂O₃ to 0.18 mW cm⁻² for Ni(OH)₂/α-Fe₂O₃ photoanode.

---

Introduction

The utilization of abundant biomass as an energy source can prevent environmental pollution and reduce the dependence on fossil resources.^1,2 The technological challenge is to sustainably capture biomass energy using green processes. Direct alcohol fuel cells (DAFCs) with potentials of high power density and being pollution-free offer a possible solution to this problem.^3,4 Although DAFCs powered by low molecular weight alcohols (usually methanol or ethanol) have been demonstrated to be practically feasible, biomass-derived compounds, such as glucose, polysaccharide, and cellulose, have not been reported as fuels for DAFCs. This is because of the low activity of catalysts in the activation of the C–C bonds and the oxidation of biomass at low temperatures.^5,6 Biofuel cells can be powered by various biomasses at low temperatures; however, low electric power output, limited lifetime and rigorous reaction conditions seriously hinder their applications.^7,8

A photofuel cell (PFC) mainly consists of a semiconductor photoanode, a cathode, electrolyte and fuels.^9–14 With PFC technology, biomass as a fuel can be oxidized on the photoanode under light irradiation and oxygen is reduced on the counter electrode through an external circuit, generating electricity. In essence, PFC is different from the fuel cells reported previously and widens the range of biomass fuels for electricity generation. Thus, various biomasses, such as glycerol, glucose, saccharides, proteins, and ammonia, can be used as fuels for PFCs with a TiO₂ photoanode. Previous work has demonstrated that PFCs can convert biomass into electricity even at low ambient temperatures. However, TiO₂ and WO₃ with band gaps of 3.2 eV and 2.8 eV, respectively, use only a small proportion of solar irradiation.¹⁵ Therefore, semiconductors with wide ranges of light absorption are highly desired for PFCs.

Hematite (α-Fe₂O₃), with an optical band gap of 2.1 eV, is a promising material for photoanodes due to its abundance, ecofriendly nature, and photochemical stability in basic electrolytes.^16–18 However, its short excited state lifetime (∼1 ps) and the small hole diffusion length (∼2 to 4 nm) significantly hinders its efficiency in charge separation and collection. To overcome these problems, various strategies have been adopted, including reducing the size,^19,20 doping^21–23 and loading cocatalysts.^24–26 Among these strategies, loading cocatalysts is the most efficient method for photoelectrochemical water splitting. However, α-Fe₂O₃ loaded with an appropriate cocatalyst that can distinctly enhance biomass oxidation has not been reported.

In this work, we present a PFC with α-Fe₂O₃ based photoanodes that directly converts biomass-derived compounds such as methanol, glucose, glycerol, cellubiose, and starch. To improve the PFC efficiency and the stability of α-Fe₂O₃ photoanode, transition metal hydroxides Ni(OH)₂, Fe(OH)₃, and Cu(OH)₂ are loaded as biomass oxidation cocatalysts. It was found that cocatalysts can remarkably enhance the photoresponse and stability of bare α-Fe₂O₃ for biomass oxidation, which demonstrates an example of stimulating a PFC by loading an efficient cocatalyst on the photoanode.

Experimental

Materials

All the chemicals were of analytical grade and were used as purchased. Solutions were prepared using high purity water (Millipore Milli-Q purification system, resistivity > 18 MΩ cm). FTO (fluorine-doped tin oxide) conductive glass was purchased from Nippon Sheet Glass Company 5 (Japan) and was ultrasonically cleaned with acetone, ethanol and deionized water for 20 min each in sequence prior to use.

Preparation of the Fe₂O₃ films

α-Fe₂O₃ films were deposited on an F-doped SnO₂ (FTO) substrate electrode using a modified chemical bath deposition method reported elsewhere.²⁷ Specifically, 0.2 mol L⁻¹ FeCl₃ (FeCl₃·6H₂O, ≥99%, Shanghai Chemical) aqueous solution (10 mL) and 0.3 mol L⁻¹ NH₂CONH₂ (98%, Shanghai Chemical) were mixed in a 50 mL glass beaker and heated at 100 °C for 4 h. FTO was placed vertically in this beaker with the conducting edge facing the wall of the beaker. After the reaction, the film formed on FTO was thoroughly rinsed with high purity water and annealed at 500 °C for 3 h to obtain the desired phase. Finally, the prepared sample was further annealed at 750 °C for 10 min.

