Influence of a solution-deposited rutile layer on the morphology of TiO₂ nanorod arrays and the performance of nanorod-based dye-sensitized solar cells

Abstract

The morphology of TiO₂ one-dimensional structures plays an important role in improving the photovoltaic performance of nanostructured solar cells. Herein we utilized a low-temperature solution-deposited rutile layer to adjust the morphology of TiO₂ nanorod (NR) arrays on SnO₂:F (FTO) conductive glass. A higher concentration of the TiCl₄ precursor solution for the deposition of the seed layer can generate more nucleation sites for the growth of the TiO₂ NRs, which raises the density of the TiO₂ NR arrays and further restrains the growth of TiO₂ in the transverse direction. The obtained TiO₂ NR arrays were successfully incorporated into dye-sensitized solar cells (DSSCs) as the photoanodes. The denser TiO₂ NR arrays, resulting from the TiCl₄ precursor solution with a higher concentration, provided a larger surface for the adsorption of dyes, and thus improved the light harvesting of solar cells. More importantly, the seed layers were proven to present an effective blocking effect in preventing electron recombination at the FTO/electrolyte interface, which increased the open-circuit voltage of the DSSCs by ∼110 mV. This is a convenient method to control the morphology of the TiO₂ NR photoanode and back electron reaction via seed layers, which could be used in other nanostructured photoelectrochemical devices.

---

Introduction

One-dimensional (1D) nanostructures have received increasing interest for photoelectric applications, such as in solar cells^1–5 and sensors,⁶ for electrochromism,⁷ and in water splitting devices,^8,9 owing to their unique electric properties. In nanostructured solar cells, photoanodes based on 1D metal-oxide nanostructures can provide a direct pathway for electron transport with reduced interfacial recombination.^10,11 In particular, TiO₂ is a general photoanode material in nanostructured solar cells, due to its appropriate electronic band levels and good stability.^12,13 Therefore, the morphology, property, and applications of TiO₂ 1D nanostructures grown on FTO were covered in a number of recent publications.^14,15 However, due to the epitaxial relationship between the FTO conductive layer and rutile TiO₂, the direct growth of rutile TiO₂ NRs on FTO was severely controlled by the size and orientation of the FTO grains. Meanwhile, the unoccupied surface of FTO provides recombination centers, which reduces the photovoltaic performance of nanostructured solar cells.

A TiO₂ seed layer could eliminate the influence of the FTO substrate and change the morphology of the TiO₂ NR arrays prepared using various synthetic methods.^16–21 A solution deposited seed layer has the advantages of being easily produced and reproduced using a simple chemical precursor solution. In addition, the introduction of a rutile seed layer to FTO substrates increases the nucleation sites for the growth of the TiO₂ NRs.^22–24 More importantly, the seed layer may act as a blocking layer to suppress the recombination reaction of the solar cell between the electrons in the FTO layers and the oxidized species in the hole transport materials.^25–27 As a result, it is important to construct an appropriate seed-layer structure to facilitate both the growth of the TiO₂ NR arrays and the photovoltaic performance of the solar cell.

Herein, we prepared a TiO₂ seed layer by deposition from TiCl₄ precursor solutions, and investigated the influence of these seed-layer structures on the growth of rutile TiO₂ NR arrays and the photovoltaic properties of DSSCs. The density of the nucleating points on the TiO₂ seed layers was easily controlled by the concentration of the precursor solutions, which was proven to vary the density and diameter of the TiO₂ NRs. Higher surface-to-volume ratios of the TiO₂ NRs array could improve the light-harvesting of the DSSCs by increasing the amount of dye loading. The effect of the seed layer on controlling the electronic recombination of DSSCs was further evaluated.

Experimental

Preparation of TiO₂ NR arrays

Synthesis of the TiO₂ NR arrays was performed using the hydrothermal method.^16,26 In a typical procedure, clean FTO substrates were immersed into an aqueous solution of TiCl₄ (0.05–0.4 M) at 70 °C for 40 min.²⁸ Then these TiCl₄-treated FTO substrates were washed thoroughly with deionized water and dried at 100 °C in air for an hour. The bare and seed-layer coated FTO substrates were then placed in Teflon-lined autoclaves (25 mL), containing 0.2 mL of tetrabutyl titanate, 8 mL of deionized water and 8 mL of concentrated hydrochloric acid. The autoclaves were then sealed and heated in an oven at 150 °C for 8 h. After this procedure, the substrates were rinsed with deionized water and dried with air flow.

