Organometal halide perovskite-based solar cells (PSCs) have been developed because of their superior optical characteristics, such as a direct band gap, significant light absorption coefficient, and long carrier diffusion length of free carriers [1–5]. Despite these remarkable advances, most high-quality and efficient PSCs have been fabricated in glove boxes with extremely limited moisture and oxygen contents [6–8]. Stability under ambient conditions in fabrication and operating environments remains an open challenge for commercialization. The properties of PSCs are strongly influenced by the device fabrication details, such as precursor preparation, annealing approach, perovskite precursor preheating, and substrate preheating [9–13]. Therefore, strategies for durable synthesis in ambient atmospheres have been suggested; however, most studies have been conducted in environments with controlled relative humidity (RH). For example, Wang et al. [7] controlled the RH by flowing air through a water-filled flask. They achieved a power conversion efficiency (PCE) of 19.5 % by preheating the substrate at a controlled 90 % RH [7]. Because this controlled environment can be another constraint in actual commercialization, confirming the durability in a general environment is necessary. In addition, the antisolvent selection and application method during the perovskite fabrication process determine the nucleation and crystallization of the perovskite film and significantly affect the performance and environmental durability of PSCs [14–17].

In this study, we fabricated a perovskite film based on an antisolvent application time point (ATP) on a preheated substrate at room temperature (RT), where the temperature and humidity were uncontrolled. Preheating the substrate promoted the nucleation and crystallization of the perovskite film, and applying early ATP increased the film thickness without affecting the perovskite grain size. The optical properties according to the increase in perovskite film thickness were confirmed through ultraviolet (UV)−visible absorbance, photoluminescence (PL), and time-resolved PL (TRPL) measurements, and the device characteristics were also evaluated. In addition, devices fabricated using substrates preheated in low- and high-humidity environments were compared.

For fabrication of PSCs, tin-doped indium oxide (ITO) substrates (8 Ω/sq, AMG) were ultrasonicated with 2 % hellmanex detergent (15 min), isopropyl alcohol (15 min), and acetone (15 min), respectively. The cleaned substrates were then treated with UV-ozone for 15 min. The diluted 7.5 % SnO₂ solution (15 %, Alfa-Aesar) was spin-coated at 3,000 rpm for 30 s and then annealed at 150 °C for 30 min. Before depositing the perovskite solution, the ITO/SnO₂ substrate was treated with UV-ozone for 15 min. Subsequently, a preheated substrate was prepared on each hotplate at 50, 70, and 90 °C to ensure sufficient temperature maintenance for 5 min. The CH₃NH₃PbI₃ (MAPbI₃) precursor solution was prepared using 159 mg of MAI (GreatCell Solar) and 461 mg of PbI₂ (Alfa Aesar) dissolved in 600 µL of DMF (Sigma- Aldrich) and 71 µL of DMSO (Sigma-Aldrich). The resultant solution was spin-coated onto the SnO₂ layer at 4,000 rpm for 30 s. To minimize the effect of the temperature drop of the preheated substrate, the sample was prepared with a constant time of less than 15 s from mounting the substrate on the spin coater to dropping the perovskite solution. To investigate the dependence of ATP on the preheated temperature, 500 µL of ethyl acetate (EA, Sigma-Aldrich) was dispensed as antisolvent using an automatic dispensing pipette (E3, Eppendorf) at various time points (5, 7, and 9 s) before the turbid point (TP) occurred. Subsequently, it was annealed on a hot plate at 150 °C for 2 min. A spiro-OMeTAD solution was prepared by dissolving 72.3 mg of spiro-OMeTAD (ELECTRONIC MATERIAL & INK) in 1 mL of chlorobenzene, then 17.5 µL of Li-TFSI (520 mg in 1 ml of acetonitrile, Sigma-Aldrich) and 28.8 µL of 4-tert-butylpyridine (Sigma-Aldrich) were added. The resulting solution was drop-cast onto the perovskite layer for 5 s after spinning at 4,000 rpm for 30 s to achieve dynamic coating. Finally, 10 nm of HAT-CN (ELECTRONIC MATERIAL & INK) and 70 nm of Ag electrode were thermally evaporated. The active area of the PSCs was 0.04 cm². Except for the solar cell devices used to measure the photovoltaic parameters, all the other samples for film characterization were coated with only the perovskite layer on a glass substrate. All the spin-coating processes were conducted under ambient conditions, where the temperature and humidity were not controlled.