Fabrication of M(OH)_x/Fe₂O₃ photoanode

Ni(OH)₂ was deposited onto α-Fe₂O₃ by a successive ionic layer adsorption and reaction method. In a typical synthesis, α-Fe₂O₃ electrodes were dipped into 0.1 mol L⁻¹ Ni(NO₃)₂ (≥98.5%, Shanghai Chemical) solution for 40 s, followed by drying with compressed air. Then, the electrodes were dipped into 1 mol L⁻¹ KOH solution for another 40 s and dried with compressed air. Next, the prepared sample was washed with ethanol and dried at 60 °C for 1 h in air. Fe(OH)₃/α-Fe₂O₃ and Cu(OH)₂/α-Fe₂O₃ samples were prepared using the same method without further heat treatment.

Characterization of the electrodes

The prepared samples were characterized by X-ray powder diffraction (XRD) on a Rigaku D/Max-2500/PC powder diffractometer using Cu Kα radiation (operating voltage: 40 kV, operating current: 20 mA, scan rate: 5° min⁻¹). The UV-visible diffuse reflectance spectra were recorded on a UV-visible spectrophotometer (JASCO V-550) and calibrated by the Kubelka-Munk method. The morphologies of the electrodes were examined using a Quanta 200 FEG scanning electron microscope (SEM) equipped with an energy dispersive spectrometer (accelerating voltage of 20 kV). Transmission electron microscopy (TEM) images were taken on a Tecnai G² Spirit (FEI company) using an accelerating voltage of 120 kV. High-resolution transmission electron microscopy (HRTEM) images were acquired on Tecnai G² F30 S-Twin (FEI company) with an accelerating voltage of 300 kV. The liquid products quantitative analysis was carried out using HPLC (Agilent 1200) with refractive index (RI) detector and ultraviolet (UV, λ = 210 nm) detector for arabinose, erythrose, glyceraldehyde, glycolaldehyde, glycolate and formate. The reactants and products were separated through an ion exclusion column (Alltech OA-1000) heated at 55 °C. The eluent was a solution of H₂SO₄ (0.005 mol L⁻¹). Products were identified through comparison with standard samples, which were obtained from Sigma-Aldrich.

Photoelectrochemical and electrochemical measurements

All the photoelectrochemical measurements were carried out in a three-electrode cell with a flat quartz window to facilitate the illumination of the photoelectrode surface. The working electrode was α-Fe₂O₃, and Hg/HgCl₂ (saturated KCl) and a Pt plate (2 cm × 4 cm) were used as a reference and counter electrode, respectively. A Nafion membrane was used to prevent the crossover between the anode and cathode. The illumination source was a 300 W Xe arc lamp and the light intensity at the surface of the electrodes was 300 mW cm⁻². The electrochemical measurements were performed on a CHI 760D electrochemical workstation (CHI, Shanghai) at room temperature. The electrolyte was 1 mol L⁻¹ KOH solution with/without 0.025 mol L⁻¹ glucose or methanol (10 vol%), glycerol (0.025 mol L⁻¹), cellubiose (0.025 mol L⁻¹) and starch (1 wt%).

A photofuel cell was constructed using a two-compartment quartz cell with a Nafion membrane as the separator. α-Fe₂O₃ and Ni(OH)₂/α-Fe₂O₃ photoanodes were used as anodes and a Pt wire was used as the cathode, while a Nafion membrane served as the separator. The distance between the anode and the cathode was 10 cm. The photovoltaic performance of the cells was measured with a Keithley 2400 source measure unit irradiated by AM 1.5 (100 mW cm⁻²).

The energy efficiency (PEC) and fill factor (FF) of such a device can be estimated by eqn (1) and (2), respectively.

PEC = P_Out/ P_hν × 100% (1)

FF = (V_max × I_max)/(V_oc × I_sc) (2)

P_Out and P_hν represent the output electrical power and the input photochemical energy, respectively; I_sc is the measured short circuit current; and V_oc is the specified open current voltage. I_max and V_max correspond to the current and voltage at the maximum power point, respectively.

Results and discussions

The XRD patterns of Fe₂O₃ film are shown in Fig. 1a. The sample shows diffraction peaks with 2θ at 35.6°and 64.0°, which correspond to the indices of the (110) and (300) planes (PDF no. 840306) of hematite phase, respectively. The UV-visible spectrum (Fig. 1b) gives the bands of α-Fe₂O₃ at 410 nm and around 600 nm, which corresponds to the direct transition of O²⁻ 2p → Fe³⁺ 3d and the transition of the spin-forbidden Fe³⁺ 3d → 3d, respectively.²⁸ Fig. 1c and d illustrate the top-view analysis and morphology of α-Fe₂O₃ film characterized by SEM. The film consists of α-Fe₂O₃ nanorod particles with a diameter of 20–30 nm and a length of 60–100 nm. The cross analysis of the film shows a thickness of about 300 nm with a uniform and continuous morphology.

