Hydrothermal synthesis of TiO₂ nanotubes and their application as an over-layer for dye-sensitized solar cells

Abstract

Different nanostructures of TiO₂ play an important role in the kinetics of dye sensitized solar cells (DSSC) and affect the overall light harvesting efficiency of the cells. This article describes that the one dimensional nanostructure of TiO₂ (nanotubes) can increase the light scattering effect, light harvesting effect and electron transport in the DSSC to improve its performance. Pure anatase TiO₂ nanotubes were synthesized by a hydrothermal method using commercial material (P25) due to which the manufacturing cost of the DSSC was enormously reduced. To enhance the power conversion efficiency of the DSSC, a new type of double layered photoanode was prepared and optimized by using TiO₂ nanoparticles as the main layer and TiO₂ nanotubes (TNT) as the over-layer. These prepared cells were analysed by optical, photovoltaic and electrochemical measurement systems. The cells having the TNT over-layer showed longer electron life time, higher BET surface area and pore volume and 40% improved light harvesting efficiency. This new and optimized structure will be concrete fundamental background towards the development of the applications of next generation dye-sensitized solar cells.

---

1. Introduction

TiO₂ is one of the most important materials and has wide applications in industry. It has been developed and used in cosmetic products, white paint, corrosion-protective coatings, gas sensors and photo catalysts. Many researchers have studied various morphologies of TiO₂ such as nanotube,¹ nanowire,^2,3 nanosphere,⁴ hybrid nanostructure^5–8 etc.^9,10 and found the novel effects due to their unique physicochemical properties and their potential applications in the industry.

Since the first invention of Gratzel's group in 1991,¹¹ dye-sensitized solar cell (DSSC) has become a promised alternate of conventional silicon solar cells due to their ease of fabrication, low-cost and wide application fields. In DSSC, TiO₂ nanoparticles with high specific surface area have been widely used as photoanode.¹² However, the high surface area of nanoparticles also results the large amount of grain boundaries between nanoparticles, leading to the presence of many electron–hole recombination centers.^13,14 Furthermore, the poor absorption of long-range wavelength light by transparent nanostructure film is also a key factor which limits the conversion efficiency.¹⁵ To improve the electron collection efficiency, many attempts have already been made to solve the mentioned problems by developing novel structures, such as one dimensional (1-D),^2,3 three dimensional (3-D)¹⁶ and core–shell nanostructures.¹⁷ The 1-D nanostructure, which has advantages of high light-scattering and fast electron transport,^18–20 is expected to enhance the performance of DSSC effectively. TiO₂ nanotubes, which have the advantages of high aspect ratio, simple preparation procedure and controllable morphology, has become the recent research focus within the 1-D nanostructure category.^21–23

TiO₂ nanotube is promising material for scattering layer in dye-sensitized solar cells due to which light scattering and electron transport rapidly increases and consequently light-harvesting efficiency is improved.²⁴ Lee et al.²⁵ synthesized TNT's by easy and rapid anodizing method, and used as scattering layer in DSSC. This scattering structure proved more effective hierarchical pore structure, better light scattering and light absorption than commercial scattering structure (large particle) but high quality titanium as precursor is very expensive and also preparation of TNT powder after anodization is too much difficult. Kim et al.²⁶ studied the effect of TiO₂ multilayer in DSSC and concluded that light scattering layer of TNT has a longer electron transfer path and higher charge collection efficiency but pure anatase form of TNT is more effective as scattering layer. Literature did not show any study for preparation of cheaper and pure anatase TNT; and its application as scattering layer in DSSC. Comparison study of standard and commercial TiO₂ powders (P25, G2 etc.) used as scattering layer is not done yet, which is very important aspect for proving that TNT's are best for scattering layer.

In this study, a pure anatase form of tubuler-TiO₂ (TNT) was synthesized by using commercial and cheap material (P25) and then TNT and large-particles (500 nm, G2) are used as scattering layer in DSSC to compare the scattering effect of TNT and commercially available G2. For the cheaper DSSC, the main layer of photoanode was also synthesized by commercially available P25. To find the optimized conditions, eight various photoanode structures were prepared and compared. This new and optimized structure will be a concrete fundamental background toward the development of the applications of the next generation dye-sensitized solar cells.

