Highly efficient and stable perovskite photovoltaics enabled by multifunctional crosslinked n⁺-type interlayer

Abstract

Tin dioxide (SnO₂) stands as a premier electron transport layer in n-i-p perovskite solar cells (PSCs), yet interfacial defect-induced carrier recombination and energy-level misalignment impede its commercialization. Herein, we engineer a multifunctional polymeric interlayer by introducing polymerizable dimethyldiallylammonium chloride (DADMAC) at the SnO₂/perovskite interface. The densely cross-linked P-DADMAC network reinforces mechanical interlocking, enhancing interfacial adhesion and stress dissipation. Concurrently, chloride ions (Cl⁻) from P-DADMAC synergistically passivate defects at both the perovskite buried interface and SnO₂ surface, inducing a graded n⁺-type band bending. This energy-band engineering reduces the heterojunction energy-level offset by 0.24 eV, thereby facilitating efficient charge extraction and minimizing non-radiative losses. Consequently, n-i-p devices achieve a power conversion efficiency (PCE) of 26.34% with a fill factor (FF) of 85.84% (certified: 26.27%), while a 25 cm² module (active area: 14 cm²) attains 22.03% PCE (FF > 80%). This work establishes a paradigm for interfacial multifunctionalization in high-performance photovoltaics.

The interface between perovskite and tin dioxide in perovskite solar cells suffers from interfacial defects and energy level mismatch that limit efficiency. Zhang et al. add a polymerized dimethyldiallylammonium chloride interlayer to passivate defects, tune band bending, and enhance charge extraction.

Introduction

In recent years, metal halide perovskite solar cells (PSCs) have emerged as a “rising star” with great potential in the photovoltaic field, owing to their remarkable power conversion efficiency (PCE), low manufacturing cost, and solution processability^1–4. Currently, the certified PCE of PSCs has reached 27%, which is on a par with that of monocrystalline silicon solar cells⁵. However, the industrialization process of PSCs still faces the challenge of excessive efficiency loss, and the interfacial issues between the electron transport layer (ETL) and the perovskite layer are particularly prominent^6–8. Tin dioxide (SnO₂) has been widely employed as an ETL material in n-i-p PSCs due to its high electron mobility, low-temperature processability, and excellent energy level matching^9–14. Nevertheless, the commonly existing oxygen vacancies (V_O) and uncoordinated Sn atoms on the surface of SnO₂ tend to form deep-level defects, thereby increasing the carrier transport barrier at the interface and affecting the charge extraction efficiency^15–17. In addition, defects at the buried interface of PSCs, such as uncoordinated Pb²⁺ and I^- vacancies, with the surface defects of SnO₂, synergistically exacerbate the interfacial energy level mismatch and influence the fill factor (FF) and open-circuit voltage (V_OC) of the devices^18–21.

The application of chlorine-containing (Cl⁻) compounds at the buried interface of PSCs has been widely acknowledged. It not only has unique advantages in passivating oxygen vacancies and enhancing the electrical conductivity of SnO_2, but also Cl⁻ can passivate the defects in perovskite films and improve lattice stability^22–25. However, traditional interfacial modification strategies mostly focus on the optimization of a single interface, making it difficult to simultaneously address the serious problems of the SnO₂/Perovskite heterojunction. Moreover, the introduced organic layers can exacerbate device degradation due to thermodynamic instability. Therefore, how to construct a multifunctional interfacial layer that combines interfacial passivation, energy level regulation, and enhanced interlayer mechanical contact is of particular significance.

To address the issues of carrier recombination induced by interfacial defects and energy level mismatch at the tin dioxide (SnO₂)/perovskite interface, this study proposes the in-situ polymerization of dimethyldiallylammonium chloride (DADMAC) at the interface to construct a densely cross-linked P-DADMAC layer. The polymer network synergistically optimizes the interface through the following three mechanisms: (1) Mechanical interlocking enhancement, the cross-linked structure improves interlayer adhesion (with interfacial shear strength increasing by >400%). (2) Dual-interface passivation, chloride ions (Cl⁻) synchronously bind to defects at both the SnO₂ surface and the buried interface of perovskite, including uncoordinated Sn⁴⁺, uncoordinated Pb²⁺, as well as oxygen vacancies and iodine vacancies, resulting in a significant reduction in defect density. (3) Energy band engineering regulation: It induces the formation of n⁺-type graded band bending, decreasing the heterojunction energy level offset from 0.76 eV to 0.52 eV and significantly lowering the carrier transport barrier. These effects result in an average reduction of the device’s series resistance (R_s) to 1.76 Ω cm² (compared to 4.63 Ω cm² in reference devices). The efficiency of the n-i-p solar cell is enhanced to 26.34% with a FF of 85.84% (certified: 26.27%). A 25 cm² module (with an active area of 14 cm²) achieves an efficiency of 22.03%, and the T₉₀ > 1050 h of unencapsulated storage. This strategy realizes multi-dimensional regulation of mechanics, electricity, and energy bands through a single polymer interface, providing a new paradigm for the design of low-loss transport pathways in perovskite photovoltaics.

Results

Mechanical properties of controllable polymer layers and device performance

The interfacial energy loss caused by the defect at the buried interface of perovskite and SnO₂ affects the performance of the device. To address this, we introduced cross-linkable DADMAC at this interface, and anticipated achieving low interfacial loss and high-performance devices with low dynamic resistance. According to the polymerization kinetics, we determined that DADMAC could undergo the cross-linking polymerization process under the initiation of Azodiisobutymidine hydrochloride (AIBA) to form the polymer (P-DADMAC) as shown in Fig. 1^26,27.

Fig. 1. Polymerization mechanism.

The cross-linking polymerization process of DADMAC under the action of AIBA initiator.

Fourier transform infrared (FT-IR) spectroscopy shows that under the action of AIBA, the stretching vibration peak of C=C (~1636.3 cm⁻¹) and the stretching vibration peak of allylic C-H (~3090.4 cm⁻¹) are significantly weakened. Meanwhile, the stretching vibration peak of saturated C-H (~2977.5 cm⁻¹) becomes significantly broader (Supplementary Fig. 1), which all confirms the thermal cross-linking process of DADMAC. Raman spectroscopy also confirms the cleavage of C=C, thus forming the cross-linked P-DADMAC (Supplementary Fig. 2).

