Formamidinium lead triiodide (FAPbI₃) perovskites have been studied for use as absorber layers in high-efficiency perovskite solar cells, owing to their high thermal stability and narrow bandgap [1–6]. However, FAPbI₃ has limited structural stability because of the phase transitions at 300 K [7,8]. In addition, it is advantageous for commercialization to fabricate solar cells under ambient conditions. However, because of the hydrophilic nature of the formamidinium cation (FA⁺), FAPbI₃ solar cells have traditionally been manufactured in highly moisture-restricted environments [9–11].

To fabricate stable FAPbI₃ with ambient conditions, the partial substitution of FA⁺ with cesium (Cs) has been identified as a useful strategy. This is because Cs-substituted perovskite (CsFAPbI₃) reduces the amount of moisture-sensitive FA⁺, and the hydrogen bonding strength is increased by rotating the [PbI₆]⁴⁻ octahedra, leading to an improved structural stability compared to that of FAPbI₃ [12,13]. However, when CsFAPbI₃ perovskite is synthesized in ambient conditions, perovskite with a substitution of Cs > 15 % leads to the formation of orthorhombic CsPbI₃ (δ-CsPbI₃), due to the presence of moisture [14,15]. To achieve high-efficiency solar cells, the formation of δ-CsPbI₃, which reduces the light absorption of CsFAPbI₃ perovskites, should be suppressed.

In a previous study, ethyl acetate (EA) was used as an anti-solvent to fabricate perovskites with high Cs content under ambient conditions, achieving the optimal reproducibility with Cs_0.22FA_0.78PbI₃ perovskite solar cells [16]. However, due to the use of dimethyl sulfoxide (DMSO) as a Lewis base additive, δ-CsPbI₃ still formed in the Cs_0.22FA_0.78PbI₃ perovskites. Therefore, in this study, N-methyl-2-pyrrolidone (NMP) is used instead of DMSO to suppress δ-CsPbI₃ formation. The perovskite synthesized with NMP forms a more stable intermediate phase than that synthesized with DMSO, thus inhibiting δ-CsPbI₃. However, owing to the incompatibility between NMP and EA, the carrier lifetime of NMP-based perovskites decreases. Despite varying concentrations of NMP, the characteristics and efficiency of the perovskite solar cells do not differ from those obtained using DMSO.

A Cs_0.22FA_0.78PbI₃ perovskite precursor solution was prepared by stirring 1 mmol each of Cs iodide, formamidinium iodide (FAI), and lead iodide (PbI2) in 1.0 mmol of DMSO and 0.6 mL of N,N-dimethylformamide (DMF) solvent. Additionally, to suppress δ-CsPbI₃ formation, another Cs_0.22FA_0.78PbI₃ precursor solution was prepared using 0.6, 1.0, and 1.4 mmol of NMP instead of DMSO.

The substrates were cleaned in an ultrasonic cleaner for 15 min using detergent, acetone, and isopropyl alcohol, followed by a 15 min ultraviolet (UV) ozone treatment. A 2.5 wt% SnO₂ solution was prepared by diluting a 15 wt% SnO₂ solution with deionized water, spincoated onto the cleaned indium tin oxide (ITO) substrates at 3000 rpm for 30 s, followed by annealing at 150 °C for 15 min. The perovskite precursor solution was spin-coated onto the substrates at 4000 rpm for 30 s. After 7 s of spin-coating, 500 μL of EA was dropped as the anti-solvent for the DMSO-based sample, and 100 μL of EA was dropped for the NMP-based sample. Following the spin-coating process, the perovskite film was fabricated through a heat treatment process at 150 °C for 2 min. For the spiro-OMeTAD layer, a solution was prepared by mixing 17.5 μL of lithium bis(trifluoromethanesulfonyl)-imide (LiTFSI) solution (520 mg of LiTFSI in 1 mL acetonitrile), 28.8 μL of 4-tert-butyl pyridine, and 72.3 mg of spiro-OMeTAD in 1 mL of chlorobenzene. This solution was then spin-coated at 4000 rpm for 30 s. Finally, Au layers were deposited at approximately 60 nm by thermal evaporation.

Surface images of the perovskite films were obtained using a scanning electron microscope (SEM) (S-4800, Hitachi). X-ray diffraction (XRD) was performed using an X-ray diffractometer (X’pert Pro MPD). The absorption spectrum was measured using an UV-visible (UV-vis) spectrophotometer (CN/mega-800). Photoluminescence (PL) measurements were performed using a 532-nm laser (CNI Optoelectronics Technology) and a silicon photodetector (Electro-Optical Systems Inc.). Time-resolved PL (TRPL) measurements were conducted with an FLS 920 spectrometer (Edinburgh Instruments), using a 656-nm picosecond-pulsed diode laser (pulse width: 70 ps) for excitation. Photovoltaic parameters were measured using a Keithley 2400 source meter and a solar simulator (Model LD50, McScience) under 1-sun simulated illumination (air mass:1.5 G).