2. Experimental

Materials

For the synthesis of TNTs, TiO₂ (P25 by Degussa Co.), NaOH (98%, Samchun Co.), hydrochloric acid (37%, Samchun Chemicals Co.) and D.I. water were used. For the preparation of three types TiO₂ pastes for photoelectrode; P25 (Degussa Co.), G2 (showadenko Co.) and TNTs (Self Prepared), ethyl cellulose (Sigma-Aldrich Co.), terpineol (Aldrich Co.), and acetic acid (99.7%, Junsei Co.) were used as received. For the photovoltaic cell assembly, FTO glass (TEC 8, Pilkington Co.), D719 (cis-diisothiocyanato-bis(2,2′-bipyridyl-4,4′-dicarboxylato) ruthenium(II) bis(tetrabutylammonium), Everlight Co.), Surlyn for spacer (60 um, Dupont Co.) and chloroplatinic acid hexahydrate (Sigma-Aldrich Co.) were used as received. 1-Butyl-3-methylimidazolium iodide (BMII), iodine (I₂), lithium iodine (LiI), 4-tert-butylpyridine (TBP) and anhydrous acetonitrile were purchased for the composition of electrolytes by Aldrich Co.

Preparation of titania nanotube

The TiO₂ nanotubes were prepared using the hydrothermal method.²⁴ 10 M NaOH aqueous solution (200 mL, in D.I. water) was placed in titanium covered PTFE vessel. The 4 g P25 powder was mixed with NaOH aqueous solution under stirring for one hour. After closing the vessel, this mixture was heated from room temperature to 120 °C under stirring at 300–350 rpm for 36 hours.

Prepared slurry was removed from vessel (using D.I. water and Teflon tools) and then neutralized and ion-exchanged by 0.1 M HCl (in D.I. water) to reduce the pH level up to 1.0 slowly and then washed three times with D.I. water through centrifugation process (500 mL × 3). After washing, the slurry was dried in freeze and vacuum-dryer under −50 °C for 1 day. To convert from amorphous phase to anatase phase, prepared powder was treated under high temperature (500 °C for 30 minutes).

Preparation of photoanode

All three types of paste were prepared for doctor-blade printing on DSSC. The final composition of all types of paste was TiO₂ nanopowder (1 g), ethyl cellulose (0.5 g), terpineol (3.3 mL) and acetic acid (0.16 mL).²⁴ Each paste was dispersed in ethanol by using zirconia ball-mill, magnetic stirrer and sonicator. After that, the final contents were concentrated by rotary evaporator and grinded with ceramic mortal. Cleaned FTO glasses (15 × 15 mm²) were immersed in TiCl₄ solution (40 mM in water) for 30 minutes at 70 °C and washed with water and ethanol. After washing, substrates were annealed at 500 °C for 30 minutes. Pre-treated FTO glasses were coated with different types of pastes with different combinations by doctor blade process using the 3M tape as shown in Fig. 1 and 2. Two sets of photoanode were prepared. In the first set, single main layer of 25 nm sized TiO₂ nanoparticles (SM) having thickness of about 7 μm was used for all the cells, and for the scattering layers; 500 nm sized larger TiO₂ particles (SMG), TNT single layer (SMT) and TNT double layers (SMTT) were used. In the second set, double main layer of 25 nm sized TiO₂ nanoparticles (DM) having thickness of about 14 μm was used for all the cells, and for the scattering layers; 500 nm sized larger TiO₂ particles (DMG), TNT single layer (DMT) and TNT double layers (DMTT) were used. The active area of cells was about 4 × 5 mm² which was measured by using software equipped camscope (ICS-305B, Sometech Co.). After printing of pastes, TiO₂ films were heated in 6 steps of 70 °C, 125 °C, 325 °C, 375 °C, 450 °C, and 500 °C for 30, 30, 5, 5, 15 and 15 minutes respectively using the high temperature furnace (Lab house Co.). In the post treatment, coated and sintered TiO₂ films were immersed in TiCl₄ solution (40 mM in water) for 30 minutes at 70 °C. After washing with water and ethanol, films were again annealed at 500 °C for 30 minutes.