A controlled polymerization process is the guarantee for achieving optimal polymer doping. Gel Permeation Chromatography (GPC) was employed to determine the optimal heating temperature and duration. Given that the theoretical decomposition temperature of AIBA is 70 °C, equal masses of DADMAC and the corresponding AIBA (0.05% by weight of DADMAC) were heated at 60 °C, 70 °C, and 80 °C for 90 s. The GPC results (Supplementary Fig. 3) reveal that the number-average molecular weight (M_n) of P-DADMAC was markedly affected by the polymerization temperature. At 60 °C, the M_n was relatively low. Upon increasing the temperature to 70 °C, the M_n reached a maximum value of approximately 28 kDa, accompanied by a narrower molecular weight distribution. This suggests a well-balanced chain initiation and propagation rate, with effective suppression of chain termination and transfer reactions. However, further increasing the temperature to 80 °C resulted in a decrease in M_n, which can be attributed to enhanced chain termination and chain transfer processes under elevated temperature conditions.

Subsequently, using the optimized temperature of 70 °C, we conducted further investigations by heating a fixed volume of the solution for durations ranging 30–150 s (Fig. 2a). It gradually increased with the extension of heating time and slowed down after 120 s. Subsequently, we compared the adhesion properties of DADMAC and P-DADMAC with different heating durations (marked as 30 s, 60 s, 90 s, 120 s, and 150 s) via tensile shear tests (Fig. 2b and Supplementary Fig. 4). The adhesion strength of P-DADMAC increased with the extension of heating duration and tended to stabilize at 120 s²⁸. Subsequently, we conducted the corresponding performance verification experiments (Fig. 2c). Statistical results indicated that devices modified with P-DADMAC polymerized under heating for 90 s achieved the optimal performance. We further investigated the origin of this discrepancy. First, atomic force microscopy (AFM) was employed to characterize the thickness of P-DADMAC films prepared with different heating polymerization durations (Supplementary Fig. 5). It was found that the film thickness increased with the prolongation of heating polymerization duration, with the thickness of polymer films obtained at 90 s and 120 s being ~21.8 nm and ~31.6 nm, respectively. Subsequently, steady-state photoluminescence (PL) spectroscopy was performed on samples modified with P-DADMAC (prepared with different heating polymerization durations) based on the ITO/SnO₂/P-DADMAC/Perovskite structure (Supplementary Fig. 6). The results showed that the sample with 90 s heating polymerization exhibited the fastest interfacial charge extraction (compared to the 0 s sample). However, the interfacial charge transfer was slightly hindered in the 120 s sample, and the 150 s sample showed significantly impeded charge transfer (compared to the 90 s sample). These findings were consistent with the statistical results of device performance. Therefore, we concluded that an excessively thick P-DADMAC film would hinder the interfacial electron extraction and transfer, thereby deteriorating the device performance. The optimal thickness of P-DADMAC was determined to be ~21.8 nm (90 s heating polymerization), and all subsequent experiments were conducted under this condition. We further explored the mechanical properties of P-DADMAC using AFM force curves. Then, we conducted AFM force curve tests on DADMAC and P-DADMAC on the ITO substrate, respectively (Fig. 2d, e), and the schematic diagram is shown in Supplementary Fig. 7. Through analysis, we found that when the probe was moving away from the material, the adsorption force of DADMAC was 0.061 μN, which was significantly smaller than that of P-DADMAC (0.074 μN). At the same time, the deformation of the probe increased from 4.2 to 7.2 nm, indicating that the attractive force between P-DADMAC and the probe was relatively stronger. Meanwhile, P-DADMAC had relatively lower hardness and obvious viscoelasticity. We further calculated the work of adhesion and found that the work of adhesion of P-DADMAC (0.990 pJ) to the probe was more than five times higher than that of DADMAC (0.185 pJ). We further conducted shear tensile tests on the SnO₂/P-DADMAC/perovskite interface (Supplementary Fig. 8). P-DADMAC can effectively enhance the mechanical interlocking and adhesive state at this interface, which is attributable to its interfacially synergistic chemical bonding capability and the viscoelasticity of the polymer. In summary, the superior mechanical properties of P-DADMAC can enhance the mechanical stability of the SnO₂/perovskite interface.

Fig. 2. Controllable crosslinking of DADMAC and its impact on devices.

a Number-average molecular weight (M_n) of P-DADMAC under different heating times. b Statistical data of shear strength for DADMAC and P-DADMAC prepared under different heating times. c Statistical performance data of perovskite devices modified with DADMAC and P-DADMAC prepared under different heating times, data are presented as mean values ± 95% confidence intervals (CI), with the dashed line showing the trend of PCE evolution, with a sample size of 8. AFM force curve of d DADMAC and e P-DADMAC with a polymerization time of 90 s. f Reverse J-V curves of champion devices with reference and P-DADMAC modification. g The relationship between FF and material band gap, including the critical boundaries of S-Q limit, 95% S-Q limit, and 90% S-Q limit, and the FF of control and P-DADMAC modified devices. h Operational stability of reference and P-DADMAC-modified devices under ISOS-L-1I protocol, including the initial PCE of each cell, showing both actual measured time and linearly extrapolated prediction.

