Suppressing Defect-Induced α-δ Phase Transition for Efficient and Stable Formamidine Perovskite Solar Cells


        Thank you for visiting Nature.com. The version of browser you are using has limited CSS support. For best results, we recommend using a newer version of your browser (or turning off compatibility mode in Internet Explorer). In the meantime, to ensure ongoing support, we are showing the site without styling or JavaScript.
        Defect passivation has been widely used to improve the performance of lead triiodide perovskite solar cells, but the effect of various defects on α-phase stability remains unclear; Here, using density functional theory, we identify for the first time the degradation pathway of formamidine lead triiodide perovskite from α-phase to δ-phase and study the effect of various defects on the phase transition energy barrier. Simulation results predict that iodine vacancies are most likely to cause degradation because they significantly lower the energy barrier for the α-δ phase transition and have the lowest formation energy at the perovskite surface. The introduction of a dense layer of water-insoluble lead oxalate onto the perovskite surface significantly inhibits the decomposition of the α-phase, preventing the migration and volatilization of iodine. Additionally, this strategy significantly reduces interfacial nonradiative recombination and increases solar cell efficiency to 25.39% (certified 24.92%). The unpackaged device can still maintain its original 92% efficiency after operating at maximum power for 550 hours under simulated 1.5 G air mass irradiation.
        The power conversion efficiency (PCE) of perovskite solar cells (PSCs) has reached a certified record high of 26%1. Since 2015, modern PSCs have preferred formamidine triiodide perovskite (FAPbI3) as a light-absorbing layer due to its excellent thermal stability and preferential bandgap close to the Shockley-Keisser limit of 2,3,4. Unfortunately, FAPbI3 films thermodynamically undergo a phase transition from a black α phase to a yellow non-perovskite δ phase at room temperature5,6. To prevent the formation of the delta phase, various complex perovskite compositions have been developed. The most common strategy to overcome this problem is to mix FAPbI3 with a combination of methyl ammonium (MA+), cesium (Cs+) and bromide (Br-) ions7,8,9. However, hybrid perovskites suffer from bandgap broadening and photoinduced phase separation, which compromise the performance and operational stability of the resulting PSCs10,11,12.
        Recent studies have shown that pure single crystal FAPbI3 without any doping has excellent stability due to its excellent crystallinity and low defects13,14. Therefore, reducing defects by increasing the crystallinity of bulk FAPbI3 is an important strategy to achieve efficient and stable PSCs2,15. However, during operation of the FAPbI3 PSC, degradation to the undesirable yellow hexagonal non-perovskite δ phase can still occur16. The process usually begins at surfaces and grain boundaries that are more susceptible to water, heat and light due to the presence of numerous defective areas17. Therefore, surface/grain passivation is necessary to stabilize the black phase of FAPbI318. Many defect passivation strategies, including the introduction of low-dimensional perovskites, acid-base Lewis molecules, and ammonium halide salts, have made great progress in formamidine PSCs19,20,21,22. To date, almost all studies have focused on the role of various defects in determining optoelectronic properties such as carrier recombination, diffusion length and band structure in solar cells22,23,24. For example, density functional theory (DFT) is used to theoretically predict the formation energies and trapping energy levels of various defects, which is widely used to guide practical passivation design20,25,26. As the number of defects decreases, the stability of the device usually improves. However, in formamidine PSCs, the mechanisms of the influence of various defects on phase stability and photoelectric properties should be completely different. To the best of our knowledge, the fundamental understanding of how defects induce the cubic to hexagonal (α-δ) phase transition and the role of surface passivation on the phase stability of α-FAPbI3 perovskite is still poorly understood.
        Here, we reveal the degradation pathway of FAPbI3 perovskite from black α-phase to yellow δ-phase and the influence of various defects on the energy barrier of α-to-δ-phase transition via DFT. I vacancies, which are easily generated during film fabrication and device operation, are predicted to be most likely to initiate the α-δ phase transition. Therefore, we introduced a water-insoluble and chemically stable dense layer of lead oxalate (PbC2O4) on top of FAPbI3 through an in situ reaction. Lead oxalate surface (LOS) inhibits the formation of I vacancies and prevents the migration of I ions when stimulated by heat, light, and electric fields. The resulting LOS significantly reduces interfacial nonradiative recombination and improves FAPbI3 PSC efficiency to 25.39% (certified to 24.92%). The unpackaged LOS device retained 92% of its original efficiency after operating at the maximum power point (MPP) for over 550 hours at a simulated air mass (AM) of 1.5 G of radiation.
