The TiO₂ main layer was prepared using the sol-gel method. Nano-TiO₂ was synthesized using titanium(IV) isopropoxide [TTIP] (Aldrich Chemical, Sigma-Aldrich Corporation, St. Louis, MO, USA), nitric acid, ethyl alcohol, and distilled water. The TTIP was mixed with ethanol, and distilled water was added drop by drop under vigorous stirring for 1 h. This solution was then peptized using nitric acid and heated under reflux at 80°C for 8 h. After this period, a TiO₂ sol was prepared. The prepared sol was dried to yield a TiO₂ powder. The TiO₂ particles were calcined in air at 450°C for 1 h using a programmable furnace to obtain the desired TiO₂ stoichiometry and crystallinity. TNTs were prepared using a hydrothermal process described in the authors' previous work. Then, 5 g of TiO₂ particles prepared by the sol-gel method were mixed with 500 ml of a 10-M NaOH aqueous solution, followed by hydrothermal treatment at 150°C (TNTs) in a Teflon-lined autoclave for 12 h. After the hydrothermal reaction, the treated powders were washed thoroughly with distilled water and 0.1 M HCl and subsequently filtered and dried at 80°C for 1 day. To achieve the desired TNT size and crystallinity, the powders were calcined in air at 500°C for 1 h 16 .

TiO₂ nanoparticles and TNTs prepared by the sol-gel and hydrothermal methods were mixed at various weight ratios (without TNT, 9:1 (10 wt.%), 8:2 (20 wt.%), 7:3 (30 wt.%), 5:5 (50 wt.%), and 100 wt.% TNTs; total weight 6 g) and ground in a mortar. Acetic acid (1 ml), distilled water (5 ml), and ethanol (30 ml) were added gradually drop by drop to disperse the TiO₂ nanoparticles and nanotubes under continuous grinding. The TiO₂ dispersions in the mortar were transferred with an excess of ethanol (100 ml) to a tall beaker and stirred with a 4-cm-long magnet tip at 300 rpm. Anhydrous terpineol (20 g) and ethyl celluloses (3 g) in ethanol were added, followed by further stirring. The dispersed contents were concentrated by evaporating the ethanol in a rotary evaporator. The pastes were finished by grinding in a three-roller mill 17 . An optically transparent conducting glass (FTO, sheet resistance 8 Ω/sq) was washed in ethanol and deionized water in an ultrasonic bath for 10 min. The FTO glass was immersed in a 40-mm-deep TiCl₄ aqueous solution at 70°C for 30 min to make good mechanical contact. A TiO₂ film with a thickness of 12 to 15 μm was deposited onto the pretreated conducting glass using the screen printing technique and sintered again at 450°C for 15 min and at 500°C for 15 min in air.

The nanoporous TiO₂ electrode films were immersed in the dye (N-719) complex for 24 h at room temperature. A counter electrode was prepared by spin-coating an H₂PTCl₆ solution onto the FTO glass and heating at 450°C for 30 min. The dye-adsorbed TiO₂ electrode and the PT counter electrode were assembled into a sandwich-type cell and sealed with a hot-melt sealant of 50-μm thick. An electrolyte solution was introduced through a drilled hole in the counter electrode. The hole was then sealed using a cover glass.

The phase of the particles obtained at various hydrothermal temperatures was examined by X-ray diffraction [XRD] using a D/MAX-2200 diffractometer with CuKα radiation (Rigaku Corporation, Shibuya-ku, Tokyo, Japan). The morphology and thickness of the prepared TNT layers were investigated by field-emission scanning electron microscopy [FE-SEM] (model S-4700, Hitachi, Chiyoda-ku, Tokyo, Japan). The absorption spectra of the TiO₂ electrode films were measured using a UV-Vis spectrometer (UV-Vis 8453, Agilent Technologies Inc., Santa Clara, CA, USA). The conversion efficiency of the fabricated DSSC was measured using an I-V solar simulator (McScience, Suwon-si, South Korea). The incident photocurrent conversion efficiency was measured using an IPCE Model Qex7 (PV Measurements, Inc., Boulder, CO, USA). The active area of the resulting cell exposed to light was approximately 0.25 cm² (0.5 cm × 0.5 cm).