Subsequently, we investigated the impact of P-DADMAC-induced improvement of interfacial contact on device performance. After optimization, it was determined that the device exhibited the best performance when 0.7 mg/mL DADMAC was heated at 70 °C for 90 s under the action of AIBA (Supplementary Fig. 9). Meanwhile, we verified that the decomposition products of AIBA and verified that neither AIBA nor its decomposition products have any adverse impact on the performance and stability of the device. (Supplementary Figs. 10–12). Compared with the reference device, the PCE of the P-DADMAC-modified device reaches 26.34% (Fig. 2f and Supplementary Table 1), with a reduced hysteresis effect (Supplementary Fig. 13). Notably, the FF reached an impressive 85.84%, approaching 95% of the Shockley-Queisser (S-Q) limit (Fig. 1g). We conducted a comprehensive statistical comparison of high PCE and FF of n-i-p PSCs reported in recent years (Supplementary Fig. 14 and Supplementary Table 2), demonstrating that both the device performance and FF in this work rank among the state-of-the-art. Furthermore, the P-DADMAC modified device achieved a certified PCE of 26.27% and a FF of 84.35% (Supplementary Fig. 15). The EQE spectrum showed better consistency between the integrated current density and current density (J_SC) (Supplementary Fig. 16). Following the ISOS-L-1I protocol, we performed maximum power point tracking (MPPT) tests (Fig. 2h). After 1002 h of continuous illumination, the control device degraded to 51.8% of its initial efficiency, while the P-DADMAC modified device retained 95.03% of its initial PCE. Linear extrapolation predicted a T₉₀ lifetime exceeding 2400 h, representing a significant advancement in operational stability. Furthermore, under the ISOS-D-3 protocol with damp heat conditions (65 °C, 85% relative humidity; Supplementary Fig. 17), the P-DADMAC modified devices exhibited exceptional stability against humidity and thermal stress.

Interface synergistic passivation strategy

The FF is directly related to the internal resistance of the device (series resistance R_s and shunt resistance R_sh), carrier transport characteristics, and interfacial properties. To explore the mechanism by which P-DADMAC improves the FF, we tested samples of different structures using photoluminescence quantum yield (PLQY) (Fig. 3a). The PLQY of P-DADMAC/PVSK (abbreviation for perovskite) is significantly higher than that of pure PVSK. We speculate that this may benefit from the passivation of the buried interface defects by P-DADMAC. Compared with SnO₂/PVSK, the increase in the PLQY value of SnO₂/P-DADMAC/PVSK represents an enhancement in the electron transport efficiency at the interface. According to the PLQY at this interface, the energy loss was quantified (Fig. 3b)^29,30. It can be found that the energy loss at this interface of the device modified by P-DADMAC decreased from 0.102 to 0.083 eV, and the voltage loss of the entire device was lower. We speculate that the improvement of FF comes from two aspects. On the one hand, the P-DADMAC polymer effectively enhances the interlayer contact at the interface; on the other hand, the Cl⁻ from P-DADMAC passivates the interfacial defects, effectively reducing the energy loss at the interface.

Fig. 3. Interface energy loss and characterization of synergistic passivation.

a Photo Luminescence Quantum Yield (PLQY) comparison of different structures and b corresponding interfacial energy loss analysis. c SEM images and d Cl signal detection by SEM-EDS of the buried interface of the reference and P-DADMAC modified perovskites. e Bonding state information between DADMAC and the uncoordinated Pb²⁺ on the surface of perovskite, as well as the binding energy information with the V_I-vacancy, including the distribution of differential charge density. f Bonding state information between DADMAC and the uncoordinated Sn⁴⁺ on the surface of SnO₂, as well as the binding energy information with the V_O-vacancy, including the distribution of differential charge density. g XPS spectra of Pb 4f at the buried interface of the reference and P-DADMAC modified perovskites, where the dashed line corresponds to the XPS characteristic peak position under reference conditions. h The magnitude of the relative stress borne by the buried interface of the reference and P-DADMAC modified perovskites. i XPS spectra of O 1s on the surface of SnO₂ covered by the reference and P-DADMAC modified perovskite films, where the dashed line corresponds to the XPS characteristic peak position under reference conditions. j XPS spectra of Sn 3d on the surface of SnO₂ covered by the reference and P-DADMAC modified perovskite films, where the dashed line corresponds to the XPS characteristic peak position under reference conditions.

To verify our conjecture, we explored the buried interface between perovskite and SnO₂. We first obtained the intact buried interface of the perovskite layer without damage using ultraviolet (UV)-curable adhesive (Supplementary Fig. 18). Scanning electron microscopy (SEM) imaging and X-ray diffraction (XRD) of the buried interface revealed that the introduction of P-DADMAC effectively suppressed excessive lead iodide (PbI₂) crystal formation while reducing grain boundary grooves, resulting in a denser and more compact crystalline morphology (Fig. 3c and Supplementary Fig. 19). A similar improvement was observed in the top-surface SEM characterization, with statistical analysis showing an increase in average grain size from 717.4 nm to 972.1 nm (Supplementary Fig. 20). To further investigate the elemental distribution, we performed SEM-energy dispersive spectroscopy (SEM-EDS) mapping measurements (Fig. 3d), which clearly demonstrated Cl⁻ enrichment at the P-DADMAC-modified perovskite buried interface.

We hypothesize that Cl⁻ can passivate defects at this interface. First, we verified this viewpoint through density functional theory (DFT) calculations (Fig. 3e, f). Cl⁻ can form Pb-Cl bonds with uncoordinated Pb²⁺ ions, and the Integrated Crystal Orbital Hamilton Population (ICOHP) is −2.322. Meanwhile, the binding energy between Cl⁻ and the iodine vacancy is −1.395 eV. From the differential charge density, it can be observed that the amount of electron transfer from the Cl⁻ of DADMAC to the perovskite is 0.181 e⁻ (Fig. 3e). Although dimethyldiallylammonium halides with heavier anions (Br⁻ and I⁻) are not yet commercially available, their potential to suppress V_I defects on FAPbI₃ and V_O defects on SnO₂ can still be assessed computationally. By comparing I⁻ and Br⁻, we found that larger halide ions elongate and weaken the metal-halide bonds, which is evidenced by longer bond lengths and lower ICOHP values (Supplementary Fig. 21 and Supplementary Table 3). The more diffuse electron clouds of Br⁻ and I⁻ enable their polymers to donate higher electron density to FAPbI₃³¹. For the SnO₂ side, the binding energy between DADMAC and the V_O on the surface of SnO₂ reaches −2.179 eV, and the ICOHP of the tin-chlorine (Sn-Cl) bond is −5.030. The differential charge density shows that the amount of charge transfer is 0.327 e⁻, which demonstrates the passivation effect of Cl⁻ (Fig. 3f). We compared the interactions of Br⁻ and I⁻ based on dimethyldiallylammonium halide with SnO₂ (Supplementary Fig. 22 and Supplementary Table 3). On the more rigid SnO₂ framework, the interaction gradually weakens down the halogen group (from top to bottom). Additionally, the defect formation energy of V_I on FAPbI₃ was then revealed to be 2.946, 2.673, and 1.674 eV for Cl, Br, and I as an anion of P‑DADMA halide; while the reference value of pristine perovskite surface was calculated as 1.268 eV (Supplementary Fig. 23). Hence, the trend suggests that P-DADMAC is the most effective species for passivating both perovskite and SnO₂.

