Post-treatment on dye-sensitized solar cells with TiCl₄ and Nb₂O₅

Abstract

Post-treatment of TiCl₄ improves the overall conversion efficiency of dye-sensitized solar cells by enhancing the photocurrent (J_SC) rather than the photovoltage (V_OC), owing to an 80 mV downward shift in the TiO₂ conduction band edge potential. Further treatment with niobium isopropoxide forms a Nb₂O₅ barrier layer that induces an upward shift in the Fermi energy and decreases the recombination rate, resulting in larger V_OC. Quite intriguingly, the incident photon to current conversion efficiency results show that this second treatment also enhances J_SC. Thus, the resulting electrode shows an improved conversion efficiency (by ca. 52%) over that of the non-treated electrode.

---

1 Introduction

Dye-sensitized solar cells (DSSCs) based on TiO₂ films have attracted considerable attention because of their low cost and high efficiency of up to 12.3%.^1–4 The electrode consists of nanoscale semiconductor colloids with nanoporous structures, resulting in a large surface area, that allows the electrode to adsorb large amounts of dye. Unfortunately, the porous structure also allows the electrolyte to penetrate throughout the electrode, which leads to a higher possibility of recombination between the injected electrons in the semiconductor and the oxidized ions in the redox mediator. To solve this problem, a post-treatment process with TiCl₄ has previously been introduced.⁵

It is well known that photocurrent and photovoltage are the two most important parameters contributing to the overall solar conversion efficiency. Although TiCl₄ treatment lowers the recombination rate, it also produces a downward shift in the TiO₂ conduction band edge potential, and does not lead to a larger photovoltage. Thus, the TiCl₄ treatment improves the overall conversion efficiency by enhancing the photocurrent rather than the photovoltage, as demonstrated by B. O'Regan et al.⁵ To reduce charge recombination in DSSCs and thereby achieve remarkably high photocurrent and photovoltage, ultrathin layers of metal oxides, carbonates, and titanates, for example, have also been applied to the surface of the TiO₂ electrode.^6,7 Coating of Nb₂O₅ on the TiO₂ surface has been reported by some researchers to form an inherent energy barrier that lowers the recombination rate and leads to enhanced performance.^8–13 However, the control experiments were conducted between coated TiO₂ and uncoated TiO₂ rather than TiCl₄-treated TiO₂. Thus, although improvements in conversion efficiency were achieved, they were still not quite satisfactory. It would therefore be of more benefit to look for a new surface modification strategy that is better than the existing TiCl₄ surface modification.

In this article, we compared four types of electrodes (“Bare”, “TiCl₄”, “Nb₂O₅”, and “TiCl₄–Nb₂O₅”) to demonstrate that treatment with niobium isopropoxide after TiCl₄ treatment is necessary. The “TiCl₄–Nb₂O₅” electrode showed the best performance of the four electrodes, with V_OC enhanced by 36 mV, J_SC by 53.5%, and η by 51.9% compared with those of the “Bare” electrode; V_OC by 30 mV, J_SC by 12.2%, and η by 13.1% compared with those of the “TiCl₄” electrode, and V_OC by 5 mV, J_SC by 19.9%, and η by 12.7% compared with those of the “Nb₂O₅”electrode.

2 Experimental

2.1 Preparation of the photoanodes

The preparation of TiO₂ particles by the hydrothermal method has been described elsewhere.¹⁴ The TiO₂ particles were spread on a fluorine-doped tin oxide (FTO) substrate (15 Ω), which had been pretreated with 0.05 M TiCl₄ solution at 70 °C for 30 min, using a screen printing method. The resulting TiO₂ films, with a thickness of about 2 μm (measured with a stylus profiler), were then annealed in air at 500 °C for 30 min using a ramp rate of 2 °C min⁻¹. As a control group, an as-fabricated TiO₂ film electrode was labeled as a “Bare” electrode. The following surface treatment processes are depicted in Fig. 1. First, the “Bare” electrode was re-immersed in the TiCl₄ solution (as described above). After washing with deionized water, the electrode was annealed again at 500 °C, and was labeled as a “TiCl₄” electrode. The “TiCl₄–Nb₂O₅” electrode was prepared by dipping a “TiCl₄” electrode into a 5 mM niobium isopropoxide solution at room temperature for 30 s, after which it was washed with isopropanol and finally sintered at 500 °C. To determine whether the improved performance of the “TiCl₄–Nb₂O₅” electrode in comparison with those of the other electrodes was caused by the outer layer of Nb₂O₅ or by a synergy of the two treatments, a “Nb₂O₅” electrode was prepared by processing a “Bare” electrode with the niobium isopropoxide treatment described above.