Figure 1a shows the XRD pattern of the sol-gel TiO₂ nanoparticles at 450°C, which indicates a mixture of the anatase and rutile phases. The XRD pattern of TiO₂ nanoparticles shows prominent anatase peaks at (101), (004), (200) and prominent rutile peaks at (110) and (101). Figure 1b shows the XRD patterns of the TNT films prepared at hydrothermal temperatures at 150°C for 12 h. The TiO₂ nanoparticles were observed to be transformed into the anatase phase by the hydrothermal method. As can be observed from the corresponding XRD patterns (Figure 1b), the TNTs possess a highly crystallized anatase structure without any impurity phase. In the TNTs, the rutile peaks indicate that the transformation to anatase is complete. FE-SEM images of the TiO₂ sol-gel nanoparticles and the TNTs prepared at hydrothermal temperatures are shown in Figure 2a. The diameter of the TiO₂ nanoparticles prepared by the sol-gel method is consistently about 25 nm. Figure 2b shows an FE-SEM image of the sample anatase TNTs which were grown at 150°C for 12 h and exhibit a pure tube-like structure. The length of the TNTs is several macrometers, their diameter is approximately 50 to 100 nm, and they are very uniform, quite clean, and smooth-surfaced. Figure 3 shows the surface morphology of the electrode film on the FTO glass. Figure 3a shows a film made from TiO₂ nanoparticles and TNT hybrids, which has a porous structure. A cross-sectional SEM image of the TiO₂ electrode film (Figure 3b) was also captured. The top part is the TiO₂ electrode film. The middle one is the FTO layer, and the lowest one is the glass substrate. The electrode is 12- to 15-μm thick in Figure 3b.

Figure 4 shows how the UV-Vis absorbance of the TNTs affects the dye-adsorbed TiO₂ films. It is known that the N-719 dye shows absorption peaks. Figure 4 shows the absorption spectrum of the N-719 dye in the 400- to 800-nm wavelength range in the flexible TiO₂ electrode film contained with various percentages of TNTs. The TNT content in the TiO₂ nanoparticles was 0, 10, 20, 30, 50, and 100 wt.%. It can be seen in Figure 4 that in the 400- to 500-nm wavelength range, the absorbance for the sample containing 10 wt.% TiO₂/TNT was the highest, and the absorbance of the sample containing 100 wt.% TNT was the lowest. The absorption of the nanoparticle film made purely of TiO₂ was slightly reduced in this region compared to the 10 wt.% and 20 wt.% TNT films. According to Lambert-Beer's law, higher absorbance means a higher dye concentration; a suitable amount of TNT in the film could provide a large surface area for dye adsorption. Therefore, the TiO₂ layer with the dye serves as the photoactive layer. It is well known that the photocurrent of a flexible DSSC is correlated directly with the number of dye molecules; the more dye molecules are adsorbed, the more incident light is harvested, and the larger is the photocurrent.

The incident photocurrent conversion efficiency [IPCE] is defined as the number of electrons in the external circuit produced by an incident photon at a given wavelength divided by the number of incident photons 18 . The IPCE spectra as a function of wavelength for the TiO₂ electrode films (10 wt.% TNT, 100 wt.% TNT, and without TNT) are shown in Figure 5. The maximum efficiency at the 510-nm wavelength coincides with the maximum absorption wavelength of the N-719 dye. The IPCE peak height at 510 nm for the 10 wt.% TiO₂/TNT cell is 53.3%, which is much higher than the values of 15.1% obtained for the 100 wt.% TNT cell and 37.2% for the cell without TNT. Furthermore, over the whole spectral region, the 10 wt.% TNT cell exhibits considerable higher IPCE values than the other two samples. Based on the experimental results and data analysis described above, the constructed TiO₂/TNT (10 wt.%) cell exhibits a combination of a relatively large amount of dye adsorption, low transfer resistance, long electron lifetime, and IPCE, all possibly leading to enhanced J _sc and η in DSSCs.

Figure 6 shows the current-voltage photovoltaic performance curves of DSSCs based on the pure TiO₂ cell, 10, 20, 30, 50, and 100 wt.% TNT cells, and cells without any TNT under AM 1.5 illumination (100 mW/cm²). One of the most important parameters of a solar cell is its photoelectric conversion efficiency, i.e., the ratio of the output power to the incident power. The energy conversion η can be estimated as:

η = V oc × J sc × FF P s ,

where V _oc is the open-circuit voltage, J _sc is the integral photocurrent density, FF is the fill factor:

V max × J max V oc × J sc ,

and P _s is the intensity of the incident light.

Table 1 summarizes the efficiency, fill factor, open-circuit voltage, and integral photocurrent for the corresponding solar cells. It can be seen that these DSSCs have a similar V _oc of 0.65 V; because these flexible DSSCs have the same compositions, it makes sense that their V _oc values are close. However, the J _sc difference increases or decreases at various weight percentages of TNTs and TiO₂ nanoparticles. A DSSC with a light-to-electric energy conversion efficiency of 4.57%, a short-circuit current density of 10.41 mA/cm², an open-circuit voltage of 0.662 V, and a fill factor of 66.17% was achieved. For the hybrid 10 wt.% TiO₂/TNT cell, the best results for conversion efficiency were obtained; the 100 wt.% TNT cell showed the worst results for conversion efficiency because the TNTs were in a random arrangement. The DSSCs based on TiO₂ nanoparticle/TNT hybrids ranging from 0 to 100 wt.% showed higher values of FF, V _oc, and J _sc, and therefore higher efficiencies η than the cell based on pure TiO₂ nanoparticles. It is obvious that the voltage of the DSSC with 10 wt.% TNTs is higher than that without TNTs.