        We first performed ab initio calculations to find the decomposition path of the FAPbI3 perovskite to transition from the α phase to the δ phase. Through a detailed phase transformation process, it is found that the transformation from a three-dimensional corner-sharing [PbI6] octahedron in the cubic α-phase of FAPbI3 to a one-dimensional edge-sharing [PbI6] octahedron in the hexagonal δ-phase of FAPbI3 is achieved. breaking 9. Pb-I forms a bond in the first step (Int-1), and its energy barrier reaches 0.62 eV/cell, as shown in Figure 1a. When the octahedron is shifted in the [0\(\bar{1}\)1] direction, the hexagonal short chain expands from 1×1 to 1×3, 1×4 and finally enters the δ phase. The orientation ratio of the entire path is (011)α//(001)δ + [100]α//[100]δ. From the energy distribution diagram, it can be found that after the nucleation of the δ phase of FAPbI3 in the following stages, the energy barrier is lower than that of the α phase transition, which means that the phase transition will be accelerated. Clearly, the first step of controlling the phase transition is critical if we want to suppress α-phase degradation.
        a Phase transformation process from left to right – black FAPbI3 phase (α-phase), first Pb-I bond cleavage (Int-1) and further Pb-I bond cleavage (Int-2, Int -3 and Int -4) and yellow phase FAPbI3 (delta phase). b Energy barriers to the α to δ phase transition of FAPbI3 based on various intrinsic point defects. The dotted line shows the energy barrier of an ideal crystal (0.62 eV). c Energy of formation of primary point defects on the surface of lead perovskite. The abscissa axis is the energy barrier of the α-δ phase transition, and the ordinate axis is the energy of defect formation. The parts shaded in grey, yellow and green are type I (low EB-high FE), type II (high FE) and type III (low EB-low FE), respectively. d Energy of formation of defects VI and LOS of FAPbI3 in the control. e I barrier to ion migration in control and LOS of FAPbI3. f – schematic representation of the migration of I ions (orange spheres) and gLOS FAPbI3 (gray, lead; violet (orange), iodine (mobile iodine)) in the gf control (left: top view; right: cross section, brown); carbon; light blue – nitrogen; red – oxygen; light pink – hydrogen). Source data is provided in the form of source data files.
        We then systematically studied the influence of various intrinsic point defects (including PbFA, IFA, PbI, and IPb antisite occupancy; Pbi and Ii interstitial atoms; and VI, VFA, and VPb vacancies), which are considered to be key factors. that cause atomic and energy level phase degradation are shown in Figure 1b and Supplementary Table 1. Interestingly, not all defects reduce the energy barrier of the α-δ phase transition (Figure 1b). We believe that defects that have both low formation energies and lower α-δ phase transition energy barriers are considered detrimental to phase stability. As previously reported, lead-rich surfaces are generally considered effective for formamidine PSC27. Therefore, we focus on the PbI2-terminated (100) surface under lead-rich conditions. The defect formation energy of surface intrinsic point defects is shown in Figure 1c and Supplementary Table 1. Based on the energy barrier (EB) and phase transition formation energy (FE), these defects are classified into three types. Type I (low EB-high FE): Although IPb, VFA and VPb significantly reduce the energy barrier to phase transition, they have high formation energies. Therefore, we believe that these types of defects have a limited impact on phase transitions since they are rarely formed. Type II (high EB): Due to the improved α-δ phase transition energy barrier, the anti-site defects PbI, IFA and PbFA do not damage the phase stability of α-FAPbI3 perovskite. Type III (low EB-low FE): VI, Ii and Pbi defects with relatively low formation energies can cause black phase degradation. Especially given the lowest FE and EB VI, we believe that the most effective strategy is to reduce I vacancies.
