Interfacial ferroelectricity unlocks stable formamidinium-based perovskites

Abstract

Pure iodide formamidinium (FA) based-perovskite has emerged as highly promising candidates for perovskite photovoltaics, but it remains challenging to achieve long-term phase-stabilized FA-based perovskites. Herein, we present a physics-driven strategy of interfacial ferroelectricity, achieved by the integration of ferroelectric CsMnBr₃ nanocrystals (NCs) into FA-based perovskites. The ferroelectric field generated by these NCs promotes FA⁺ cation ordering, modulates Pb–I framework, and enhances the structural regulation of the perovskite lattice. This synergistically increases the kinetic barrier for the undesired Pb-I octahedral transformation and raises the energy barrier for ion migration. The resulting perovskite materials exhibit high structure stability, enabling perovskite solar cell (PSC) minimodule to retain 99% of its initial efficiency after 1000 hours’ stability testing under 85% relative humidity at 85 °C. Owing to the improvement at the interface, the PSCs yield an efficiency of 26.62% (certified 26.40%), and the minimodules reach 24.67% (certified 23.23%). This work presents an effective approach to achieving high-performance, long-term stable perovskite optoelectronic devices through interfacial ferroelectric engineering.

Introduction

Formamidinium (FA)-based perovskites, particularly FA-rich compositions such as FA_1-xCs_xPbI₃, have emerged as highly promising light-absorbing layers for solar cells, owing to their optimal optoelectronic properties^1,2,3,4. However, their widespread applications are hindered by their intrinsic structure instability from the photoactive black phase (cubic/tetragonal, corner-sharing octahedra) to the photoinactive yellow phase (hexagonal, face-sharing octahedra)^5,6,7. These undesired transitions of Pb-I octahedral configuration considerably degrade the electronic and optical properties of FA-based perovskites, posing a major challenge to achieve stability⁸.

In retrospect to the development of stable and efficient FA-based perovskite materials and devices, studies have emphasized the importance of the precise structural manipulation, such as tilted structure^8,9,10,11. The tilted octahedra in FA-rich perovskites act as a natural barrier against the formation of undesired hexagonal face-sharing octahedral structure—a concept increasingly supported by recent research^12,13,14. To realize such a goal, extensive strategies have been explored to stabilize the perovskite lattice, including additive screening, dimensional manipulation, and composition engineering^{6,15,16,17,18}. However, these chemical approaches often involve trade-offs between stability and device efficiency. The chemical compounds may affect the material features such as conductivity, bandgap and defect dynamics, etc^12,19,20,21. For example, ethylenediaminetetraacetic acid was first found to guide the growth of octahedral-tilted-stabilized FAPbI₃ films, but unfortunately, it also impairs carrier dynamics, leading to substantial losses in device performance¹². More recent approaches emphasize crystal structure management—such as lattice strain engineering, suppression of octahedral tilting, regulation of hydrogen-bonding networks, and defect regulation through crystal symmetry control—to further stabilize FA-based perovskites, suppress ion migration, and enhance both efficiency and durability^8,22,23. These results highlight the critical role of structure engineering in overcoming stability challenges of FA-based perovskites.

Herein, we present a fundamental, physics-driven method based on the interfacial ferroelectric field to enhance the intrinsic structure ordering and achieve stable, pure iodide FA-based perovskites. By integrating ferroelectric CsMnBr₃ nanocrystals (NCs) at the surface/interface of FA-based perovskites, the ferroelectric field caused by these NCs promotes FA⁺ cation ordering, suppresses dynamic fluctuations, and strengthens Pb-I lattice connectivity. This structural reinforcement effectively inhibits the transition of Pb-I octahedra from corner-sharing to face-sharing and increases the energy barrier for ion migration. Meanwhile, the enhanced structure raises the energy barrier for ion migration. These improvements enable the corresponding PSC minimodule to achieve T₉₉ stability over 1000 hours under 85% relative humidity (RH) at 85 °C. Additionally, the interfacial ferroelectric field mitigates defect states at the surface/interface and increases the charge-carrier lifetime. The resulting PSCs achieve a power conversion efficiency (PCE) of 26.62%, and the minimodules reach 24.67%.

Results

Integrating a ferroelectric layer into FA-based perovskites

Ferroelectric CsMnBr₃ NCs were synthesized through a simple injection method, as illustrated in Fig. S1. Transmission electron microscopy (TEM) images reveal that the resulting CsMnBr₃ NCs exhibit a well-defined hexagonal shape with an average size of ~20 nm (Fig. S2). These NCs crystallize in a hexagonal phase with a P₆₃/mmc space group, consistent with the reference pattern (PDF#26-0387), as confirmed by the powder X-ray diffraction (XRD) patterns in Fig. 1a. The measured interplanar spacing of 3.06 Å corresponds to the (201) crystal plane, further validating the hexagonal crystal structure of CsMnBr₃. The CsMnBr₃ NCs show a distinct absorption edge ~625 nm (Fig. S3) and under ultraviolet illumination conditions, emit red emission in agreement with previous result²⁴.