Fig. 1 (a) XRD pattern and (b) UV-vis absorption spectra of Fe₂O₃ film. (c) Top view and (d) side view SEM images of Fe₂O₃ film.

Fig. 2 shows the dark and photocurrent densities of an α-Fe₂O₃ photoanode under light illumination. The dark response is negligible up to 0.6 V vs. SCE for both 1 mol L⁻¹ KOH electrolyte and 1 mol L⁻¹ KOH containing glucose. Above 0.6 V vs. SCE, an anodic current is formed due to water or glucose oxidation. This suggested that the α-Fe₂O₃ photoanode has equal oxidation ability for glucose and water under these conditions. In KOH, the onset potential is shifted to −0.43 V vs. SCE, and the photocurrent is due to water oxidation. When glucose is added to the aqueous KOH electrolyte, the photo-onset potential shifts to a lower value of −0.6 V vs. SCE and the photocurrent density obviously increases. These results indicate that glucose is preferentially oxidized over water and more photo-generated electrons and holes are used for O₂ reduction and glucose oxidation. These results also suggest that the α-Fe₂O₃ photoanode can efficiently oxidize biomass in a photochemical process, which implies that sunlight utilization of PFC could be extended to the visible light region.

Fig. 2 Dark and photocurrent densities for α-Fe₂O₃ photoanode in 1 mol L⁻¹ KOH electrolyte and in 1 mol L⁻¹ KOH electrolyte containing 0.025 mol L⁻¹ glucose; light source: 300 W Xe lamp; scanning rate: 20 mV s⁻¹.

To improve the activity and stability of the α-Fe₂O₃ photoanode, transition metal hydroxides Ni(OH)₂, Fe(OH)₃, and Cu(OH)₂ were deposited on α-Fe₂O₃ film by a successive ionic layer adsorption and reaction method. SEM elemental mapping was carried out to detect the state of the Ni(OH)₂ modification on Fe₂O₃. As shown in Fig. S1,† Ni is distributed randomly on Fe₂O₃ and the mass ratio of Ni to Fe is ca. 0.24, as detected by EDS. To further study the microstructure of Ni(OH)₂/α-Fe₂O₃, HRTEM analyses were conducted. HRTEM images of a representative Ni(OH)₂/α-Fe₂O₃ photoanode are shown in Fig. 3a and b. The HRTEM image confirms the presence of Ni species on the edge of α-Fe₂O₃. The magnified HRTEM image in Fig. 3b exhibits fringes with a lattice spacing of ca. 0.37 nm, which corresponds to the (012) plane of α-Fe₂O₃, while the lattice fringes with a spacing of 0.218 and 0.154 nm are consistent with the (103) and (300) planes of the 3Ni(OH)₂·2H₂O (α-Ni(OH)₂). The EDS linescan performed across the α-Fe₂O₃ nanorod shows the distinct spatial profiles of Fe and Ni, confirming that Ni species were deposited on the Fe₂O₃ surface (Fig. S2, ESI†). These results reveal that α-Ni(OH)₂ nanoparticles adhere to the surface of α-Fe₂O₃, forming heterojunctions between the two components.

Fig. 3 (a) HRTEM image of Ni(OH)₂/α-Fe₂O₃ and (b) magnified HRTEM image of the selected frame from image (a).

Fig. 4a shows the linear sweep voltammetric (LSV) curves of the α-Fe₂O₃ and Ni(OH)₂/α-Fe₂O₃ photoanodes in 1 mol L⁻¹ KOH electrolyte with glucose under chopped light illumination. After loading Ni(OH)₂ cocatalyst, the photocurrent for glucose oxidation on the α-Fe₂O₃ electrode is significantly increased. The Ni(OH)₂ film itself shows no photoresponse for glucose oxidation (Fig. S3, ESI†). In 1 mol L⁻¹ KOH electrolyte without glucose, the photocurrent curve of the Ni(OH)₂/α-Fe₂O₃ electrode is different from that of the α-Fe₂O₃ electrode (Fig. S4, ESI†), which may be because of the oxidation of Ni(OH)₂ in the photoelectrochemical process. This suggests that Ni(OH)₂ acts as an active site for glucose oxidation.