        To reduce VI, we developed a dense layer of PbC2O4 to improve the surface of FAPbI3. Compared to organic halide salt passivators such as phenylethylammonium iodide (PEAI) and n-octylammonium iodide (OAI), PbC2O4, which contains no mobile halogen ions, is chemically stable, insoluble in water, and easily deactivated upon stimulation. Good stabilization of surface moisture and electric field of perovskite. The solubility of PbC2O4 in water is only 0.00065 g/L, which is even lower than that of PbSO428. More importantly, dense and uniform layers of LOS can be softly prepared on perovskite films using in situ reactions (see below). We performed DFT simulations of the interfacial bonding between FAPbI3 and PbC2O4 as shown in Supplementary Figure 1. Supplementary Table 2 presents the defect formation energy after LOS injection. We found that LOS not only increases the formation energy of VI defects by 0.69–1.53 eV (Figure 1d), but also increases the activation energy of I at the migration surface and exit surface (Figure 1e). In the first stage, I ions migrate along the perovskite surface, leaving VI ions in a lattice position with an energy barrier of 0.61 eV. After the introduction of LOS, due to the effect of steric hindrance, the activation energy for the migration of I ions increases to. 1.28 eV. During the migration of I ions leaving the perovskite surface, the energy barrier in the VOC is also higher than in the control sample (Fig. 1e). Schematic diagrams of I ion migration pathways in control and LOS FAPbI3 are shown in Figure 1 f and g, respectively. The simulation results show that LOS can inhibit the formation of VI defects and the volatilization of I, thereby preventing the nucleation of the α to δ phase transition.
        The reaction between oxalic acid and FAPbI3 perovskite was tested. After mixing the solutions of oxalic acid and FAPbI3, a large amount of white precipitate was formed, as shown in Supplementary Figure 2. The powder product was identified as pure PbC2O4 material using X-ray diffraction (XRD) (Supplementary Figure 3) and Fourier transform infrared spectroscopy (FTIR) (Supplementary Figure 4). We found that oxalic acid is highly soluble in isopropyl alcohol (IPA) at room temperature with a solubility of approximately 18 mg/mL, as shown in Supplementary Figure 5. This makes subsequent processing easier since IPA, as a common passivation solvent, does not damage the perovskite layer beyond short time29. Therefore, by immersing the perovskite film in oxalic acid solution or spin-coating the oxalic acid solution onto the perovskite, thin and dense PbC2O4 can be quickly obtained on the surface of the perovskite film according to the following chemical equation: H2C2O4 + FAPbI3 = PbC2O4 + FAI +HI. FAI can be dissolved in IPA and thus removed during cooking. The thickness of LOS can be controlled by reaction time and precursor concentration.
        Scanning electron microscopy (SEM) images of control and LOS perovskite films are shown in Figures 2a,b. The results show that the perovskite surface morphology is well preserved, and a large number of fine particles are deposited on the grain surface, which should represent a PbC2O4 layer formed by the in-situ reaction. The LOS perovskite film has a slightly smoother surface (Supplementary Figure 6) and a larger water contact angle compared to the control film (Supplementary Figure 7). High-resolution transverse transmission electron microscopy (HR-TEM) was used to distinguish the surface layer of the product. Compared to the control film (Fig. 2c), a uniform and dense thin layer with a thickness of about 10 nm is clearly visible on top of the LOS perovskite (Fig. 2d). Using high-angle annular dark-field scanning electron microscopy (HAADF-STEM) to examine the interface between PbC2O4 and FAPbI3, the presence of crystalline regions of FAPbI3 and amorphous regions of PbC2O4 can be clearly observed (Supplementary Figure 8). The surface composition of the perovskite after oxalic acid treatment was characterized by X-ray photoelectron spectroscopy (XPS) measurements, as shown in Figures 2e–g. In Figure 2e, the C 1s peaks around 284.8 eV and 288.5 eV belong to the specific CC and FA signals, respectively. Compared to the control membrane, the LOS membrane exhibited an additional peak at 289.2 eV, attributed to C2O42-. The O 1s spectrum of LOS perovskite exhibits three chemically distinct O 1s peaks at 531.7 eV, 532.5 eV, and 533.4 eV, corresponding to deprotonated COO, C=O of intact oxalate groups 30 and O atoms of the OH component (Fig. 2e ). )). For the control sample, only a small O 1s peak was observed, which can be attributed to oxygen chemisorbed on the surface. The control membrane characteristics of Pb 4f7/2 and Pb 4f5/2 are located at 138.4 eV and 143.3 eV, respectively. We observed that the LOS perovskite exhibits a shift of the Pb peak of about 0.15 eV towards higher binding energy, indicating a stronger interaction between the C2O42- and Pb atoms (Fig. 2g).