Fabrication of DSSCs

The as-prepared TiO₂ NR arrays were annealed in air at 450 °C for 30 min in order to remove any residual organics. After cooling, the TiO₂ NR arrays were loaded with a dye monolayer by immersing them in an ethanolic solution (0.3 mM) of N719 (Solaronix SA) at room temperature for 24 h. A platinum-coated FTO substrate made by the thermal deposition of H₂PtCl₆ was used as the counter electrode. An acetonitrile/valeronitrile (85 : 15, v/v) electrolyte containing 0.6 M 1-propyl-3-methylimidazolium iodide, 0.03 M I₂, 0.1 M guanidinium thiocyanate, and 0.5 M 4-tert-butylpyridine was injected into the space between the dye-coated electrode and counter electrode to complete the cell assembly.

Characterizations

The sample morphology was observed via a FEI Quanta 250 scanning electron microscope (SEM). X-ray diffraction (XRD) patterns were recorded on a Rigaku D/max-2500 X-ray diffractometer to determine the crystal phase of the samples. Raman spectra were collected with a Jobin-Yvon HR800 Raman microscope. UV-Vis absorption spectra were collected on a UH 4150 spectrophotometer. Dye loading measurements were conducted by desorbing dye molecules from the TiO₂ film in 0.05 M NaOH aqueous solution. The amount of dye molecules was determined by measuring the absorbance of the desorbed dye solution. Current–voltage (J–V) characteristics of the solar cell were measured with a Keithley 2400 source meter under AM 1.5 simulated solar irradiation (100 mW cm⁻²) supplied by a Sun 2000 solar simulator (ABET Technology). The active area of the solar cell was 0.126 cm². Open-circuit photovoltage decay (OCVD) of the solar cell was recorded with a Keithley 2400 source meter, after the illuminated cell reached a steady photovoltage close to its open-circuit value under AM 1.5 simulated solar light. A home-made high power white LED array equipped with a fast electronic shutter was used in the OCVD measurements. Electrochemical impedance spectroscopy (EIS) was recorded over a frequency range from 50 mHz to 1 MHz with an ac amplitude of 10 mV, using a Princeton PARSTAT 2273 advanced electrochemical system. The impedance measurements were carried out under a forward bias of −0.78 V in the dark.

Results and discussion

We deposited TiO₂ seed layers on FTO surfaces from TiCl₄ precursor solutions for the further growth of the TiO₂ NRs. The Raman spectrum of the TiO₂ seed layer was assigned to E_g, A_1g and combination modes of the TiO₂ rutile phase (Fig. S1†).²⁹ As shown in the SEM images (Fig. 1), all of the solution-deposited TiO₂ seed layers on FTO were uniform without large aggregations. When the concentration of the TiCl₄ precursor solution was 0.05 M, the TiO₂ seed layer was ultra-thin, so we could still observe the large crystal of fluorine-doped tin oxide in the top-view SEM image of the TiO₂ seed layer (Fig. 1b). As the concentration of the TiCl₄ precursor solution increased, the TiO₂ seed layer became thicker and totally covered the FTO surfaces. To study the blocking effect of the TiO₂ seed layers on preventing the electron transfer reaction across FTO for nanostructured solar cell applications, we compared the values of the oxidation peaks of bare FTO and those coated by seed layers in cyclic voltammograms. An aqueous solution of Fe(CN)₆^3−/4− was chosen as the model redox system since it is a pH-independent redox probe with a simple one-electron-transfer reaction and with sufficiently positive potential to the flat-band potential of TiO₂.^30,31 The small values of the oxidation peaks in the cyclic voltammograms of FTO covered with TiO₂ seed layers (Fig. 2) suggested that the TiO₂ seed layers covered the FTO surface completely, even when deposited from the TiCl₄ precursor solution with a low concentration.

Fig. 1 Top-view SEM images of (a) bare FTO and TiO₂ seed layers deposited from TiCl₄ precursor solutions with concentrations of (b) 0.05, (c) 0.2, and (d) 0.4 M.

Fig. 2 Cyclic voltammograms of bare FTO and that covered with different seed layers. The electrolyte solution was 0.01 mM K₄Fe(CN)₆ + 0.001 mM K₃Fe(CN)₆.

