Formamidinium-based metal halide perovskite (FAPbI₃) is a highly desirable material for high-power conversion efficiency (PCE) perovskite solar cells owing to its exceptional optoelectronic properties and high thermal stability [1]. However, FAPbI₃ typically maintains a photo-inactive δ-phase at room temperature (RT) while retaining the photo-active α-phase only at temperatures exceeding 300 K [2]. The photo-inactive phase refers to a phase with a band gap greater than 2 eV; therefore, it cannot absorb light in the entire visible spectrum. The photo-active phase refers to a phase with a band gap that can absorb light in the visible spectrum. In solar cells, it is essential to absorb light in the visible spectrum, because a significant amount of light in the visible range reaches the Earth’s surface. Therefore, maintaining a photo-active phase that can absorb light over a broader range of the visible spectrum is crucial as opposed to a photo-inactive phase with a larger bandgap that cannot absorb light in the visible spectrum.

The phase instability of FAPbI₃ can be improved by the addition of appropriate amounts of other A-site cations. For example, in the case of perovskite (MAFAPbI₃) mixed with formamidinium [CH(NH₂)₂⁺, FA⁺] and methylammonium cations [CH₃NH₃⁺, MA⁺], FAPbI₃ maintained the α-phase at RT because of the increased hydrogen bonding between hydrogen and iodine, which can be attributed to the higher dipole moment of the MA⁺ cation than the FA⁺ cation [3]. However, MA cations are relatively volatile, leading to the formation of MA⁺ vacancies and thermal instability, which can negatively impact PCE [4, 5]. Substituting FA⁺ cations with inorganic cesium cations (Cs⁺) helps maintain the α-phase at RT and enhances the thermal stability compared to MAFAPbI₃ [6–8]. Owing to these advantages, CsFAPbI₃ has exhibited an improved PCE compared to that of FAPbI₃ [8–10]. For example, in a study by Chen et al. [8], the efficiency of Cs_0.09FA_0.91PbI₃ reached 20.7 %, which was higher than the 19.9 % efficiency of FAPbI₃. In most studies, the optimal Cs concentration for Cs_xFA_1−xPbI₃ was found to be between 10 and 17 % (x = 0.10 – 0.17), regardless of the synthesis space (glovebox or ambient conditions) [9,10]. This finding could be attributed to the fact that with the heavy addition of Cs, moisture leads to phase separation by forming a photo-inactive orthorhombic CsPbI₃ phase (δ-CsPbI₃) owing to the moisture-absorbing property of dimethyl sulfoxide (DMSO) [11]. Therefore, when synthesizing CsFAPbI₃ under ambient conditions that involve moisture as a variable, the optimal Cs concentration of CsFAPbI₃ could be different because of the formation of δ-CsPbI₃, which causes a reduction in PCE.

While synthesizing CsFAPbI₃ under ambient conditions, using ethyl acetate as an antisolvent could be an effective method to suppress the formation of δ-CsPbI₃ because ethyl acetate absorbs moisture during the perovskite layer deposition process, allowing the synthesis of perovskites even in high-humidity environments. In a study by Troughton et al. [12], ethyl acetate was used to synthesize pinhole-free MAPbI₃ perovskite in an environment with a relative humidity of 75 %. Devices using ethyl acetate showed improved PCE under humid conditions compared to those using other antisolvents, such as diethyl ether, toluene, and chlorobenzene which are commonly used in many studies [12–15]. In our previous study, methyl acetate led to pinhole and void formation owing to its high crystallization rate. In contrast, ethyl acetate enables the formation of pinhole-free MAPbI₃ perovskites, resulting in enhanced PCE [16]. For CsFAPbI₃, Cs concentrations below 33 % can maintain the CsFAPbI₃ phase because of the theoretically calculated negative Gibbs free energy, which indicates the formation of a stable mixture [2]. Therefore, when synthesizing CsFAPbI₃ using ethyl acetate under ambient conditions, the optimal Cs concentration of CsFAPbI₃ may be higher than that reported in other studies [9,10].

In this study, we synthesized Cs_xFA_1−xPbI₃ perovskites with varying Cs concentrations under ambient conditions to investigate the effects of Cs addition and optimize the Cs concentration. All Cs_xFA_1−xPbI₃ perovskite solar cell fabrication processes were carried out at approximately 40 % relative humidity and at approximately 27 °C (except for the HAT-CN and Ag thermal evaporation process). We deposited Cs_xFA_1−xPbI₃ perovskites on glass substrates to analyze their structural and optical properties, and fabricated n-i-p structured Cs_xFA_1−xPbI₃ perovskite solar cells to examine their characteristics.

