1 Introduction

Dye sensitized solar cell (DSSC) is considered as one of the most feasible alternative to the conventional silicon-based solar cell devices due to the easiness of fabrication and the lower material cost. Ever since the groundbreaking discovery by Grätzel et al. [1], the energy conversion efficiency of the state-of-the-art DSSC has long been glued to around 10 % [2, 3, 4]. Unlike the conventional silicon-based solar cells, DSSC could easily be fabricated into a flexible form, and it is once again considered as a distinguished advantage. Together with the world-wide efforts to increase the energy harvesting efficiency, many tries to expand the application range of DSSC have also been performed including the preparation of flexible devices [5–9].

In most of the cases, where glass substrate is employed, a high temperature preparation method at such as 500 °C is frequently used to fabricate the TiO2 photoanode. The high temperature sintering is to induce the necking of TiO2 particles while maintaining the porous structure, which provides the electron diffusion pathway [10, 11]. However, when plastic substrate such as polyethylene naphthalate (PEN) or polyethylene terephthalate (PET) is used for the flexible DSSC, the photoanode deposited on the substrate should be processed under 150 °C [12–14]. The low-temperature processing of photoanode inevitably produces poor consolidation of particles, consequently sacrificing the photoconversion efficiency of the device. This is, in many cases, a big problem blocking prompt application of flexible DSSC to daily devices.

To tackle this low-temperature processing problem, sintering aids such as binder is frequently used, which helps the necking of TiO2 particles even after low-temperature treatment. Pasquier demonstrated that a sol–gel prepared TiO2 sol holds nanoparticles together after a heat treatment at around 130 °C [15]. Park et al. [16] showed a possibility of using binder-free TiO2 paste for fabricating the photoanode just by drying at 150 °C. CaCO3 coating also helped the TiO2 nanoparticles to be cemented with each other eventually to produce higher efficiency from the device [17, 18]. In some cases, the post treatment with Ti, In, Zr oxide precursor solution was tried to induce hydrolysis of the precursor after the TiO2 photoanode is deposited on substrate [19]. As a novel trial, the anodized nanotube array was transferred to the flexible substrate to induce the efficiency improvement by the nanostructuring [6]. The aerosol deposition method was also tried to deposit TiO2 layer without undergoing thermal treatment at elevated temperatures [20]. Not only the low-temperature deposition but also high-temperature sintering of TiO2 on the surface of stainless steel mesh was tried to elucidate the possibility as the photoanode without transparent conductive electrode [21]. However, in spite of all these efforts, the flexible DSSC still needs more improvement in efficiency for the successful daily application.

In this study, we employed a precursor of TiO2, titanium isopropoxide, as the binder for low-temperature consolidation of TiO2 nanoparticles. Binder incorporated TiO2 coating paste was prepared for the application of dip-coating method. Three sets of coating paste were prepared to investigate the effect of titanium isopropoxide (TIPP) content on photoconversion efficiency. To utilize the reactivity, the precursor molecules were added to the paste without prior hydrolysis, by which better chance of bonding among particles was intended.

2 Experiments

Commercially available TiO2 nanopowder (Degussa, Aeroxide® P25) was used as the main material for the preparation of active electrode. The used nanoparticle is of 21 nm in average primary particle size and 80 wt% anatase: 20 wt% rutile in crystallographic phase. Titanium isopropoxide (TIPP, 98+ %, ACROS) was used as the precursor of sol–gel chemical synthesis to form binding bridges among the TiO2 particles. All the chemicals and materials were used as received without further purification.

Coating paste was prepared as follows. 10 g of TiO2 nanopowder was first dispersed into an ethanolic solution (1.4 g of D.I. water and 86.8 g of ethanol) importantly by using ultrasonication for 20 min. A simple stirring with either magnetic bar or mechanical screw was not enough to obtain stable dispersion of TiO2 particles for at least 7 days, which is frequently requested under the mass production environment. Instead a high power (1 kW) ultrasonication is indispensable. Into the dispersion, 1.8 g of TIPP was added dropwise. Then the mixture was kept under mild stirring for about 5 h to achieve the further homogenization. For finding the optimized processing condition, the composition of the coating paste was varied in terms of molar ratio of TIPP:P25 powder, and the detailed compositions are given in Table 1.

Table 1 The composition of each prepared coating pastes
Full size table

Using the above prepared coating pastes, TiO2 active electrode were dip-coated on flexible substrate, which is, in this study, ITO-coated PEN film (20–25 Ω/□, OIKE Co., Japan). To make the films thick enough for the operation as the active electrode for flexible DSSC, dipping was done 4 times so as the thickness of the prepared films to reach up to a certain extent. The deposited films were dried in air for about 15 min, and then heat treated in a convection oven at 120 °C for 1 h. The microstructure of the prepared films was investigated by using scanning electron microscopy (SEM).