Figures 1(a)–(d) illustrate SEM images of the perovskite films fabricated using a precursor containing DMSO (1.0 mmol) and NMP (0.6, 1.0, and 1.4 mmol). As illustrated in Fig. 1(a), rod-shaped grains were exhibited on the surface, indicating δ-CsPbI₃ [14,15]. δ-CsPbI₃ formed because DMSO interacted with moisture instead of forming an intermediate phase [17]. However, δ-CsPbI₃ was not observed in the perovskite films with NMP, as illustrated in Figs. 1(b)–(d), indicating that NMP suppressed the formation of δ-CsPbI₃. The formation energy of the intermediate phase for FAI⋅PbI2⋅NMP was lower than that of FAI⋅PbI2⋅DMSO, implying that the intermediate phase of FAPbI₃ with NMP was more stable [18]. This phenomenon suggests that the interaction between NMP and moisture was more effectively suppressed than with DMSO, leading to the suppression of δ-CsPbI₃. Therefore, the use of NMP facilitated the homogeneous growth of CsFAPbI₃ under ambient conditions. The grain size of the perovskite film with NMP was larger than that of the film with DMSO because of the homogeneous grain growth.

Figure 1. (a)–(d) Top-view and cross-section SEM images of the Cs_0.22FA_0.78PbI₃ perovskite films fabricated using DMSO and varying concentrations of NMP (scale bar: 500 nm). The thickness of the perovskite is indicated in the cross-section SEM images.

Figure 2(a) illustrates the XRD spectra of the Cs_0.22FA_0.78PbI₃ films coated with a perovskite precursor containing DMSO and NMP. The diffraction peaks at approximately 9.9 and 13.1° correspond to δ-CsPbI₃ and the peak observed at 12.7° is attributed to the presence of residual PbI2 in the perovskite film. δ-CsPbI₃peaks did not appear in the perovskite film with NMP, which was consistent with the SEM images illustrated in Fig. 1. The strong peak at 14° corresponds to the (100) plane of the Cs_0.22FA_0.78PbI₃ perovskite phase. Figure 2(b) illustrates the full width at half maximum (FWHM) and the diffraction angles of the (100) plane of the perovskite film. The FWHM of the (100) plane for the perovskite film with NMP was smaller than that of the film with DMSO. According to Scherrer’s equation, a narrow FWHM suggests an increase in grain size [19]. The perovskite film with NMP exhibited a larger grain size and narrow FWHM than that of the film with DMSO because the formation of δ-CsPbI₃ was suppressed. However, the FWHM did not change significantly with the NMP concentration. The diffraction angle of the perovskite film with NMP was higher than that of the film with DMSO. According to Bragg’s law, an increase in the diffraction angle indicates a decrease in the unit cell volume. As the Cs concentration increased in Cs_xFA1_−xPbI₃, the unit cell volume decreased owing to the rotation of [PbI₆]⁴⁻ octahedra [13]. For the perovskite film with NMP, the unit cell volume decreased because the formation of δ-CsPbI₃ was suppressed, indicating that Cs was successfully incorporated into the CsFAPbI₃ mixed phase. The decreased unit cell volume led to an increase in the bandgap energy because of the enhanced coupling of the atomic orbitals [20]. Therefore, the bandgap energy of the perovskite film with NMP was higher than that of the film with DMSO, as illustrated in Figs. 2(c) and 2(d).

Figure 2. (a) XRD spectra of the Cs_0.22FA_0.78PbI₃ perovskite films fabricated using DMSO and varying concentrations of NMP. (b) XRD peak position (black circles) and FWHM (red circles) of the (100) plane direction of the Cs_0.22FA_0.78PbI₃ perovskites. (c) UV–vis absorption spectra of the Cs_0.22FA_0.78PbI₃ perovskite films fabricated using DMSO and varying concentrations of NMP. (d) Bandgap energy derived from Tauc plots of the UV–vis absorption spectra.