Fig. 1: Structural analyses of CsMnBr₃ and FACs-based perovskite with CsMnBr₃ integration.

a XRD patterns for synthesized CsMnBr₃ NCs (light blue line) and standard CsMnBr₃ perovskite (red line, PDF#26-0387). The inset shows photographs of the CsMnBr₃ NCs solution under ambient light and ultraviolet illumination. b Phase and amplitude switching spectroscopy loops for a CsMnBr₃ film, demonstrating ferroelectric-like hysteresis. c FA⁺, Pb²⁺, Mn²⁺ distribution throughout IF-FACs perovskite, a. u., arbitrary units. d SEM images for FACs and IF-FACs perovskite films, scale represents 1 μm. e TEM morphologies for the IF-FACs perovskite crystal, scale represents 20 nm, the insert images are HRTEM lattice fringes. f Phase mapping images of FACs- and IF-FACs- based films, scale represents 600 nm.

Our further characterization confirms the ferroelectric properties of CsMnBr₃ NCs. Piezoresponse force microscopy (PFM) measurements of CsMnBr₃ NCs film reveal robust ferroelectric-like behavior, featuring distinct polarization domains and reversible switching under applied electric fields (Figs. 1b and S4). The topographic image shows the granular morphology, and the PFM phase map highlights well-defined domains, indicating stable polarization ordering. A clear 180° phase shift in the phase–voltage hysteresis loop (Fig. 1b), and a characteristic butterfly-shaped amplitude–voltage response confirm ferroelectric switching actions in these NCs²⁵. In fact, octahedral distortions in CsMnBr₃ can break local inversion symmetry transiently or in confined regions, thereby enabling polarization switching under an external electric field, which differs from conventional ferroelectricity arising from intrinsic polar crystal symmetry.

We managed to integrate the ferroelectric CsMnBr₃ NCs into FA-based perovskite films. The reference FA_0.95Cs_0.05PbI₃ is denoted as FACs, while FACs with CsMnBr₃ NCs integration is denoted as IF-FACs. Such integration process has neglectable effect on the key features of perovskite films, such as absorption edge, crystallinity supported by ultraviolet-visible (UV-vis) absorption spectra (Fig. S5) and XRD result (Fig. S6).

To determine the spatial distribution of the ferroelectric NCs, time-of-flight secondary ion mass spectrometry (TOF-SIMS) reveals a uniform distribution of Mn element throughout the perovskite layer, similar to the distributions of FA⁺ and Pb²⁺ (Figs. 1c and S7), confirming the homogeneous integration of CsMnBr₃ NCs. Scanning electron microscope (SEM) images exhibit numerous small CsMnBr₃ NCs on the surface and at the grain boundaries of the perovskite film (Fig. 1d). In order to further investigate their positioning and interaction with perovskites, IF-FACs perovskite powders were carefully exfoliated from the ITO substrate and then dispersed into a chlorobenzene solution, for high resolution transmission electron microscope (HRTEM) analysis. The lattice fringes of CsMnBr₃ NCs (marked in red) and FACs perovskite (marked in blue) confirm their distinct crystalline structures, as shown in Figs. 1e and S8. CsMnBr₃ NCs are observed to spontaneously adsorb onto the surface of the perovskite crystal, forming a homogeneous dispersion. HRTEM images show well-defined lattice fringes of CsMnBr₃ NCs with the (201) planes interplanar spacing of 3.06 Å and FACs perovskites with the (200) planes interplanar spacing of 3.21 Å. These results provide compelling structural insights into the integration of CsMnBr₃ NCs into the FACs perovskite matrix.

Benefitting from the integration of CsMnBr₃ NCs, FACs-based perovskite exhibits significantly enhanced phase structure stability, as shown in Fig. S9. Conventional FACs-based perovskite films undergo a complete phase transition to the undesirable yellow δ-phase (face/edge-sharing octahedra) within only 200 hours under ambient air conditions (85% RH–25 °C), while IF-FACs perovskite films maintain their black phase (corner-sharing octahedra) integrity, exhibiting negligible degradation even after an extended 2000-hour test. Similarly, prolonged thermal stress testing under 85% RH at 85 °C for 2000 hours leads to the emergence of δ-phase and lead iodide in FACs perovskite films. In contrast, IF-FACs perovskite films preserve their pure-phase structure, demonstrating remarkable resistance to moisture and thermal stress. Similarly, with CsMnBr₃ integration, FAPbI₃ exhibits significant phase structure stability enhancement (Fig. S10). These findings highlight the crucial role of CsMnBr₃ NCs in reinforcing the intrinsic structure stability of perovskite materials.

