Preparation of Nanocrystalline TiO₂ with Mixed Template and Its Application for Dye-Sensitized Solar Cells

Conducting glass plates (indium tin oxide, ITO, Geomatec Co. Ltd.) were used as the substrate for depositing nanocrystalline electrodes. Tetraisopropylorthotitanate (TIPT), acetylacetone (ACA), LiI, CTAB (Wako Chemicals), F127 (BASF), cis-di(thiocyanato)--bis(2, 2^'-bipyridyl-4-carboxylate-4^'-tetrabutyl ammonium carboxylate) ruthenium(II), (N719; Solalonix), tert-buthylpyridine (TBP), and dimethylpropylimidazolium iodide (DMPII) (Shikoku kasei) were used as received.

A mixed template was composed of copolymer F127 and surfactant CTAB. The gel for electrode preparation contained the hydrolyzation and condensation of TIPT. First, 6.8 g of tetraisopropyl orthotitanate (TIPT) was mixed with 2.4 g of acetylacetone (ACA) to form a stock solution, which was added to 60 mL mixed template solution with a fixed amount of F127 (10 wt %), and various concentrations of CTAB. The final solution was stirred at 40°C until a uniform transparent solution was obtained. Then the resulting solution was kept at 80°C in a closed container for three days and formed a bright yellow gel. Finally some of the gel was directly calcined at 450°C for 2 h under ambient atmosphere to remove the templates and the thus formed white nanoscale was used for taking measurements such as X-ray diffraction (XRD) and transmission electron microscopy (TEM). For the preparation of photoelectrodes, several drops of gel were spread onto the conducting glass substrate using a glass rod with adhesive tape (Scotch, ∼40 μm thick) as the spacer. We fabricated thick films by repetitive coating and calcining at 450°C for 10 min. At last, the film was calcined at 450°C for 1 h before DSC fabrication and film thickness measurement.

After sintering at 450°C for 1 h, the porous electrodes were cooled to about 80°C, and immersed in an ethanol solution of a ruthenium complex (N719) over 24 h for dye adsorption. The dye-adsorbed electrodes were assembled like a sandwich with a counter electrode (a platinum-sputtered ITO glass) held by clamps. A drop of electrolyte solution (0.1 M LiI, 0.6 M DMPII, 0.05 M and 0.5 M TBP in methoxyacetonitoryl) was introduced into the clamped electrodes.

The crystalline phase of and its particle size were determined by XRD (Rigaku Goniometer PMG-A2-CN2155D2) with Cu Kα radiation at a scan rate of The average crystallite size was calculated according to the Scherrer equation. The morphology of samples was examined by TEM (JEOL 200CX, operated at 200 kV). The specimen for TEM imaging was prepared by suspending 2 mg of the powder sample with ethanol and sonicating it in an ultrasonic bath for 30 min. The Brunauer-Emmett-Teller (BET) surface area and pore size distribution of the samples were determined using a BEL SORP 18 PLUS nitrogen adsorption apparatus with all samples degassed at 200°C for 2 h before the actual measurements. The morphology and thickness of the mesoporous electrodes were characterized by scanning electron micrograph (SEM, JEOL, JSM-5510). Photocurrent-voltage measurements were performed using an AM 1.5 solar simulator (100 mW/cm², Oriel, 1000W9112). The cell size was 0.25 cm². The cells made of P25, a commercial -containing anatase and rutile phase, were also prepared for comparison.