Fig. 1 Schematic diagram depicting of the experimental processes (H-TiO₂ and T-TiO₂ represent TiO₂ particles obtained from hydrothermal method and TiCl₄ treatment, respectively).

2.2 Fabrication of DSSCs

The four types of working electrodes, i.e. the “Bare” electrode, the “TiCl₄” electrode, the “Nb₂O₅” electrode and the “TiCl₄–Nb₂O₅” electrode, were immersed in a 0.3 mM cis-diisothiocyanato-bis(2,20-bipyridyl-4,40-dicarboxylato) ruthenium(II) bis(tetrabutylammonium) (N719, Dalian HeptaChromaSolarTech, China) solution for 24 h. Then, the four types of working electrodes and their Pt-coated FTO counter electrodes were sealed together with a hot-melt Surlyn spacer of approximately 25 μm in thickness. A I⁻/I₃⁻ based liquid electrolyte was introduced through holes in the reverse of the counter electrodes, and then the holes were sealed using the same hot-melt Surlyn spacer. The total active electrode area was 0.3 (0.75 cm × 0.4 cm) cm².

2.3 Characterizations and photoelectrochemical measurement

The crystallinities of the four types of working electrodes were investigated using X-ray diffraction (XRD). For ensuring the components, the samples were subjected to X-ray photoelectron spectroscopy (XPS). The optical transmittance of the four types of working electrodes before immersed in the dye (N719) was measured at room temperature with an UV-vis spectrophotometer, using a blank substrate as a reference. The transmittance versus wavelength data were used to calculate the band gap of the electrodes according to the method adopted by Tandon and Gupta. The photovoltaic performance of the DSSCs was measured under AM 1.5 simulated sunlight, produced by a 300-W Oriel Solar Simulator (model, 91160) with the illumination intensity being 100 mW cm⁻². An electrochemical analyzer was used to record the information of photocurrent and photovoltage. The incident photon to current conversion efficiency (IPCE) spectra were measured as a function of wavelength on the basis of a monochromator. The electrochemical impedance spectra (EIS) were conducted with an electrochemical workstation (VoltaLab 40) with the frequency range from 0.1 Hz to 100 kHz in the dark condition.

3 Results and discussions

3.1 Characteristics

The XPS results of the four types of electrodes are collectively shown in Fig. 2(a). The spectra were dominated by the photoelectron lines of Ti and O elements. The C 1s line, which is generally believed to have arisen from the adventitious carbon due to the exposure to the atmosphere, was also found in the spectra of all the samples and used as the standard for calibration during data processing. The energies of the Ti 2p_1/2 and 2p_3/2 lines were determined after calibration with the energy of the C 1s line set at 284.6 eV and were in accordance with our precious work.¹⁵ Therefore, the XPS results confirmed that the chemical composition of the four types of electrodes were all titanium dioxide.

Fig. 2 (a) Results of the XPS analysis, (c) XRD patterns of the four types of electrodes; (b) XPS full spectrum of Nb for the “TiCl₄–Nb₂O₅” electrode (“*” represents anatase; “#” represents FTO substrate).