To confirm the specific role of ferroelectric NCs on the enhanced stability of IF-FACs perovskite films, we synthesize non-ferroelectric CsPbBr₃ NCs as a control to investigate the impact of NC form. TEM images show that the CsPbBr₃ NCs have a similar particle size to CsMnBr₃ NCs (Fig. S11), but lack the ferroelectric properties (Fig. S12). While both types of NCs have a limited effect on the optical and structural features of perovskite film (Fig. S13), their effects on stability differ drastically (Fig. S14). Such results rule out the influence of NC form and highlight the contribution of ferroelectricity caused by CsMnBr₃ NCs, suggesting that the ferroelectric interaction is a key mechanism for stabilizing the FA-based perovskite lattice, in contrast to prior reports where ferroelectrics primarily enhanced built-in fields and carrier transport^26,27.

PFM analyses reveal a pronounced polarization response in IF-FACs perovskite upon integration of CsMnBr₃ NCs (Figs. 1f and S15). The PFM phase map reveals well-defined domains, indicative of long-range polarization ordering. Notably, a significant 180° phase shift in the phase–voltage hysteresis loop, together with a characteristic butterfly-shaped amplitude–voltage curve, highlights the strong ferroelectric switching when integrating these NCs into perovskite. Additionally, applying electric writing to CsMnBr₃ and IF-FACs films, the contrast between bright and dark areas in the PFM image indicates ferroelectric domain switching (Fig. S16). Such results suggest that CsMnBr₃ NCs modify the local electric field and potentially influence the structural properties of FACs perovskites.

Ferroelectric field-driven ordering structure for stabilizing FA-based perovskite

The ferroelectricity of CsMnBr₃ inherently produces a spontaneous ferroelectric polarization field to alter the local ionic chemical environment, which is confirmed by XPS results (Fig. S17). To probe the structural and chemical effects of this field, we employ solid-state nuclear magnetic resonance (NMR) and Fourier transform infrared spectroscopy (FTIR) (Fig. 2a, b). Upon CsMnBr₃ NCs integration, the ¹H NMR spectra of IF-FACs exhibit two obvious features: a significant upfield shift (Δδ ~ 0.25 ppm for -CH and ~0.10 ppm for -NH₂), indicating an enhanced shielding effect attributed to altered hydrogen bonding and redistribution of electron density around FA⁺ cations. Meanwhile, a narrowing of peak linewidths, suggesting reduced dynamic disorder and enhanced rotational ordering of FA⁺ cation^20,28. These results indicate that the ferroelectric polarization field of CsMnBr₃ NCs locally enhances dipole-dipole coupling and suppresses FA⁺ cation rotational entropy. Complementary FTIR spectroscopy reveals blue shifts in the N–H bending (δ_N−H, Δ = +4.7 cm⁻¹) and C–N stretching (ν_C−N, Δ = +2.4 cm⁻¹) modes, confirming polarization-induced stiffening of these bonds^29,30. This should be attributed to polarization-induced electron density redistribution and consequently enhanced bond force constant. This aligns with NMR results, collectively demonstrating that CsMnBr₃ integration reorganizes FA⁺ cations in the local environment via electronic redistribution and promotes long-range cation rotational ordering, as evidenced by the synergistic narrowing of NMR peaks and hardening of vibrational modes.

Fig. 2: Interface ferroelectricity-driven ordering crystal structure.

a ¹H and ²⁰⁷Pb spectra of FACs- and IF-FACs- based samples. b FTIR spectra for FACs- and IF-FACs- based samples. Wavelet transformations-EXAFS images of (c) FACs and d IF-FACs. e Schematic illustration of the effect of interfacial ferroelectricity on the crystal structure of FAPbI₃ perovskite and the energy barrier of the transition of Pb-I octahedral configuration from corner-sharing to face-sharing.