Table I shows the synthesis conditions and the corresponding physicochemical properties of the as-prepared particles. It clearly shows that the surface area of particle was very large, and varied with the concentration of CTAB in the mixed templates in which the concentration of F127 was fixed at 10 wt %. The as-prepared particles were very small with an average size of 3-5 nm, depending upon the concentration of CTAB in the template. The particle with the smallest average size of 3.58 nm was obtained in 0.1 M CTAB. Figure 1b shows a typical TEM image of particles prepared in the template system. It reveals that the nanocrystals have a relatively spherical shape with a size distribution of 3-5 nm. Figure 1a showed the XRD patterns of the particles calcined at 450°C for 2 h. Clear and intensive anatase patterns according to JCPDS card 21-1272 are observed in different systems, and the shape of the patterns are similar to each other. The diffraction intensity gradually increased with the concentration of CTAB in the 10 wt % F127 mixed template, which suggested increasing crystallinity in these materials. The size of particles calculated by Scherre equation is below 5 nm and are shown in Table I which agrees well with the results of TEM observation. It indicates that the particles are primary particles with excellent dispersibility and stability even with heat-treatment. Figure 1c is a high-resolution TEM image of sample B. It shows that all the primary particles are highly crystallized and the crystal growth according to the mechanism of oriented attachment¹³ does not occur during calcinations. That is due to the use of the mixed template which restricted the attachment and aggregation of primary particles and kept the growth of crystals. These results indicated that the template drastically affects the size, surface area, and crystallinity of particles. In the present case, copolymer F127 promoted the homogeneity of the mixture solution and hindered the aggregation and recombination of the neighboring crystalline nucleus in the formation and calcination processes, just as reported in the literature.¹⁴ However, the size of the resultant particles is below 5 nm, which is smaller than the micelle size formed in single copolymer F127 without CTAB. It is expected that surfactant CTAB has modified the micelle structure and resulted in the formation of smaller nanocrystals, which is in accordance with the reports in the literature.^{10 11 12} In these studies, Chari et al.¹⁰ and Griffiths et al.¹¹ found that some selected surfactant could strongly interact with the polymer to form micelles of smaller size. In our work, the CTAB surfactant and copolymer F127 are expected to have been incorporated with each other which leads to the reduction of the size of the produced particle in the mixed template system. Because of its high surface area and high crystallinity, all the as-prepared samples have a quantum confinement effect and high photocatalystic activity. Considering the surface area and size of particles, we selected sample B with the largest surface area and the smallest size to prepare film electrodes.

Table I. 

Synthesis condition and physicochemical properties of particles.
Sample	10%	Crystallite size (nm)		Pore volume(mL g−1)	Pore size (nm)
A	0.05	4.62	150.12	0.542	4.32
B	0.1	3.58	185	0.677	4.97
C	0.25	4.13	167.69	0.604	5.22
D	0.5	4.62	143.98	0.205	6.58

Zoom In Zoom Out Reset image size

Figure 2 shows the SEM images of the surface morphology and the cross section of film with different coating times after calcinations at 450°C for 1 h. The thickness of porous film with the first coating is about 1.1 μm (Fig. 2a). film with 3.2 μm thickness is obtained after three coatings (Fig. 2b). Eight coatings makes a film with a thickness of 8.2 μm. It indicated that almost the same thickness of film is fabricated with each coating. Figure 3 shows the thickness as a function of the number of coating times. The film thickness was increased monotonically with the coating times in the two film series. Curves a and b correspond to different concentrations of Ti in the gel, the former is 0.4 M and the latter 0.2 M. However, the increasing rate is slightly deaccelerated over the six coating process. Moreover, it is clear that no crack or gap can be observed between each coating layer, and the surface of each layer is very smooth and homogeneous without cavities and cracks, although the film was not made by a one-time coating process. (Fig. 2f-h). Figure 2j shows a high magnification SEM image of surface morphology with the eight coating processes which reveals that no clear change can be observed in the size of particles. These results indicated that using a mixed template effectively avoided cracks in the film and maintained the homogeneity and also prevented the aggregation of particles in coating and calcination. Also the burning and decomposition of the mixed templates in calcinations makes the film a porous structure which is suitable for the solar cell electrode (Fig. 2f-j). From the comparison of transmittance between the coated electrode and the substrate, we can show that the film is transparent in the visible light range even with eight coatings (shown in Fig. 3 inset). As is well known, light scattering and light absorption are both factors that decrease optical transmission. The transmittance decrease in the ultraviolet range is caused by the absorption of Light scattering in our system is avoided by forming a homogeneous phase film with small size particles stuck together and phase separation is prohibited. In our work, the electrode can be formed with perfect transparency although it may not be the best condition for high-efficiency DSC, which is illustrated in the end of this paper.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

All the SEM images show a fine crack-free porous electrode without a distinct cavity which can be attributed to the perfect coating characteristics of the gel and also the coating technique. From the second layer coating, the gel is coated on the still hot substrate, the temperature is about 80°C which is the formation temperature of gel in the preparation. Because the mixed template of F127 modified by CTAB keeps excellent gelation properties at this temperature, the gel is immediately dried and tightly attached into the underlayer, which avoids the formation of gaps between two layers and crack or cavities with each coating in comparison with the traditional dispersed technique in which the crack and cavities are often formed due to incomplete dispersion and unsuitable composition of the paste. Meanwhile, the removal of the mixed template by evaporation followed by calcinations produced a finely porous structure in the film, which can be observed in the SEM images even with thick film (Fig. 2f-j). The SEM images also indicate that the porous structure is formed with each repetative times coating, drying, and sintering and does not vary with the increasing thickness of the film. Of course, the size of the pores would shrink with increasing heat-treatment time and temperature as reported.¹⁵ In our work, when the coating is repeated over six times, the thickness of each coating is decreased which is expected to be due to the shrinkage of the porous structure in the underlayer with continuous heat-treatment. However, the homogeneous gel with very small size pores prepared with the mixed template has avoided the cracks in the film and gives a perfect porous electrode. With this property, the gel can be used to make film with large thickness depending on the number of coatings.

