TiO2 treatment using ultrasonication for bubble cavitation generation and efficiency assessment of a dye-sensitized solar cell

Graphical abstract


Introduction
Solar cells are clean and renewable energy sources that do not cause environmental pollution. Depending on their constituent chemicals, solar cells are divided into silicon solar cells [1,2], thin film solar cells [3,4,5], organic solar cell [6,7], dye-sensitized solar cells (DSSCs) [8,9], and perovskite solar cells [10,11]. Among them, silicon solar cells are the most commonly used because they show a high efficiency of 25% or more [1]. However, they are expensive because of their complicated manufacturing process and high energy consumption [2]. To compensate for these disadvantages, many researchers have developed new solar cells to replace silicon solar cells, and one of them is the DSSC.
The DSSC was first introduced by Grätzel in 1991. It generates electricity by irradiating light energy to a dye adsorbed on a semiconductor photoanode, and the excited electrons trigger a redox reaction in the electrolyte [12]. TiO 2 , which is a metal oxide semiconductor, is economical and has high physical and thermal stabilities, making it the most suitable photoanode for DSSCs [8,12]. Although the DSSC using TiO 2 has low cost and high stability, it has low dye adsorption and high electron recombination rates, which pose as obstacles to the efficiency improvement [9].
One of the methods of increasing the TiO 2 photoelectrode efficiency is doping with a transition metal, such as Cu [13], Zn [14], Zr, and W [15]. In our previous studies, doping with various transition metals also improved the TiO 2 efficiency. Park et al. extended the reactive region of the photoelectrode to the visible region by Cu doping and achieved a high efficiency of up to 11.35% [16]. Meanwhile, Kim et al. improved the dye adsorption by the post-treatment of Zr oxide in TiO 2 and obtained an efficiency of approximately 7.03% [17]. Lee et al. improved the specific surface area and dye adsorption by doping TiO 2 with SiO 2 and N. They enhanced the energy conversion efficiency by up to 8.68% [45]. On the contrary, current studies have encountered difficulties when adding complex heat treatment processes, such as the hydrothermal method, for doping transition metal, which consumes much time to manufacture. Therefore, a new process for improving the TiO 2 efficiency must be developed.
The methods of preparing TiO 2 into a photoanode are spin coating [18], dip coating [19], screen printing [20], and doctor blade method [20], among others. Spin coating can smoothly control the voids and thickness of the photoelectrode, but the paste may aggregate at the edges due to a centrifugal force [21]. In dip coating, the photoelectrode thickness is uniform and thin; however, as a limitation, the surface is formed non-uniformly when the paste concentration is high [22]. The screen-printing technology is advantageous for the large-scale production of paste due to uniform thickness. However, in this method, paste is thickly applied, consequently impairing the DSSC efficiency [20]. The doctor blade method is a general photoanode manufacturing method, in which TiO 2 particles are uniformly distributed inside the paste, and particle loss is small; however, when high-concentration paste is applied, the thickness is not constant, thereby needing improvement [20].
In this study, we will try to improve the energy conversion efficiency of the DSSC by adding a simple and new "sonicated" process to the currently used doctor blade preparation method. Sonication is one of the processes used to form nanoparticles. It leads to surface activation by forming defects on the particle surface as a cavitation reaction [23,24,25]. The cavitation reaction is defined as a process in which bubbles in a fluid are formed, grown, and collapsed [26,27]. It was confirmed that the cavitation reaction is a physical and chemical reaction that oxidizes and activates the catalyst surface [28,29,30]. Hafeez et al. reported that the high heat and pressure accompanying the cavitation reaction cause defects and deformation on the particle surface, which contribute to the catalyst activation [23]. Narakaew attempted to improve the activation and reactiveness of the TiO 2 surface by the cavitation phenomenon [36]. In addition, Stucchi et al. improved the dispersion of the active area by dispersing the particles through ultrasonication [35,36]. Thus, ultrasonic treatment could results in surface activation and improves the performance of catalysts and photocatalysts by uniformly dispersing materials. However, the sonication process has rarely been applied in the DSSC studies reported so far. Therefore, in the present study, a DSSC was manufactured by applying ultrasonicated TiO 2 to the photoelectrode. Its applicability and superiority are then investigated. The effect of TiO 2 treated with ultrasonic cleaner and ultrasonic horn on the DSSC efficiency is also studied.