Fig. 4 (a) LSV curves of Fe₂O₃ and Ni(OH)₂/α-Fe₂O₃ photoanodes under chopped light illumination; (b) the possible process of glucose oxidation over Ni(OH)₂/Fe₂O₃ photoanode; (c) and (d) LSV curves of α-Fe₂O₃ photoanodes modified with Fe(OH)₃ and Cu(OH)₂; reaction condition: 1 mol L⁻¹ KOH aqueous solution with 0.025 mol L⁻¹ glucose; light source: 300 W Xe lamp; scanning rate: 20 mV s⁻¹.

The enhanced photocurrent observed for the Ni(OH)₂/α-Fe₂O₃ sample in Fig. 4b is possibly due to the fact that Ni(OH)₂ traps photo-generated holes, forming Ni³⁺ species, which efficiently catalyzes glucose oxidation while Ni³⁺ is simultaneously reduced back to Ni²⁺.^29–31 Notably, the cocatalyst Ni(OH)₂ can enhance the photo-generated charge separation of α-Fe₂O₃ and provide catalytic sites for glucose oxidation.

Some metal components (Fe and Cu) with multiple oxidation states were demonstrated to be excellent electrocatalysts for glucose sensors.^32,33 Therefore, Fe(OH)₃ and Cu(OH)₂ were also used to modify the α-Fe₂O₃ photoanode for glucose oxidation. As shown in Fig. 4c and d, the major phenomenological observation is that the modification of α-Fe₂O₃ with Fe(OH)₃ and Cu(OH)₂ can noticeably enhance the photocurrent density. The abovementioned results show that Fe(OH)₃ and Cu(OH)₂ play a similar role as that of Ni(OH)₂. Specifically, they can improve the hole transfer from Fe₂O₃ to glucose and provide active sites for glucose oxidation.

The α-Fe₂O₃ based photoanodes with cocatalysts, such as Ni(OH)₂, reported in this study can photooxidize various biomass molecules in combination with an O₂-reducing cathode. Fig. 5 shows the photocurrent density generated by methanol, glycerol, cellubiose and starch. The photocurrent densities of Ni(OH)_2/α-Fe₂O₃ photoanodes can reach 4.7 and 5.0 mA cm⁻² at 0 V vs. SCE fueled by methanol and glycerol, respectively. Whereas cellubiose and starch with relatively high molecular weights give photocurrent densities of approximately 4.0 and 3.5 mA cm⁻², respectively, at 0 V vs. SCE with Ni(OH)₂/α-Fe₂O₃ photoanodes. It was noted that the photocurrent density produced by starch was lower than that obtained from cellubiose solution under similar conditions probably because of the slow hydrolysis of starch. The results illustrate that all these biomass materials are promising fuels for PFCs and loading cocatalysts is an efficient method for improving the PFC performance.

Fig. 5 LSV curves of α-Fe₂O₃ (solid) and Ni(OH)₂/α-Fe₂O₃ (dashed) photoanodes in 1 mol L⁻¹ KOH electrolyte with different biomass-derived compounds under chopped light illumination; methanol (10 vol%); glycerol (0.025 mol L⁻¹); cellubiose (0.025 mol L⁻¹) and starch (1 wt%). Light source: 300 W Xe lamp; scanning rate: 20 mV s⁻¹.

A simple PFC consisting of an α-Fe₂O₃ (or Ni(OH)₂/α-Fe₂O₃) photoanode and a Pt plate cathode in glucose solution was fabricated. Fig. 6 illustrates the voltage–current density and power density–current density curves, and Table 1 displays the photovoltaic parameters for both the PFCs. The device with the α-Fe₂O₃ photoanode yields an open-circuit voltage (V_oc) of 0.38 V, a short-circuit current (I_sc) of 1.17 mA cm⁻², and a fill factor (FF) of 0.18, resulting in an overall power conversion efficiency (PCE) of 0.082%. In contrast, the photovoltaic device containing the Ni(OH)₂/α-Fe₂O₃ photoanode presents a V_oc of 0.43 V, an I_sc of 1.97 mA cm⁻² and an FF of 0.21. As a result, the Ni(OH)₂/α-Fe₂O₃ based cell reaches a conversion efficiency of 0.18%, outperforming that of the Fe₂O₃-based device. As shown in power density–current density curves, the device with the α-Fe₂O₃ photoanode yields a maximum power density of 0.082 mA cm⁻², while the maximum power density (P_max) of Ni(OH)₂/α-Fe₂O₃ is 0.18 mA cm⁻². These results reveal that with an α-Fe₂O₃ based PFC, biomass can be directly converted into electrical energy and the efficiency can be enhanced by modifying the α-Fe₂O₃ photoanode with cocatalysts.