        a SEM images of control and b LOS perovskite films, top view. c High-resolution cross-sectional transmission electron microscopy (HR-TEM) of control and d LOS perovskite films. High-resolution XPS of e C 1s, f O 1s and g Pb 4f perovskite films. Source data is provided in the form of source data files.
        According to the DFT results, it is theoretically predicted that VI defects and I migration easily cause the phase transition from α to δ. Previous reports have shown that I2 is rapidly released from PC-based perovskite films during photoimmersion after exposing the films to light and thermal stress31,32,33. To confirm the stabilizing effect of lead oxalate on the α-phase of perovskite, we immersed the control and LOS perovskite films in transparent glass bottles containing toluene, respectively, and then irradiated them with 1 sunlight for 24 h. We measured the absorption of ultraviolet and visible light (UV-Vis). ) toluene solution, as shown in Figure 3a. Compared with the control sample, much lower I2 absorption intensity was observed in the case of LOS-perovskite, indicating that compact LOS can inhibit the release of I2 from the perovskite film during light immersion. Photographs of aged control and LOS perovskite films are shown in the insets of Figures 3b and c. The LOS perovskite is still black, while most of the control film has turned yellow. The UV–visible absorption spectra of the immersed film are shown in Figs. 3b, c. We observed that the absorption corresponding to α in the control film was clearly decreased. X-ray measurements were performed to document the evolution of the crystal structure. After 24 hours of illumination, the control perovskite showed a strong yellow δ-phase signal (11.8°), while the LOS perovskite still maintained a good black phase (Figure 3d).
        UV-visible absorption spectra of toluene solutions in which the control film and LOS film were immersed under 1 sunlight for 24 hours. The inset shows a vial in which each film was immersed in an equal volume of toluene. b UV-Vis absorption spectra of control film and c LOS film before and after 24 h of immersion under 1 sunlight. The inset shows a photograph of the test film. d X-ray diffraction patterns of control and LOS films before and after 24 h of exposure. SEM images of control film e and film f LOS after 24 hours of exposure. Source data is provided in the form of source data files.
        We performed scanning electron microscopy (SEM) measurements to observe the microstructural changes of the perovskite film after 24 hours of illumination, as shown in Figures 3e,f. In the control film, large grains were destroyed and turned into small needles, which corresponded to the morphology of the δ-phase product FAPbI3 (Fig. 3e). For LOS films, the perovskite grains remain in good condition (Figure 3f). The results confirmed that the loss of I significantly induces the transition from the black phase to the yellow phase, while PbC2O4 stabilizes the black phase, preventing the loss of I. Since the vacancy density at the surface is much higher than in the grain bulk,34 this phase is more likely to occur at the surface of the grain. simultaneously releasing iodine and forming VI. As predicted by DFT, LOS can inhibit the formation of VI defects and prevent the migration of I ions to the perovskite surface.
        Additionally, the effect of the PbC2O4 layer on the moisture resistance of perovskite films in atmospheric air (relative humidity 30-60%) was studied. As shown in Supplementary Figure 9, the LOS perovskite was still black after 12 days, while the control film turned yellow. In XRD measurements, the control film shows a strong peak at 11.8° corresponding to the δ phase of FAPbI3, while the LOS perovskite retains the black α phase well (Supplementary Figure 10).
        Steady-state photoluminescence (PL) and time-resolved photoluminescence (TRPL) were used to study the passivation effect of lead oxalate on the perovskite surface. In Fig. Figure 4a shows that the LOS film has increased PL intensity. In the PL mapping image, the intensity of the LOS film over the entire area of ​​10 × 10 μm2 is higher than that of the control film (Supplementary Figure 11), indicating that PbC2O4 uniformly passivates the perovskite film. The carrier lifetime is determined by approximating the TRPL decay with a single exponential function (Fig. 4b). The carrier lifetime of the LOS film is 5.2 μs, which is much longer than the control film with a carrier lifetime of 0.9 μs, indicating reduced surface nonradiative recombination.