Moreover, the ²⁰⁷Pb NMR spectra (Fig. 2a, right) of IF-FACs display a significantly narrower and symmetric profile compared to that of IF-FACs, indicating a more uniform electrostatic environment around Pb centers and reduced structural disorder^31,32. To gain deeper insight into the polarization-induced structural modulation, synchrotron-based extended X-ray absorption fine structure (EXAFS) spectroscopy was conducted (Figs. 2c, d and S18). The R-space EXAFS spectra reveal that the characteristic Pb–I bonds for the pristine FACs sample appear at ~3.0 Å, corresponding to the neighboring Pb-I coordination. As to the IF-FACs perovskite, this peak shifts to a 0.05 Å longer radial distance, indicating a slight elongation of the Pb-I bond. This structural relaxation is likely a result of polarization-induced local field effects, which reduce the disorder of oriented FA⁺ cations and modulate the electrostatic potential landscape around Pb²⁺, thereby alleviating bond strain. Additionally, the internal electric fields originating from the ferroelectric CsMnBr₃ domains can partially screen the coulombic attraction between Pb²⁺ and I⁻, leading to a decrease in effective bond strength and elongation of the Pb–I bond distance.

These spectral results, corroborated by the PFM results of IF-FACs perovskite films (Fig. 1f), highlight the spontaneous polarization of CsMnBr₃ ferroelectric phase as the key driver of lattice structure modification in FA-based perovskites through inducing charge density redistribution around lattice ions (Fig. 2e). Theoretical investigation (Figs. S19–21) further confirms that the internal ferroelectric polarization field of CsMnBr₃ induces substantial redistribution of charge density along the crystallographic a, b, c-axis. Notably, the most pronounced charge density fluctuation occurs along the c-axis, where the induced dipole moment reaches –4.824 D, compared to 2.878 D and 1.536 D along the a- and b-axes, respectively.

These structural modifications, driven by interfacial ferroelectric polarization domains, not only reorganize the local lattice environment but also serve as the fundamental mechanism behind the enhanced phase stability observed in FA-based perovskites. Our DFT calculations demonstrate that the resulting crystal structure caused by interfacial ferroelectricity dramatically increases the transition energy barrier (ΔE_b) from −0.255 eV to 1.259 eV for the octahedral reconfiguration from corner-sharing to face-sharing octahedra. This enhancement arises from the synergistic effects of (i) FA⁺ cation ordering, (ii) local Pb-I structural relaxation, and (iii) strengthened interactions between Pb-I and FA⁺ cation. The resulting enhanced structure kinetically stabilizes Pb-I configuration—consistent with our experimental observations of delayed phase transition onset.

Together, these experimental and theoretical insights provide compelling evidence that the integration of ferroelectric CsMnBr₃ introduces a built-in ferroelectric polarization field that reshapes the local coordination environment through charge redistribution and bond reconfiguration. This enables a tunable pathway to manipulate perovskite lattice and stabilize perovskite lattice.

Interfacial ferroelectricity unlocks stable PSCs

The enhanced crystal structure induced by the interfacial ferroelectric layer can effectively stabilize PSCs. Long-term maximum power point (MPP) tracking under simulated 1-sun illumination at 30 °C demonstrates good stability in IF-FACs-based devices, which retains nearly 100% of their initial efficiency over 3000 hours, in stark contrast to the rapid degradation observed in control FACs-based PSCs, which drop to ~22% of their original efficiency (Fig. 3a). Moreover, perovskite minimodules incorporating IF-FACs demonstrate notable durability under 85% RH at 85 °C, IF-FACs based minimodules sustain 99% of their initial performance after 1000 hours, while FACs counterparts show substantial decay (Fig. 3b).

Fig. 3: Stability of IF-FACs-based PSCs.

a The long-term operational stability of the encapsulated FACs and IF-FACs-based PSCs. b PCE evolution during 1000 h stability test for the encapsulated IF-FACs-based mini-module under 85% RH at 85 °C. A 5 × 5 μm² two-dimensional I⁻ distribution at ~300 nm depth in the perovskite layer of PSCs based on (c) FACs and d IF-FACs perovskites was obtained by TOF-SIMs after a 1000 h stability test at their MPP. e Density of mobile ions (top panel) and the energy barrier for the ion migration in FACs and IF-FACs films (bottom panel).

Such enhanced stability could originate from the suppression of ionic migration, as facilitated by the ferroelectric layer. Depth-resolved ion mapping after 1000 hours MPP testing reveals obvious differences in iodide distribution: FACs light-absorbing layer exhibits pronounced iodide (I⁻) clustering and phase heterogeneity (Fig. 3c), whereas IF-FACs light-absorbing layer maintains uniform ion dispersion (Fig. 3d), indicative of inhibited ion mobility. Quantitative ionic transport analysis further supports this conclusion: ionic concentration (N_ion) is dramatically reduced from 4.32 × 10¹³cm⁻³ in FACs to just 0.28 × 10¹³cm⁻³ in IF-FACs (Figs. 3e and S22), while the ion migration activation energy (E_a) increases from 0.46 ± 0.02 eV in FACs to 0.57 ± 0.06 eV in IF-FACs, signifying a higher energy barrier for ion transport. Time-dependent migration rate analysis in Fig. S23 further confirms that IF-FACs maintain a low ion mobility throughout extended testing, highlighting their long-term stability. These results indicate that interfacial ferroelectricity enhances the interaction between ions and their surroundings and reduces their mobility through enhancing structure ordering, thereby increasing the energy barrier for migration and restricting ion diffusion pathways, and finally contributing to significantly enhanced stability.