For the comparison of energy conversion efficiency of the low-temperature processed active electrodes, unit cells were also assembled using the obtained films. The photoanodes were kept in the dye solution of 0.3 mM N719 in ethanol for 21 h for the adsorption of dye onto the surface of porous TiO2 films. The dye-adsorbed TiO2 films were then washed with ethanol and subsequently dried in air at 60 °C for 30 min. The preparation of the unit cell followed standard preparation process; the dye-coated photoanodes were sealed (Surlyn, Dupont) into the unit cells together with the counter electrode (Pt coated on the ITO-coated PEN) and the injected electrolyte (0.6 M BMII, 0.05 M I2, 0.1 M LiI, and 0.5 M TBP in methoxypropionitrile). Photoconversion efficiency was investigated by the J–V measurement method under 1 Sun (Air Mass 1.5) illumination condition by a solar simulator (SS-200XIL, EKO Instruments Co.).

3 Results and discussion

For the comparison of the effect of TIPP:P25 ratio on TiO2 film properties, the pastes were prepared at varying TIPP:P25 ratios as given in Table 1. The hydrolysis and condensation of TIPP were carried out without the addition of catalyst such as HCl or HNO3, since the reactivity of TIPP was fast enough at room temperature. The TEM image of the TiO2 particles in the paste of TIPP:P25 = 0.1:1 is given in Fig. 1. TiO2 network that was formed by the hydrolysis of TIPP with the water in the paste and condensation into solid phase was also seen as in the inset of Fig. 1. As a whole, the paste showed comparably stable dispersion after the ultrasonication regardless of given TIPP:P25 ratio. The addition of TIPP into the TiO2 paste produced stable sols and the formation of bridging network between the particles was verified by the SEM microscopy.

Fig. 1
figure 1

TEM image of TiO2 particles dispersed in the coating paste (TIPP:P25 = 0.1:1)

Full size image

Apparently, the pastes with TIPP:P25 = 0.05:1 and 0.1:1 exhibited stable dispersion as in Fig. 2a and b, respectively, while the sample with TIPP:P25 = 0.3:1 showed agglomeration of particles that were observed on the glass tube wall as in Fig. 2c. The excess amount of TIPP was found to produce agglomeration of P25 particles, due to an extended TiO2 network formation among the particles [22]. This type of agglomeration was also reported when the paste was stored certain period of time [23]. The film made by the paste with TIPP:P25 = 0.1:1 showed highest quality as shown in Fig. 2b. The quality of the deposited films were compared by the UV/VIS spectra as given in Fig. 2e. The film by the paste with TIPP:P25 = 0.1:1 showed highest transparency, which could be attributed to the uniform dispersion of particles that was observed in the paste as in Fig. 2a. The transparent film also showed noticeable flexibility as in Fig. 2d.

Fig. 2
figure 2

Prepared TiO2 pastes with different molar ratios of TIPP:P25 and corresponding film samples; a TIPP:P25 = 0.05:1, b TIPP:P25 = 0.1:1 and c TIPP:P25 = 0.3:1. d Flexibility of TiO2 film casted on the flexible substrate with the paste in b. e The UV/VIS spectra of the films a, b and c

Full size image

The other samples with lesser or higher amount of TIPP than the one in Fig. 2b showed much different properties. The film with lesser amount of binder (TIPP) showed poor adhesion as in Fig. 2a showing exfoliation at some parts of the sample. The adjusted TIPP:P25 ratio at 0.05:1 simply meant insufficient addition of binder in the film. On the contrary, the excess amount of TIPP (TIPP:P25 = 0.3:1) produced translucency of the obtained film. This is because of the particles agglomeration that was observed from the paste (Fig. 2c) [22].

Surface microstructure of the films was investigated by SEM and the micrographs are given in Fig. 3. As intended, porous nature of films were observed for all samples, however, the microstructure was somehow different according to the TIPP:P25 ratio. It could be easily noticed that the film with TIPP:P25 = 0.1:1 has most uniform surface without much apparent cracks as in Fig. 3c and d. Uniform dispersion of particles in the paste and the high transparency of the film that was observed with bare eyes were once again supported by the most uniform SEM microstructure of the prepared film. In contrast, the sample with lower (Fig. 3a, b) or higher amount of TIPP addition (Fig. 3f, g) showed much cracks on the surface. Insufficient addition of TIPP could produce cracks because of the poor binding of the particles, and the extended formation of TiO2 network by the excess amount of TIPP could also cause the cracks because of the stiffness of film layer [24]. This phenomenon of crack formation with the excess addition of sol–gel precursor could be understood by a development of the surface tension of pore fluid during the gel formation and the subsequent drying of gel [25]. The incorporated precursor molecules are ready to form the gel network via the hydrolysis and condensation reaction, which is subject to a drying after the film deposition. During the evaporation of the solvent from the gel network in ambient atmosphere inevitably produces stress in the network because of the surface tension, which is strong enough to break the network. To reduce the surface tension, many works employed the treatments such as addition of surfactant or dispersant [26, 27]. In the case of nanoparticle binding, the stress built-up in the TiO2 film is of course proportionally increasing with the increasing portion of gel network by the addition of TIPP. The crack formation in the TiO2 film with the excess amount of TIPP addition as in Fig. 3f could be attributed to this stress built up in the binder network. In contrast, the proper amount of TIPP (in our experiment, TIPP:P25 = 0.1:1) that produced endurable amount of stress gave uniform crack-free film as in Fig. 3c. From the observation, we could select TIPP:P25 = 0.1:1 as the optimum composition of paste. Thickness of the films was around 5 μm within the selected range of TIPP addition, as was shown in the cross-sectional view (Fig. 3e) of the film with TIPP:25-0.1:1.