Fig. 1 Fabrication sequence of dye sensitized solar cell.

Fig. 2 Two sets of photoanodes with various TiO₂ structures.

Fabrication of photovoltaic cell

After cooling less than 100 °C in furnace, annealed TiO₂ films were rapidly immersed in D719 solution (0.5 mM in ethanol) and kept in airtight glass-container at room temperature under dark and dry conditions for 20–24 hours. Dye-adsorbed photoanodes were washed to remove non-adsorbed dye by using tap ethanol and then dried under nitrogen flow. Counter electrodes were prepared by drop coating method, using 10 mM H₂PtCl₆ solution (in isopropanol) on one-holed FTO glass and heated at 400 °C for 20 minutes using the same furnace.

The dye-adsorbed photoelectrodes and counter electrodes were assembled using ionomer-Surlyn by hot-press at 80 °C. After assembling, the electrolyte solution (composed of 0.6 M BMII, 0.05 M I₂, 0.1 M LiI, and 0.5 M TBP in acetonitrile) was injected into the cells through one-holed FTO glass by capillary effect under vacuum and the holes were sealed with cover-glasses using the same Surlyn. Indium was coated at the edge of electrodes to make a good current collector. A black mask aperture was placed on the front electrode for better analysis of photovoltaic characteristics.²⁷

Characterization

To check the crystal structure of the titania nanotube, wide angle X-ray diffraction (WAXD) method was used on a Rigaku Denki X-ray generator (Rigaku, D/MAX-2500), using CuKα (λ = 1.5418 Å) radiation operated on 40 kV and 60 mA. The scan angle covered 5°< 2θ < 50° (2θ is the scattering angle, θ is the Bragg angle) at a speed of 5° min⁻¹. Morphology and structures of photoanode were investigated by Field Emission-Scanning Electron Microscopy (FE-SEM, JEOL JSM-6700F). BET specific surface area and pore volume were determined by N₂ adsorption method using a Quantachrome Autosorb-6 Sorption System (USA). UV-vis-spectrum was obtained by using UV-1650PC spectrophotometer (Shamadzu Co.) in the wavelength range of 350 nm to 800 nm. Photocurrent–voltage measurement was performed by using K101-Lab20 (Mac Science Co.) source measuring unit. Solar simulator with 160 W xenon arc lamp was used as light source satisfying AAA class (spectral match; 0.75–1.25, non-uniformity of irradiance; ≤±2%, temporal instability; ≤±2%). The light intensity was calibrated with a KIER-calibrated Si solar cell (Mc science Co.). The IPCE was measured using a lock-in amplifier with a current pre-amplifier (K3100, Mac Science Co.) under short circuit current state with illumination of monochromatic light. Electrochemical impedance spectrum of the cells was measured by using ultimate electrochemical workstation (BioLogic Co.) in the frequency range of 0.05 Hz to 500 kHz under 100 mW cm⁻².

3. Results and discussion

Characterization of nanostructure TiO₂ layer

Fig. 3 shows X-ray diffraction pattern of self-synthesized 20 nm sized pure anatase, P25 and TNT, indicating that P25 is anatase and rutile mixed structure and synthesized TNT is highly crystallized anatase structure without impure phase. The intensities of XRD patterns are different due to the difference in crystal sizes which could be calculated with Scherer equation. This highly crystallized anatase phase TiO₂ nanomaterial could be synthesized by hydrothermal reaction, at high concentration of NaOH in water.

Fig. 3 X-ray diffraction pattern for pure anatase, P25 and TNT nanostructures.

Morphology and size of TNT's, observed by SEM and TEM are shown in Fig. 4. SEM and TEM images illustrate the rod and tubular shape of TNT's with inner diameter of about 2 nm and outer diameter of 10–20 nm. SEM images of different patterns of TiO₂ structures fabricated on the FTO glass are shown in Fig. 5. Stacking patterns and difference of thickness of all layers could be observed from the Fig. 5.

