Enhanced Indoor Conversion Efficiency of Dye-Sensitized Solar Cells by Optimizing Ball-Milling Process of TiO₂ Paste

Abstract

The rapid spread of the Internet of Things (IoT) along with the development of innovative low-power electronic devices has also driven the development of indoor photovoltaics. In this paper, we propose a simple and economically feasible solution that can improve the efficiency of dye-sensitized solar cells (DSSCs) under indoor light conditions by ~112%, without requiring a complex TiO₂ photoanode architecture or the design of new dyes. The ball milling process of the TiO₂ paste was optimized for indoor light conditions for the first time, both in terms of efficiency and production costs, by developing a rapid preparation method that can be used industrially for the application of DSSCs. A simple use of 12 mm diameter balls caused beneficial structural modifications, decreasing the size of the crystallites, and leading to a high OH generation on the TiO₂ surface responsible for the improvement of energy conversion efficiency.

1. Introduction

In recent decades, electronic devices have become indispensable tools for society and modern life due to the rapid development and spread of the Internet of Things (IoT) along with the development of innovative, low-power electronic devices, such as mobile gadgets and sensors [1]. A global estimate shows that there were approximately 11.3 billion connected IoT devices at the end of 2020, and by 2025, there should be more than 27 billion [2]. Given the increasing scale of the IoT, significant maintenance costs for a periodic battery replacement must be taken into consideration [3]. Indoor photovoltaics (IPVs) are a potential energy-supply substitute due to their ability to collect ambient light and convert it into electricity [4]. The narrow emission spectrum of the commonly utilized indoor light sources, for example, light-emitting diodes and fluorescent lamps, establishes that the appropriate band gap of indoor light absorber materials is 1.9 eV [5,6]. Thus, the best efficiency of silicon solar cells (a-Si:H IPV) tested in indoor light conditions has reached up to 21% [7], but for commercial a-Si:H IPV, the efficiency is between 4.4% and 9.2% [8].

Recently, dye-sensitized solar cells (DSSCs) have received a lot of interest in the research and development of next generation solar cells, which are a promising alternative to conventional silicon solar cells due to their sufficient power conversion efficiency and projected cost effectiveness [9,10,11,12].

Due to the low incident light intensity (0.1–10 W m⁻²) in indoor environments, the visible spectrum is predominantly 200 lux in home areas, and 300–2000 lux in workspaces and other indoor spaces [13,14].

Until now, most researchers have focused on optimizing the dye in DSSC to improve effective light harvesting for indoor light conditions. Therefore, Liu et al. reported a new porphyrin dye Y1A1 with 19.5% PCE under indoor LED light intensity (350 lux) [15]. Reddy et al. fabricated SK7, another porphyrin dye with 19.72% PCE under a T5 light source (6000 lux) [16]. Wang et al. reported organic anthracene-based dyes (AN1, AN3, AN5, AN7, AN8) with 5.45% PCE for the AN3 dye when integrated in a DSSC module with a 0.16 cm² cell area under artificial light sources (1000 Lux) [17]. To increase the absorption extinction parameters with 27.76%, 28.74%, and 30.45% under a radiation intensity of 600, 1000, and 2500 lux, Jiang et al. produced Y-shaped sensitizers by integrating quinoxaline or quinoxalinoid moieties [18]. Freitag et al. coupled two sensitizers, D35 and XY1, and utilized a copper complex serving as the redox shuttle. In low fluorescent illumination settings (1000 lux), the PCE was 28.9% [19]. Likewise, Cao et al. combined the Y123/XY1b sensitizers with a Cu electrolyte, and the PCE was 31.8% under 1000 lux fluorescent light intensity [20]. Tanaka et al. combined a low-cost 5T with XY1 dye under artificial fluorescent lighting (1000 lux), and the PCE was 29.2% [21]. Michaels et al. combined a small organic L1 dye with XY1 dye, and the results showed an unprecedented PCE of 34% at 1000 lux, 32.7% at 500 lux, and 31.4% at 200 lux using FL irradiation [22]. Zhang et al. obtained a record efficiency for DSSC (1000 lux) in ambient light with a PCE of 34.5% using MS5/XY1b as dyes and the [Cu(tmby)₂]^2+/+ redox couple [23].

In addition to the optimization of dyes for indoor conditions presented above, photoanode optimization also plays a key role in improving the performance of DSSCs [24,25], ensuring the transport of photogenerated electrons and active sites for dye adsorption [26,27].