Photovoltaic performances of the PSCs

In addition to improving the stability of FACs-based perovskite materials and PSCs, the interface ferroelectric field significantly improves carrier dynamics in the perovskite films. To rule out the effect of organic ligands, here, we also prepare two other references, i.e., spin-coating only OA or OAm solutions onto the surface of FACs perovskite films (denoted as OA-FACs and OAm-FACs). Steady-state photoluminescence (PL) spectra in Fig. S24 revealed that OA-FACs and OAm-FACs exhibit stronger luminescence than FACs, indicating certain passivation effects of the organic ligand. Yet, their PL intensities are far lower than IF-FACs, indicating the dominant role of the interface ferroelectric field caused by CsMnBr₃. Time-resolved PL (TRPL) results further corroborate this improvement with prolonged carrier lifetimes in IF-FACs-based perovskite films (Table S1). Moreover, according to PL mapping results (Fig. 4a), IF-FACs perovskite film shows a significantly enhanced luminescence intensity and spatial uniformity. This notable difference provides evidence for suppressed non-radiative recombination within IF-FACs perovskite films. The defect physics of these samples are further investigated. Thermal admittance spectroscopy (TAS) is performed to probe the trap density and the energy depth of trap states in the device. The tDOS–Eω analysis shows a successive reduction in defect density in the order of FACs > OA-FACs > OAm-FACs > IF-FACs (Fig. S25). These results indicate a moderate passivation effect of OA and OAm passivation agents. The significantly reduced defects in IF-FAC films can be attributed to the interface ferroelectric field.

Fig. 4: Photovoltaic and device characterization.

a PL mapping image detected by 475 nm excitation wavelength of the FACs and IF-FACs-based perovskite films, scale represents 1 μm. b The J–V curves of the champion devices of FACs and IF-FACs PSCs with 0.09 cm² effective cell area. c Steady-state efficiency of IF-FACs-based PSCs. d J–V characteristics of PSC minimodule based on FACs, and IF-FACs with 10.86 cm² aperture cell area under simulated AM 1.5 G solar illumination of 100 mW cm⁻² in the reverse scan.

Because of the remarkable perovskite quality, IF-FACs-based PSCs exhibit significantly enhanced performance. As shown in Fig. 4b, the current density-voltage (J-V) curves give a PCE of 26.62% for the champion IF-FACs-based PSC, with an independently certified efficiency of 26.40% (Fig. S26). This enhanced efficiency is attributed to hthe igh open-circuit voltage (V_oc) of 1.182 V and fill factor (FF) of 0.859—significantly higher than conventional FACs-based PSCs, which achieve a PCE of only 24.42%. The external quantum efficiency (EQE) spectra confirm a consistency between the integrated photocurrent and the J_sc obtained from J-V measurements (Fig. S27). Additionally, a statistical analysis of 32 devices reveals improved reproducibility, as shown in Fig. S28. Notably, IF-FACs-based PSCs exhibit minimal hysteresis (Fig. S29 and Table S2), leading to a stabilized output power of 26.38% (Fig. 4c), indicating improved charge transport dynamics and reduced ion migration. To further evaluate the scalability of the interface ferroelectricity strategy, we fabricated 10.86 cm²-aperture mini-modules (Fig. S30). These IF-FACs-based mini-modules deliver an efficiency of 24.67% (Fig. 4d), with a V_oc of 7.17 V and FF of 84.5%. Furthermore, an independently certified efficiency of 23.23% is achieved (Fig. S31).

Discussion

In summary, we present a physics-driven strategy of interfacial ferroelectricity to achieve efficient and stable pure-iodide FA-based perovskites via the integration of a Mn-based ferroelectric layer. By integrating ferroelectric CsMnBr₃ NCs into FA-based perovskites, a strong ferroelectric field is generated, which can enhance crystal structure order through promoting FA cation order and manipulating the Pb-I framework. The elevated energy barrier for the undesired phase transition and ion migration collectively contribute to the significantly enhanced stability of perovskite materials and devices. This physical concept, which utilizes interfacial ferroelectricity to stabilize halide perovskite structures, offers a promising pathway for the development of high-performance, long-term stable perovskite-based optoelectronic devices.