Fig. 4 FE-SEM images of (a) P25 nanoparticles, (b) self-synthesized nanotubes and FE-TEM images of (c) self-synthesized nanotubes.

Fig. 5 FE-SEM cross section images of two sets of photoanodes with various TiO₂ nanostructures (×3000).

P25 showed a BET surface area of 62.61 m² g⁻¹ and pore volume of 0.3743 cm³ g⁻¹ and the prepared anatase TiO₂ TNT's showed a BET surface area of 325.08 m² g⁻¹ and pore volume of 0.9248 cm³ g⁻¹ (5.2 times and 2.5 times higher respectively as shown in Fig. 6). Surface area and pore volume of TNTs is higher due to their physical structure (inner and outer surface area of nanotubes) but only large surface area cannot provide higher number of sites for adhesion of dye molecule on TiO₂ surface. Fig. 7 shows the light absorbance pattern to compare the amount of adsorbed dye on TiO₂ surfaces of different structures (P25 nanoparticle, TNT nanotube and G2 large particle). Nanoparticle structure provided highest absorption peak and nanotube structure provided slightly less absorption peak. Large particles structure provided very small peak. Nanoparticle structure have offered more sites for adhesion of dye molecules on TiO₂ surface than prepared tubular structure (TNT) because inner diameter (2 nm) of nanotubes is too narrow as compared to the size of dye molecule (1 nm²) due to which dye molecules cannot penetrate easily inside of tube and consequently only outer surface of nanotubes can make bonds with dye molecules. So the active surface area of nanotubes for adhesion of dye molecules is less than nanoparticles.

Fig. 6 N₂ adsorption isotherms at 77 K for anatase TiO₂ nanotubes (TNT) compared with P25.

Fig. 7 Light absorbance pattern to compare the amount of adsorbed dye on the surface of different structures of TiO₂ (P25 nanoparticle, TNT nanotube and G2 large-particle).

Photovoltaic performance of DSSCs

In order to investigate and compare the photovoltaic properties of DSSC's having scattering layers of TNT and larger nanoparticles (G2); both morphologies were formed on the TiO₂ main layer (single & double). Table 1 and Fig. 8 comprise the photovoltaic properties of all types of cells, without and with different scattering layers.

Table 1 Photovoltaic performance and EIS interfacial resistances of DSSC having different structures of photoanode

	VOC [V]	JSC [mA cm^−2]	FF [%]	PCE [%]	Rs [Ohm cm^2]	R1 [Ohm cm^2]	C1 [F cm^−2]	R2 [Ohm cm^2]	C2 [F cm^−2]	R3 [Ohm cm^2]	C3 [F cm^−2]
SM	0.767	11.133	73.45	6.27	1.076	0.478	7.248 × 10^−7	4.094	5.569 × 10^−3	1.122	0.143
SMG	0.767	12.987	73.02	7.27	1.075	0.420	8.842 × 10^−7	3.242	6.034 × 10^−3	1.190	0.133
SMT	0.736	13.819	73.95	7.52	1.101	0.432	6.802 × 10^−7	3.212	8.994 × 10^−3	0.394	0.166
SMTT	0.724	14.771	74.60	7.98	1.161	0.346	6.711 × 10^−7	2.227	1.198 × 10^−2	0.971	0.380
DM	0.739	13.794	73.29	7.47	1.026	0.284	1.189 × 10^−7	2.282	1.147 × 10^−2	0.537	1.157
DMG	0.733	14.549	73.20	7.81	1.012	0.281	7.654 × 10^−7	1.665	1.201 × 10^−2	1.026	0.181
DMT	0.718	16.075	72.75	8.40	0.985	0.325	6.431 × 10^−4	1.556	4.785 × 10^−2	0.824	3.035 × 10^−2
DMTT	0.718	16.627	73.48	8.77	1.060	0.284	6.932 × 10^−7	1.566	1.782 × 10^−2	0.906	0.416

---

Fig. 8 I–V curves of DSSC having different structures of photoanode under simulated AM 1.5 global sunlight (1 Sun, 100 mW cm⁻²).