X-ray photoelectron spectroscopy (XPS) analysis of the Cl 2p and N 1 s core levels (Supplementary Fig. 24a, b) confirmed the presence of DADMAC within the perovskite film. Through the XPS spectrum of the Pb 4f orbital (Fig. 3g), we found that the peak position shifted towards the lower binding energy, and the electron cloud density around the Pb atom increased, indicating that Cl⁻ can passivate the uncoordinated Pb²⁺ by forming Pb-Cl bonds. To confirm that this interaction is indeed distributed in the buried interface region of the perovskite, we prepared a perovskite film under MACl-free conditions and conducted depth-profiling XPS analysis (Supplementary Fig. 25). Compared with the bulk phase, the characteristic peak of the Pb 4f orbital at the buried interface shows a significant shift, confirming that this interaction is mainly concentrated at the buried interface. The peak shift of the I 3d orbital indicates that Cl⁻ in P-DADMAC can effectively fill the shallow I⁻ vacancy defects on the surface, and at the same time, the N cations on the polymer main chain can have electrostatic interactions with I, thereby maintaining and stabilizing the lattice configuration (Supplementary Fig. 24c), consistent with the DFT calculations. These findings validate our hypothesis that Cl⁻ effectively passivates defects at the perovskite buried interface. To better observe the improvement in the quality of the perovskite buried interface, we conducted PL mapping tests on the perovskite buried interface obtained from the ITO/P-DADMAC/Perovskite samples (Supplementary Fig. 26). Compared with the reference film, the P-DADMAC-modified film exhibits a more uniform and stronger fluorescence distribution, which indicates fewer overall defects and higher quality at the buried interface.

Due to the difference in the thermal expansion coefficients between the substrate and the perovskite film, a huge tensile strain is imposed on the film at the initial stage of its growth, which affects the performance and stability of the device. We analyzed the strain state and stress magnitude of the buried interface of the perovskite film with and without the P-DADMAC interfacial layer through grazing incidence X-ray diffraction (GIXRD) measurements (Supplementary Fig. 27). We found that as the tilt angle of the instrument increased, the (001) crystal plane at the buried interface of the reference perovskite exhibited a shift toward lower angles, accompanied by the splitting of diffraction peaks. This indicates that the perovskite crystals have relatively severe tensile strain and show a tendency to grow in different sub-grain orientations related to the (001) crystal plane. After modification with P-DADMAC, the characteristic peak shifted toward higher angles, which demonstrates that P-DADMAC can effectively relieve the tensile strain at the buried interface and convert it into a slight compressive strain (Supplementary Fig. 27a–c)³². We further analyzed the change in interplanar spacing (Supplementary Fig. 27d) and found that for the buried interface modified by P-DADMAC, the interplanar spacing decreased with increasing depth—this signifies significant compression of the perovskite crystals. Supplementary Fig. 27e, f and Fig. 3 show the relative strain degree and relative stress magnitude of the interface with and without the P-DADMAC interfacial layer, respectively. Through calculation and analysis, it is known that the pristine perovskite film endured a relative tensile stress of 44.9 MPa, while the perovskite film modified by P-DADMAC endured a relative compressive stress of 19.3 MPa. The tensile stress at the buried interface of the reference perovskite film arises from the mismatch in thermal expansion coefficients between perovskite and SnO₂ and accumulates at this interface region. The incorporation of the P-DADMAC layer enables efficient transfer of the interfacial tensile stress to the polymer layer via strong interfacial adhesion, thereby suppressing its accumulation within the perovskite crystal lattice. Meanwhile, the flexible polymer chains of P-DADMAC can dissipate a portion of the stress through chain slipping, elongation and network deformation. This synergistic stress modulation mechanism ultimately realizes the stress inversion at the buried perovskite/SnO₂ interface. Slight compressive strain is of positive significance for the phase stability of perovskite, which can effectively improve the performance and stability of the device.

The abundant oxygen vacancies (V_O) on the surface of SnO₂ act as non-radiative recombination centers for carriers. To address this issue, we analyzed the ratio of lattice oxygen (Sn-O) to V_O (I_Sn-O/I_VO) on the SnO₂ surface via X-ray photoelectron spectroscopy (XPS). After treatment with P-DADMAC, the (I_Sn-O/I_VO ratio increased from 3.5704 to 6.4555. This result indicates that Cl⁻ effectively reduces the content of V_O through the formation of Sn-Cl bonds (Fig. 3i), thereby significantly suppressing the generation of defect-induced non-radiative recombination centers³³. Electron paramagnetic resonance (EPR) measurements further confirmed that Cl⁻ ions can efficiently occupy the V_O in SnO₂ (Supplementary Fig. 28)³⁴. Additionally, a distinct signal of quaternary ammonium N⁺ was detected on the SnO₂ surface, which is derived from the polymer backbone (Supplementary Fig. 29). We also observed a shift of lattice oxygen to higher binding energy (0.11 eV), implying a reduction in the electron density around O atoms. Thus, we propose that N⁺ on the polymer backbone can interact with lattice oxygen through long-range electrostatic interactions and inductive effects (Fig. 3i). DFT calculations were performed to verify this interaction. The results revealed a decrease in the electron density of O atoms on the SnO₂ surface, with electrons transferring to the polymer backbone (Supplementary Fig. 30). Subsequently, conductivity measurements demonstrated an enhancement in the electrical conductivity of SnO₂ after P-DADMAC modification, which further corroborates the aforementioned conclusions (Supplementary Fig. 31).