Methods

Materials

Formamidinium iodide (FAI) was purchased from Greatcell Solar Materials Pty Ltd. Lead (II) iodide (PbI₂, 99.99%) was purchased from TCI. Cesium iodide (CsI) was purchased from Thermo Fisher Scientific Co., Ltd. Methylamine hydrochloride (MACl, 99.995%), lead (II) bromide (PbBr₂, 99.99%), Cesium Bromide (CsBr, 99.99%), (4-(7H-dibenzo [c, g] carbazol-7-yl) butyl) phosphonic acid (4PADCB), C₆₀ and BCP (>99% sublimed) were purchased from Xi’an Polymer Light Technology Corp. N, N-dimethylformamide (DMF, anhydrous, 99.8%), dimethyl sulfoxide (DMSO, anhydrous, 99.7%), isopropanol alcohol (IPA, 99.5%) and chlorobenzene (anhydrous, 99.8%) were purchased from J&K Scientific Ltd. oleic acid (OA, 90%, from Aladdin reagent), oleylamine (OAm, 80–90%), toluene (anhydrous, 99.5%), 1-octadecene (ODE, 90%), cesium carbonate (Cs₂CO₃, 99%), manganese acetate tetrahydrate (MnAc₂·4H₂O, 99.0%) and bromotrimethylsilane (TMSBr, ≥90%) were purchased from Aladdin reagent. Ethyl acetate (C₄H₈O₂, ≥99%) was purchased from Macklin.

Synthesis of CsMnBr₃ NCs

Cs₂CO₃ (65 mg), MnAc₂·4H₂O (98 mg), OA (1 mL), and ODE (2 mL) were added to a 20 mL glass bottle, and the solution was stirred at 90 °C until all the precursors were completely dissolved. The mixture was cooled down to room temperature, and dried OAm (0.1 mL) was subsequently injected to solubilize the solution. TMSBr (0.5 mL) was then swiftly injected, and the reaction lasts for 10–60 s. After being precipitated by ethyl acetate, the turbid liquid was centrifuged at 12,000 rpm for 5 minutes. The precipitate was dispersed into toluene and centrifuged at 12,000 rpm for 5 min to remove non-dispersible crystal particles. The supernatant containing intense red fluorescence was collected for future use.

Synthesis of CsPbBr₃ NCs

The precursor solution was obtained by dissolving PbBr₂ (0.4 mmol) and CsBr (0.4 mmol) in DMF (10 mL). OAm (0.2 mL) and OA (0.4 mL) were then added into the mixture above. The mixture was stirred for half an hour to obtain a clear solution. 0.5 mL precursor solution was quickly added into 10 mL toluene under vigorous stirring at 1500 rpm for 20 s. The solution was centrifuged at 12,000 rpm for 5 min to obtain the precipitate. All above-mentioned experimental process was performed at room temperature in air.

Perovskite precursor solution

1.63 M perovskite precursor solution was prepared by dissolving 266.3 mg FAI, 21.2 mg CsI, 751 mg PbI₂, and 20 mg MACl in 1 mL mixed DMF and DMSO solvents (volume ratio: 9:1) and stirred in a N₂ glovebox at room temperature overnight before film fabrication.

Perovskite solar cell fabrication

The ITO glass was washed by ultrasonication with water, acetone, and IPA sequentially and then treated with UV-Ozone for 15 min before use. 4PADCB IPA solution with a concentration of 0.3 mg/mL was spin-coated on the ITO glass at 5000 rpm for 30 s, followed by annealing at 100 °C for 10 min. In all, 70 μL perovskite precursor was then spin-coated on the ITO/4PADCB substrate by a two-consecutive step program at 1000 rpm for 5 s and 5000 rpm for 30 s. During the second step, 200 μL CB was dripped on film 14 seconds before the end of the program. Subsequently, the samples were annealed on a hotplate at 120 °C for 30 min. Finally, C₆₀(40 nm), and Au (60 nm) were sequentially deposited on top of the above perovskite by thermal evaporation. Before depositing Au electrode, a 20 nm layer of SnO₂ was sequentially deposited on top of the C₆₀ layer using atomic layer deposition. For the FA-based perovskite with NCs incorporation, the solution of CsMnBr₃ NCs (1 mM in 1 mL CB) was deposited and spin-coated at 5000 rpm for 30 sec on the ITO/4PADCB/perovskite sample, followed by heat treatment at 80 °C for 5 min, while other fabrication procedures were kept the same. For mini-module fabrication, the precursor solution is first applied to the substrate surface, followed by spin coating for 5 seconds to remove excess solution. The resulting film is then subjected to vacuum flashing at 0.1 Pa for 1 minute, and subsequently annealed at 120 °C for 30 minutes.