Materials
The DSSC used in this experiment was prepared using

Preparation of the TiO 2 (commercial TiO 2 ) paste
The TiO 2 paste was prepared through the sol-gel method with 25 ml ethanol, 2.15 ml α-terpinol, and 0.6 ml distilled water added to 2 g of TiO 2 and stirred at 120 • C at 300 rpm for 15 min. Subsequently, 0.3 g of ethyl cellulose was added and stirred until the paste was completed. The complete paste was applied to the FTO plate by the doctor blade method, which is a method of coating with a constant thickness using a blade, and the process is as follows. In order to prepare a photoelectrode having an area of 5 mm × 5 mm on FTO glass, the other areas were masked. After this process, a certain amount of paste was repeatedly coated on glasses with 5 to 10 times using a blade, and the TiO 2 photoelectrode was manufactured by a calcination process at 450 • C for 2 h at a temperature increase rate of 5 • C/min. The photoelectrodes were immersed in an N719 dye (0.5 mM) in ethanol at room temperature for 24 h.

Preparation of the ultrasonic cleaner-TiO 2 paste
The ultrasonic cleaner-TiO 2 paste manufacturing added an indirect ultrasonic treatment process to the same method for the TiO 2 paste preparation. After adding 2 g of TiO 2 to 50 ml of ethanol, ultrasonic treatment (intensity high) was indirectly performed in an ultrasonic cleaner bath for 30 min to prepare an ultrasonic cleaner-TiO 2 solution.
α-Terpinol (2.15 ml) and 0.6 ml distilled water were added to the ultrasonic cleaner-TiO 2 solution and stirred at 120 • C at 300 rpm for 15 min. After this process, 0.3 g ethyl cellulose was added and stirred until the paste was completed. An ultrasonic cleaner-TiO 2 photoelectrode was then manufactured in the same manner as that of TiO 2 .

Manufacture of the ultrasonic horn-TiO 2 paste
The ultrasonic horn-TiO 2 paste was manufactured in the same manner as the ultrasonic cleaner-TiO 2 paste. Ultrasonic treatment was directly performed. An ultrasonic horn-TiO 2 solution was prepared by directly sonicating a solution, in which 2 g of TiO 2 was added to 50 ml of ethanol with an ultrasonic horn for 30 min. Thereafter, an ultrasonic horn-TiO 2 photoelectrode was manufactured in the same manner as the ultrasonic cleaner-TiO 2 .

Counter electrode manufacturing and joining
The counter electrode was ultrasonically cleaned in ethanol for 15 min after drilling two holes through which the electrolytes could be injected in the FTO glass. A H 2 PtCl 6 solution was coated on the washed FTO glass by spin coating and then calcined at 350 • C for 1 h to prepare a counter electrode. After joining the prepared counter electrode and the photoanode, Iodolyte AN-50 was injected as an electrolyte, and the hole was closed to complete the DSSC.

Characterization
The scanning electron microscopy (SEM, ×3000), transmission electron microscopy (TEM, ×250000), Brunauer-Emmett-Teller (BET), and X-ray diffraction (XRD, scan rate 5 • /min), and X-ray photoelectron spectroscopy (XPS) analyses were performed in powder form. XRD and XPS were performed to understand the shape and structure of the manufactured TiO 2 . The specific surface area was measured through BET. The surface properties were analyzed through SEM and TEM analyses. The prepared DSSC was analyzed with photo-electrochemical data using 2400 source (Keithley Instruments) under AM 1.5 illumination (100 mW/cm 2 ). The electron mobility of the electrode was analyzed by EIS. The chemical capacitance and the recombination resistance were calculated using the EIS and efficiency data.