Dynamics at the Interface

When the crystal lattice is subjected to tensile strain, the bond lengths between the internal atoms will change, and the atomic arrangement will change accordingly, which can affect the morphology and surface potential of the perovskite at the buried interface. To verify the influence of the improvement of the tensile strain on the morphology and surface potential of the perovskite buried interface, we conducted a combined atomic force microscope-Kelvin probe force microscope (AFM-KPFM) test on the buried interface (Fig. 4a, b). We observed that the root-mean-square (RMS) roughness of the buried interface of the P-DADMAC-modified perovskite film increased to 32.0 nm (compared with 27.3 nm for the reference sample). This change stems from the nanoscale protrusions and voids on the surface of the cross-linked polymer, where the protrusions act as heterogeneous nucleation sites to reduce the activation energy for perovskite crystal formation, and the voids slow down the lateral diffusion of perovskite precursor molecules, thus facilitating the ordered arrangement of the crystal lattice. Meanwhile, the corresponding surface contact potential difference (CPD) decreased from 32.1 to 13.7 mV, which decrease in CPD is attributed to the effective passivation of the V_I and uncoordinated Pb²⁺ at the buried interface by Cl⁻, which in turn improves the strain state and electrical properties of the buried interface and enhances the optoelectronic and mechanical stability of the buried interface.

Fig. 4. Impact of P-DADMAC on carrier transport at the SnO₂/perovskite interface.

a AFM images and b KPFM images of the buried interface of the reference and P-DADMAC modified perovskites. c PL spectra of the reference and P-DADMAC modified perovskite films on the glass substrate, with a schematic diagram of the sample structure above. d PL spectra of the reference and P-DADMAC modified perovskite films on the glass/ITO/SnO₂ substrate, with a schematic diagram of the sample structure above. e PL spectra of the reference perovskite film on the glass/ITO/SnO₂ substrate for different incident surfaces, with a schematic diagram of the sample structure above. f PL spectra of the P-DADMAC modified perovskite film on the glass/ITO/SnO₂ substrate for different incident surfaces, with a schematic diagram of the sample structure above. g TRPL spectra of samples with and without P-DADMAC modification based on the ITO/perovskite structure, the laser is incident on the film surface. h TRPL spectra of samples with and without P-DADMAC modification based on the ITO/SnO₂/perovskite structure, the laser is incident on the film surface.

We analyzed the influence of P-DADMAC on carrier transport and non-radiative recombination in the bulk phase and at the interface through steady-state PL spectroscopy. Firstly, we conducted PL tests on the upper surface of the perovskite film with and without the P-DADMAC interfacial layer on a pure glass substrate (Fig. 4c, the schematic diagram of the sample structure is shown above). The PL peak intensity of the perovskite film modified by P-DADMAC increased significantly, which can be attributed to the improvement in the film quality. Subsequently, we carried out PL tests on the top surface of the perovskite film with/without the P-DADMAC interfacial layer on the ITO/SnO₂ substrate (Fig. 4d, the schematic diagram of the sample structure is shown above). The sample modified by P-DADMAC exhibited obvious PL quenching, which indicates that the introduction of P-DADMAC has improved the extraction of carriers at the SnO₂/Perovskite interface³⁵.

We further conducted steady-state PL tests in two different directions, namely from the glass substrate side and the perovskite film side (Fig. 4e, f, and the sample structure diagrams are located above the corresponding PL spectra). The reference film showed inconsistent intensities (the intensity on the glass side was lower than that on the film side), which can be attributed to the synergistic effect of shallow defects at the perovskite buried interface and SnO₂ surface defects, resulting in relatively serious interfacial non-radiative recombination losses at the buried interface. For the film modified by P-DADMAC, the intensities of the PL emission peaks on both the glass side and the film side were negligible, and the PL on the glass side exhibited a slight blue shift, indicating the passivation of P-DADMAC on the defects at the SnO₂/Perovskite interface, which would have a positive impact on the improvement of the FF³⁶. We also confirmed the passivation effect of P-DADMAC on the buried interface of perovskite and the enhancement of interfacial carrier transport through Time-Resolved Photoluminescence (TRPL) spectra of different device structures (Fig. 4g, h and Supplementary Table 4). We analyzed the degree of current leakage induced by internal defects through dark current tests (Supplementary Fig. 32). The dark current of the device modified by P-DADMAC was nearly two orders of magnitude lower than that of the control device, indicating a reduction in non-radiative recombination losses, which is beneficial for improving the V_OC and FF of the device.

Intrinsic mechanism for the enhancement of fill factor

To further confirm this result, we applied the space charge limited current (SCLC) method to devices with only electrons (ITO/SnO₂/Perovskite/PCBM/Ag) to evaluate the density of defect states (N_trap) in the perovskite film (Supplementary Fig. 33). Through calculation and statistics (Equation 1, Supporting Information), we found that the N_trap of the electron-only device decreased from 9.46 × 10¹⁵ to 7.68 × 10¹⁵ cm⁻³.