Characterizations

The morphology was obtained by scanning electron microscopy (Thermo Scios 2 Dual Beam). XRD pattern was obtained using a D8 Discover (Bruker, Germany) diffractometer with Cu Kα (λ  =  1.54 Å) radiation. TOF-SIMS analysis was measured by TOF SIMS 5-100 (ION-TOF GmbH, Germany). The EQE spectra were performed by the Enlitech QE-3011 system. UV-vis spectra were collected using a Cary 3600i Plus UV-vis spectrophotometer (Shimadzu, Japan) in air ambient environments. PFM measurements were performed with an MFP-3D Asylum Research microscope (Oxford Instrument Co.) using the silicon (n-type) cantilevers with Pt coating tips (Olympus OMCL-AC240TS). The PFM amplitude and phase images of the poled ferroelectric domain switching were measured through an application of a  ± 9-V bias to a probe on the surface of the films. Steady PL and TRPL were measured by FLS1000 Photoluminescence Spectrometer (Edinburgh Instruments Ltd.) with a 445 nm excitation laser. The PL mapping was characterized by a confocal Raman microscope (WITec) with a 532 nm excitation laser, and an emission signal was collected in the wavelength range from 780 to 830 nm with a center wavelength of 805 nm. Solid-State NMR experiments were performed on a Bruker Avance NEO 600 equipment.

GIWAXS measurements were performed at the BL14B1 beamline of the Shanghai Synchrotron Radiation Facility (SSRF) with a beam wavelength of 0.12398 nm. Pb L3-edge analysis was performed with Si (111) crystal monochromators at the BL11B beamlines at the SSRF. Before the analysis at the beamline, samples were pressed into thin sheets with 1 cm in diameter and sealed using Kapton tape film. The XAFS spectra were recorded at room temperature using a 4-channel Silicon Drift Detector Bruker 5040. Pb L3-edge EXAFS spectra were recorded in transmission mode. Negligible changes in the line shape and peak position of Pb L3-edge XANES spectra were observed between two scans taken for a specific sample. The XAFS spectra of the standard sample (Pb foil) were recorded in transmission mode. The spectra were processed and analyzed by the software codes Athena and Artemis. The wt-EXAFS spectra were obtained using the Hama software tool with k-space data from 0 to 9 Å⁻¹.

The J-V curves were measured with a Keithley 2401 source meter with a scan rate of 0.05 V s⁻¹ under the simulated AM 1.5 G illumination (100 mW cm⁻²) using an Enlitech 3 A light source. The current density-voltage (J–V) characteristics of the fabricated devices were acquired utilizing the Keithley 2400 Source Meter at room temperature. Both reverse scan (1.2 → 0.01 V, step: 0.01 V, delay time: 10 ms) and forward scan (0.01 → 1.2 V, step: 0.01 V, delay time: 10 ms) were performed. A mask of 0.09 cm² was employed during the testing process. The power output of the lamp was calibrated using a standard Si reference cell with a KG1 window.

Stability test

All PSCs were encapsulated using a laminator under the following conditions: a temperature of 100 °C and a lamination time of 10 minutes. A polyolefin elastomer film was employed as the encapsulation material, while the edges were sealed with butyl rubber to ensure a robust and moisture-resistant barrier.

Long-term operating stability tests were evaluated according to the protocol of ISOS-L-1, which involves maximum-power-point tracking under continuous light under continuous LED illumination. For the long-term stability test, a custom-built, multi-channel MPP tracking system with high-precision source meters (current resolution of 1 μA; 5 voltage resolution of 1 mV) was employed to monitor MPP of multiple solar cells simultaneously. The ambient temperature within the chamber was continuously monitored using an infrared thermometer during device operation. The LED’s intensity was calibrated by adjusting the power to generate a 1-sun-equivalent current density in the device. A silicon photodiode was used to monitor the power change of the lamp, and no apparent changes were observed during MPP tracking. Additionally, the stability tests were further assessed using the protocol of ISOS-L-3, which includes heating and humidity conditions (85% RH at 85 °C), with the device illuminated by LED light.

Transient ion drift (TID) tests

TID test is performed with a LCR meter (Keysight E4980A) and a customized heating stage. The frequency for test is 40 kHZ with an AC voltage of 20 mV. TID measurements measure the capacitance of PSCs. The capacitance of PSCs can be described through the equation.

$$C=A\sqrt{\frac{e{\varepsilon }_{0}\varepsilon (N\pm {N}_{{{\rm{ion}}}})}{2({V}_{{{\rm{bi}}}}-V)}}$$

(1)

where N_ion is the concentration of the mobile ions; N is the doping density of perovskites; V is the external voltage. Here, a forward bias of 0.8 V for some time is applied onto PSCs to drive the accumulated ions at the interface back to the bulk of the perovskite films. A bias of 0 V for some time is applied onto PSCs to drive the mobile ions in the bulk perovskite to accumulate at the interface of perovskite layers and charge-transporting layers.