Apart from this, the photoanode made from a nanocrystalline porous TiO₂ film electrode plays an essential part since the surface characteristics of the cell, including surface area, roughness, porosity, and film thickness, significantly impact the total conversion efficiency of the cell [28,29]. Therefore, it is necessary to develop a rapid preparation for the TiO₂ paste that can be used industrially for the application of DSSCs to increase efficiency and lower production costs. Using the high energy planetary ball mill in the preparation of the TiO₂ paste to improve DSSC efficiency compared with untreated TiO₂ pastes; (i) leads to decreasing the size of the particles [30], (ii) breaks down the bonds between molecules, and (iii) helps to improve dye adsorption due to the high-force impact kinetic energy on the material, as well as the synchronized alternation of centrifugal forces [31,32].

Until now, most studies have focused on the ball-milling process used for DSSCs tested in outdoor conditions. Optimizing the manufacturing process for the TiO₂ paste using ball milling has never been studied, even less for the optimization of the paste for DSSCs used in artificial light conditions, which is much more demanding due to the low intensity of light and its capture by the dye.

Most of the studies using the ball-milling process in the manufacture of the outdoor DSSC paste have not optimized this process in terms of energy consumption and DSSC efficiency. Therefore, Seigo Ito et al. prepared porous electrodes with two different methods to disperse the nanocrystalline TiO₂ powder (ball-milling route and mortal-grinding route) using 0.5 mm ceramics balls with 8.27% PCE using N-719 dye [33]. Using the same dye, Niu Huang et al. prepared a viscous TiO₂ paste by ball milling, using commercially available P25 powder with ethylene glycol and citric acid for 48 h obtaining an efficiency of 7.36% [34]. A 7.98% efficiency was noted by Chang Kook Hong et al. while preparing green phosphors LaPO4:Ce, Tb, or (Mg, Zn)Al11O19:Eu into TiO₂ paste using the Planetary Mono Mill for 10 h at 300 rot/min [35]. Yasuhiro Yamamoto et al. prepared a TiO₂ paste using the bead mill method between 1 and 7 h, the optimum dispersion conditions providing an efficiency of 10.1% using the N719 dye. Moreover, increasing the dispersion time under ball milling has caused a decrease in particle size and in transparency of the coated film prepared using the screen printing method [36]. Hadja Fatima Mehnane et al. prepared a Sr doped TiO₂ paste using a ball mill for 12 h with 7.88% PCE using the N719 dye [37]. Recently, Huy Hao Nguyen et al. prepared a blue TiO₂ paste using the ball-milling method for 20 h for DSSC with 6.18% PCE using the N719 dye [38]. N.A. Abdullah et al. prepared TiO₂ nanoparticles as a paste for DSSC using the ball-milling technique for 8 h with 1.18% PCE immersed in the N719 dye. It is noted that the use of the ball-milling technique leads to an increase in the band gap from 3.30 eV (without milling) to 3.71 eV (8 h milling), and a decrease in the crystallite size [39]. A. Nurlaela et al. prepared TiO₂ nanoparticles as paste for DSSC from a microcrystalline powder using the high-energy ball-milling technique (0.5, 1, and 1.5 h of ball-milling time) and made a comparison with untreated TiO₂ pastes. They found that the particle size decreases after the milling process, and the efficiency of DSSC increases from 0.0988% (for unmilled TiO₂ paste) to 0.1715% (for TiO₂ paste milled for 0.5 h) [30].

Therefore, all studies presented above highlight only the effect of the ball-milling process on DSSC efficiency without optimizing this process in terms of energy consumption. The paste-preparation process is carried out over a very long period of time, which requires a high electricity consumption and higher maintenance of the ball mill.

To our knowledge, this is the first time that the TiO₂ paste ball-milling process was optimized in order to enhance the conversion efficiency of DSSC under indoor light conditions. Using the N719 photosensitizer, this study aimed to elucidate the ball size effect and ball-milling time for the preparation of the TiO₂ paste in terms of energy consumption and DSSC efficiency.

The obtained results revealed that the DSSC obtained from TiO₂ paste mortared for 3 h with 12 mm balls had an increase in efficiency of ~38% (21.19% PCE) compared to the paste obtained with 5 mm balls (15.36% PCE) under 1000 lux. At the same time, it was noted that, with the increase in the mortaring time from 3 h to 12 h using 12 mm balls, the increase in DSSC efficiency was only 7.7% (22.7% of PCE) under 1000 lux. This could be seen at J_sc, which increased by only 7 μA cm⁻², and FF increased by only ~3.7%, insignificant from the point of view of the production costs of the paste.

