A rapid cooling (quenching) step has been introduced in fabrication of TiO₂ photoelectrodes for dye-sensitized solar cells (DSSCs). The quenching process, studied at a fixed sintering temperature, decreased particle size but increased surface roughness without any substantial change in the crystal structure or oxidation state of TiO₂ films. Therefore, the change in the DSSC performance induced by the quenching was related closely to the microstructural and morphological changes in the TiO₂ films. Smaller particle size and the rough surface of TiO₂ films facilitated dye adsorption and increased the number of active reaction sites. In particular, the enlarged number of active reaction sites produced by the quenching process promoted the charge transfer reaction at the TiO₂-dye-electrolyte interface, resulting in overall performance improvement of DSSCs. The conversion efficiency of the furnace cooled- and quenched-TiO₂ films at 500 ^oC were 4.588% and 5.797%, respectively

Keywords: Dye-sensitized solar cells, Photoelectrode, TiO₂, Sintering process, Quenching method

Dye-sensitized solar cells (DSSCs) are a promising technological advancement due to their conversion efficiency, low-cost fabrication process, and diverse applicability compared to conventional silicon-based solar cells [1-3]. The structure of a DSSC is composed of photoelectrode, dye sensitizer, electrolyte, counter electrode, and transparent conducting substrate. Electrons are generated by excited-dye molecules adsorbed on the surface of the photoelectrode with incident light and move toward the photoelectrode or electrolyte [4-7]. Various efforts such as electrode surface morphology control [8-12], dopants in the electrode [13-15], develop- ment of advanced dyes [16-18], and electrolytes [19, 20] have aimed to improve the performance of DSSCs.
The photoelectrode in a DSSC plays a key role in enhancing the diffusion of electrons and adsorbing a large number of dye molecules at the electrode surfaces [21-23]. Therefore, the photoelectrode layer usually is fabricated in the form of a mesoporous film with a large surface area to promote efficient light harvesting and a large amount of photo-generation [24]. Meanwhile, TiO₂ is one of the most suitable materials to satisfy photoelectrode conditions in DSSCs because of its large band gap (anatase: 3.2 eV, rutile: 3.0 eV), suitable band edge levels for charge injection/extraction, non-toxicity, and low cost [25-28].
Generally, it is advantageous to reduce the particle size of TiO₂ in the photoelectrode because more dye molecules can be adsorbed at the extended surface area [7]. However, the reduction of particle size leads to an increase in defects at the boundary [29, 30]. Cao et al. reported that particle size and surface area in the mesoporous films strongly affect electron-hole generation and charge transport from dye to TiO₂ particles in the photoelectrode [31, 32]. Yanagida et al. reported that particle size and surface area of TiO₂ are major factors that affect the electron diffusion coefficient and recom- bination lifetime [33, 34]. Yang et al. reported that particles with a hollow hemisphere structure can increase the amount of dye adsorption and electron collection efficiency by inducing the light scattering [35]. Meanwhile, the dispersed micro/nano-sized noble metal particles (Au, Ag etc.) in metal oxide nano- particles can improve the performance of DSSCs by plasmonic resonance effect of electrons on the material surface with respect to incident light [36-38]. In this regard, it is important to control the microstructure, such as particle size and surface morphology, of TiO₂ photoelectrode to improve the performance of DSSCs. The microstructure can be controlled by manufacture process such as deposition methods for the fabrication of mesoporous films (doctor blading, screen printing, spin coating, etc.), sintering process (temperature, holding time, heating/cooling rate). It is known that the rapid cooling (quenching) process can maintain high temperature phase even at room temperature and suppress the grain growth [39-41]. Especially, it has been reported that TiO₂ photoelectrode with small grain size showed much better DSSC performance than that with large grain size [42-45].
In this work, we introduced a rapid cooling (quen- ching) step into the conventional sintering process to modify the microstructure and surface morphology of the TiO₂ photoelectrode. The quenching process at various temperatures was conducted to investigate the changes in microstructure and surface morphology. The correlation between performance and change in the microstructure driven by the quenching process was investigated. Unlike the conventional sintering process, the quenching treatment in the sintering process increased the surface area of TiO₂ and yielded a larger number of adsorption sites during dye infiltration, resulting in performance improvement of the DSSCs.