Moreover, two Nb lines could be found in the spectra of the “Nb₂O₅” electrode and “TiCl₄–Nb₂O₅” electrode, suggesting that the surfaces of the two electrodes had been successfully modified with the Nb element. To confirm the formation of Nb element, the XPS full spectrum of Nb for the “TiCl₄–Nb₂O₅” electrode is shown in Fig. 2(b). Peaks assignable to Nb 3d_3/2 and Nb 3d_5/2 were detected at 209.6 and 206.8 eV, respectively, indicating that the product of the niobium isopropoxide treatment was niobium oxide.¹⁰

Fig. 2(c) shows the XRD patterns of the four types of electrodes. All the diffraction peaks of all samples were consistent with the anatase phase (JCPDS no. 21-1272). However, it can be found that the four types of electrodes are almost pure anatase structure, which may be due to the Nb₂O₅ barrier layer being thin. Notably, there were still some differences in peak intensity and full width at half maximum of the peaks at around 25° among the four types of electrodes. These can be ascribed to two reasons. First, different preparation methods may lead to different XRD peak intensities for samples with the same components, such as the main TiO₂ layer synthesized by the hydrothermal method and the thin TiO₂ layer prepared by the TiCl₄ treatment. Second, the Nb₂O₅ barrier layer as the shell may have shielded the intensity of the TiO₂ peaks.^16,17

3.2 DSSCs performances

Fig. 3 shows the J–V characteristics of the four types of working electrodes, including the “Bare” electrode, the “TiCl₄” electrode, the “Nb₂O₅” electrode and the “TiCl₄–Nb₂O₅” electrode, which were all 2 μm-thick. The corresponding photovoltaic performance parameters are listed in Table 1. The conversion efficiency of the “Bare” electrode was 2.33%, which is in accordance with previous literature.⁵ Considering the thickness of the TiO₂ films,¹⁸ ∼2 μm, higher efficiencies may be achievable with thicker TiO₂ films. Evidently, both the “TiCl₄” electrode and the “Nb₂O₅” electrode were better than the “Bare” electrode, and the “TiCl₄–Nb₂O₅” electrode exhibited the best performance of the four types electrodes. In comparison with the performance of the “Bare” electrode, the open circuit voltage V_OC of the “TiCl₄” electrode was enhanced by 6 mV, the short circuit current density J_SC by 36.9%, and the cell efficiency η by 34.3%. Thus, the main contribution to the increase in overall cell efficiency was because of the increase in J_SC, and bore little relation to V_OC or the side effects of the fill factor, which is in agreement with the previous literature.^5,19 However, the V_OC, J_SC, and η of the “TiCl₄–Nb₂O₅” electrode were enhanced by 36 mV, 53.5%, and 51.9% compared with those of the “Bare” electrode, and boosted by 30 mV, 12.2%, and 13.1%, respectively, when compared with those of the “TiCl₄” electrode. This indicates that further treatment with the niobium isopropoxide solution caused the performance of the “TiCl₄–Nb₂O₅” electrode to be improved beyond that achieved with the TiCl₄ treatment. The “TiCl₄–Nb₂O₅” electrode exhibited a V_OC enhanced by 5 mV, J_SC by 19.9%, and η by 12.7% compared with those of the “Nb₂O₅” electrode, which indicates that the improved performance of the “TiCl₄–Nb₂O₅” electrode in comparison with the other electrodes is not because of the outer Nb₂O₅ layer alone, but by a synergy of the two treatments.

Fig. 3 J–V plots of the four types of electrodes with 2 um-thick TiO₂ films.

Table 1 Photovoltaic performance parameters of the four types of electrodes and the corresponding values of the band gaps

Samples/2 μm-thick	JSC/mA cm^−2	VOC/V	FF	η/%	Eg/eV
“Bare”	5.10	0.735	0.62	2.33	3.39
“TiCl4”	6.98	0.741	0.60	3.13	3.31
“Nb2O5”	6.53	0.766	0.63	3.14	3.43
“TiCl4–Nb2O5”	7.83	0.771	0.59	3.54	3.34

---

3.3 Effect on photovoltage

For regenerative photoelectrochemical systems, the open-circuit photovoltage (V_OC), determined by the difference between the electron quasi-Fermi level in the TiO₂ film under illumination and the redox electrolyte level,²⁰ is a crucial parameter for the overall conversion efficiency. It was described by the following expression:²¹
where I_inj is the current generated by injected electrons. This is determined by many factors, such as the electronic coupling between the electron-donating orbital of the dye and the electron-accepting orbital of the semiconductor, the density of states in the semiconductor coupled to the dye and the difference between the conduction band edge of the semiconductor and the excited state of the dye.^5,22,23 n_cb is the concentration of electrons at the TiO₂ surface, which may be of sufficient magnitude to change the potential drop across the Helmholtz layer, resulting in conduction band edge movement.²⁴ k_et is the rate constants for triiodide reduction due to the recombination at the TiO₂/electrolyte interface, whose decrease will lead to an increase in the V_OC of the cell.²⁰ In conclusion, the conduction band edge and the recombination rate are the two vital factors for V_OC.