The 1.49 M FAPbI₃ (Cs 0 %) precursor solution was prepared by mixing 172 mg (1 mmol) of formamide iodide (FAI) and 461 mg (1 mmol) of lead iodide (PbI₂) in 71 µL of DMSO and 0.6 mL of N, N- dimethylformamide. Cs_xFA_1−xPbI₃ (x = 0.19, 0.22, 0.25, and 0.28) precursor solutions were prepared by stirring an appropriate amount of cesium iodide instead of FAI during preparation of the FAPbI₃ solution. The substrates were cleaned in an ultrasonic cleaner for 15 min using detergent, acetone, and isopropyl alcohol. The cleaned substrates were treated with ultraviolet ozone for 15 min. A 15 wt% SnO₂ solution was diluted in deionized water to prepare a 7.5 wt% SnO₂ solution. The diluted solution was spin-coated on cleaned indium tin oxide (ITO) substrates at 3,000 rpm for 30 s, followed by annealing at 150 °C for 30 min. Each prepared precursor solution was spin-coated on substrates at 4,000 rpm for 30 s. Ethyl acetate (500 µL), an antisolvent, was added dropwise 7 s after the spin coating began. Thereafter, the perovskite films were fabricated by annealing at 150 °C for 2 min. Spiro-OMeTAD solution was created by dissolving 72.3 mg of spiro- OMeTAD, 28.8 µL of 4-tert-butyl pyridine, and 17.5 µL of lithium bis(trifluoromethanesulfonyl)imide solution (520 mg/mL of acetonitrile) in 1 mL of chlorobenzene. The spiro-OMeTAD solution was spin-coated onto the perovskite layer at 4,000 rpm for 30 s. Finally, HAT-CN (10 nm) and Ag (60 nm) layers were sequentially deposited by thermal evaporation.

X-ray diffraction (XRD) measurements were performed using an X’ Pert Pro MPD X-ray diffractometer (PANalytical). The layer morphology was determined using scanning electron microscopy (SEM) (S-4800, Hitachi). Photoluminescence (PL) measurements were conducted using a Si photodiode (Electro Optical Systems Inc.) with the excitation source emanating from a 532-nm diode-pumped solid-state laser (CNI Optoelectronics Technology). The PL signals were accumulated using an SR510 lock-in amplifier (Stanford Research Systems). Time-resolved PL (TRPL) spectra were obtained using an FLS 920 spectrometer (Edinburgh Instruments) with a 656 nm picosecond pulsed diode laser (pulse width: 70 ps) as the excitation source. The photocurrent density–voltage (J-V) curves were measured using a Keithley 2400 source meter. Photovoltaic parameters were recorded using a solar simulator (Model LD50, McScience) under 1-sun simulated illumination (air mass:1.5 G).

As shown in Figs. 1(a)-(e), top-view SEM images were obtained to investigate the morphology of the Cs_xFA_1−xPbI₃ perovskite films with varying Cs concentrations. An increase in Cs concentration led to the formation of photo-inactive orthorhombic CsPbI₃ (δ-CsPbI₃), which appeared as rod-like grains on the surface in top-view SEM images [10,17]. The formation of δ-CsPbI₃ is attributed to the moistureabsorbing property of DMSO. During the spin coating process under ambient conditions, few DMSO molecules interact with water instead of Pb, leading to the formation of δ-CsPbI₃ [11]. Applying perovskite films with such a rough morphology to solar cells could interfere with the contact between the perovskite film and hole-transport layer. In the cross-sectional SEM images in Figs. 1(f)-(j), there is no change in the film thickness. However, with increasing Cs concentration, numerous small grains were formed. Cs-induced perovskite seeds serve as nucleation sites for subsequent crystal growth, resulting in a more uniform grain formation [17,18]. Therefore, as the Cs concentration increased, more nucleation sites were formed, reducing the grain size.

Figure 1. (a)–(e) Top-view and (f)–(j) cross-section SEM images of the Cs_x FA_1−xPbI₃ perovskite films.