TiO₂ NR arrays were grown on bare FTO and that coated with TiO₂ seed layers via the hydrothermal method. The TiO₂ NR arrays were named FTO-NR for the sample grown on bare FTO, and 0.05 M-NR, 0.2 M-NR, and 0.4 M-NR for the samples grown on seed-layers from solutions with concentrations of 0.05, 0.2, and 0.4 M, respectively. XRD patterns (Fig. 3) of the TiO₂ NR arrays all present peaks which were in good agreement with the tetragonal rutile TiO₂ (PDF#21-1276), as well as the peaks associated with the tetragonal SnO₂. The two main peaks at 36.69° and 63.44° were assigned to the (101) and (002) planes of rutile TiO₂. Our previous research has demonstrated that the enhanced (002) diffraction peak and other diminished peaks implied that the TiO₂ NRs grow preferentially along the (001) direction, namely vertical to the FTO substrate.^26,32 Therefore, we can investigate the alignment of rutile TiO₂ NRs as a function of the concentrations of the TiCl₄ solutions for seed-layer preparation through comparing the intensity ratio of the (002) peak to the (101) peak. The higher intensity of the (101) peak of rutile TiO₂ on bare FTO suggests a random alignment of the TiO₂ NRs. As the concentration of the TiCl₄ solution for seed-layer preparation increased, the intensity of the (002) peak of rutile TiO₂ was enhanced, indicating a good alignment of rutile TiO₂ NRs.

Fig. 3 XRD patterns of TiO₂ NRs grown on bare FTO glass and those substrates coated by different seed layers. For comparison, standard diffraction patterns of SnO₂ (PDF#46-1088) and rutile TiO₂ (PDF#21-1276) are included.

The influence of the solution-deposited seed layer on the morphology of the TiO₂ NR arrays was further investigated using SEM (Fig. 4). The average diameter, length, and density of the TiO₂ NRs grown on bare FTO and on different seed layers were collected in Fig. 5 and Table S1.† It was found that the length of the TiO₂ NRs basically remained constant (∼1.3 μm) for all samples. However, as the concentration of the TiCl₄ precursor solutions increased from 0 (bare FTO) to 0.4 M, the density of the TiO₂ NRs rose from 24 to 99 per μm⁻², which may be related to the amount of nucleation sites on the TiO₂ seed layers. To prove this hypothesis, we compared the morphology of the TiO₂ NR arrays at the initial growth stage (Fig. S2†). After 30 min hydrothermal reaction, the growing points of the TiO₂ were well-observed to increase with the concentration increase of the TiCl₄ precursor solution from 0 to 0.4 M. As a result, the number of nucleation sites for growth of the TiO₂ NRs could be easily controlled by the concentration of TiCl₄ precursor solution for seed layer deposition. Meanwhile, we also observed that the diameter of the TiO₂ NRs decreased from 82 to 52 nm. It was easily understood that the growth of the TiO₂ NRs in the transverse direction was restrained when the number of nucleation sites was large.³⁰

Fig. 4 Top-view and cross-sectional SEM images of the TiO₂ NR arrays grown on bare FTO (a and b) and TiO₂ seed layers which were deposited from TiCl₄ precursor solutions with concentrations of 0.05 (c and d), 0.2 (e and f), and 0.4 M (g and h). The scale bars are 500 nm.

Fig. 5 Evolution of the average diameter (left axis, black) and density (right axis, red) of the TiO₂ NRs with the concentration of TiCl₄ precursor solutions for the preparation of the seed layer.

We utilized these TiO₂ NR arrays as photoanodes to fabricate DSSCs, incorporating N719 sensitizer, iodine electrolyte, and platinum counter electrodes. The J–V curves of the DSSCs (Fig. 6) were measured at 100 mW cm⁻², under simulated AM 1.5 conditions, and the detailed photovoltaic parameters are compiled in Table 1. The short-circuit current density (J_sc) values of these DSSCs increased from 2.51 to 6.15 mA cm⁻² when the concentration of the TiCl₄ precursor solution for seed-layer preparation rose from 0 to 0.4 M. This rise of J_sc was easily attributed to the light-harvesting enhancement of the DSSCs, since the dye loading amounts were measured to have increased from 1.7 to 6.1 nmol cm⁻². The denser TiO₂ NR arrays, which grow on the seed layer deposited from the TiCl₄ precursor solution with a higher concentration, possess a larger surface-to-volume ratio and further provide more sites for the adsorption of dye molecules.

Fig. 6 J–V curves of DSSCs based on different TiO₂ NR photoanodes, measured under AM 1.5G solar simulator illumination.