Preparation of TiO₂ photoelectrode
Fluorine-doped tin oxide (FTO)-coated glass (TEC 8, Pilkington, UK) with a sheet resistance of 6-9 Ω/□ was cleaned using acetone (99.5%, Daejung Chemicals and Metals Co., Korea), 2-propanol (99+%, Alfa Aesar, USA), and DI-water for an interval of 10 min per solvent and then dried using an air gun. A screen printing method was used with TiO₂ paste (Ti-Nanoixde T/SP, Solaronix, Switzerland) for deposition of mesoporous TiO₂ films. The screen-printed samples were sintered at 500 ^oC for 30 min in a tube furnace in an ambient atmosphere as shown in Fig. 1. The heating rate was adjusted by 10 ^oC/min. In conventional cooled (named to “furnace cooled”), The furnace was set to drop from 500 ^oC to room temperature in 3 hours. Unlike the conventional furnace cooling step, in the quenching step, the sliding furnace was pushed to expose to room temperature atmosphere after finishing the sintering step. The final thickness of TiO₂ films was approximately 9 μm. Meanwhile, the sintering temperature was varied at 400, 450, and 500 ^oC to investigate the sintering tem- perature effect.

Fabrication of DSSCs
The prepared mesoporous TiO₂ films were immersed in a prepared 0.3 mM (bis(tetrabutylammonium)-cis-di(thiocyanato)-N,N’–bis(4-carboxylato-4’-carboxylic acid-2,2-bipyridine) ruthenium (Ⅱ) (N719, Solaronix, Switzerland) ethanol-based solution in a dark field at room temperature for 24 h. For the counter electrode, Pt was deposited on indium-doped tin oxide (ITO)-coated glass (STN 10, UID, Korea) with a sheet re- sistance of 6-9 Ω/□ using DC sputtering equipment (E-1030, Hitachi, Japan) at 25 mA for 30 s. The thickness of the deposited Pt was approximately 5 nm.
The DSSC cells were assembled by attaching the photoelectrode and the counter electrode such that they faced one another. A 60-μm-thick Surlyn film (Solaronix, Switzerland) was used as a sealant. An electrolyte with an iodide/tri-iodide (I⁻/I³⁻) redox couple in acetonitrile solvent (AN-50, Solaronix, Switzerland) was injected through the holes on the counter electrode. The active area of the TiO₂ photoelectrode and electrolyte was 0.25 cm² and 1.2 cm², respectively.

Characterization and performance evaluation
X-ray diffraction (XRD) was carried out using an X-ray diffractometer (X’Pert Pro Powder, PANalytical, Netherlands) with Cu Kα radiation to analyze phases and crystal structure. Diffraction patterns were recorded at a scan rate of 4^o/min over a 2θ range of 20 to 70^o. X-ray photoelectron spectroscopy (XPS) was performed using a surface analysis spectrometer (AXIS Nova, Kratos Analytical Ltd., UK) with a monochromatic Al Kα source to verify the valence state of Ti in the samples. The microstructure and surface morphology of TiO₂ thin films were analyzed by a high resolution-scanning electron microscope (HR-SEM, SU8230, Hitachi, Japan) and an atomic force microscope (AFM, Nanoscope Ⅲ, Veeco Instrument Inc., USA), respectively. Ultraviolet-visible (UV-Vis) spectroscopy was performed by UV-vis spectrometry (HP 8453, Hewlett Packard, USA) to analyze the dye absorption according to the cooling method change.
Current-voltage (I-V) characteristics of DSSCs were obtained under a standard illumination condition (100 mW/cm² AM 1.5G) using a source meter (Keithely 2400, Tekronix Inc., USA) in the range of ±1 V, at a scan step of 0.02 V and delay time of 0.50 s. Electro- chemical impedance spectroscopy (EIS) analysis was carried out using a potentiostat (SP150, Biologic SAS, France) with a frequency range of 9 mHz to 100 kHz to confirm the change of the electrochemical component in the DSSCs. The results of the I-V curve and EIS spectra were analyzed with the average values of at least six individual cells.