Figure 2(a) shows the XRD spectra of the Cs_xFA_1−xPbI₃ perovskite films with varying Cs concentrations. The diffraction peak near 9.9° corresponds to the δ-CsPbI₃ [19,20]. The peak around 11.7° represents the photo-inactive hexagonal FAPbI₃ (h-FAPbI₃), and the peak at approximately 14° corresponds to the photo-active α-CsFAPbI₃ in the (100) direction [20]. With the addition of Cs in the films, the diffraction peak of h-FAPbI₃ disappears, and only the α-CsFAPbI₃ diffraction peak near 14 ° is observed. The addition of Cs favors the thermodynamically stable α-phase perovskite [21]. Figure 2(b) shows the positions and intensities of the diffraction peaks at approximately 14°. As the Cs concentration increases by 19 %, the diffraction angle increases from 14.02 to 14.10°. According to Bragg’s law, an increase in the diffraction angle implies a decrease in unit cell volume. From this perspective, the increase in the diffraction angle of the (100) plane indicates that the Cs cation, with an ionic radius smaller than that of the FA cation, was successfully replaced. However, as the Cs concentration increased by more than 22 %, the diffraction angle decreased from 14.10 to 14.05 °, forming δ-CsPbI₃. Additionally, for Cs at a concentration of 28 %, the intensity of the (100) plane diffraction peak decreased sharply, indicating a reduction in crystallinity. This finding suggests that the addition of excess Cs caused phase separation into δ-CsPbI₃.

Figure 2. (a) XRD patterns of the Cs_x FA_1−xPbI₃ perovskite films. (b) XRD peak position (red line) and peak intensity (black line) near 14 degree.

The occurrence of phase separation with δ-CsPbI₃ was confirmed through PL and absorbance measurements. Figure 3(a) shows the normalized PL spectra of the Cs_xFA_1−xPbI₃ perovskite films. For perovskites, the PL spectra indicate a band-to-band recombination process because of their weak exciton binding energy [22,23]. The addition of Cs led to a blue shift of the PL peak from 813 nm (Cs 0 %) to 800 nm (Cs 19 %) as the bandgap increased. However, when Cs concentration exceeded 22 %, the PL peak was redshifted from 800 nm (Cs 19 %) to 804 nm (Cs ≥ 22 %), similar to the XRD diffraction angle tendency because of phase separation. Figure 3(b) shows the absorbance spectra in the visible region. The increase in the baseline of the absorption spectra beyond 850 nm was attributed to light scattering [24]. Therefore, as shown in Fig. 3(b), the baselines of Cs 0 % and Cs 19 % were adjusted to match that of Cs 22 % by lowering it by approximately 0.1. By substituting with Cs 19 %, h-FAPbI₃ with a bandgap of 2.43 eV is suppressed in the XRD spectra, resulting in enhanced absorption in the visible region. However, adding Cs over 22 % leads to phase separation, forming δ-CsPbI₃ with a bandgap of 2.8 eV, reducing visible region absorption compared to the Cs 19 % film. Figures 3(c) and 3(d) show the PL spectra and TRPL decay curves. The TRPL decay curves were fitted using the bi-exponential function $y=A_1 e^{-t/{\tau }_1}+ A_2 e^{-t/{\tau }_2}$. The average carrier lifetime (τ_avg) was calculated using the equation ${\tau }_{avg}=\left(A_1{\tau }_1^2+A_2{\tau }_2^2\right)/\left(A_1{\tau }_1+A_2{\tau }_2\right)$. The fitting parameters are listed in Table I. τ₁ represents the defect-related recombination process [25]. τ₂ represents the band-to-band recombination process and exhibits a similar tendency as the PL intensity. In Table I, τ₁ (~ 2.5 ns) and A1(~1%) were very low in all films, indicating the formation of low trap densities in all films [25]. An increase in the Cs concentration results in an increase in the PL intensity and charge-carrier lifetime, indicating that nonradiative recombination is suppressed and radiative recombination increases. The addition of Cs suppresses the vibration of FA cations because of the increased hydrogen bonding between the FA cations and iodine, consequently suppressing nonradiative recombination [6,26]. The carrier lifetime increases because of the reduction in defect density with the inclusion of Cs [27].

Table 1 . Fitting parameters of TRPL decay curves fitted with bi-exponential function for the Cs_xFA_1−xPbI₃ perovskite films..

Cs concentration (%)	τ1(ns)	τ2(ns)	A1 (%)	A2 (%)	τavg(ns)
0	1.60	31.35	2.48	97.52	31.31
19	2.58	56.81	1.69	98.31	56.76
22	1.83	68.80	0.83	99.17	68.79
25	1.89	72.93	0.82	99.18	72.91
28	1.81	70.71	0.97	99.03	70.69

Figure 3. (a) Normalized PL spectra, (b) ultraviolet–visible absorption spectra, (c) PL spectra, and (d) TRPL decay curves of the Cs_x FA_1−xPbI₃ perovskite films measured at RT.