        Steady-state PL and b-spectra of temporary PL of perovskite films on glass substrates. c SP curve of the device (FTO/TiO2/SnO2/perovskite/spiro-OMeTAD/Au). d EQE spectrum and Jsc EQE spectrum integrated from the most efficient device. d Dependence of light intensity of a perovskite device on the Voc diagram. f Typical MKRC analysis using an ITO/PEDOT:PSS/perovskite/PCBM/Au clean hole device. VTFL is the maximum trap filling voltage. From these data we calculated the trap density (Nt). Source data is provided in the form of source data files.
        To study the effect of the lead oxalate layer on the device performance, a traditional FTO/TiO2/SnO2/perovskite/spiro-OMeTAD/Au contact structure was used. We use formamidine chloride (FACl) as an additive to the perovskite precursor instead of methylamine hydrochloride (MACl) to achieve better device performance, since FACl can provide better crystal quality and avoid the band gap of FAPbI335 (see Supplementary Figures 1 and 2 for detailed comparison). ). 12-14). IPA was chosen as the antisolvent because it provides better crystal quality and preferred orientation in perovskite films compared to diethyl ether (DE) or chlorobenzene (CB)36 (Supplementary Figures 15 and 16). The thickness of PbC2O4 was carefully optimized to well balance defect passivation and charge transport by adjusting the oxalic acid concentration (Supplementary Figure 17). Cross-sectional SEM images of the optimized control and LOS devices are shown in Supplementary Figure 18. Typical current density (CD) curves for control and LOS devices are shown in Figure 4c, and the extracted parameters are given in Supplementary Table 3. Maximum power conversion efficiency (PCE) control cells 23.43% (22.94%), Jsc 25.75 mA cm-2 (25.74 mA cm-2), Voc 1.16 V (1.16 V) and reverse (forward) scan. The fill factor (FF) is 78.40% (76.69%). Maximum PCE LOS PSC is 25.39% (24.79%), Jsc is 25.77 mA cm-2, Voc is 1.18 V, FF is 83.50% (81.52%) from reverse (forward Scan to). The LOS device achieved a certified photovoltaic performance of 24.92% in a trusted third-party photovoltaic laboratory (Supplementary Figure 19). The external quantum efficiency (EQE) gave an integrated Jsc of 24.90 mA cm-2 (control) and 25.18 mA cm-2 (LOS PSC), respectively, which was in good agreement with the Jsc measured in the standard AM 1.5 G spectrum (Fig. .4d). ) . The statistical distribution of measured PCEs for control and LOS PSCs is shown in Supplementary Figure 20.
        As shown in Figure 4e, the relationship between Voc and light intensity was calculated to study the effect of PbC2O4 on trap-assisted surface recombination. The slope of the fitted line for the LOS device is 1.16 kBT/sq, which is lower than the slope of the fitted line for the control device (1.31 kBT/sq), confirming that LOS is useful for inhibiting surface recombination by decoys. We use space charge current limiting (SCLC) technology to quantitatively measure the defect density of a perovskite film by measuring the dark I-V characteristic of a hole device (ITO/PEDOT:PSS/perovskite/spiro-OMeTAD/Au) as shown in the figure. 4f Show. The trap density is calculated by the formula Nt = 2ε0εVTFL/eL2, where ε is the relative dielectric constant of the perovskite film, ε0 is the dielectric constant of vacuum, VTFL is the limiting voltage for filling the trap, e is the charge, L is the thickness of the perovskite film (650 nm). The defect density of the VOC device is calculated to be 1.450 × 1015 cm–3, which is lower than the defect density of the control device, which is 1.795 × 1015 cm–3.