Characteristics analysis of TiO 2
The SEM analysis was performed to determine the dispersion of TiO 2 , ultrasonic cleaner-TiO 2 , and ultrasonic horn-TiO 2 (Fig. 1). The particles in TiO 2 (Fig. 1 a) were agglomerated, whereas those in the ultrasonically treated ultrasonic cleaner-TiO 2 and ultrasonic horn-TiO 2 were dispersed ( Fig. 1 b and c). The dispersion degree of the directly applied ultrasonic horn-TiO 2 was relatively higher than that of the indirectly applied ultrasonic cleaner-TiO 2 , which was consistent with the results of the previous studies showing that ultrasonic waves affect the particle dispersion [36]. The BET analysis was performed to determine the change in the specific surface area before and after the sonication of each TiO 2 ( Table 1). The specific surface areas were measured to be approximately 87.50 m 2 /g for TiO 2 , approximately 88.15 m 2 /g for ultrasonic cleaner-TiO 2 , and approximately 87.44 m 2 /g for ultrasonic horn-TiO 2 . Although the ultrasonic waves affected the particle dispersion, they were thought to not contribute to the changes in the specific surface area.
The previous analysis showed the external change of the TiO 2 particles by ultrasonication. The changes in the properties of the inside of the particles were confirmed by the results of the TEM, XRD, and XPS analyses. Fig. 1 (e-h) depict the TEM images of each sample measured at 250,000 × magnification. Accordingly, TiO 2 particles with a diameter ranging from 30 nm to 50 nm were observed. The roughness of the outside of the particle was increased by the ultrasonication, which is more clearly shown in the insets describing TiO 2 (Fig. 1 e) and ultrasonic horn-TiO 2 (Fig. 1 f). It was assumed that the ultrasonic wave was applied to the liquid medium (ethanol) mixed with TiO 2 , and the generated cavitation bubble formed a defect on the particle surface. The TiO 2 surface defects occurred in the form of an oxygen vacancy or Ti 3+ [31,32] and did not significantly affect the surface area of the particle itself. It may, however, contribute to the particle surface activation [33]. This is related to the XPS analysis results, which will be described later. The TiO 2 surface activation caused by the defects improved the photocatalytic performance [34]. Fig. 2 shows the XRD analysis results comparing the ultrasonic cleaner/ultrasonic horn-TiO 2 and TiO 2 . Three spectra with almost identical diffraction peaks were identified, regardless of ultrasonication. The distinct diffraction peaks at 25 • , 38 • , 48 • , 54 • , and 55 • were consistent with the intrinsic spectrum of the anatase phase TiO 2 (JCPDS no.: 88-1175 and 84-1286) [37,38]. This indicates that no change occurred in the crystal compositions of the three TiO 2 used herein.
The XPS analysis was performed to analyze the oxidation state of each TiO 2 surface. Fig. 3 depicts the obtained results. All XPS analyses were corrected for C 1 s (284.6 eV). In the Ti 2p spectrum, two peaks corresponding to Ti 4+ (i.e., Ti 4+ 2p 1/2 and Ti 4+ 2p 3/2 ) and one peak representing Ti 3+ (i.e., Ti 3+ 2p 1/2 ) were analyzed. The Ti 3+ 2p 1/2 peak originated from the oxygen vacancy on the TiO 2 surface caused by ultrasonication, indicating that the ultrasonic wave activated the TiO 2 surface [39,40]. The proportion of Ti 3+ on each TiO 2 surface was the highest in the ultrasonic horn-TiO 2 (8.38%). Those of the ultrasonic cleaner-TiO 2 and TiO 2 were 6.76% and 5.22%, respectively. These results demonstrate that ultrasonic treatment induced the TiO 2 surface activation.
The O 1 s spectrum showed the oxygen species present on the TiO 2 surface. The peaks corresponding to the lattice oxygen (bridging O, 529-530 eV) and active oxygen (central O, 530-532 eV) of TiO 2 were found in the O 1 s spectrum [41][42][43]. As shown in Fig. 3, the active oxygen in the ultrasonic horn-TiO 2 was 33.96%; that in the ultrasonic cleaner-TiO 2 was 31.27%; and that in C-TiO 2 was 27.89%. These results were consistent with those of the Ti 2p spectrum, in which the ultrasonic treatment improved the active oxygen content of TiO 2 and activated the TiO 2 surface.