Table 1 Summary of photovoltaic parameters, simulated values of resistances and ideality factor, as well as dye loading amounts of DSSCs with TiO₂ NR arrays grown on bare FTO and different seed layers

Solar cells	Voc (mV)	Jsc (mA cm^−2)	FF	PCE (%)	Rs (Ω cm^−2)	Rsh (Ω cm^−2)	n	Dye loading (nmol cm^−2)
FTO-NR	680	2.51	0.53	0.9	3.4	2700	3	1.7
0.05 M-NR	790	3.53	0.61	1.7	3.8	3000	1.9	2.8
0.2 M-NR	780	4.26	0.58	1.9	5.0	3800	1.7	3.6
0.4 M-NR	770	6.15	0.61	2.9	5.4	4100	1.7	6.1

---

The seed layer influenced the open-circuit voltage (V_oc) in different ways. The V_oc was observed to increase by 110 mV when a thin seed layer was coated on FTO, and then slightly decreased along with the rise of the concentration of the TiCl₄ precursor solution. The increase in the V_oc of the DSSCs with seed layers, compared to those with bare FTO, was mainly due to the blocking effect of the seed layer in preventing charge recombination between electrons in the FTO and the oxidized species in the electrolyte, and is well consistent with the smaller value of the oxidation peak in the cyclic voltammograms of FTO coated with seed layers. We fitted the J–V curves under AM 1.5G illumination and extracted the device parameters (Table 1) to understand the roles of the seed layer in the photovoltaic performance of the solar cells.^33–35 The series resistances (R_s) of these DSSCs slightly increased from 3.4 to 5.4 Ω cm⁻² with the rise in concentration of the TiCl₄ precursor solution. The larger R_s of the 0.4 M-NR cells should be attributed to the thicker TiO₂ seed layer when the concentration of the TiCl₄ precursor solution became higher. Meanwhile, the shunt resistances (R_sh) of these DSSCs were calculated to increase from 2700 to 4100 Ω cm⁻² with the rise in concentration of the TiCl₄ precursor solution, which suggests a blocking effect of the TiO₂ seed-layer to reduce the back reactions between the electrons in FTO and the oxidized species in the electrolyte. The decreased ideality factors (n) of the DSSCs with TiO₂ seed layers,³⁶ compared with those based on bare FTO, may also derive from the blocking effect of the seed layers.

The electron recombination processes of these DSSCs were further investigated through OCVD analysis which was a valid method to calculate the electron lifetime (τ_n).^32,37,38 From Fig. 7a, it can be observed that the V_oc values of all cells were stable under illumination and continued to decay after the interruption of illumination. However, the decay times of all cells were different. The electron lifetime can be directly estimated through the processes of V_oc decay, since the excess electrons undergo recombination under open-circuit conditions. As a result, the τ_n is calculated according to eqn (1):³⁷

where k_B is the Boltzmann constant, T is the absolute temperature, and e is the elementary charge. As shown in Fig. 7b, the τ_n of the DSSCs based on bare FTO was shorter than those of the DSSCs with TiO₂ seed layers at the same V_oc. This was consistent with the observed smaller V_oc in the J–V test and the larger value of the oxidation peak in the cyclic voltammogram based on bare FTO. This experimental phenomenon further proves the blocking effect of the TiO₂ seed layers. However, the reduction of τ_n along with the concentration increase of the TiCl₄ precursor solution confirms that the obtained TiO₂ NR arrays with a large surface-to-volume ratio produce more sites for charge recombination. As a consequence, a slight reduction of the V_oc for the DSSCs with an enlarged surface-to-volume ratio of the TiO₂ NR arrays has been witnessed.

Fig. 7 (a) Open-circuit photovoltage decay curves and (b) the electron recombination lifetime of typical DSSCs fabricated using different TiO₂ NR photoanodes.

EIS measurement was further carried out under dark conditions to unveil the slight variation of V_oc of the DSSCs with TiO₂ seed layers. Fig. 8 shows the Nyquist curves of three DSSCs with seed layers under a forward bias of 780 mV, which is near the V_oc value of DSSCs. The diameter of the arc in the medium frequencies has been well established to represent interfacial charge transfer resistance which reflects the charge recombination process at the sensitizers/TiO₂/electrolyte interfaces.³⁹ The corresponding equivalent circuit is shown in the inset and the estimated resistance parameters are listed in the ESI, Table S2.† As shown in Table S2,† the interfacial recombination resistance (R₂) of the DSSCs was reduced with the increase in concentration of the TiCl₄ precursor solution, which was well consistent with the change of electron lifetime estimated from the OCVD measurements.