Crystallographic phases of mesoporous TiO₂ films sintered at 500 ^oC are shown in Fig. 2. The TiO₂ anatase phase (JCPDS 73-1764) and SnO₂ phase (JCPDS 46-1088) were observed in both the furnace-cooled and quenched samples. The SnO₂ phase comes from the FTO substrate. Based on the XRD patterns, no signi- ficant change in characteristic diffraction peaks was created due to the rapid cooling, which indicates that the quenching step during the sintering process did not induce any substantial change in the crystal structure of TiO₂.
XPS analysis was performed to confirm the valence states of Ti in the TiO₂ films because dye adsorption can be affected by the electronic structure of the TiO₂ photoelectrode. The Ti2p core-level spectra consisted of stoichiometric doublet peaks from spin-orbit splitting corresponding to Ti2p_3/2 and Ti2p_1/2 states at 458.8 and 464.5 eV, respectively, as shown in Fig. 3. Upon fitting by a Gaussian method, the shoulder peak of Ti³⁺2p1/2 at 459.1 eV as well as stoichiometric doublet peaks of Ti⁴⁺2p_3/2 and Ti⁴⁺2p_1/2 were observed. This indicates that TiO₂ and Ti₂O₃ coexist in the as-deposited TiO₂ films. Moreover, no peak shift was observed, and the peak areas were similar between the furnace-cooled sample and the quenched sample. Thus, the XPS analysis showed that the surface oxidation state and chemical bonding of TiO₂ films are not affected by a quenching step during the sintering process.
The HR-SEM images of the furnace-cooled and quenched TiO₂ films at various sintering temperatures are shown in Fig. 4. Porous TiO₂ films with nanometer-sized spherical particle necking were formed. Unlike the XRD and XPS data, microstructural changes in the TiO₂ film were observed according to presence of the quenching process. The quenched TiO₂ films had smaller particles compared to the furnace-cooled TiO₂ films.
The ImageJ processing program was used to quantify the distribution of TiO₂ particle size from the HR-SEM images. Fig. 5 shows the particle size distributions of the furnace-cooled and quenched TiO₂ films at various sintering temperatures, and the calculated average particle diameters are listed in Table 1. The average particle size of the samples increased with sintering temperature, as shown in Fig. 5 and Table 1. However, the quenched TiO₂ films had a smaller average particle diameter than did the furnace-cooled TiO₂ films at the same sintering temperature. This indicates that it is possible to induce an increase in the nano-sized void while maintaining the overall photoelectrode structure by introduction of a quenching process. Therefore, decrease in particle size of TiO₂ can lead to an ex- pansion of the adsorption site of the dye molecules and consequently increase the amount of dye loading.
AFM images of the furnace-cooled and quenched TiO₂ films at 500 ^oC are shown in Fig. 6. The quenched TiO₂ film had a relatively rough surface with higher root-mean-square roughness (RMS) values than the furnace-cooled TiO₂ film. The rough surface can enhance the active reaction area by expanding the interface between the electrolyte and the photoelectrode.
I-V characteristic measurement confirmed the perfor- mance according to change in the sintering process (Fig. 7), and the parameters are listed in Table 2. All the quenched samples showed better performance than the furnace-cooled samples at the same sintering tem- perature. The sample quenched at 500 ^oC exhibited the highest conversion efficiency of 5.797%. This per- formance is comparable with those of several studies (Table 2) [46-49]. The short-circuit current density (I_SC) of the quenched samples increased significantly com- pared to that of the furnace-cooled samples. Generally, dyes are a key component for improvement of DSSC performance because they generate electrons when sunlight is incident on the device. Therefore, the amount of dye adsorbed on mesoporous TiO₂ films is an impor- tant factor in generation and movement of electrons, which directly affects the current density and efficiency. Meanwhile, since the electricity generation mechanism of DSSCs is related closely to the electrochemical reactions, the charge transfer reaction at the TiO₂-dye-electrolyte interface, which is defined as a triple phase boundary, is a critical performance factor. The electro- chemical charge transfer reaction strongly depends on the number of active reaction sites. As the number of active reaction sites increases, the charge transfer reaction proceeds more actively. Therefore, the small particle size and rough surface morphology according to intro- duction of the quenching process, as shown in Table 1 and Fig. 6, can increase the amount of dye loading and promote the charge transfer reaction, thereby improving the performance.