For the set of single main layer, there is an increasing trend of cell efficiency from bare main layer (SM) to with scattering layer of larger particles (SMG) and then more increase for TNT single (SMT) and TNT double (SMTT) scattering layers. The light conversion efficiency is increased from 6.27% to 7.27%, 7.52% and 7.98% respectively corresponding to 16%, 20% and 27% improvement. This increase of efficiency is due to the increase in short-circuit current (J_SC) but there is also decreasing trend of open-circuit voltage (V_OC). The higher J_SC in SMTT is related to the higher amount of adsorbed dye and better light scattering.²⁴ As we have already discussed in explanation of Fig. 6 and 7 that TNT's adsorb more dye than G2, so SMT and SMTT cells are producing more current than SMG cells but adsorbed dye on TNT's also increases the chances of recombination between excited dye on TNT's and I₃⁻ in electrolyte due to which V_OC of SMT and SMTT cells is less than SMG cells. SM and SMG cells showed the same V_OC because less amount of dye molecules are adsorbed on G2 larger particles and consequently chances of back recombination are very less.

For the set of double main layer, there is also an increasing trend of cell efficiency from bare main layer (DM) to with scattering layer of larger particles (DMG) and then more increase for TNT single (DMT) and TNT double (DMTT) scattering layers. The light conversion efficiency is increased from 7.47% to 7.81%, 8.40% and 8.77% respectively corresponding to ca. 4.6%, 12.5% and 17.5% improvement. This increase of efficiency is due to the increase in short-circuit current (J_SC) but there is also decreasing trend of open-circuit voltage (V_OC). Reasons for increase of J_SC and decrease of V_OC have already been discussed in the last section.

In comparison between the both sets of cells, second set of cells having double main layer showed better performance due to more dye adsorption and less interfacial resistances.²⁸

External quantum efficiency (EQE) or IPCE is one of the important photovoltaic parameter used to analyse the quality of the solar cell. EQE describes the conversion efficiency of cells, illuminated under different wavelength of lights. EQE can be defined as:

EQE spectra of all types of cells (SM, SMG, SMT, SMTT, DM, DMG, DMT, and DMTT) are shown in Fig. 9. The double main layer cells demonstrated a significant increase in EQE over the entire wavelength compared to the single main layer cells. The EQE results are in fine agreement with the J_SC value and efficiency of prepared cells shown in Table 1. However, cells having scattering layers G2 and TNT showed different patterns of EQE for both sets due to difference of their morphology and properties. Cells having TNT scattering layers showed better EQE under shorter wavelength light region but the cells having G2 scattering layers showed better EQE under longer wavelength light region. It means that reflection behaviour of TNT's is more convenient for shorter wavelength light due to rod like fine structure of TNT's but the reflection behaviour of G2 is more convenient for longer wavelength light due to its round and large size structures. Scattering efficiency of small and fine structures is always better than large structures.

Fig. 9 External quantum efficiency (EQE) spectra of DSSC having various photoanode structures.