We further analyzed the changes in the bandgap and energy levels at the SnO₂/Perovskite interface through ultraviolet-visible absorption spectroscopy (UV-Vis) and ultraviolet photoelectron spectroscopy (UPS). First, we tested the ultraviolet absorption spectrum and UPS spectrum of P-DADMAC with the optimal polymerization time, and determined its band gap, HOMO, and LUMO energy level information (Supplementary Fig. 34a). After modification with the P-DADMAC interfacial layer, there was no obvious shift in the bandgap of the SnO₂ surface (3.705 eV) (Supplementary Fig. 35). However, a significant shift occurred in the bandgap at the buried interface of the perovskite film, changing from 1.551 to 1.564 eV (Supplementary Fig. 37), which is attributed to the migration of Cl⁻ to the perovskite film, leading to the change in the bandgap of the perovskite film^22,37,38. Cl⁻ do not migrate into the film bulk or toward the top under stimuli such as light irradiation, thus causing no adverse effects on device performance and long-term stability. This is ascribed to the dual confinement effects from the electrostatic binding of the polymer backbone and the effective physical barrier of the cross-linked polymer network (Supplementary Fig. 38). Based on the results of UPS (Supplementary Figs. 34b, 36, and 39), we plotted the energy band diagram (Fig. 5a). The degree of energy band bending of the control device was limited, and there was an interfacial electron transport barrier of 0.76 eV between the conduction band minimum (CBM) of SnO₂ and that of the perovskite. After modification with P-DADMAC, the work function of the buried interface of perovskite is reduced compared with that of the bulk phase. This benefit stems from the passivation of defects at the perovskite buried interface by Cl⁻, which eliminates the Fermi level pinning effect induced by deep-level interface defects. The microscopic manifestation is the variation in the energy band relative to the bulk phase, exhibiting more pronounced n-type semiconductor characteristics. Correspondingly, the CBM of SnO₂ increased from −4.49 to −4.32 eV, and the valence band maximum (VBM) increased from −8.21 to −8.04 eV; the CBM of the perovskite buried interface decreased from −3.73 to −3.81 eV, and the VBM decreased from −5.28 to −5.36 eV. The highest occupied molecular orbital (HOMO, −4.04 eV) and lowest unoccupied molecular orbital (LUMO, −7.29 eV) of P-DADMAC lie between those of SnO₂ and perovskite, enabling energy level-assisted alignment at the interface and reducing the charge transport barrier. This improvement brought by P-DADMAC allows carriers to cross the interface more efficiently and migrate toward the charge transport layers (CTLs), which is not only beneficial for improving the FF but also plays a positive role in promoting the enhancement of device performance^39,40.

Fig. 5. Energy level structure at the interface and FF loss analysis.

a Schematic diagrams of the energy levels of the reference device and the device modified by P-DADMAC. b Potential curve at the interface between perovskite and SnO₂ without P-DADMAC modification. c Potential curve at the interface between perovskite and SnO₂ with P-DADMAC modification. The insets are the corresponding KPFM images, where PVSK refers to perovskite. d Light intensity-dependent open-circuit voltage curves of the reference and P-DADMAC modified devices. e Distribution diagram of the FF loss of the reference and P-DADMAC modified devices. f Statistical comparison diagram of the series resistance (R_s) and shunt resistance (R_sh) corresponding to the J-V curves of a total of 34 reference and P-DADMAC modified devices. g Tangent analysis near the maximum power point voltage, corresponding to the internal dynamic resistance of the device. h Comparison diagram of pseudo J-V and measured J-V Curves. The inset in the figure shows the shaded area, corresponding to the MPP region, to better observe the improvement in fill factor. i Comparison chart of theoretical fill factor and PCE.

To deeply investigate the effect of P-DADMAC interface modification on the surface potential of the perovskite/SnO₂ heterojunction, we characterized the surface potential distribution of the interface regions with and without P-DADMAC modification via KPFM (Fig. 5b, c). It can be seen that the surface potential difference between perovskite and SnO₂ is 376.6 mV, and the surface potential difference between perovskite and SnO₂ decreases to 252.2 mV after P-DADMAC modification. Meanwhile, the surface potential fluctuation in the perovskite region is reduced, and the potential transition at the interface is smoother. This indicates that the introduction of P-DADMAC effectively regulates the interfacial electronic structure, making the interfacial energy levels more matched, realizing better assisted energy level alignment, and thus promoting efficient charge transport.

To gain a deeper understanding of the reasons for the increase in FF from the perspective of device physics, we measured the change in V_OC with the light intensity (Fig. 5d). The ideality factor was used to evaluate the defect-assisted Shockley-Read-Hall (SRH) single-molecule recombination state of the device⁴¹. Through calculation using Eq. 3 (Supporting Information), the ideality factor of the control device was 1.51, and that of the device modified by P-DADMAC (1.04) was closer to 1, which indicates that P-DADMAC suppresses the defect-assisted SRH single-molecule recombination. The defect-assisted non-radiative recombination loss and charge transfer loss are the main reasons for a lower FF than the S-Q limit. Based on the actually measured FF (FF_measure), we calculated the maximum FF (FF_max) using Eqs. 4 and 5 (Supporting Information), which were 85.51% (Control) and 89.28% (With P-DADMAC), respectively (Fig. 5e). The differences between FF_max and FF_S-Q limit were 4.69 (Control) and 0.92 (With P-DADMAC), respectively, and the differences between FF_measure and FF_max were 6.96 (Control) and 3.45 (With P-DADMAC), respectively. Smaller differences indicate that the introduction of P-DADMAC will significantly suppress the defect-assisted non-radiative recombination loss, thereby reducing the carrier transport loss at the interface^42,43.

As is known, the R_s and R_sh can directly reflect the charge transport in the device. We conducted statistics and found that compared with the control device, the R_s of the device modified by P-DADMAC was smaller and the R_sh was larger (Fig. 5f). We further analyzed the position of V_MPP on the current-voltage (J-V) curves of the control and the device modified by P-DADMAC (Fig. 5g). The slope of the tangent line of the J-V curve at the corresponding position of V_MPP reflects the internal dynamic resistance of the device at this voltage. The dynamic resistance is closely related to the FF, and a smaller dynamic resistance will reduce the energy loss inside the cell, which is conducive to improving the FF. After calculation, the slope of the tangent line decreased from 0.7656 (Control) to 0.4946 (With P-DADMAC).

We calculated the pseudo J-V curves and compared them with the actually measured J-V curves (Fig. 5h). The pseudo J-V curve represents the ideal J-V curve of the device under the state without transmission loss (only reflection loss and heat loss). Its J_SC is determined by 95% J_S-Q (1.55 eV)⁴⁴, V_OC is V_QFLS (obtained in Fig. 3b), and FF is FF_max (Eqs. 4 and 5, Supporting Information). Statistical results showed (Fig. 5i) that the theoretical PCE of the P-DADMAC modified device without transmission loss was 28.90% (27.51% for the reference device).