2. Materials and Methods

To set up the FTO glass substrate (7 Ω/sq from Sigma Aldrich, Merck KGaA, Darmstadt, Germany) for the DSSC preparation, it was washed for 30 min using water, ethanol, and acetone at each stage. This was succeeded by a 30-min UV–Ozone exposure (Ossila B.V., Leiden, The Netherlands). The preparation of the TiO₂ paste was conducted in accordance with our prior studies [40,41]. Firstly, 100 μL glacial acetic acid and 50 μL of H₂O were mixed, then 300 mg ethyl cellulose was dissolved into 15 mL ethanol, and 2 mL terpineol was added. The paste was ultrasonicated for 30 min and placed in a rotary evaporator to obtain the desired viscosity. The paste obtained was introduced into a high energy planetary ball mill (XQM-0.4A, Tencan, Changsha, China) using zirconium jars and balls with different sizes (5 mm, 8.5 mm, 10 mm, and 12 mm in diameter, and 0.445, 2.02, 3.24, and 5.38 g in weight, respectively) at 90% of the critical speed using different times (3–12 h).

For the photoanode fabrication, a single layer of paste was applied onto the previously cleaned FTO glass substrate, using the doctor blade technique. As a masking material, the same tape was used during the deposition process, ensuring uniform film thickness across all samples. The as-obtained films were calcinated at 500 °C for 60 min with 1 °C/min heating rate to avoid substrate cracking. Finally, the photoelectrodes were immersed in a 40 mM TiCl₄ solution for 30 min at 70 °C, and calcined at 450 °C with a heating rate of 1 °C/min for 60 min.

The abbreviation of the photoanodes prepared using different diameter balls and different times is as follows: a mm_b h, where a is the ball diameter, and b is the milling time. The results were compared with the reference photoanode-containing TiO₂-NP paste [41], untreated with a high energy planetary ball mill followed by a TiCl₄ treatment (referred to as TiO₂-NP).

The photoanodes were immersed in a 0.1 mM N719 dye mixture in 100% ethanol for five hours. The photoanodes and platinized counter electrode (CE) (H₂PtCl₆ on FTO substrate at 400 °C/30 min in air) were sealed at 134 °C with a Meltonix (1170-60) thick spacer. An electrolyte was injected in the area within the electrodes. It involved a solution of 0.03 M I₂, 0.10 M guanidinium thiocyanate, 0.5 M 4 tert-butylpyridine, 0.6 M 1-butyl-3-methyl-immidazolium iodide, and 3-methoxypropylonitrile (MPN).

The absorbance spectra of the photoanodes were obtained from the diffuse reflectance spectra (DSR) in the wavelength spectrum of 300–800 nm at ambient temperature using a UV/Vis/NIR (Lambda 950 UV-Vis-NIR from Perkin Elmer, Waltham, CT, USA) spectrophotometer with a 150 mm integrating sphere.

The I-V characteristic curves of DSSCs were obtained using a Keithley 2450 SourceMeter inside a homemade measurement case with black paint on every component and a LED lighting source (Lexman, E14, 1.8 W, 250 lm, Ronchin, France) lamp holder integrated at the top with 3.00678 W/m² at 1005.34 lux using a BTS2048-LV spectroradiometer (Gigahertz Optik GmbH, Munich, Germany); inside the box, the light intensities were calibrated using a Mastech MS6610 light meter [42] (Mastech, London, UK).

The electrochemical impedance spectroscopy (EIS) examination of the DSSCs obtained from paste mixing for 3 h using different balls in a Hi-Energy Planetary Ball Mill under indoor LED illumination (1000 lux) using a different bias potential of DSSCs was conducted with a Autolab potentiostat (unit PGZ 402, Metrohm, Herisau, Switzerland).

A portion of each of the pastes obtained was sintered separately at 500 °C to study the structural properties following the milling process using a high-energy planetary ball mill and different ball sizes. The products’ compositions were identified using powder X-ray diffraction (XRD, Malvern Panalytical, Malvern, UK), PW 3040/60 X’Pert PRO with Cu-Kα radiation (λ = 1.5418 Å), within the range of 2θ = 10–80°, at room temperature. Using the KBr pellet method, FTIR spectra were acquired in the 4000–400 cm⁻¹ domain using a JASCO-430 Fourier transform spectrometer (Jasco Co., Tokyo, Japan).