The scanning electron microscope (SEM) images of the perovskite films were obtained using an S-4800 microscope (Hitachi). UV-Vis absorption spectra were collected using a Ubi-490 (MicroDigital) spectrophotometer. The PL signal was recorded using a Si photodetector (S/IGA-025, Electro-Optical Systems) with a lock-in amplifier (SR510, Stanford Research System) and a 532-nm diode-pumped solid-state laser (Changchun New Industries Optoelectronics Tech.) was used as the excitation source. TRPL measurements were performed using an FLS 920 fluorescence spectrometer (Edinburgh Instruments) and a 656-nm picosecond pulsed diode laser (EPL-655, Edinburgh Instruments). J−V curves were collected using a Keithley 2400 under AM 1.5 G 1 sun illumination from a solar simulator (SimuLight SS-LD50S, McScience). The AM 1.5 G 1 sun illumination was calibrated using a standard Si solar cell (LSRS-01, LiveStrong Optoelectronics).

During the perovskite film deposition process, if spin coating is continued without applying an antisolvent, the surface of the spun transparent perovskite wet film becomes hazy. This is called TP, and the surface of the perovskite film after TP is unintentionally rough and does not guarantee full coverage, which affects the formation of defects in the film [18]. Table I shows the TP as a function of the preheating temperature of the substrate under different ambient conditions (temperature and RH). TP was measured in a pure ambient environment, where the perovskite film deposition environment was not arbitrarily controlled. Based on the data in the first row of the table, changes in both RT and RH affect the TP of the unpreheated substrate (as-grown). Specifically, when RT increases, TP appears to occur more quickly, whereas an increase in RH seems to slow TP. At elevated temperatures, the prepared perovskite precursor solution and substrate maintain thermal equilibrium with the ambient atmosphere, which increases the volatilization rate of the solvent in the perovskite wet film. Rapid solvent volatilization increases the nucleation and crystallization rates and decreases the TP. At high humidity, a relatively heavy and humid atmosphere inhibits volatilization of the solvent in the perovskite wet film and delays supersaturation [7]. Consequently, the concentration of precursors required for nucleation is insufficient, and the TP slows. As the preheating temperature of the substrate increased, TP decreased as shown in Table I. This is due to an increase in the volatilization rate of the solvent, as described above. Note that preheating the substrate reduced TP’s deviation due to changes in ambient conditions. This suggests that the high-temperature substrate further accelerated the solvent volatilization of the perovskite wet film and ensured the durability of nucleation and crystallization against humidity. In addition, because the application of an antisolvent plays a critical role in the nucleation and crystallization of the perovskite, the correlation between the preheating of the substrate and ATP of an antisolvent was confirmed. As shown in Table I, the TP of the substrate preheated to 50 °C does not show a significant difference compared to unpreheated substrate. Conversely, for the substrates preheated to 90 °C, the range of ATP for tunability was narrow owing to the rapid attainment of TP. Therefore, we will focus mainly on the ATP at the substrate preheated to 70 °C.

Table 1 . TP in terms of the preheating temperature of the substrate at different ambient temperatures and RH.