Fig. 8 Nyquist plots of DSSCs with TiO₂ NRs arrays grown on different TiO₂ seed layers. The inset is the equivalent circuit applied to fit the resistance data.

Conclusions

In conclusion, a solution-deposited rutile layer was utilized as the seed layer to adjust the morphology of TiO₂ NR arrays and the blocking layer of DSSCs to prevent interface electron recombination at the same time. The density of the nucleating points on the TiO₂ seed layers was easily controlled by the concentration of the TiCl₄ precursor solutions. More nucleating points on the seed layer, generated from the higher concentration of the TiCl₄ precursor solution, were proven to raise the density of the TiO₂ NR arrays, but reduced the diameter of the TiO₂ NRs. These TiO₂ NR arrays could be successfully incorporated into DSSCs as the photoanodes, and the denser TiO₂ NR arrays provided more surface area for the loading of dye sensitizers, thus improving the light-harvesting and J_sc of DSSCs. Furthermore, several photoelectrical experiments revealed that these rutile layers can serve as an effective blocking layer to suppress charge recombination at the FTO/electrolyte interface and therefore improve the V_oc of DSSCs. The solution-deposited rutile layer would facilitate the modification of the TiO₂ NR arrays for their application in different photoelectrochemical devices.

Acknowledgements

This work was supported by the Natural Science Foundation of China (Grant no. 91233204, 51372036, 51202025, 11304035 and 21301041), the Key Project of Chinese Ministry of Education (no. 113020A), the Scientific and Technological Development Scheme of Jilin Province (no. 20140520096JH), the Specialized Research Fund for the Doctoral Program of Higher Education (2012004311002), the National Basic Research Program (2012CB933703), the Open project program key Laboratory of UV-Emitting Materials and Technology of Ministry of Education (130014546), the Fundamental Research Funds for the Central Universities (2412015KJ010), and the 111 project (no. B13013).