        The unpackaged device was tested at the maximum power point (MPP) under full daylight under nitrogen to examine its long-term performance stability (Figure 5a). After 550 hours, the LOS device still maintained 92% of its maximum efficiency, while the control device’s performance had dropped to 60% of its original performance. The distribution of elements in the old device was measured by time-of-flight secondary ion mass spectrometry (ToF-SIMS) (Fig. 5b, c). A large accumulation of iodine can be seen in the upper gold control area. The conditions of inert gas protection exclude environmentally degrading factors such as moisture and oxygen, suggesting that internal mechanisms (i.e., ion migration) are responsible. According to ToF-SIMS results, I- and AuI2- ions were detected in the Au electrode, indicating the diffusion of I from the perovskite into Au. The signal intensity of I- and AuI2- ions in the control device is approximately 10 times higher than that of the VOC sample. Previous reports have shown that ion permeation can lead to a rapid decrease in the hole conductivity of spiro-OMeTAD and chemical corrosion of the top electrode layer, thereby deteriorating the interfacial contact in the device37,38. The Au electrode was removed and the spiro-OMeTAD layer was cleaned from the substrate with a chlorobenzene solution. We then characterized the film using grazing incidence X-ray diffraction (GIXRD) (Figure 5d). The results show that the control film has an obvious diffraction peak at 11.8°, while no new diffraction peak appears in the LOS sample. The results show that large losses of I ions in the control film lead to the generation of the δ phase, while in the LOS film this process is clearly inhibited.
        575 hours of continuous MPP tracking of an unsealed device in a nitrogen atmosphere and 1 sunlight without a UV filter. ToF-SIMS distribution of b I- and c AuI2- ions in the LOS MPP control device and aging device. The shades of yellow, green and orange correspond to Au, Spiro-OMeTAD and perovskite. d GIXRD of perovskite film after MPP test. Source data is provided in the form of source data files.
        Temperature-dependent conductivity was measured to confirm that PbC2O4 could inhibit ion migration (Supplementary Figure 21). The activation energy (Ea) of ion migration is determined by measuring the change in conductivity (σ) of the FAPbI3 film at different temperatures (T) and using the Nernst-Einstein relation: σT = σ0exp(−Ea/kBT), where σ0 is a constant, kB is the Boltzmann constant. We obtain the value of Ea from the slope of ln(σT) versus 1/T, which is 0.283 eV for the control and 0.419 eV for the LOS device.
        In summary, we provide a theoretical framework to identify the degradation pathway of FAPbI3 perovskite and the influence of various defects on the energy barrier of the α-δ phase transition. Among these defects, VI defects are theoretically predicted to easily cause a phase transition from α to δ. A water-insoluble and chemically stable dense layer of PbC2O4 is introduced to stabilize the α-phase of FAPbI3 by inhibiting the formation of I vacancies and the migration of I ions. This strategy significantly reduces the interfacial non-radiative recombination, increases the solar cell efficiency to 25.39%, and improves the operating stability. Our results provide guidance for achieving efficient and stable formamidine PSCs by inhibiting the defect-induced α to δ phase transition.
        Titanium(IV) isopropoxide (TTIP, 99.999%) was purchased from Sigma-Aldrich. Hydrochloric acid (HCl, 35.0–37.0%) and ethanol (anhydrous) were purchased from Guangzhou Chemical Industry. SnO2 (15 wt% tin(IV) oxide colloidal dispersion) was purchased from Alfa Aesar. Lead(II) iodide (PbI2, 99.99%) was purchased from TCI Shanghai (China). Formamidine iodide (FAI, ≥99.5%), formamidine chloride (FACl, ≥99.5%), methylamine hydrochloride (MACl, ≥99.5%), 2,2′,7,7′-tetrakis-(N , N-di-p) )-methoxyaniline)-9,9′-spirobifluorene (Spiro-OMeTAD, ≥99.5%), lithium bis(trifluoromethane)sulfonylimide (Li-TFSI, 99.95%), 4-tert -butylpyridine (tBP, 96%) was purchased from Xi’an Polymer Light Technology Company (China). N,N-dimethylformamide (DMF, 99.8%), dimethyl sulfoxide (DMSO, 99.9%), isopropyl alcohol (IPA, 99.8%), chlorobenzene (CB, 99.8%), acetonitrile (ACN). Purchased from Sigma-Aldrich. Oxalic acid (H2C2O4, 99.9%) was purchased from Macklin. All chemicals were used as received without any other modifications.