3.3.1 Band edge movement. Fig. 4(a) shows the UV-vis transmission spectra of the four types of electrodes. All of the electrodes had the lowest transmittance in the UV region (below about 350 nm) owing to the intrinsic optical properties of these semiconductors. A significant shift in the onset transmission can also be seen from Fig. 4(a). The “TiCl₄” electrode had an obvious red shift compared with the spectrum of the “Bare” electrode, which indicates that the “TiCl₄” electrode had a narrower band gap than the “Bare” electrode, i.e., TiCl₄ treatment made the band gap decrease, which is in agreement with the previous literature.^5,19 Meanwhile, the “TiCl₄–Nb₂O₅” electrode had clear blue shifts relative to the spectrum of the “TiCl₄” electrode, which suggests that the “TiCl₄–Nb₂O₅” electrode had a broader band gap than the “TiCl₄” electrode. Similarly, the same conclusion can be found by comparing the spectra of the “Nb₂O₅” electrode and “Bare” electrode. That is, niobium isopropoxide treatment broadened the band gap relative to that of the corresponding untreated electrode.^25–27 Specifically, TiCl₄ treatment leads to a narrow band gap, while the niobium isopropoxide treatment gives rise to a broad band gap.

Fig. 4 (a) UV-Vis transmission spectra; (b) (ahν)^1/2 vs. photo energy (hν) plots of the four types of electrodes with 2 μm-thick TiO₂ films.

The method adopted by Tandon and Gupta was used to obtain the values of the optical band gaps. Fig. 4(b) shows the plots of (ahν)^1/2 vs. photo energy hν. In these plots, the optical band gap can be determined by the intercept of the x-axis from the extrapolation of the linear region of the plot.²⁸ The corresponding band gap values of the four types of electrodes are listed in Table 1. Obviously, the relative values are in good agreement the above analysis, but are bigger than that of the bulk TiO₂ (3.2 eV),²⁹ which can be interpreted as the quantum confinement effect of the fine particle size. From Table 1, it is clear that the TiCl₄ treatment led to a narrow band gap relative to that of the “Bare” electrodes, while the niobium isopropoxide treatment gave rise to a broader band gap relative to the untreated corresponding electrodes.

Furthermore, owing to the small electron effective mass and valence band pinning, most of the band gap change is seen as a shift in the conduction band.^30,31 Therefore, the TiCl₄ treatment led to a downward shift of the conduction band edge potential of about 80 mV, while the niobium isopropoxide treatment led to an upward shift of the conduction band edge potential by about 30–40 mV. Comparing the “Bare”, the “TiCl₄”, and the “TiCl₄–Nb₂O₅” electrodes, it can be found that although treatment with the niobium isopropoxide solution could not offset the downward shift of the conduction band produced by the TiCl₄ treatment completely, it has offset it by more than one third. Moreover, because the conduction band edge potential of Nb₂O₅ is about 100 mV more negative than that of TiO₂, which is located in between the conduction band of TiO₂ and the excited states of the dye, the treatment with niobium isopropoxide formed an energy barrier at the surface of the “TiCl₄” electrodes.^{10–13,25–27} This difference in conduction band edge potential indicates that the “TiCl₄–Nb₂O₅” electrode was able to accumulate more electrons in the conduction band because of the Nb₂O₅ energy barrier layer, which suggests a more negative Fermi energy and larger V_OC. Thus, the further treatment of TiCl₄ treated electrodes with niobium isopropoxide is necessary because it is not only raises the conduction band edge, but also raises the Fermi energy level owing to greater accumulation of electrons, finally leading to larger V_OC.