Figures 4(a)-(d) show the photovoltaic parameters of the ITO/SnO₂ /Cs_xFA_1−xPbI₃/spiro-OMeTAD/HAT-CN/Ag device with respect to Cs concentration. More than ten devices were fabricated. The photovoltaic parameters of these devices are listed in Table II. With the addition of Cs, the short-circuit current (J_SC) increased with a longer carrier lifetime, despite the formation of δ-CsPbI₃. The open-circuit voltage (V_OC) increases with an increase in the bandgap of the PL spectra. In the Cs 19 % devices, the fill factor (FF) increased compared with that in the Cs 0 % devices. The FF showed a decreasing tendency with an increasing Cs concentration of more than 22 % because of the rough morphology, which increased the number of rodlike grains because of the formation of δ-CsPbI₃, as observed in the top-view SEM images as shown in Figs. 1(c)-(e). Moreover, the formation of small grains observed in the SEM images can impact J_SC and FF because it results in a lower carrier mobility compared with larger grains [28]. In the devices with excessive Cs addition (Cs 28 %), the increase in δ-CsPbI₃ observed in SEM and XRD, along with the decreased crystallinity of α-CsFAPbI₃, led to a decrease in the values of all parameters, despite the long carrier lifetime. As shown in Fig. 4(d), the Cs 19 % devices exhibit the highest average PCE of 12.82 %. However, the Cs 19 % devices exhibited the lowest reproducibility, with only 66.7 % of the devices working. Cs 22 % and Cs 25 % devices exhibited the highest PCE of 14.31 and 14.47 %, respectively, for all operating devices. This result could be attributed to the fact that as the Cs concentration increased from 0 to 25 %, the decreasing Gibbs free energy allowed for the formation of a more stable α-CsFAPbI₃. Furthermore, as the Cs concentration increases from 25 to 28 %, the Gibbs free energy increases, leading to a decrease in the stability of α-CsFAPbI₃ formation [2]. Therefore, the reproducibility and PCE of the device with Cs 28 % decreased compared to those with Cs 25 % in our results, with only 75 % of the devices working in the Cs 28 %. Additionally, ethyl acetate facilitated the formation of the more stable CsFAPbI₃ at high Cs concentrations. Because ethyl acetate absorbs moisture during spin coating, it inhibits the formation of δ-CsPbI₃. Therefore, the Cs 22 % devices exhibited a higher average PCE than the Cs 25 % devices, with less formed δ-CsPbI₃ on the surface as shown in the SEM images [Figs. 1(c) and 1(d)], which negatively affected the PCE. Therefore, we confirmed that the optimal Cs concentration for synthesis under ambient conditions was 22 %.

Table 2 . Average values and standard deviation of photovoltaic parameters for the Cs_xFA_1−xPbI₃ perovskite solar cells..

Cs (%)	VOC (mV)	JSC (mA/cm2)	FF (%)	PCE (%)
0	1,000.69 ± 10.64	20.20 ± 0.69	54.32 ± 6.27	11.02 ± 1.52
19	1,045.31 ± 5.24	21.55 ± 0.41	56.73 ± 9.88	12.82 ± 2.37
22	1,038.13 ± 9.75	21.67 ± 0.18	54.46 ± 5.09	12.26 ± 1.25
25	1,047.76 ± 15.47	21.31 ± 0.19	53.35 ± 5.72	11.92 ± 1.37
28	1,015.67 ± 23.38	18.18 ± 2.32	49.84 ± 6.39	9.36 ± 2.33

Figure 4. Statistical (a) J_SC, (b) V_OC, (c) FF, and (d) PCE parameters of the Cs_x FA_1−xPbI₃ perovskite solar cells.

Cs_xFA_1−xPbI₃ perovskite solar cells were synthesized and analyzed under ambient conditions. By adding Cs to FAPbI₃, the formation of δ-FAPbI₃ was suppressed and only structurally stable α-CsFAPbI₃ was formed, resulting in a high-quality perovskite film. However, excessive Cs addition led to phase separation of δ-CsPbI₃, resulting in a reduced PCE. As the Cs concentration increased, nonradiative recombination was suppressed, leading to an increased carrier lifetime. The solar cells with Cs 22 % exhibited the best reproducibility and high PCE. The optimal Cs concentration for Cs_xFA_1−xPbI₃perovskite synthesized under ambient conditions was 22 %. These findings provide valuable insights into the future development of high-efficiency solar devices synthesized under ambient conditions.

This research was supported by the Basic Science Research Program through the National Research Foundation of Korea, funded by the Ministry of Education, Science, and Technology (NRF-2019R1A2C-1086813). This study has been worked with the support of a research grant of Kangwon National University in 2023. Time-resolved photoluminescence measurements were performed at the central lab of Kangwon National University.

The authors declare no conflicts of interest.