Preheating temperature	TP (s) at ambient temperature / RH (°C / %)
	17.2 / 25	19.6 / 26	19.7 / 31	19.6 / 37	20.7 / 66
as-grown	17 s	13 s	14 s	15 s	18 s
50 °C	13 s	12 s	12 s	13 s	14 s
70 °C	10 s	9 s	10 s	10 s	11 s
90 °C	6 s	5 s	5 s	6 s	6 s

In Fig. 1, both the top-view and cross-sectional SEM images of the perovskite film were displayed. The antisolvent was applied at different time points and, for convenience, the samples are named in parentheses; 7 s before TP for the unpreheated substrate (as-grown), 5 s before TP (S5), 7 s before TP (S7), and 9 s before TP (S9) for the substrate preheated at 70 °C. All samples showed similar surface morphologies and grain sizes, regardless of when the antisolvent was applied or whether the substrate was preheated. However, the perovskite films deposited on the preheated substrates were thicker than those fabricated at unpreheated substrates. The thickness of the perovskite film increased significantly from 505 ± 13 (S5) to 679 ± 11 nm (S9) as the ATP moved away from the TP. This implied that the perovskite film became thicker when the antisolvent was applied earlier. The early application of the antisolvent provided sufficient time for rapid supersaturation and nucleation of the perovskite wet film. The abundant nucleation sites provide a driving force for crystal growth during the subsequent annealing process. As shown in the photograph of the perovskite film immediately after antisolvent application without annealing in Fig. 1 inset, the samples with the early application show a darker appearance (S9). The crystal growth of the thick perovskite film of early ATP suggests that growth in the thickness direction is preferred over the substrate direction.

Figure 1. Top-view (a–d) and cross-sectional (e–h) SEM images of perovskite films deposited at unpreheated substrate (as-grown) and at different antisolvent ATPs on preheated substrates at 70 °C. Insets are a photographic image of each sample before (a-d) and after annealing (e-h). Scale bars correspond to 500 nm (a–d) and 1 µm (e–h).

The influence of the optical properties of the perovskite film on the ATP of the preheated substrates was investigated. As the ATP increased, the thickness of the perovskite film increased, as shown in the SEM images [Figs. 1(b)–(d)]. As demonstrated in Fig. 2(a), the light absorption of the perovskite films increased, and the absorbance of the sample in which the antisolvent was applied at the fastest ATP was high in the entire absorption region. The PL results measured at RT also showed trends similar to those of the UV-Vis absorbance. The PL peak appeared at 772 nm for all samples, and the PL intensity increased in the order of S5, S7, and S9 as the perovskite film thickness increased shown in Fig. 2(b). This result is attributed to the increased recombination of internal carriers owing to increased light absorption. To analyze the carrier dynamics according to the change in the perovskite film thickness, TRPL measurements were performed, and the PL decay curve and corresponding fitting parameters are shown in Fig. 2(c). The PL decay traces can be fitted by a biexponential decay function $I={\displaystyle {\sum }_{i=1}^2{A}_i\exp }(-\frac{t}{{\tau }_i})$, where τ_i is the time constant with the corresponding amplitute A_i. The fast-decay components τ₁ have been ascribed to defect-related non-radiative recombination and slow-decay component τ₂ related radiative recombination, respectively. The average decay time was calculated using fitting parameters [19]. As the perovskite film thickness increases, the fast decay time τ₁ (19.2 to 13.1 ns) and the amplitude A₁ (2.33 to 0.54 %) associated with defects decrease, while the slow decay components τ₂ (147.1 to 248.1 ns) and A₂ (97.67 to 99.46 %) increase. Consequently, the average decay time increased from 146.7 to 248.0 ns. This suggests that rapid nucleation and crystallization have a beneficial effect on the defect reduction in perovskite films. In addition, the unpreheated sample (as-grown) exhibited lower absorbance, weaker PL intensity, and a faster decay time than the preheated samples, demonstrating that the deposition of perovskite films on preheated substrates improves the optical properties.

Figure 2. Optical characterization of annealed perovskite films prepared on substrates preheated at 70 °C with different antisolvent ATPs. (a) UV-Vis absorption spectra, (b) PL spectra, and (c) TRPL decay curves and fitting parameters (inset). For comparison, the absorption, PL, and TRPL spectra of perovskite films grown on unpreheated substrates (as-grown) are also displayed.