3.3.2 Recombination. Plots of the dark current vs. voltage were measured to determine the extent of recombination in the four types of electrode. Although such a plot cannot express directly the process of recombination, it could be used to obtaining comparative information for the four electrodes.^25,27 As shown in Fig. 5(a), at the same voltage, the “TiCl₄–Nb₂O₅” electrode had the lowest dark current, while the “Bare” electrode had the highest dark current, that is, the “TiCl₄–Nb₂O₅” electrode had the lowest recombination rate of the four electrodes, while the “Bare” electrodes had the highest. Electrochemical impedance spectroscopy (EIS) of DSSCs based on “TiCl₄” electrode and “TiCl₄–Nb₂O₅” electrode were performed to gain better insight into the dynamics of the charge transfer process affected by the niobium isopropoxide treatment, as shown by the Nyquist plots in Fig. 5(b). The large semicircles in the Nyquist plots are attributed to the electron recombination at the TiO₂/dye/electrolyte interface.^10,13,32 The recombination resistance of the DSSCs based on the “TiCl₄–Nb₂O₅” electrode was larger than that for the DSSCs based on “TiCl₄” electrode, indicating that the niobium isopropoxide treatment weakened the electron recombination process. Therefore, treating the “TiCl₄” electrode with the niobium isopropoxide solution further decreased the rate of recombination, thus increasing its V_OC.

Fig. 5 (a) Dark current density vs. voltage plots of the four types of electrodes with 2 μm-thick TiO₂ films; (b) Nyquist plots from electrochemical impedance spectra of the “TiCl₄–Nb₂O₅” electrode and “TiCl₄” electrode based DSSCs measured in the dark.

Fig. 6 displays a schematic diagram of V_OC affected by the conduction band edge movement and the recombination caused by the TiCl₄ and niobium isopropoxide treatments. Fig. 6(a) indicates that V_OC is determined by the difference between the Fermi level of TiO₂ film and the redox potential of the electrolyte. Fig. 6(b) shows the energy diagram after the TiCl₄ treatment. This treatment decreases the rate of recombination and leads to more electrons accumulating in the TiO₂ conduction band, which gives rise to more negative level of Fermi energy. While this should resulted in a higher V_OC, the downward shift in the TiO₂ conduction band edge potential by this treatment, as discussed above, results in only slightly higher V_OC. In short, the downward shift in the TiO₂ conduction band edge potential lessens the rise in V_OC caused by the decreased recombination rate. Therefore, the main contribution to the increase in overall cell efficiency for the “TiCl₄” electrode bears little relation to V_OC (as listed in Table 1). Thus, the subsequent treatment by the niobium isopropoxide solution is necessary (as shown in Fig. 6(c)). Fig. 4(a) and (b) show that the niobium isopropoxide treatment improves the conduction band edge potential of the “TiCl₄” electrode, but cannot move it upward to match the original position of the “Bare” electrode completely. However, as discussed above, the niobium isopropoxide treatment forms a barrier layer, which, because more electrons are ultimately accumulated in the conduction band, results in the higher Fermi energy of the “TiCl₄–Nb₂O₅” electrode and the larger V_OC value. Simultaneously, similar with TiCl₄ treatment, the niobium isopropoxide treatment decreases the rate of recombination, which further leads to more a negative Fermi energy level and larger V_OC value. In short, the further treatment with niobium isopropoxide, which resulted in the more negative Fermi energy and lower recombination rate, enlarged the V_OC value beyond that achieved with the TiCl₄ treatment.

Fig. 6 Energy schematic diagram depicting the effects of the movement of the TiO₂ conduction band and recombination, caused by the TiCl₄ and niobium isopropoxide treatments, on the open-circuit photovoltage.

3.4 Effect on photocurrent

The total photocurrent can be calculated by the following expression:³³
where q is the electron charge, F(λ) is the incident photon flux density at wavelength λ, r(λ) is the incident light loss due to the conducting glass, and IPCE(λ) is the incident photon to current conversion efficiency. Therefore, to analyze in more detail the enhanced J_SC, IPCE measurements have been conducted for the four types of electrodes. The results are presented in Fig. 7.