3. Results and Discussion

XRD diffraction patterns were used to study the effect of a high energy planetary ball mill using different ball sizes and ball-milling time (3 and 12 h) on the phase structure and the phase composition of TiO₂. For the paste prepared at 3 h, a decrease in the intensity of the anatase polymorph (1 0 1) crystal plane with increasing ball size could be noted due to the higher mechanical energy of the zirconia balls. This intensity decreased by approximately 11% for the paste made with 5 mm balls, and 23% for 8.5 mm balls, respectively, compared to TiO₂_NP. For the paste made with 10 mm and 12 mm balls, no effect on the intensity of the (1 0 1) crystal plane was observed.

According to Figure 1a, each of the diffraction peaks corresponded to JCPDS No. 00-004-0477 as TiO₂ crystallized in anatase form, with a tetragonal crystal structure and a I41/amd space group. In addition to this, no additional phases or impurities were detected.

For each sample, the dimensions of the crystallite were calculated using the Williamson–Hall (W–H) equation [43]:

$$\mathsf{\beta }\cos \theta =\frac{0.9\lambda }{D}+4\epsilon \sin \theta$$

(1)

In Equation (1), the notations have the following meaning: β—full width at half maximum (FWHM), θ—diffraction angle of the diffraction peaks, λ—wavelength of X-ray radiation, D—effective crystallite size, and $\epsilon$—lattice strain. The value of D, using linear fitting, was given from plotting βcos θ against 4sin θ. In

Table 1. Peak (1 0 1) intensity and crystallite size for Ti_NP and paste using 5, 8.5, 10 and 12 mm of balls.

Sample	Peak (1 0 1) Intensity		Crystallite Size
	3 h [%]	12 h [%]	3 h [nm]	12 h [nm]
Ti_NP	100	100	16.3	16.3
5 mm balls	89	60	13.4	13.6
8.5 mm balls	77	50	11.7	10.4
10 mm balls	45	50	11.3	10.3
12 mm balls	45	45	10.3	9.5

, the results of the W–H plots show that the crystallite size continuously decreased from 16.3 nm for Ti_NP to 10.3 nm for the paste made with 12 mm balls. The reduced crystallite size correlates strongly with the higher mechanical energy because of the zirconia balls inside the paste.

Figure 2 shows the SEM images of the TiO₂ photoanode using a high-energy planetary ball mill with 5 mm balls at a 3 h rotation. During the deposition process, the same tape was used for all photoanodes ensuring uniform film thickness across all samples. From the cross-section (Figure 2a), the thickness of the photoanode was 0.85 µm, and from the top section of the obtained photoanodes (Figure 2b), a uniform deposition without cracks could be noted.

For the 12 h paste (Figure 1b), it could be noted that the intensity of the (1 0 1) crystal plane suddenly dropped by 40% for the paste made with 5 mm balls and continued to decrease to 50% for the paste made with 8.5 mm and 10 mm balls. Even if the grinding time increased from 3 h to 12 h, it could be noted that the intensity of the peak was still 45% for the paste made with 12 mm balls compared to the intensity of the peak for Ti_NP. All of the diffraction peaks were indexed with JCPDS No. 00-004-0477 as TiO₂ crystallized in anatase form, and no other phases or impurities were observed. The (1 0 1) peak intensity and crystallite size of the Ti_NP paste using 5 mm balls, 8.5 mm balls, 10 mm balls, and 12 mm balls are shown in Table 1. A similar evolution of the crystalline size determined the value of 9.5 nm for the paste made with 12 mm balls.

Figure 3 shows the optical density spectra of the TiO₂ photoanodes using a high-energy planetary ball mill with different ball sizes and different rotation times (3 and 12 h) to prepare the paste. From the optical density spectra of the photoanodes loaded with the N719 dye (Figure 3), it can be noted that the absorption of the dye was improved with an increase in ball size, especially for the paste obtained with the 10 and 12 mm balls. This is due to the fact that decreasing the size of the crystallites confirmed by the XRD analysis allowed the absorption of a much larger amount of dye on the surface of the photoanode, and implicitly led to an increase in the efficiency of DSSC under indoor light conditions [44].