Fig. 3
figure 3

SEM images of porous TiO2 films prepared with the pastes of different molar ratios; a and b TIPP:P25 = 0.05:1, c–e TIPP:P25 = 0.1:1 and f and g TIPP:P25 = 0.3:1. e is the cross-sectional view of the film prepared with the paste of TIPP:P25 = 0.1:1

Full size image

XRD analysis was done on the prepared photoanodes, and the results are given in Fig. 4. Both anatase and rutile phase that are the phase found in the P25 TiO2 powder (80 wt% anatase : 20 wt% rutile) were also detected in our experiment. Since the photoanodes were prepared by binding the P25 powder with TIPP through the low-temperature sol–gel process, the crystallographic phase of the photoanodes followed that of source powder, and the addition of TIPP does not exert any remarkable change in phase.

Fig. 4
figure 4

XRD patterns of the prepared photoanodes with different TIPP:P25 ratios; a TIPP:P25 = 0.05:1, b TIPP:P25 = 0.1:1 and c TIPP:P25 = 0.3

Full size image

The unit cells with a dimension of 0.6 mm × 0.6 mm were built using the prepared porous TiO2 active electrodes. The photoconversion efficiency of the cells was investigated by the J–V measurement method using the unit cells having the sandwich type structure as in Fig. 5a, and the results are given in Fig. 5b, and the efficiency from the flexible unit cell was also compared with the rigid device, where ITO-coated glass used as the substrate. The photoconversion efficiencies of the TiO2 electrodes with different binder contents, and the fill factors (FF), open circuit voltages (Voc) and short circuit currents (Jsc) are compared in Table 2. The unit cell assembled with the TiO2 film of TIPP:P25 = 0.1:1 with the thickness ~5 µm exhibited the highest efficiency of 1.96 %, which is noticeably low when compared with the rigid cell (2.93 %). This lowered efficiency could be attributed to the higher sheet resistance of the ITO-PEN (20–25 Ω/□) than ITO-glass (~5 Ω/□). The sheet resistance of the ITO PEN leads to high series resistance and hence to a lower Jsc which is shown in Table 2 (4 mA/cm2). From this it is suitable to give a future R&D route to reduce the ITO-PEN sheet resistance to reach higher current density. Also, we think that the TiO2 photoanode contains micro cracks, which increase the recombination component in the TiO2 layer. The Jsc is still lower in this study regarding values obtained in other research works (14–20 mA/cm2, Voc = 0.8 Volts, η = 7.29 %) [28–30].

Fig. 5
figure 5

a The unit cell structure prepared in this study, and b the J–V curve of the unit cells comprising the porous TiO2 film prepared with the paste of TIPP:P25 = 0.1:1

Full size image
Table 2 Efficiency of DSSC unit cells (0.6 × 0.6 cm)
Full size table

FF for all the unit cells prepared was around 60 % and Voc was in the same range, while Jsc differs with the variation of binder content; from 3.14 mA/cm2 for TIPP:P25 = 0.3:1 to 4.44 mA/cm2 for TIPP:P25 = 0.1:1. It was shown that the enhancement of efficiency for TIPP:P25 = 0.1:1 mainly comes from the enhancement of short circuit current, which was induced from the optimized binding among the TiO2 particles providing better pathway for the diffusion of injected electron.

In this study, the binder (TIPP addition into the paste) effect on the low-temperature-prepared active electrode of flexible DSSC was demonstrated. By comparing the photoconversion efficiency of the prepared unit cells, we found that the appropriate composition of binder (TIPP:P25 = 0.1:1), which exhibited the best dispersion of particles in the coating paste and the most transparent film with uniform microstructure, resultantly produced the best unit cell performance.

4 Conclusions

We investigated the effect of TIPP addition on the photoconversion efficiency as the binder for the TiO2 particles in a low-temperature processed photoanode for flexible DSSC. By a direct addition of TIPP, a stable TiO2 coating paste of low-temperature curable photoanode was prepared. The formation of bridging network between the particles was observed by TEM, and the crack-free uniformly porous film was obtained with the optimized composition of the paste as TIPP:P25 = 0.1:1. The photoconversion efficiency obtained by using the optimized film of thickness ~5 µm was 1.96 %.