The photovoltaic parameters for ATP on the preheated substrates are shown in Fig. 3. The device structure was ITO/SnO₂/MAPbI₃/spiro-OMeTAD/HAT-CN/Ag. For comparison with our previous fabrication conditions, the photovoltaic parameters of the device with the antisolvent applied 7 s before TP at unpreheated substrate (as-grown) were added [20,21]. All the photovoltaic parameters improved when the antisolvent was applied earlier. In particular, the average short-circuit current density (J_SC) showed an improvement in PCE owing to the increase in the thickness of the preheated substrate compared to the sample without preheating (as-grown). However, the J_SC increase of 22.47 mA/cm² for S9 compared to 22.24 mA/cm² of S5 is far from the significant improvement in the properties expected from the SEM, UV-Vis, PL, and TRPL results. This result can be attributed to enhanced light absorption in the thicker perovskite layers. However, a thicker absorber layer may result in a loss of charge extraction efficiency and limit device efficiency due to increased bimolecular recombination [22].

Figure 3. Photovoltaic performance of the fabricated PSCs at different substrate temperatures and antisolvent ATPs. (a) V_OC, (b) J_SC, (c) FF, and (d) PCE.

Figure 4 shows the highest J−V curves of the PSCs with an antisolvent applied 7 s before the TP at unpreheated and 70 °C preheated substrates to prove the durability of the fabrication environment of preheating the substrate in a high-humidity environment (66 % RH). The average photovoltaic parameters and standard deviations of each sample are shown in the insets. As shown in Fig. 3, preheating the substrate did not have a significant effect on the improvement of the PSC characteristics, except for a slight increase in J_SC. In addition, no improvement was observed, even in a high-humidity environment. Based on these results, two factors were inferred. First, our manufacturing recipe has already been optimized for a high-humidity environment. Initially, we selected EA, an antisolvent with high durability, reproducibility, and efficiency in a high-humidity environment, and confirmed the optimal application time for its optimization [20,21]. The choice of a suitable antisolvent and application method can minimize the effects of moisture on the manufacturing environment. Second, the temperature of the preheated substrate should be maintained. We mounted the preheated substrate onto a spin coater and dropped the perovskite precursor solution within 15 s. The preheated substrate was exposed to a relatively low ambient temperature during perovskite film deposition as it was mounted onto the spin coater. This environment led to a change in the temperature of the preheated substrate. In our case, after 15 s, the 50 °C substrate was 41 °C, the 70 °C substrate was 58 °C, and the 90 °C substrate was 74 °C. The higher the preheating temperature, the larger the temperature drop. Specifically, it is difficult to maintain the prepared substrate temperature in advance. As shown in Fig. 3, the standard deviation of the preheated substrate sample is larger than that of the as-grown sample. This indicates that the reproducibility of the preheated samples is degraded. Heating the precursor solution or the spin coater can partially solve this problem [13,23]; however, because the change in the fabrication temperature affects the crystallization rate of the perovskite, detailed optimization will be required.

Figure 4. J−V curves of the highest efficiency PCSs fabricated on (a) unpreheated substrate and (b) preheated substrate at 70 °C under low and high humidity conditions. Photovoltaic parameters corresponding to each condition (inset).

The optical properties and device characteristics of the perovskite film were confirmed in terms of the ATP on the preheated substrate. Preheating the substrate accelerated perovskite nucleation and crystallization, resulting in a faster TP and less influence from RH. In addition, the thickness of the perovskite film could be controlled by changing the ATP of antisolvent. The thick perovskite film improved the film quality by increasing the absorbance, PL intensity, and carrier decay time owing to the improved light absorption ability and reduced defects. However, in actual device applications, a thicker film showed a slight increase in J_SC and PCE, but the effect was found to be insignificant owing to internal recombination. Excellent durability was confirmed in a high-humidity environment regardless of substrate preheating; however, it is expected that maintaining the substrate temperature will contribute to improving the durability and efficiency of the perovskite device.

This research was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF), funded by the Ministry of Education, Science, and Technology (NRF-2019R1A2-C1086813). TRPL measurements were performed at the Central Lab of Kangwon National University.

The authors declare no competing financial interest.