Fig. 6 I–V and I–P curves of PFCs for α-Fe₂O₃ and Ni(OH)₂/α-Fe₂O₃ photoanodes operating with light irradiation in 1 mol L⁻¹ KOH electrolyte containing 0.025 mol L⁻¹ glucose under irradiation with the AM 1.5G simulated solar light (100 mW cm⁻²).

Table 1 Photovoltaic parameters of the light-assisted glucose fuel cells consisting of α-Fe₂O₃ and Ni(OH)₂/α-Fe₂O₃ photoanodes^a

Anode	Intensity (mW cm^−2)	VOC (V)	ISC (mA cm^−2)	FF	P (mW cm^−2)	PEC (%)
a See Fig. 6 for the test condition.
α-Fe2O3	100	0.38	1.17	0.18	0.082	0.08
Ni(OH)2/α-Fe2O3	100	0.43	1.98	0.21	0.18	0.18

---

The stabilities of α-Fe₂O₃ based photoanodes and the intermediates in the reaction were tested in 1 mol L⁻¹ KOH electrolyte containing glucose. Fig. 7 illustrates the amperometric I–t curves of the α-Fe₂O₃ and Ni(OH)₂/α-Fe₂O₃ electrodes under continuous illumination at 0 V vs. SCE. The photocurrent of bare α-Fe₂O₃ after 4 h illumination is 86% of the initial value; however, in the case of the Ni(OH)₂/α-Fe₂O₃ electrode, the photocurrent can be maintained at 96%. Additionally, XPS results (Fig. S5, ESI†) show that the valence state of Ni is maintained during the reaction in 4 h, suggesting that Ni²⁺ in the Ni(OH)₂/α-Fe₂O₃ electrode can be rapidly regenerated. This suggests that the cocatalyst is rather effective for improving the steady state of the α-Fe₂O₃ photoanode.

Fig. 7 Amperometric I–t curves of α-Fe₂O₃ and Ni(OH)₂/α-Fe₂O₃ photoanodes at 0 V vs. SCE in a 1 mol L⁻¹ KOH electrolyte containing 0.025 mol L⁻¹ glucose. Light source: 300 W Xe lamp. The area of α-Fe₂O₃ film is 1.58 cm².

The products from glucose with the Ni(OH)₂/α-Fe₂O₃ photoanode were detected by HPLC after 6 h of reaction. These intermediates were identified as arabinose, erythrose, glyceraldehyde, glycolaldehyde, glycolate and formate, and the carbon balance is ca. 71% (Table S1 and Fig. S6, ESI†). The carbon loss of 29% is mainly attributed to the release of CO₂, which is produced from the cleavage of the C–C bonds of glucose and its derivatives. This result implies that α-Fe₂O₃ based photoanodes can efficiently cleave the C–C bond in biomass and oxidize it to CO₂ in the PFC at low temperature.

Conclusion

PFCs consisting of an α-Fe₂O₃ based photoanode and an O₂-reducing cathode were fabricated. The sunlight utilization of PFCs was extended to the visible light region. We initially reported that transition metal hydroxides (Ni, Fe and Cu) are excellent cocatalysts of α-Fe₂O₃ for biomass oxidation. The performance of PFCs can be noticeably enhanced by loading these cocatalysts on an α-Fe₂O₃ photoanode. Moreover, compared with traditional DAFCs, PFCs can efficiently break the C–C bonds of the biomass, and thus PFCs can be directly powered with natural biomass molecules.

Acknowledgements

This work was financially supported by the National Natural Science Foundation of China (NSFC, grant no. 21090340, 21373209).