The energy barrier for (E_a) ion migration is extracted based on the Arrhenius equation (reference: Li, B. et al. The influence of A-site dipole moment on iodine migration in perovskite films revealed by TID³³.

$${{\mathrm{ln}}}\left(\frac{T}{\tau }\right)={{\mathrm{ln}}}\left(\frac{{e}^{2}N{D}_{0}}{{\varepsilon }_{0}\varepsilon {k}_{{{\rm{B}}}}}\right)-\frac{{E}_{{{\rm{a}}}}}{{k}_{{{\rm{B}}}}T}$$

(2)

where D₀ is the pre-exponential factor. k_B is the Boltzmann constant. T is temperature. τ is the time constant for the capacitance transient, which is extracted by fitting the capacitance transient with the equation (reference: Li, B. et al. The influence of A-site dipole moment on iodine migration in perovskite films revealed by TID³³.

$$\Delta C(t)={\Delta C}_{\infty }\exp (-\frac{t}{\tau })$$

(3)

where ∆C_∞ equals C(₀)–C_R. C(₀) is the capacitance when t = 0 while C_R is the capacitance when it reaches equilibrium. t is the time. Then, an Arrhenius curve can be plotted using the τ at different temperatures. The slope of the Arrhenius curve gives the E_a of mobile ions, while the intercept of the curve gives D₀. Moreover, the density of mobile ions can be calculated through the equation (reference: Li, B. et al.³³. The influence of A-site dipole moment on iodine migration in perovskite films revealed by TID.)

$${N}_{{{\rm{ion}}}}\cong 2N\left(\frac{{\Delta C}_{\infty }}{{C}_{{{\rm{R}}}}}\right)$$

(4)

TAS measurements.

The AC frequency scanning is from 20 Hz to 1000 KHz with a δV of 20 mV. Temperature-dependent capacitance measurements were carried out via a customized temperature-controlling stage (WUHAN CONGTICAL TECHNOLOGY Co., LTD).

According to the theory of TAS, the frequency at the peak (ω₀) is related to emission rates of a trap state as described in

$${\omega }_{0}={2v}_{0}\exp (-\frac{{E}_{{{\rm{d}}}}}{{k}_{{{\rm{B}}}}T})$$

(5)

where v₀ is the attempt-to-escape frequency, T is temperature, k_B is Boltzmann constant. E_d is the energy depth of a trap state. We note that v₀ can be described by ξ₀T², where ξ₀ is a temperature-independent constant. Based on Eq. (5), an Arrhenius equation can be derived as described in

$${{\mathrm{ln}}}\left(\frac{{\omega }_{0}}{{T}^{2}}\right)={{\mathrm{ln}}}(2\xi _{0})-\frac{{E}_{{{\rm{d}}}}}{{k}_{{{\rm{B}}}}T}$$

(6)

Therefore, we can calculate v₀ based on the intercept of the linear fitting line of the ln(ω₀/T²) – 1/T plot. Based on Eq. (5), a demarcation energy (E_ω) can be defined

$${E}_{{{\rm{\omega }}}}={k}_{{{\rm{B}}}}T{\mathrm{ln}}\left(\frac{{2v}_{0}}{\omega }\right)$$

(7)

In addition, the density states of the trap state (tDOS) are described by

$${{\rm{tDOS}}}=-\frac{{V}_{{{\rm{bi}}}}}{{Aq}{k}_{{{\rm{B}}}}{Td}}\frac{{{\rm{d}}}C}{{{\rm{dln}}}(\omega )}$$

(8)

where A is the device area, d is the thickness of perovskites, which is ~700 nm for the fabricated films, V_bi is the built-in potential of the device. Then, the peak in the tDOS plot is integrated to get the density of defects (N_t).

Computational details

All of the calculations are performed in the framework of the spin-polarized density functional theory with the projector augmented plane-wave method, as implemented in the Vienna ab initio simulation package^34,35. The generalized gradient approximation proposed by Perdew, Burke, and Ernzerhof is selected for the exchange-correlation potential^36,37. The long-range van der Waals interaction is described by the DFT-D3 approach³⁸. The cut-off energy for the plane wave is set to 450 eV. The energy criterion is set to 10 − 5 eV in the iterative solution of the Kohn-Sham equation. All the structures are relaxed until the residual forces on the atoms have declined to less than 0.02 eV/Å. Data analysis and visualization are carried out with the help of the VASPKIT code and VESTA^39,40. To avoid interlaminar interactions, a vacuum spacing of 20 Å is applied perpendicular to the slab.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.