        ITO or FTO substrates (1.5 × 1.5 cm2) were ultrasonically cleaned with detergent, acetone, and ethanol for 10 min, respectively, and then dried under a nitrogen stream. A dense TiO2 barrier layer was deposited on an FTO substrate using a solution of titanium diisopropoxybis(acetylacetonate) in ethanol (1/25, v/v) deposited at 500 °C for 60 min. The SnO2 colloidal dispersion was diluted with deionized water in a volume ratio of 1:5. On a clean substrate treated with UV ozone for 20 minutes, a thin film of SnO2 nanoparticles was deposited at 4000 rpm for 30 seconds and then preheated at 150 °C for 30 minutes. For the perovskite precursor solution, 275.2 mg FAI, 737.6 mg PbI2 and FACl (20 mol%) were dissolved in DMF/DMSO (15/1) mixed solvent. The perovskite layer was prepared by centrifuging 40 μL of perovskite precursor solution on top of the UV-ozone-treated SnO2 layer at 5000 rpm in ambient air for 25 s. 5 seconds after the last time, 50 μL of MACl IPA solution (4 mg/mL) was quickly dropped onto the substrate as an antisolvent. Then, the freshly prepared films were annealed at 150°C for 20 min and then at 100°C for 10 min. After cooling the perovskite film to room temperature, H2C2O4 solution (1, 2, 4 mg dissolved in 1 mL IPA) was centrifuged at 4000 rpm for 30 s to passivate the perovskite surface. A spiro-OMeTAD solution prepared by mixing 72.3 mg spiro-OMeTAD, 1 ml CB, 27 µl tBP and 17.5 µl Li-TFSI (520 mg in 1 ml acetonitrile) was spin-coated onto the film at 4000 rpm within 30 s. Finally, a 100 nm thick Au layer was evaporated in vacuum at a rate of 0.05 nm/s (0~1 nm), 0.1 nm/s (2~15 nm) and 0.5 nm/s (16~100 nm) . ).
        The SC performance of the perovskite solar cells was measured using a Keithley 2400 meter under solar simulator illumination (SS-X50) at a light intensity of 100 mW/cm2 and verified using calibrated standard silicon solar cells. Unless otherwise stated, SP curves were measured in a nitrogen-filled glove box at room temperature (~25°C) in forward and reverse scan modes (voltage step 20 mV, delay time 10 ms). A shadow mask was used to determine an effective area of ​​0.067 cm2 for the measured PSC. EQE measurements were carried out in ambient air using a PVE300-IVT210 system (Industrial Vision Technology(s) Pte Ltd) with monochromatic light focused on the device. For device stability, testing of non-encapsulated solar cells was carried out in a nitrogen glovebox at 100 mW/cm2 pressure without a UV filter. ToF-SIMS is measured using PHI nanoTOFII time-of-flight SIMS. Depth profiling was obtained using a 4 kV Ar ion gun with an area of ​​400×400 µm.
        X-ray photoelectron spectroscopy (XPS) measurements were performed on a Thermo-VG Scientific system (ESCALAB 250) using monochromatized Al Kα (for XPS mode) at a pressure of 5.0 × 10–7 Pa. Scanning electron microscopy (SEM) was performed on a JEOL-JSM-6330F system. The surface morphology and roughness of the perovskite films were measured using atomic force microscopy (AFM) (Bruker Dimension FastScan). STEM and HAADF-STEM are held at the FEI Titan Themis STEM. UV–Vis absorption spectra were measured using a UV-3600Plus (Shimadzu Corporation). Space charge limiting current (SCLC) was recorded on a Keithley 2400 meter. Steady-state photoluminescence (PL) and time-resolved photoluminescence (TRPL) of carrier lifetime decay were measured using an FLS 1000 photoluminescence spectrometer. PL mapping images were measured using a Horiba LabRam Raman system HR Evolution. Fourier transform infrared spectroscopy (FTIR) was performed using a Thermo-Fisher Nicolet NXR 9650 system.
        In this work, we use the SSW path sampling method to study the phase transition path from α-phase to δ-phase. In the SSW method, the motion of the potential energy surface is determined by the direction of the random soft mode (second derivative), which allows a detailed and objective study of the potential energy surface. In this work, path sampling is performed on a 72-atom supercell, and more than 100 initial/final state (IS/FS) pairs are collected at the DFT level. Based on the IS/FS pairwise data set, the path connecting the initial structure and the final structure can be determined with the correspondence between atoms, and then the two-way movement along the variable unit surface is used to smoothly determine the transition state method. (VK-DESV). After searching for the transition state, the path with the lowest barrier can be determined by ranking the energy barriers.