Scalability of the strategy

To verify the scalability of this strategy in large-scale production, we fabricated a 25 cm² perovskite module (active area = 14 cm²). Figure 6a, b shows the device structure diagram and the photo of the perovskite module, respectively. After optimization with P-DADMAC, the PCE of the module increased from 19.98% (V_OC = 11.04 V, J_SC = 2.36 mA/cm², FF = 76.70%) to 22.03% (V_OC = 11.43 V, J_SC = 2.40 mA/cm², FF = 80.32%), mainly manifested in the increase of V_OC and FF, which is consistent with the trend of the miniature devices, demonstrating the feasibility and commercial value of this strategy in device expansion (Fig. 6c). We conducted stability tests on unencapsulated control and P-DADMAC modified modules under the ISOS-D-1 protocol (T = 25 °C, RH ≈ 20%) (Fig. 6d). The P-DADMAC modified module maintained 95.4% of its initial efficiency after 482 h, and the linearly extrapolated predicted T₉₀ exceeded 1050 h. In contrast, the control module had decayed to 73.3% of its initial efficiency after 482 h.

Fig. 6. Performance and stability of large-area modules.

a Schematic diagram of the device structure of the large-area module. b Outdoor display photo of the 25 cm² module, with an effective area of 14 cm². c J-V curves of the reference and the best 25 cm² module modified by P-DADMAC, with an effective area of 14 cm². d ISOS-D-1 stability of the reference and P-DADMAC modified modules, including the initial PCE of each cell.

Discussion

In this work, we introduced polymerizable DADMAC to the SnO₂/perovskite interface. This implementation enhanced the interfacial contact and mechanical stability, simultaneously passivated the defects at the buried interface of perovskite and the surface of SnO₂, and reduced the energy loss at the interface. An auxiliary energy band structure with an energy gradient was formed on both sides of the interface, effectively enhancing the interlayer electrical matching, reducing the series resistance, and improving the carrier injection efficiency. Eventually, based on the n-i-p type devices, a PCE as high as 26.34% was achieved, with a FF reaching 85.84% (certified: 26.27%). For the fabricated 25 cm² module (active area = 14 cm²), the FF exceeded 80% and the efficiency exceeded 22%. This interfacial strategy is of great significance for the development of devices with low interfacial energy loss and the large-scale preparation of perovskite devices.

Methods

Materials

All materials used in the experiment were purchased from the drug supplier and were used directly without any further processing. The ITO and FTO substrates were purchased from Advanced Election Technology Co., Ltd. SnO₂ (15% aqueous colloidal dispersion, Alfa Aesar), Lead (II) iodide (PbI₂, 99.99%, TCI), Cesium iodide (CsI, >99.0%, TCI), Lead (II) bromide (PbBr₂, >98%, TCI) were purchased from Beijing Innochem Science & Technology Co., Ltd. Formamidine hydroiodide (FAI, 99.99%), Methylamine chloride (MACl), Methylammonium iodide (MAI), 4-Methoxyphenethylammonium Iodide (MeO-PEAI), and 2,2’,7,7’-tetrakis[N,N-di(4-methoxyphenyl)amino]-9,9’-spirobifluorene (Spiro-OMeTAD, >99.2%) were purchased from Advanced Election Technology Co., Ltd. Dimethyldiallylammonium chloride (DADMAC, 60% in water), Azodiisobutymidine hydrochloride (AIBA) was purchased from Beijing Innochem Science & Technology Co., Ltd. All reagents were used as purchased samples without further purification. Isopropanol (IPA, 99.8%, Acros), N, N-dimethylformamide (DMF, 99.8%, Acros), ethyl acetate (EA, 99.9%, Acros), chlorobenzene (CB, ≥99.5%, Aladdin), acetonitrile (CAN, 99.9%, Acros), dimethyl sulfoxide (DMSO, >99.7%, Acros), Lithium bis (trifluoromethanesulphonyl) imide (Li-TFSI, 99.95% trace metals basis, Sigma-Aldrich), and 4-(tert-butyl)picolinonitrile (4-tBP, ≥95%, Aladdin) were all purchased from Beijing Innochem Science & Technology Co., Ltd. PCBM was purchased from Xi’an Yuri Solar Co., Ltd.

Device fabrication

Wash the FTO glass substrate with acetone, detergent, deionized water, and isopropanol through ultrasonic treatment for 20 min, then dry and use. Then treat the substrate in a plasma cleaning machine for 20 min. A dense TiO₂ layer is deposited on the substrate by atomic layer deposition (ALD), and then sintered in ambient air at 500 °C for 30 min. Tin (IV) oxide was diluted to 2.67 wt% with deionized water and stirred for 10 min. Spin coat the final solution at 3000 rpm for 30 s and anneal in air at 150 °C for 30 min. Subsequently, DADMAC with different concentrations and AIBA at corresponding concentrations, both dissolved in aqueous solution, were spin-coated onto the prepared SnO₂. Then, heating was conducted at 70 °C for 90 s to remove the solvent and promote the cross-linking of DADMAC. A perovskite precursor solution was prepared according to the stoichiometric composition of Cs_0.07FA_0.90MA_0.03Pb(I_0.975Br_0.025)₃. Dissolve 6.4 mg of MAI, 4.2 mg of FABr, 24.6 mg of CsI, 32.0 mg of MACl, 12.4 mg of PbBr₂, 203.2 mg of FAI, and 625.2 mg of PbI₂ in 1 mL of co-solvent of DMF: DMSO (the ratio of volume is 4:1) to the perovskite precursor solution. The precursor solution is stirred thoroughly until dissolved and then filtered through a 0.22 μm PTFE membrane for later use. The perovskite precursor was spin-coated through a two-step process at 1000 rpm for 10 s and 3000 rpm for 30 s. During the second step, 300 μL of EA was dropped on the spinning substrate 10 s before the end of the second step. All perovskite films were annealed at 120 °C for 40 min. Quickly transfer the film into an N₂ glove box. Then, spin-coat 3 mg/mL of MeO-PEAI dissolved in IPA onto the perovskite film to prepare the passivation layer. The Spiro-OMeTAD was prepared by dissolving 72.3 mg of Spiro-OMeTAD in 1 mL of PhCl, to which 28.8 μL of 4-tertbutylpyridine (96%, Sigma-Aldrich) and 17.5 μL of lithium bis(trifluoro-methanesulfonyl) imide (Li-TFSI, 98%, Sigma-Aldrich) solution (520 mg Li-TFSI in 1 mL acetonitrile (99.8%, Sigma-Aldrich) was added. The Spiro-OMeTAD hole-transporting layer was obtained by being spin-coated at 4000 rpm for 20 s, then annealed at 60 °C for 5 min. Finally, vacuum evaporates the metal electrode Au at a thermal evaporation rate of 0.1–0.7 A/s, with a thickness of approximately 80 nm.