Fig. 7 IPCE curves of the four types of electrodes.

The IPCE plots showed that the “TiCl₄” electrode and the “Nb₂O₅” electrode exhibited an upward shift in the wavelength region from 450 to 750 nm compared with the “Bare” electrode. Simultaneously, the “TiCl₄–Nb₂O₅” electrode exhibited a further upward shift in IPCE in the same wavelength range when compared with both the “TiCl₄” electrode and the “Nb₂O₅” electrode. That is, the “TiCl₄” electrode and the “Nb₂O₅” electrode had larger J_SC than the “Bare” electrode, and the “TiCl₄–Nb₂O₅” electrode had the largest J_SC, which was in accordance with results shown in Fig. 3 and Table 1.

4 Conclusions

In conclusion, the best performance of “TiCl₄–Nb₂O₅” electrode among the four types of electrodes indicated that the present niobium isopropoxide treatment was necessary. Both the previous literature and this paper showed that while TiCl₄ treatment enhanced the overall efficiency of the solar cell, the V_OC bore little relation to the increased overall cell efficiency owing to an 80 mV downward shift in the TiO₂ conduction band edge potential caused by the TiCl₄ treatment. Therefore, niobium isopropoxide treatment was used to optimize the TiCl₄ treatment. The duplicate effect of the TiCl₄ and niobium isopropoxide treatments, which suppressed the recombination rate, resulted in a more negative Fermi energy for the “TiCl₄–Nb₂O₅” electrode and enlarged its V_OC. Furthermore, the IPCE results showed that the “TiCl₄–Nb₂O₅” electrode had the highest photon to current conversion efficiency because of the duplicate effect of the two treatments, which gave rise to the largest J_SC value among the different electrodes. For these reasons, the “TiCl₄–Nb₂O₅” electrode showed the best performance of the four electrodes, with V_OC enhanced by 36 mV, J_SC by 53.5%, and η by 51.9% compared with those of the “Bare” electrode, and V_OC by 30 mV, J_SC by 12.2%, and η by 13.1% compared with those of the “TiCl₄” electrode.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (Nos. 61171023 and 61306079) and the National Key Basic Research Program of China (Nos. 2013CB933604 and 2010CB934203).