The photovoltaic characteristics of the DSSC, obtained from the TiO₂ paste using the high-energy planetary ball mill with different ball sizes (5, 8.5, 10, and 12 mm) and rotation times (3 and 12 h) under 1000 lux illumination, are shown in Figure 4. For the DSSC obtained with the paste prepared at 3 h of ball-milling time (Figure 4a), the J_SC increased from 112.81 μA cm⁻² for the paste made with 5 mm balls to 133.83 μA cm⁻² for the paste made with 8.5 mm balls, to 154.19 μA cm⁻² for the paste made with 10 mm balls, and to 154.39 μA cm⁻² for the paste made with 12 mm balls. The results are correlated with the XRD peak (1 0 1) intensity and absorption spectra. For the DSSC obtained with paste using 12 h of ball-milling time (Figure 4b), the J_SC increased from 118.88 μA cm⁻² made with 5 mm balls to 129.39 μA cm⁻² made with 8.5 mm balls, 160.33 μA cm⁻² made with 10 mm balls, and 160.91 μA cm⁻² made with 12 mm balls. For the DSSC using the TiO₂-NP paste untreated with the high energy planetary ball mill, a decrease to 81.91 μA cm⁻² in the J_SC value could be noted, due to the high crystallite size compared to the paste treated with the high-energy planetary ball mill, and implicitly an increase of particles size [45].

According to

Table 2. Photovoltaic performances of DSSC of untreated TiO₂-NP and using different ball sizes (5, 8.5, 10 and 12 mm) and different time (3 and 12 h) to prepare the paste and under 1000 lux illumination conditions ^a.

Nr. Crt.	Time	Materials	JSC [µA cm−2]	VOC [mV]	FF [%]	H [%]
1	-	TiO2_NP	81.91 ± 0.01	549.9 ± 0.1	66.3 ± 0.1	9.96 ± 0.02
2	3 h	5 mm balls	112.81 ± 0.01	605.6 ± 0.1	67.6 ± 0.1	15.38 ± 0.04
3		8.5 mm balls	133.83 ± 0.01	602.7 ± 0.2	67.8 ± 0.1	18.22 ± 0.03
4		10 mm balls	154.19 ± 0.02	613.6 ± 0.2	67.2 ± 0.1	21.20 ± 0.03
5		12 mm balls	154.39 ± 0.02	613.6 ± 0.2	67.0 ± 0.1	21.15 ± 0.02
6	12 h	5 mm balls	118.88 ± 0.01	630.5 ± 0.1	67.5 ± 0.1	16.85 ± 0.03
7		8.5 mm balls	129.39 ± 0.01	596.7 ± 0.1	66.9 ± 0.1	17.21 ± 0.03
8		10 mm balls	160.33 ± 0.02	596.7 ± 0.2	69.1 ± 0.1	22.09 ± 0.02
9		12 mm balls	160.91 ± 0.02	613.6 ± 0.2	69.7 ± 0.1	22.92 ± 0.02

^a The standard deviation data for each DSSC are obtained based on three cells.

, the energy conversion efficiency (η) was improved with the size of the ball used during the milling process of the TiO₂ paste of a photoanode that reliably improved the dye loading capacity. Therefore, for the paste using 3 h of ball-milling time, the energy conversion efficiency has improved from 15.38% with 5 mm balls to 18.22% with 8.5 mm balls, 21.20% with 10 mm balls, and 21.15% with 12 mm balls. The energy conversion efficiency for paste using 12 h of ball milling time was improved from 16.86% with 5 mm balls to 17.21% with 8.5 mm balls, 22.02% with 10 mm balls, and 22.93% with 12 mm balls. For the TiO₂-NP paste untreated with high-energy planetary ball mill, the energy conversion efficiency was 9.96%. In terms of production costs (energy consumption and maintenance of high-energy planetary ball mill), the 8.4% increase in energy conversion efficiency does not justify the 12 h used for the ball-milling process compared to 3 h for 12 mm balls.

For a deeper understanding of the effect of the ball-milling time and ball sizes on the energy conversion efficiency, the FTIR spectra of the molecular absorption and functional groups of the TiO₂ surface were analyzed in correlation with the UV-Vis absorption spectra. All FTIR spectra revealed similar absorption bands with only one exception (Figure 5). Thus, the typical absorption bands at 497–531 cm⁻¹ corresponded to the presence of the Ti–O metal oxide bonding [46]; 1631 cm⁻¹ was attributed to the bending vibration of –OH, and 3423 cm⁻¹ to the stretching mode of the –OH bond [47]. The absorption bands observed at 1380 and 2922 cm⁻¹ were also attributed to the C–H bond of the residual organic group of the TiO₂ paste after sintering at 500 °C. The intensity of the C–H peak decreased with an increase in ball size in a high-energy planetary ball mill [48,49].