References

1. P. McKendry, Bioresour. Technol., 2002, 83, 37–46.
2. J. Potočnik, Science, 2007, 315, 810–811.
3. E. Antolini, ChemSusChem, 2013, 6, 966–973.
4. E. H. Yu, U. Krewer and K. Scott, Energies, 2010, 3, 1499–1528.
5. A. Serov and C. Kwak, Appl. Catal., B, 2010, 97, 1–12.
6. A. Brouzgou, A. Podias and P. Tsiakaras, J. Appl. Electrochem., 2013, 43, 119–136.
7. S. D. Minteer, B. Y. Liaw and M. J. Cooney, Curr. Opin. Biotechnol., 2007, 18, 1–7.
8. F. Ahmad, M. N. Atiyeh, B. Pereira and G. N. Stephanopoulos, Biomass Bioenergy, 2013, 56, 179–188.
9. M. Kaneko, J. Nemoto, H. Ueno, N. Gokan, K. Ohnuki, M. Horikawa, R. Saito and T. Shibata, Electrochem. Commun., 2006, 8, 336–340.
10. R. L. Chamousis and F. E. Osterloh, ChemSusChem, 2012, 5, 1–7.
11. P. J. Barczuk, A. Lewera, K. Miecznikowski, P. Kulesza and J. Augustynskiz, Electrochem. Solid-State Lett., 2009, 12, B165–B166.
12. M. Antoniadou, D. Kondarides, D. Labou, S. Neophytides and P. Lianos, Sol. Energy Mater. Sol. Cells, 2010, 94, 592–597.
13. M. Antoniadou and P. Lianos, Catal. Today, 2009, 144, 166–171.
14. Y. Yan, J. Fang, Z. Yang, J. Qiao, Z. Wang, Q. Yu and K. Sun, Chem. Commun., 2013, 49, 8632–8634.
15. S. D. Tilley, M. Cornuz, K. Sivula and M. Grätzel, Angew. Chem., Int. Ed., 2010, 49, 6405–6408.
16. K. Sivula, F. Le Formal and M. Grätzel, ChemSusChem, 2011, 4, 432–449.
17. D. A. Wheeler, G. Wang, Y. Ling, Y. Li and J. Z. Zhang, Energy Environ. Sci., 2012, 5, 6682–6702.
18. K. Sivula, R. Zboril, F. Le Formal, R. Robert, A. Weidenkaff, J. Tucek, J. Frydrych and M. Grätzel, J. Am. Chem. Soc., 2010, 132, 7436–7444.
19. S. K. Mohapatra, S. E. John, S. Banerjee and M. Misra, Chem. Mater., 2009, 21, 3048–3055.
20. J. Zhu, Z. Yin, D. Yang, T. Sun, H. Yu, H. E. Hoster, H. H. Hng, H. Zhang and Q. Yan, Energy Environ. Sci., 2013, 6, 987–993.
21. S. Saremi-Yarahmadi, K. G. U. Wijayantha, A. A. Tahir and B. Vaidhyanathan, J. Phys. Chem. C, 2009, 113, 4768–4778.
22. I. Cesar, A. Kay, J. A. G. Martinez and M. Grätzel, J. Am. Chem. Soc., 2006, 128, 4582–4583.
23. Y. Hu, A. Kleiman-Shwarsctein, A. J. Forman, D. Hazen, J. Park and E. W. McFarland, Chem. Mater., 2008, 20, 3803–3805.
24. S. D. Tilley, M. Cornuz, K. Sivula and M. Grätzel, Angew. Chem., Int. Ed., 2010, 49, 6405–6408.
25. I. Cesar, A. Kay, J. A. G. Martinez and M. Grätzel, J. Am. Chem. Soc., 2006, 128, 4582–4583.
26. J. Y. Kim, G. Magesh, D. H. Youn, J. Jang, J. Kubota, K. Domen and J. S. Lee, Sci. Rep., 2013, 3, 2681.
27. H. K. Mulmudi, N. Mathews, X. C. Dou, L. F. Xi, S. S. Pramana, Y. M. Lam and S. G. Mhaisalker, Electrochem. Commun., 2011, 13, 951–954.
28. L. Armelao, R. Bartoncello, L. Crociani, G. Depaoli, G. Granozzi, E. Tondello and M. Benttinelli, J. Mater. Chem., 1995, 5, 79–83.
29. P. R. Martins, M. A. Rocha, L. Angnes, H. E. Toma and K. Araki, Electroanalysis, 2011, 23, 2541–2548.
30. J. Nai, S. Wang, Y. Bai and L. Guo, Small, 2013, 9, 3147–3152.
31. S. Xie, T. Zhai, W. Li, M. Yu, C. Liang, J. Gan, X. Lu and Y. Tong, Green Chem., 2013, 15, 2434–2440.
32. X. Cao and N. Wang, Analyst, 2011, 136, 4241–4246.
33. S. Sun, X. Zhang, Y. Sun, S. Yang, X. Song and Z. Yang, ACS Appl. Mater. Interfaces, 2013, 5, 4429–4437.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c4ra07498j

‡ These authors contributed equally to this work.

---