Electrochemical properties of the DSSCs

The internal kinetics of DSSCs, such as the essence of holes-ions transport processes and the behaviours of interfacial charge transport, could be understand by measuring the electrochemical impedance spectrum (EIS) of solar devices.^29–32 All the EIS spectra were simulated using EC-Lab software according to the equivalent circuit shown in Fig. 10. The interfacial resistance and capacitance values are tabulated in Table 1. The measured Nyquist and Bode plots are shown in Fig. 11. In Nyquist plot, the complex impedance of a typical DSSC is the sum of each of the components, given as R_S, Z₁, Z₂ and Z₃. R_S represents the contact impedance between substrate (FTO) and metal ohmic contact, which is the only real part of impedance. Z₁, Z₂ and Z₃ are complex parts of impedance (real and imaginary part) which are described by using semicircles. First semicircle shows Pt-catalysed counter electrode impedance, middle semicircle shows complex impedance (interface among the semiconductor, dye molecules and the redox shuttle) and the last semicircle shows diffusion of triiodide ions related (Warburg) impedance. In general, three semicircles are observed clearly in the measured frequency range of mHz to MHz scale. Larger semicircles occur in the lower-frequency range and a smaller semicircle occurs in the higher-frequency range under illumination. Nyquist plots of all types of cells are shown in Fig. 11(a). All of the start points and first circles (R_S and Z₁) are similar due to the same substrate and metal electrodes. For the both sets of cells, the second semicircle showed clear decreasing trend from SM to SMG, SMT and SMTT; and from DM to DMG, DMT and DMTT. Second semicircle is much shorter for second set of cells (having double main layer) as compared to first set of cells (single main layer) because of lesser interfacial charge resistance among the semiconductor, dye molecules and the redox shuttle. Thus it is evident from the shorter semicircles that more efficient electron generation and transport is being done in the DSSC's having TNT scattering layer. This efficient electron transport is due to the (1) higher number of efficient electron transport pathways in the nanotube (TNT) network than large particles (G2) as shown in Fig. 4 and 5, and (2) increased overall conductivity of the film due to the better conductive behaviour of the nanotube structure. It is well-known fact that the 1-D nanostructure shows much faster charge transport characteristics than randomly packed nanoparticle structure.^2,3 This rapid charge transport and better conductive behaviour could be explained by the Bode plots shown in Fig. 11(b). It is clear that for both the sets having single main layer and double main layer, the effective electron life time for TNT based cells is longer than its counterpart because (1) 1-D structure provides less electron traps and rapid charge transport, (2) more dye adsorption produces more electrons and consequently electron traps are reduced. In comparison between single main layer and double layer, double main layer showed a longer electron life time due to the more thickness of main layer and consequently more dye adsorption, higher electron production and less electron traps.²⁸ The electron life time is obtained from the Bode plot peaks in the middle-frequency range (0.1–100 Hz), by using following relation³³ in which ‘f’ is the frequency of peak.

Fig. 10 Equivalent circuit of the cell; consisting of metal substrate (FTO), counter electrode/redox electrolyte interface, Photoanode/redox electrolyte interface and electrolyte diffusion.

Fig. 11 Electrochemical analysis (EIS) of the DSSC's based on various photoanode structures under full-illumination ((a) Nyquist plot and (b) Bode plot).

EIS results proved that faster electron transport in TNT based cells is responsible for the enhanced J_SC value as compared to their counterpart.

4. Conclusions

In this paper, an easy and cheaper hydrothermal method for preparation of pure anatase TNT's from commercial TiO₂ is introduced and pure TNT's and commercially available larger particles (G2) were used as scattering layer in DSSC to compare the photovoltaic properties of both structures. Pure anatase TNT's showed much higher BET surface area and pore volume than G2 larger particles. The DSSC having TNT functional upper layer provided effective light absorption, better light scattering, longer electron life time and rapid electron transfer and consequently J_SC and light harvesting efficiency are improved. The proposed anode structures (P25/TNT double layer) produced enhanced photovoltaic and electrochemical properties and the optimized structure showed about 40% improvement in photo conversion efficiencies. This structure would be concrete fundamental background toward the development of the applications for the next generation dye-sensitized solar cells.

Acknowledgements

This work was supported by the Industrial Strategic Technology Development Program (10038599, Human Activity Based Green Energy Harvesting and High Efficiency Power Transmission System), funded by the Ministry of Knowledge Economy, Republic of Korea.