Characteristics analysis of DSSC
The previous analysis results confirmed that ultrasonic treatment activates the TiO 2 surface. UV-vis, EIS, J-V graph analyses were employed to prove how the surface activation of TiO 2 affected the DSSC efficiency. First, the ultraviolet-visible (UV-vis) spectrum was used to check the dye adsorption, which was the main electron source of the DSSC herein, to confirm the increase in the dye adsorption (Fig. 4). To evaluate the amount of dye adsorption, the absorbance was measured by adsorbing dye to the photoelectrode for 24 h, then desorbing it with 1 mM KOH solution [44]. At 500 nm wavelength, the absorbance of the ultrasonic horn was significantly higher than that of the other two samples. The amount of dye adsorbed on the photoelectrode surface was observed in the following order: ultrasonic horn-TiO 2 > ultrasonic cleaner-TiO 2 > TiO 2 .
When compared to other two treatments, the ultrasonic horn-TiO 2 exhibited a higher dye adsorption caused by the particle surface activation, which was due to the cavitation bubbles with particle dispersion [45]. The high dye adsorption of the ultrasonic cleaner-TiO 2 and the ultrasonic horn-TiO 2 acted as a key factor for improving the energy conversion efficiency of the DSSC by increasing the current density   Table 3 Parameters for calculating the chemical capacitance and the recombination resistance. ΔE c means the difference in conduction band energy (E c ) between each material. Since these samples have the same chemical composition, there is no change in the E c value.
Samples ΔE c j 0 (mA/cm 2 ) R 0 (Ω cm 2 ) R s (kΩ)  [ 46,47]. Through the EIS analysis, impedance was measured to determine the electron transfer resistance. Fig. 5 depicts the EIS data observed under dark conditions. The semicircle in Fig. 5 depicts the charge transfer resistance measured at approximately 15 kΩ, with TiO 2 being the highest. The semicircle size of the ultrasonic cleaner-TiO 2 and the ultrasonic horn-TiO 2 was approximately 7-8 kΩ, confirming that the charge transfer resistance was 50% lower than that of TiO 2 . The resistance received by the electrons was low when the charge transfer resistance was low; thus, the current flowed well and may act as a factor for improving the DSSC efficiency.
The energy conversion efficiency of each TiO 2 photoanode was evaluated. Fig. 6 shows the J-V graph measuring the energy conversion efficiency. Table 2 lists the parameters. The energy conversion efficiency of the ultrasonic cleaner-TiO 2 was 2.82%, which improved by approximately 20% compared to TiO 2 (2.35%). The ultrasonic horn-TiO 2 showed an efficiency of 3.35%, which was 43% higher than that of TiO 2 (2.35%) and approximately 19% higher than that of the ultrasonic cleaner-TiO 2 . The amount of dye adsorption was increased by the dispersion effect of ultrasonication and the activation effect of the TiO 2 surface, which increased the current density. In conclusion, sonication contributed to the increase of the DSSC efficiency by lowering the electron transfer resistance and increasing the current density.
The chemical capacitance (C μ ) and the recombination resistance (R r ) were calculated to confirm the electrical characteristics of the DSSCs prepared using TiO 2 as a photoelectrode. The two calculations were based on the method proposed by Bisquret and the results of our previous study [44,48,49,50]. Table 3 shows the parameters required for the calculation. In the table, ΔE c denotes the difference in E c between each TiO 2 . We found no change in the E c value because the materials were chemically identical to each other. The R 0 value is a parameter that determines the effect on the recombination process (Eq. (4)). The R s value indicates the series resistance of the DSSC and obtained herein using the EIS data.
In Eq. (1), L denotes the TiO 2 film thickness; p denotes the photoelectrode porosity; and q denotes the elementary charge.

g(E Fn
Eq. (2) presents the Fermi level (E Fn ), Boltzmann constant (k B ), total density of the bandgap states (N L ), and exponent of the electrons below the conduction band. It is calculated using the constant (α) for the parameter for the exponential trap distribution of electrons under the conduction band.
The recombination resistance (R r ) is calculated using Eq. (3). The variable V F (Fermi level voltage) important for the R r calculation is calculated using Eq. (5). V appl represents the applied voltage.
The equivalent common conduction band voltage (V ecb ) of the V F value had the same value as the V F because the material used for each TiO 2 photoanode is chemically the same. Therefore, the chemical capacitance and the recombination resistance data using V ecb were not presented herein.
The chemical capacitance in Fig. 7(a) showed similar C μ values because each DSSC applied to the calculation is chemically the same material. Meanwhile, the recombination resistance in Fig. 7(b) illustrated values of approximately 200 Ω cm 2 for TiO 2 and approximately 450 Ω⋅cm 2 for ultrasonic horn-TiO 2 at V F 0.5 V. The R r values increased more than twice because the open-circuit voltage was increased by the effect of the ultrasonic treatment. The increase in the recombination resistance was one of the factors that increased the energy conversion efficiency along with the increases of the dye adsorption and the current density.

Conclusions
In this study, a TiO 2 photoelectrode was fabricated through a simple ultrasonication treatment. The surface area change of TiO 2 was analyzed by performing SEM and TEM analyses. Ultrasonic treatment greatly improved the particle dispersion and activated the TiO 2 surface. The XRD analysis confirmed that the anatase structure was maintained, even after ultrasonic treatment. The structural TiO 2 surface activation was analyzed by XPS, and the results showed that the cavitation bubble increased the active area of TiO 2 . The effect of the TiO 2 surface activation on the DSSC efficiency was investigated through UV-vis and EIS analyses. The UV-vis analysis revealed that sonication significantly increased the dye adsorption. Meanwhile, the EIS analysis confirmed that sonication reduced the charge transfer resistance. Although the chemical capacitance of each TiO 2 was calculated, the materials used in each DSSC were chemically identical, showing similar capacities. The measurement of the energy conversion efficiency of the DSSC depicted that the DSSC consisting of ultrasonic horn-TiO 2 had an energy conversion efficiency of 3.35%, displaying approximately 45% of improvement compared to TiO 2 (2.35%). In addition, the recombination resistance was calculated using the open-circuit voltage and current density values of the DSSC. The results confirmed that ultrasonic treatment contributed to efficiency improvement by increasing the recombination resistance. In conclusion, the simple and effective sonication described in this paper will be very helpful in improving the DSSC efficiency.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.