The performance of the DSC can be characterized by measuring the current-voltage curves of the prepared DSC from which the light-to-electricity conversion efficiency can also be achieved. Figure 4a shows the dependence of the light-to-electricity conversion efficiencies (η) on the repetitive coating times of films. Each measurement was carried out using six electrodes to confirm the reproducibility. The conversion efficiency of the film at the one-time coating was 3.2-3.6%. The η of DSC increased monotonically with the thickness of the film till the six coatings, and decreased from seven coatings. At the six coatings η became 8.3%, which is about 2.5 times higher than that of the one-time coating process. Figure 5 shows the current-voltage characteristics of the dye-sensitized solar cell with a film coated six times (about 6.1 μm thickness). The open-circuit photovoltage the short-circuit photocurrent the fill factor, and η were 760 mV, 17.2 mA/cm², 0.635, and 8.3%, respectively. The efficiency was far higher than mesoporous electrodes made from commercial P25 with the same thickness. Also the latter can not be coated many times due to the cracks and gaps produced between or in the layers.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figure 4b shows the dependence of the short-circuit current on the thickness of the film prepared with the gel prepared in this work and also commercial P25. It is clear the of the DSC prepared in our work is far higher than that of P25 electrodes with the same thickness. However, when the thickness of the film is over 7 μm, the efficiency and photocurrents are generally decreased. These results are different from some other reports, in which the thick is an ideal candidate for high conversion efficiency in DSC.^{1 2 3 4 5} The amount of dye adsorbed in the film is increased monotonically even over the thickness of 7 μm (Fig. 4c). The difference between our work and others must be due to the size of our prepared particles and the porous structure of the prepared film. The particles prepared in this work are much smaller than those produced in previous work,^{1 2 3 4 5} so in our DSC the can adsorb more dye than that produced with larger size particles with the same film thickness. Thus, when the thickness is over 6 μm, the film might have adsorbed enough dye so that the quantity of adsorbed dye is not the decisive factor for efficiency any more. Because of the relatively small size of the particle, the electron transfer impedance can be much larger due to more sticking points between particles which hinder the electron diffusion in the nanoporous film electrode. Moreover the thick film increased the diffusion resistance of the electrolyte, especially the large ions, which increased the opportunity for the recapture of an electron by and resulted in decreased efficiency. As we have shown, the average pore size of film depended on the average size of the aggregation crystals that have been formed after calcinations. In our work, the average pore size is very small due to small particle size (3-5 nm) and less aggregation during calcinations which is due to the mixed template used. From the SEM image and adsorption measurement, the pore size of the porous film is just about 4-6 nm. In such small pores, after the dye has been adsorbed in the porous film, 3 nm are occupied by the dye molecules (molecular diameter of 1.5 nm) that adsorbed on the pore walls, which leaves an aperture of only 1-2 nm for the diffusion of the electrolyte. This distance is very similar to the size of the ion which makes the diffusion of electrolyte become the limiting step under this condition. This has been verified by Gratzel et al. 's research and has been illustrated in detail.¹⁵ Considering the mechanism of the DSC, we can conclude that the thickness of the film has an opposite bifunctional character as increasing the dye adsorption as well as increasing the diffusion resistance of the electron and electrolyte. The DSC is a rather complicated system including such processes in series as light adsorption, charge injection, charge collection, and electrolyte diffusion. There are many factors that influence the resultant light-electricity conversion efficiency of DSC. So the high efficiency cannot be achieved by optimization of only one or two factors and processes. The DSC must be considered as a whole in order to obtain a high efficiency DSC. Using our method we can make a highly transparent electrode with adequate thickness. But this does not mean that high transmittance is the best condition for DSC because it increases the loss of light energy. In our work, the high transmittance electrode may be one reason for not having obtained higher efficiency DSCs. It has been suggested that adding a top scattering layer, which can sufficiently scatter and use the limited light may be a better way to increase the efficiency of our DSC. The effect of a scattering layer will be investigated in further work.