References

1. K. Huo, B. Gao, J. Fu, L. Zhao and P. K. Chu, RSC Adv., 2014, 4, 17300.
2. X. J. Feng, K. Shankar, O. K. Varghese, M. Paulose, T. J. Latempa and C. A. Grimes, Nano Lett., 2008, 8, 3781.
3. J. Qiaoying, J. Nanosci. Nanotechnol., 2011, 11, 9846.
4. Y. J. Kim, M. H. Lee, H. J. Kim, G. Lim, Y. S. Choi, N. G. Park, K. Kim and W. I. Lee, Adv. Mater., 2009, 21, 3668.
5. S. Yodyingyong, Q. Zhang, K. S. Park, C. S. Dandeneau, X. Zhou, D. Triampo and G. Cao, Appl. Phys. Lett., 2010, 96, 073155.
6. J. B. Baxter and E. S. Aydil, Sol. Energy Mater. Sol. Cells, 2006, 90, 607.
7. F. Sauvage, F. D. Fonzo, A. L. Bassi, C. S. Casari, V. Russo, G. Divitini, C. Ducati, C. E. Bottani, P. Comte and M. Graetzel, Nano Lett., 2010, 10, 2562.
8. C. Y. Jiang, X. W. Sun, G. Q. Lo, D. L. Kwong and J. X. Wang, Appl. Phys. Lett., 2007, 90, 263501.
9. M. K. Song, P. Rai, K. J. Ko, S. H. Jeon, B. S. Chon, C. H. Lee and Y. T. Yu, RSC Adv., 2014, 4, 3529.
10. Y. J. Kim, H. J. Kim, M. H. Lee, G. I. Lim, H. Y. Song, Y. S. Choi, N. G. Park, C. Lee and W. I. Lee, J. Nanosci. Nanotechnol., 2010, 10, 340.
11. B. O. Regan and M. Gratzel, Nature, 1991, 353, 737.
12. P. Roy, D. Kim, I. Paramasivam and P. Schmuki, Electrochem. Commun., 2009, 11, 1001.
13. Z. Z. Yang, T. Xu, Y. Ito, U. Welp and W. K. Kwok, J. Phys. Chem. C, 2009, 133, 20521.
14. B. F. Martinson, J. W. Elam, J. Liu, M. J. Pellin, T. J. Marks and J. T. Hupp, Nano Lett., 2008, 8, 2862.
15. X. B. Chen and S. S. Mao, Chem. Rev., 2007, 107, 2891.
16. D. K. Hwang, B. H. Lee and D. H. Kim, RSC Adv., 2013, 3, 3017.
17. C. O. Avellaneda, A. D. Goncalves, J. E. Benedetti and A. F. Nogueira, Electrochim. Acta, 2010, 55, 1468.
18. J. J. Gong, C. J. Lin, M. D. Ye and Y. K. Lai, Chem. Commun., 2011, 47, 2598.
19. F. Hao, H. Lin, C. Zhou, Y. Z. Liu and J. B. Li, Phys. Chem. Chem. Phys., 2011, 35, 15918.
20. W. G. Yang, Y. L. Wang and W. M. Shi, CrystEngComm, 2012, 1, 230.
21. G. K. Mor, O. K. Varghese, M. Paulose, K. Shankar and C. A. Grimes, Sol. Energy Mater. Sol. Cells, 2006, 90, 2011.
22. J. M. Macak and P. Schmuki, Electrochim. Acta, 2006, 52, 1258.
23. J. M. Macak, H. Tsuchiya, A. Ghicov and P. Schmuki, Electrochem. Commun., 2005, 7, 1133.
24. K. C. Sun, S. H. Yun, C. H. Yoon, H. H. Ko, S. C. Yi and S. H. Jeong, J. Nanosci. Nanotechnol., 2013, 13, 7938.
25. K. S. Lee, J. Kwon, J. H. Im, C. R. Lee, N. G. Park and J. H. Park, ACS Appl. Mater. Interfaces, 2012, 4, 4164.
26. Y. Kim and M. S. Kang, J. Photochem. Photobiol., A, 2012, 233, 20.
27. S. Ito, M. K. Nazeeruddin, P. Liska, P. Comte, R. Charvet, P. Pechy, M. Jirousek, A. Kay, S. M. Zakeeruddin and M. Gratzel, Prog. Photovoltaics, 2006, 14, 589.
28. 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.
29. Y. Zhao, J. Zhai, J. L. He, X. Chen, L. Chen, L. B. Zhang, Y. X. Tian, L. Jiang and D. B. Zhu, Chem. Mater., 2008, 20, 6022.
30. X. Li and M. X. Wan, Cryst. Growth Des., 2006, 6, 2661.
31. M. Gratzel, Nature, 2001, 414, 338.
32. Y. Zhao, J. Zhai, T. X. Wei, L. Jiang and D. B. Zhu, J. Mater. Chem., 2007, 17, 5084.
33. M. Adachi, M. Sakamoto, J. Jiu, Y. Ogata and S. Isoda, J. Phys. Chem. B, 2006, 110, 13872.

---