Module fabrication

For fabricating modules, the P1 etching process was pre-patterned on FTO glass (5 × 5 cm²) with a 1064 nm fiber laser (Han’s laser). The laser power ratio, laser duty cycle, and laser frequency were 30%, 5%, and 50 kHz, respectively. Then, patterned FTO substrates were cleaned and treated by UV Ozone Cleaner for 15 min. TiO₂ was prepared via atomic layer deposition (ALD), while the SnO₂, DADMAC, and Spiro-OMeTAD layers were all fabricated using the spin-coating method. For the P2 etching process, the laser used was a 532 nm laser with a laser power ratio of 65%, a laser duty cycle of 5%, and a laser frequency of 100 kHz. 80 nm thick Au electrodes were thermally evaporated under a vacuum to complete the module’s fabrication. Finally, P3 etching used the same laser as P2 with a laser power ratio of 50%, a laser duty cycle of 5%, and a laser frequency of 100 kHz. P4 is an etching procedure for cleaning the edges of the modules. The laser used in P4 is the same as P1, with a laser power ratio of 40%, a laser duty cycle 396 of 10%, and a laser frequency of 100 kHz.

Characterization

Film characterization

Fourier transform infrared (FTIR) spectroscopy was characterized by Nicolet iS50 (Thermo Scientific) under purging with N₂. The FTIR measurement samples were prepared by spin coating pure Apronal and Apronal-modified perovskite films before and after heating on the substrate. Scrape the powder from the substrate and laminate using sample: KBr = 1:100. Top-view, cross-sectional SEM images were obtained with a field-emission scanning electron microscope (S-4800, HITACHI). Glancing incident angle X-ray diffraction (GIXRD) was performed on Rigaku SmartLab SE. The XRD pattern was measured by Ultima IV of Rigaku with Cu Kα radiation (1.5406 Å). The UV-vis absorbance was measured by a UV/Vis spectrometer (Ocean Optics). Steady PL was recorded on a fluorometer (Ocean Optics) excited at 460 nm. XPS spectra were performed for films on ITO using a Thermo Fisher ESCALAB 250 Xi. Curve fitting was performed using the Thermo Avantage software. The curves were corrected based on the C1s peak at 284.8 eV. The UPS equipped with a He-I source (hʋ = 21.22 eV) (AXIS ULTRA DLD, Kratos, UK) was used to determine the valence band energy and Fermi level. The Fermi level of the samples was referred to that of Au, which was in electrical contact with a sample in UPS measurements. The external quantum efficiency (EQE) spectra were carried out using a Sciencetech SF150 xenon arc lamp and a PTI monochromator, with the monochromatic light intensity calibrated by a Si photodiode (Newport, 818-UV).

Device performance test

The J-V characteristics of the device were recorded using a solar simulator (PV Measurements Co., Ltd) under AM 1.5 G illumination at 100 mW cm⁻². The devices were measured by masking the devices with a metal mask with an aperture area of 0.09 cm² to measure the J-V curves of all devices. The light intensity of the solar simulator is calibrated using standard silicon solar cells certified by the National Renewable Energy Laboratory. All measurements were conducted in an ambient atmosphere.

Device stability test

ISOS-L-1I

The operational stability was performed using a stability setup (LC Auto-Test 24, Shenzhen Lancheng Technology Co., Ltd), tested under continuous light illumination and maximum power point tracking. All devices were subjected to unencapsulation treatment. The light source consisted of an array of white LEDs (SLS-LED-80, Qingdao Solar Scientific Instrument High-tech Co., Ltd) powered by a constant current. Equivalent sun intensities were calibrated using a calibrated Si-reference cell.

ISOS-D-3

Dark storage stability was evaluated by storing the devices in the specified environment (~65 °C, ~85%RH) and conducting periodic performance tests, All devices were subjected to encapsulation treatment. The light source was a solar simulator (PV Measurements Co., Ltd) with a light intensity of AM 1.5 G illumination at 100 mW cm⁻². Equivalent sun intensities were calibrated using a calibrated Si-reference cell.

ISOS-D-1

Dark storage stability was evaluated by storing the devices and conducting periodic performance tests. All devices were subjected to encapsulation treatment. The light source was a solar simulator (PV Measurements Co., Ltd) with a light intensity of AM 1.5 G illumination at 100 mW cm⁻². Equivalent sun intensities were calibrated using a calibrated Si-reference cell.

Author contributions

Z.Z. supervised the project. J.Z. and Z.Z. proposed the research idea. J.Z. fabricated the PSCs and conducted various characterization measurements. W.Y. provided assistance during the device fabrication process. Z.L. and C.P. fabricated the mini-modules. M.Z. carried out DFT calculations and analyses. H.J. and H. Sun helped to perform Infrared spectroscopy, Raman spectroscopy measurements, and stability tests of perovskite modules. M.L., M.M., and W.W. assisted in the fabrication of PSCs and post-data processing. Y.L. and C.Y. helped with scanning electron microscopy measurements and UPS measurements. J.Z. wrote the first draft of the manuscript. H. Su analyzed the polymerization mechanism of the polymer. All authors contributed to the analysis of results and revision of the manuscript.

Peer review

Peer review information

Nature Communications thanks Li-Qiang Xie, Xiong Li, and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

The data that support the plots within this paper and other findings of this study are available from the corresponding author upon reasonable request. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-72793-y.