Figure 3(a) illustrates the PL spectra of the Cs_0.22FA_0.78PbI₃/glass film coated with the precursor solution containing DMSO and NMP. The PL peak of the perovskite film was blue-shifted from 806 nm (with DMSO) to 785 nm (with NMP) because of the increased bandgap energy, which was consistent with the XRD and UV-vis absorption results illustrated in Fig. 2. The PL intensity increased in the perovskite film with NMP because the formation of δ-CsPbI₃ was suppressed. However, the PL intensity of the perovskite with 1.4 mmol of NMP is decreased compared to the perovskite with DMSO because the thickness of the perovskite coated with 1.4 mmol of NMP was decreased compared to that of the perovskite with DMSO, as shown in Figs. 1(a) and 1(d). Figure 3(b) presents the TRPL decay curves of the Cs_0.22FA_0.78PbI₃ films. TRPL decay parameters were obtained by fitting to the equation $I(t)=I_0+A_1\text{ exp}(-t/{\tau }_1)+A_2\text{ exp}(-t/{\tau }_2)$, and the average decay time (carrier lifetime) was calculated using ${\tau }_{\text{avg}}=\left(A_1{\tau }_1^2+A_2{\tau }_2^2\right)/\left(A_1{\tau }_1+A_2{\tau }_2\right)$. The decay parameters are summarized in Table I. Despite the increased PL intensity of the perovskite film with NMP, its carrier lifetime was shorter than that of the film with DMSO, which was attributed to an increase in non-radiative recombination due to the increased trap density in the perovskite film [21,22]. During the spin-coating process, the anti-solvent extracts only the host solvent (DMF) to allow the formation of an intermediate phase with the additive solvent (DMSO or NMP). According to the Hansen solubility distance (R_a) between EA and the solvents (DMF, NMP, and DMSO), while NMP (R_a = 8.3) has a stronger chemical reaction with EA than with DMF (R_a = 9.9), DMSO (R_a = 12.6) has a weaker chemical reaction with EA than with DMF [23]. Therefore, because EA extracts NMP more readily than DMF, a significant amount of the host solvent (DMF) remained in the perovskite films, limiting the function of the anti-solvent. Consequently, the trap density within the perovskite film with NMP increased more than that within the film with DMSO.

Figure 3. (a) PL spectra and (b) TRPL decay curves of the Cs_0.22FA_0.78PbI₃ perovskite films fabricated using DMSO and varying concentrations of NMP.

Table I. Fitting parameters of the TRPL decay curves fitted with a bi-exponential function for the Cs_0.22FA_0.78PbI₃ perovskite films..

	τ1(ns)	τ2(ns)	A1 (%)	A2 (%)	τavg(ns)
DMSO	36.95	202.54	4.76	95.24	201.0438
NMP 0.6	31.95	126.71	6.30	93.70	125.1313
NMP 1.0	29.89	146.66	4.89	95.11	145.4538
NMP 1.4	15.35	102.20	3.85	96.15	101.6782

The perovskite solar cells based on DMSO and NMP were fabricated with an ITO/SnO₂/Cs_0.22FA_0.78PbI₃/Spiro-OMeTAD/Au structure. Figures 4(a)–(d) illustrate the statistical parameters and their averages for each solar cell. The perovskite film with DMSO exhibited the highest efficiency, with an open-circuit voltage (V_OC) of 1.09 V, short-circuit current density (J_SC) of 21.6 mA/cm², fill factor (FF) of 76 %, and power conversion efficiency (PCE) of 17.85 %. Although the perovskite film with NMP had an increased PL intensity and higher bandgap, the PCE of the perovskite solar cells with NMP did not improve, which may be attributed to the increased non-radiative recombination due to the increased trap density. The perovskite solar cells with NMP had a higher J_SC than that of the solar cells with DMSO, because of the increased absorption and suppression of δ-CsPbI₃. Despite the suppression of δ-CsPbI₃, the FF of the perovskite solar cells with NMP remained unchanged as shown in Fig. 4(c), owing to the increased trap density. Consequently, the average PCE of the perovskite solar cells incorporating NMP was similar to that of the solar cells with DMSO. Although δ-CsPbI₃ formation was suppressed by NMP, the interaction between EA and NMP was less favorable than that with DMSO, leading to no significant change in performance for the perovskite solar cells.

Figure 4. Statistical (a) V_OC, (b) J_SC, (c) FF, and (d) PCE parameters of the Cs_0.22FA_0.78PbI₃ perovskite films fabricated using DMSO and varying concentrations of NMP.

In this study, the properties of Cs_0.22FA_0.78PbI₃ perovskite solar cells synthesized using DMSO and NMP under ambient conditions were analyzed. The use of NMP led to the formation of a more stable intermediate phase compared to DMSO, resulting in the suppression of δ-CsPbI₃ formation. The perovskite film with NMP exhibited an increased PL intensity and bandgap owing to the successful incorporation of Cs into the mixed phase. However, the carrier lifetime of the perovskite film with NMP decreased more than that of the film with DMSO, which was attributed to the increase in trap density owing to the difference in the interaction between NMP and EA. Therefore, despite the suppression of δ-CsPbI₃, there was no significant difference in the solar cell performance between the perovskite films with NMP and DMSO. Further research on the interaction between the anti-solvent and additive solvent is required to improve the performance of solar cells.

This work was supported by the Research Program through the National Research Foundation of Korea and funded by the Ministry of Education, Science, and Technology (NRF-2019R1A2C1086813). TRPL measurements were performed at the Central Laboratory of Kangwon National University.

The authors declare no conflicts of interest.