UV-visible spectroscopy analysis was performed to compare the adsorption amount of N719 dye according to sintering process of the TiO₂ photoelectrode (Fig. 8). The quenched TiO₂ film showed higher absorbance in the 525-535 nm wavelength range corresponding to the N719 dye molecule compared with the furnace-cooled TiO₂ film. This indicates that the amount of dye ad- sorption increased by introduction of the quenching process.
EIS analysis was used to confirm the relationship between performance of the DSSCs and electrochemical reaction. AC impedance spectra of DSSCs with the furnace-cooled and quenched TiO₂ films at various temperatures are shown in Fig. 9. In general, there are typically three distinct semicircles on the Nyquist plots of DSSCs. The first semicircle in the high-frequency range, the second semicircle in the middle-frequency range, and the third semicircle in the low-frequency range are designated to redox reaction of 3I^-/I^3- at the Pt counter electrode (R_CE), charge transfer reaction at the interface between TiO₂/N719 and electrolyte (R_ct), and Warburg diffusion process of iodide (R_w), respectively [50-52]. The electrochemical parameters obtained by simulation of fitting between equivalent circuit and impedance spectra from Fig. 9 are listed in Table 3. The measured impedance data (blank dots) are well matched with the simulated data (lines) calculated by the equivalent circuit. Therefore, the calculated electro- chemical parameters could be meaningful. As shown in Fig. 9, the size of the second semicircle for all the quenched samples decreased significantly compared with the furnace-cooled samples at the same sintering temperature. Similarly, the resistance values of charge transfer (R_ct) and diffusion process (R_w) for all the quenched samples decreased compared with those of the furnace-cooled samples (Table 3). This result is consistent with the tendency of I_SC and efficiency in I-V characteristics. Namely, introduction of the quenching process might facilitate the charge transfer reaction and diffusion process. It is also reported that the second semicircle decreases when the number of electrons generated increases [52]. As mentioned on the I-V characteristics section, since the electrochemical charge transfer reaction is related closely to the number of active reaction sites, the small particle size and rough surface morphology induced by quenching increased the number of active reaction sites and decreased the resistance of the charge transfer reaction, leading to improvement of overall DSSC performance.

The effect of the quenching process on the performance of a TiO₂ photoelectrode for DSSCs was investigated. Introduction of the quenching step during sintering did not induce any change in the crystal structure or surface oxidation state of TiO₂ films. However, the quenching process significantly affected the microstructure and the surface morphology of TiO₂ films. The quenched TiO₂ films had smaller particles and rougher surface morphology compared to the furnace-cooled TiO₂ films. These microstructural and morphological changes were related to the overall DSSC performance. Based on UV-visible spectroscopy analysis, the amount of dye adsorption on the quenched TiO₂ films increased com- pared with that of the furnace-cooled TiO₂ films. In terms of the electrochemical reaction, microstructural and morphological changes such as small particle size and rough surface morphology caused by quenching increased the number of active reaction sites and consequently promoted the charge transfer reaction and diffusion process, improving overall DSSC performance.

This research was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (No. 2020R1A2C1004580). This work was also supported by the Technology Develop- ment Program to Solve Climate Changes of the National Research Foundation (NRF) grant funded by the Korea government (Ministry of Science and ICT) (2017M1A2A2044930). This work was partly supported by Korea Institute of Energy Technology Evaluation and Planning (KETEP) grant funded by the Korea government (MOTIE) (20213030040110).