A special absorption band was detected at 3748 cm⁻¹ that may be associated with the hydroxyl group (Ti–OH) of stretching mode of vibration for TiO₂ nanoparticles [50]. With ball sizes increasing from 5 to 12 mm, the TiO₂ surfaces were highly hydroxylated, improving the presence of the Ti–OH groups. The surface of the Ti–OH groups is amphoteric, and their ionization equilibrium can be written as (Equations (2) and (3)) [51]:

$$\mathsf{T}\mathsf{i}–\mathsf{O}{\mathsf{H}}_2^{+}\leftrightarrow \mathsf{T}\mathsf{i}–\mathsf{O}\mathsf{H}+{\mathsf{H}}^{+}$$

(2)

$$\mathsf{T}\mathsf{i}–\mathsf{O}\mathsf{H}\leftrightarrow \mathsf{T}\mathsf{i}–{\mathsf{O}}^{-}+{\mathsf{H}}^{+}$$

(3)

Hirose et al. have shown that OH generation enhances N719 chemisorption on the TiO₂ surface and thus leads to an increase in efficiency [52]. In this study, it can be noted that the large generation of OH due to the increase in ball size is in good correlation with the dye-loading capacity of the photoanode is highlighted by the UV-Vis absorption spectra.

In order to better understand the charge dynamics involved in our DSSCs and the impact of TiO₂ paste preparation, a study using electrochemical impedance spectroscopy (EIS) was carried out at V_OC and under 1000 lux light exposure. The equivalent circuit is shown in the insets of Figure 6, consisting in: Rs intrinsic resistance of the assembled cells, R_CE charge transfer resistance at the Pt counter electrode/electrolyte interface [53] and R_ct charge transfer resistance at the TiO₂ photoelectrode/dye/electrolyte interface [54].

Table 3. EIS parameters obtained from fitting the Nyquist plots of DSSCs for untreated TiO₂ paste (TiO₂_NP) and using different ball sizes (5, 8.5, 10, and 12 mm) at 3 h rotation time, under 1000 lux illumination intensity.

DSSC Type	Light Intensity [lux]	RS [Ω]	RCE [Ω]	Rct [Ω]
TiO2_NP	1000	11.6	5.6	413
5 mm balls	-	11.5	6.5	345
8.5 mm balls	-	11.3	4.3	327
10 mm balls	-	11.6	7.2	292
12 mm balls	-	11.5	5.2	268

summarizes the fitted Rs, R_CE, and R_ct values from the equivalent circuit using the Z-view 4 software. The values were in accordance with the photovoltaic parameters of the DSSCs (Table 2).

The intrinsic resistances of the assembled cells (R_S) for all DSSCs varied little between 11.3 and 11.6 Ω, having no negative effect on the photovoltaic parameters of the DSSCs. The value of R_ct decreased for all cells; for example, by ~35.1% for the DSSC based on the TiO₂ paste untreated at high energy planetary ball mill, and by ~22.3% for the DSSC based on the TiO₂ paste obtained using a ball size of 12 mm compared to the DSSC based on the TiO₂ paste obtained with a ball size of 5 mm, suggesting a decrease in the charge recombination process within the cell, and were reflected in the increase in J_SC from the J-V curves. For every DSSC, the R_CE values were comparable, indicating that the recombination of charges through the electrolyte and CE was consistent.

4. Conclusions

In conclusion, the ball-milling process of the TiO₂ paste was optimized for indoor light conditions for the first time, both in terms of efficiency and production costs, by developing a rapid preparation method that can be adopted industrially for the application of DSSCs. Due to the high impact force on the material and the kinetic energy which breaks down the bonds between molecules, a simple use of 12 mm diameter balls caused beneficial structural modifications, decreasing the size of the crystallites, and increasing OH generation on the TiO₂ surface for indoor light conditions.

Therefore, the obtained results revealed that the DSSC obtained from the TiO₂ paste mortared for 3 h with 12 mm balls highlighted an efficiency increase of ~37.5% (21.15% PCE) compared to the paste obtained with 5 mm balls (15.38% PCE) and 112.3% compared to the untreated paste (9.96% PCE), respectively, at 1000 lux. In terms of production costs, the 8.4% increase in energy conversion efficiency does not justify the 12 h used for the ball-milling process compared to 3 h for 12 mm balls.

Our paper highlights a simple and economically feasible solution that can improve the efficiency of the DSSC under indoor light conditions by ~112.3%, without requiring a complex TiO₂ photoanode architecture or the design of new dyes.