        All DFT calculations were performed using VASP (version 5.3.5), where the electron–ion interactions of C, N, H, Pb, and I atoms are represented by a projected amplified wave (PAW) scheme. The exchange correlation function is described by the generalized gradient approximation in the Perdue-Burke-Ernzerhoff parametrization. The energy limit for plane waves was set to 400 eV. The Monkhorst–Pack k-point grid has a size of (2 × 2 × 1). For all structures, lattice and atomic positions were fully optimized until the maximum stress component was below 0.1 GPa and the maximum force component was below 0.02 eV/Å. In the surface model, the surface of FAPbI3 has 4 layers, the bottom layer has fixed atoms simulating the body of FAPbI3, and the top three layers can move freely during the optimization process. The PbC2O4 layer is 1 ML thick and is located on the I-terminal surface of FAPbI3, where Pb is bound to 1 I and 4 O.
       For more information about the study design, see the Natural Portfolio Report Abstract associated with this article.
        All data obtained or analyzed during this study are included in the published article, as well as in the supporting information and raw data files. The raw data presented in this study are available at https://doi.org/10.6084/m9.figshare.2410016440. Source data is provided for this article.
        Green, M. et al. Solar Cell Efficiency Tables (57th ed.). program. photoelectric. resource. application. 29, 3–15 (2021).
        Parker J. et al. Controlling the growth of perovskite layers using volatile alkyl ammonium chlorides. Nature 616, 724–730 (2023).
        Zhao Y. et al. Inactive (PbI2)2RbCl stabilizes perovskite films for high-efficiency solar cells. Science 377, 531–534 (2022).
        Tan, K. et al. Inverted perovskite solar cells using dimethylacridinyl dopant. Nature, 620, 545–551 (2023).
        Han, K. et al. Single crystalline formamidine lead iodide (FAPbI3): insights into structural, optical and electrical properties. adverb. Matt. 28, 2253–2258 (2016).
        Massey, S. et al. Stabilization of the black perovskite phase in FAPbI3 and CsPbI3. AKS Energy Communications. 5, 1974–1985 (2020).
        You, JJ, et al. Efficient perovskite solar cells through improved carrier management. Nature 590, 587–593 (2021).
        Saliba M. et al. Incorporation of rubidium cations into perovskite solar cells improves photovoltaic performance. Science 354, 206–209 (2016).
        Saliba M. et al. Triple-cation perovskite cesium solar cells: improved stability, reproducibility and high efficiency. energy environment. the science. 9, 1989–1997 (2016).
        Cui X. et al. Recent advances in FAPbI3 phase stabilization in high-performance perovskite solar cells Sol. RRL 6, 2200497 (2022).
        Delagetta S. et al. Rationalized photoinduced phase separation of mixed halide organic-inorganic perovskites. Nat. communicate. 8, 200 (2017).
        Slotcavage, DJ et al. Light-induced phase separation in halide perovskite absorbers. AKS Energy Communications. 1, 1199–1205 (2016).
        Chen, L. et al. Intrinsic phase stability and intrinsic bandgap of formamidine lead triiodide perovskite single crystal. Anjiva. Chemical. internationality. Ed. 61. e202212700 (2022).
        Duinsti, E.A. etc. Understand the decomposition of methylenediammonium and its role in the phase stabilization of lead triiodide formamidine. J. Chem. Bitch. 18, 10275–10284 (2023).
        Lu, H.Z. et al. Efficient and stable vapor deposition of black perovskite solar cells FAPbI3. Science 370, 74 (2020).
        Doherty, T.A.S. etc. Stable tilted octahedral halide perovskites suppress the localized formation of phases with limited characteristics. Science 374, 1598–1605 (2021).
        Ho, K. et al. Mechanisms of transformation and degradation of formamidine grains and cesium and lead iodide perovskites under the influence of moisture and light. AKS Energy Communications. 6, 934–940 (2021).
        Zheng J. et al. Development of pseudohalide anions for α-FAPbI3 perovskite solar cells. Nature 592, 381–385 (2021).