Notes and references

1. H. S. Kim, J. W. Lee, N. Yantara, P. P. Boix, S. A. Kulkarni, S. Mhaisalkar, M. Grätzel and N. G. Park, Nano Lett., 2013, 13, 2412–2417.
2. T. Kawawaki, H. B. Wang, T. Kubo, K. Saito, J. Nakazaki, H. Segawa and T. Tatsuma, ACS Nano, 2015, 9, 4165–4172.
3. X. X. Chen, Q. W. Tang, Z. Y. Zhao, X. H. Wang, B. L. He and L. M. Yu, Chem. Commun., 2015, 51, 1945–1948.
4. H. Wang, Y. S. Bai, H. Zhang, Z. H. Zhang, J. H. Li and L. Guo, J. Phys. Chem. C, 2010, 114, 16451–16455.
5. Y. L. Zhao, X. Q. Gu and Y. H. Qiang, Thin Solid Films, 2012, 520, 2814–2818.
6. Y. C. Wang, J. Tang, Z. Peng, Y. H. Wang, D. S. Jia, B. Kong, A. A. Elzatahry, D. Y. Zhao and G. F. Zheng, Nano Lett., 2014, 14, 3668–3673.
7. S. P. Liu, X. T. Zhang, P. P. Sun, C. H. Wang, Y. A. Wei and Y. C. Liu, J. Mater. Chem. C, 2014, 2, 7891–7896.
8. I. S. Cho, M. Logar, C. H. Lee, L. L. Cai, F. B. Prinz and X. L. Zheng, Nano Lett., 2014, 14, 24–31.
9. S. Zhang, X. Q. Gu, Y. L. Zhao and Y. H. Qiang, J. Electron. Mater., 2016, 45, 648–653.
10. J. Y. Liao, B. X. Lei, H. Y. Chen, D. B. Kuang and C. Y. Su, Energy Environ. Sci., 2012, 5, 5750–5757.
11. K. Zhu, N. R. Neale, A. Miedaner and A. J. Frank, Nano Lett., 2007, 7, 69–74.
12. D. Y. Qi, L. Z. Wang and J. L. Zhang, RSC Adv., 2015, 5, 74557–74561.
13. Y. C. Choi, D. U. Lee, J. H. Noh, E. K. Kim and S. Seok, Adv. Funct. Mater., 2014, 24, 3587–3592.
14. A. Wisnet, K. Bader, S. B. Betzler, M. Handloser, P. Ehrenreich, T. Pfadler, J. Weickert, C. Scheu and J. A. Dorman, Adv. Funct. Mater., 2015, 25, 2601–2608.
15. H. Y. Li, H. R. Bala, G. Y. Yuan, X. L. Fu, W. Y. Fu, L. Lin, X. D. Wang, G. Sun and Z. Y. Zhang, Mater. Lett., 2015, 160, 572–575.
16. B. Liu and E. S. Aydil, J. Am. Chem. Soc., 2009, 131, 3985–3990.
17. X. J. Feng, K. Zhu, A. J. Frank, C. A. Grimes and T. E. Mallouk, Angew. Chem., Int. Ed., 2012, 51, 2727–2730.
18. M. Q. Lv, D. J. Zheng, M. D. Ye, J. Xiao, C. J. Lin and J. Zuo, Energy Environ. Sci., 2013, 6, 1615–1622.
19. P. D. Christy, N. Melikechi, N. S. Nirmala Jothi, A. R. Suganthi and P. Sagayaraj, J. Nanopart. Res., 2010, 12, 2875–2882.
20. J. H. Noh, J. H. Park, H. S. Han, D. H. Kim, B. S. Han, S. Lee, J. Y. Kim, H. S. Jung and K. S. Hong, J. Phys. Chem. C, 2012, 116, 8102–8110.
21. J. Shi, C. L. Sun, M. B. Starr and X. D. Wang, Nano Lett., 2011, 11, 624–631.
22. N. G. Park, G. Schlichthorl, J. V. D. Lagemaat, H. M. Cheong, A. Mascarenhas and A. J. Frank, J. Phys. Chem. B, 1999, 103, 3308–3314.
23. E. H. Kong, Y. H. Yoon, Y. J. Chang and H. M. Jang, Mater. Chem. Phys., 2014, 143, 1440–1445.
24. M. Q. Lv, D. J. Zheng, M. D. Ye, J. Xiao, W. X. Guo, L. Sun, C. J. Lin and J. Zuo, Energy Environ. Sci., 2013, 6, 1615–1622.
25. I. S. Cho, Z. Chen, A. J. Forman, K. R. Kim, P. M. Rao, T. F. Jaramillo and X. Zheng, Nano Lett., 2011, 11, 4978–4984.
26. P. P. Sun, X. T. Zhang, X. P. Liu, L. L. Wang, C. H. Wang, J. K. Yang and Y. C. Liu, J. Mater. Chem., 2012, 22, 6389–6393.
27. E. H. Kong, Y. H. Yoon, Y. J. Chang and H. M. Jang, Mater. Chem. Phys., 2014, 143, 1440–1445.
28. A. Yella, L. P. Heiniger, P. Gao, M. K. Nazeeruddin and M. Grätzel, Nano Lett., 2014, 14, 2591–2596.
29. M. J. Yang, B. Ding, S. W. Lee and J. K. Lee, J. Phys. Chem. C, 2011, 115, 14534–14541.
30. D. Wang, X. T. Zhang, P. P. Sun, S. Lu, L. L. Wang, Y. A. Wei and Y. C. Liu, Int. J. Hydrogen Energy, 2014, 39, 16212–16219.
31. L. Kavan, N. Tétreault, T. Moehl and M. Grätzel, J. Phys. Chem. C, 2014, 30, 16408–16418.
32. P. P. Sun, X. T. Zhang, C. H. Wang, Y. A. Wei, L. L. Wang and Y. C. Liu, J. Mater. Chem. A, 2013, 1, 3309–3314.
33. C. F. Zhang, J. C. Zhang, Y. Hao, Z. H. Lin and C. X. Zhu, J. Appl. Phys., 2011, 110, 064504.
34. L. Y. Han, N. Koide, Y. Chiba and T. Mitata, Appl. Phys. Lett., 2004, 84, 2433–2435.
35. S. Sarker, H. W. Seo and D. M. Kim, J. Power Sources, 2014, 248, 739–744.
36. S. Raj, A. K. Sinha and A. K. Panchal, J. Renewable Sustainable Energy, 2013, 5, 033105.
37. A. Zaban, M. Greenshtein and J. Bisquert, ChemPhysChem, 2003, 4, 859–864.
38. W. Q. Wu, Y. F. Xu, H. S. Rao, C. Y. Su and D. B. Kuang, J. Am. Chem. Soc., 2014, 136, 6437–6445.
39. F. F. Santiago, J. Bisquert, E. Palomares, L. Otero, D. B. Kuang, S. M. Zakeeruddin and M. Grätzel, J. Phys. Chem. C, 2007, 111, 6550–6560.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra24888d

‡ These authors contributed equally to the work.

---