Notes and references

1. A. Yella, H. W. Lee, H. N. Tsao, C. Y. Yi, A. K. Chandiran, M. K. Nazeeruddin, E. W. G. Diau, C. Y. Yeh, S. M. Zakeeruddin and M. Gratzel, Science, 2011, 334, 629.
2. V. Galstyan, A. Vomiero, E. Comini, G. Faglia and G. Sberveglieri, RSC Adv., 2011, 1, 1038.
3. K. J. Hwang, C. Im, D. W. Cho, S. J. Yoo, J. W. Lee and W. G. Shim, RSC Adv., 2012, 2, 3034.
4. S. Ramkumar and S. Anandan, RSC Adv., 2013, 3, 21535.
5. B. C. O'Regan, J. R. Durrant, P. M. Sommeling and N. J. Bakker, J. Phys. Chem. C, 2007, 111, 14001.
6. S. T. Wang, X. Zhang, G. Zhou and Z. S. Wang, Phys. Chem. Chem. Phys., 2012, 4, 816.
7. F. Lenzmann, J. Krueger, S. Burnside, K. Brooks, M. Gratzel, D. Gal, S. Ruhle and D. Cahen, J. Phys. Chem. B, 2001, 105, 6347.
8. E. Barea, X. Q. Xu, V. Gonzalez-Pedro, T. Ripolles-Sanchis, F. Fabregat-Santiago and J. Bisquert, Energy Environ. Sci., 2011, 4, 3414.
9. R. Ghosh, M. K. Brennaman, T. Uher, M. R. Ok, E. T. Samulski, L. E. McNeil, T. J. Meyer and R. Lopez, ACS Appl. Mater. Interfaces, 2011, 3, 3929.
10. J. Liang, J. Yang, G. M. Zhang and W. T. Sun, Electrochem. Commun., 2013, 37, 80.
11. K. S. Ahn, M. S. Kang, J. K. Lee, B. C. Shin and J. W. Lee, Appl. Phys. Lett., 2006, 89, 013103.
12. K. Sayama, H. Sugihara and H. Arakawa, Chem. Mater., 1998, 10, 3825.
13. P. F. Du, L. X. Song, J. Xiong, Y. F. Yuan, L. C. Wang, Z. Q. Xi, D. L. Jin and J. J. Chen, Electrochem. Commun., 2012, 25, 46.
14. C. Y. Huang, Y. C. Hsu, J. G. Chen, V. Suryanarayanan, K. M. Lee and K. C. Ho, Sol. Energy Mater. Sol. Cells, 2006, 90, 2391.
15. J. Liang and G. M. Zhang, ACS Appl. Mater. Interfaces, 2012, 4, 6053.
16. Z. H. Zhang, X. H. Zhong, S. H. Liu, D. F. Li and M. Y. Han, Angew. Chem., Int. Ed., 2005, 44, 3466.
17. N. Zhang, S. Q. Liu, X. Z. Fu and Y. J. Xu, J. Phys. Chem. C, 2011, 115, 9136.
18. S. Ito, S. M. Zakeeruddin, R. Humphry-Baker, P. Liska, R. Charvet, P. Comte, M. K. Nazeeruddin, P. Pechy, M. Takata, H. Miura, S. Uchida and M. Gratzel, Adv. Mater., 2006, 18, 1202.
19. P. M. Sommeling, B. C. O'Regan, R. R. Haswell, H. J. P. Smit, N. J. Bakker, J. J. T. Smits, J. M. Kroon and J. A. M. van Roosmalen, J. Phys. Chem. B, 2006, 110, 19191.
20. N. Kopidakis, N. R. Neale and A. J. Frank, J. Phys. Chem. B, 2006, 110, 12485.
21. M. K. Nazeeruddin, A. Kay, I. Rodicio, R. Humphrybaker, E. Muller, P. Liska, N. Vlachopoulos and M. Gratzel, J. Am. Chem. Soc., 1993, 115, 6382.
22. K. Hara, T. Horiguchi, T. Kinoshita, K. Sayama, H. Sugihara and H. Arakawa, Sol. Energy Mater. Sol. Cells, 2000, 64, 115.
23. J. B. Asbury, E. Hao, Y. Q. Wang, H. N. Ghosh and T. Q. Lian, J. Phys. Chem. B, 2001, 105, 4545.
24. G. Schlichthorl, S. Y. Huang, J. Sprague and A. J. Frank, J. Phys. Chem. B, 1997, 101, 8141.
25. Y. Diamant, S. G. Chen, O. Melamed and A. Zaban, J. Phys. Chem. B, 2003, 107, 1977.
26. J. Z. Ou, R. A. Rani, M. H. Ham, M. R. Field, Y. Zhang, H. D. Zheng, P. Reece, S. Zhuiykov, S. Sriram, M. Bhaskaran, R. B. Kaner and K. Kalantar-zadeh, ACS Nano, 2012, 6, 4045.
27. S. Morita, T. C. Wei, M. Ikegami and T. Miyasaka, J. Power Sources, 2013, 240, 753.
28. M. Zhang, G. Q. Lin, C. Dong and L. S. Wen, Surf. Coat. Technol., 2007, 201, 7252.
29. M. Gratzel, Nature, 2001, 414, 338.
30. X. F. Gao, W. T. Sun, G. Ai and L. M. Peng, Appl. Phys. Lett., 2010, 96, 153104.
31. P. V. Kamat, J. Phys. Chem. C, 2008, 112, 18737.
32. W. Q. Wu, H. S. Rao, Y. F. Xu, Y. F. Wang, C. Y. Su and D. B. Kuang, Sci. Rep., 2013, 3, 1892.
33. Z. S. Wang, H. Kawauchi, T. Kashima and H. Arakawa, Coord. Chem. Rev., 2004, 248, 1381.

---